Owon AG1012F Arbitrary Waveform Generator

[TomC]

I bought one of these AWGs from Saelig during cyberweek on impulse. The price for the AG1012F was $239, which is about the maximum I was willing to spend for my hobby. I didn't really take the time to properly research other AWGs in the same price range, mainly, because I already own an Owon SDS7102 DSO and I wanted the ability of playing back waveforms captured on it on the AWG. Since I didn't have much time before  the sale ended and it didn't seem to me that this would be possible without going through major conversion gymnastics if I bought a different make, I went ahead and bought the Owon. Now, for good or for bad, my AWG for the foreseeable future is going to be the Owon AG1012F!

I've been playing with my new unit for about a week, and since there is no other EEVblog thread on this unit, I gave it a new thread to document my experiences and give others a chance to contribute or comment. The attachments came straight out of the manual.

As can be seen on the specs, the AG1012F is the low end of the series. However, compared to the next one up (AG1022F), which was quite a bit more expensive, the main difference is that it can only output Sine waves up to 10MHz instead of 25MHz. I already have other RF generators that can go > 100MHz. Also, if I wanted to get higher sine wave frequencies out of the AG1012F, I think it may be possible. For example, I could capture 3 cycles of a sine wave on the SDS7102 and then play it back on the AG1012F. Although not perfect, I think when I play this back at 10MHz the output frequency should be 30MHz. So my output frequency should be the playback frequency times 3 when using this arbitrary waveform.

Anyway, I plan to document this and other experiments on this thread as I go along! The topics I would like to cover include:

1. The Function Generator.

2. The Built in waveforms.

3. The frequency Counter.

4. Modulation.

5. Editable waveforms.

6. Storage.

7. Ultrawave PC software.

[TomC]

First Impression
-----------------

On this post I wanted to list the salient points of my observations during my initial informal testing and fiddling with the unit. During this time I've briefly tried just about all the AG1012F advertised features. In general, it seems to be able to do everything that is claimed with some caveats. More about the caveats further down. Refer to the front panel image in my previous post for the numbers in parentheses.

User interface:

I found the user interface mostly intuitive and easy to learn, so my opinion is that for the most part it can be described as user friendly. Most waveform parameters can be changed via the knob (4), or directly entered via the keypad (3). When the keypad is used you choose the units via the F keys (2). One feature I found useful is that for the standard waveforms, sine, square, and ramp, you have a choice of pp or rms for the amplitude unit. Another interesting feature is the ability to copy all the waveform parameters from CH1 to Ch2 or viceversa. This is accessed by pressing the knob (4) and choosing the copy direction by pressing F1 or F2 (2).

On the other hand, things go downhill when creating or editing waveforms. The process is time consuming and the firmware doesn't seem to work correctly consistently.

Creating a waveform from scratch is possible, even if thousands of points are involved. For example, if you specify 8000 points and you are creating something like a square wave or a trapezoidal wave, it's not necessary to specify the voltage for each individual point. If, for example, you set point 500 to 1V and you want the voltage to stay the same for the next 300 points, then the next step is to set point 800 to 1V. The 0V points in between automatically become 1V. However, when I made a mistake and wanted to go back and correct it, I couldn't find an "undo" feature or procedure other than starting over or editing each individual point.

Editing previously created waveforms, or waveforms obtained from my Owon SDS7102 DSO is possible sometimes. Changing the voltage of a point often results in all subsequent points also changing automatically. For example, if you were editing the third peak of a sine wave with a total of 10 peaks, the last 7 cycles may become a diagonal line. Again, no "undo", starting over won't work unless you reboot the AG1012F first. Even then, if it's a waveform with many points, say more than 1000, it may not work anyway.

Fortunately, I had better, but not perfect results using the Ultrawave PC software, more on that in one of my next posts.

Standard waveforms:

The sine, square, ramp, and noise waveforms seem to work exactly as advertised. I will test these more extensively and report the results in a later post. In contrast, the pulse waveform doesn't quite do what I would expect. The AG1012F screen often lies and reports a pulse width different from what is actually being output. Although it displays pulse widths in increments of 1ns, the output signal changes in increments of 40ns. For example, if you set the pulse width to 40ns (minimum setting), the output signal is 40ns wide. As you increase the pulse width to about 79ns, the output signal remains only 40ns wide. When you set the pulse with between 80ns and 119ns the output signal is 80ns wide. I also noticed that at some frequencies, for example 3MHz, there is no output at all if the pulse width is set to 40ns. However, increasing the setting slightly, say 42ns, restores the output signal.

Built in arbitrary waveforms:

There is an interesting collection of 45 built-in waveforms. I tried each one of them and the display on my SDS7102 was as expected. Had to fiddle a little with the heart waveform to see the hearts on my scope.

Frequency Counter:

I tried the counter with several input frequencies and it seems to work as advertised. The one caveat is the input connector which is on the back of the unit.

[whipman]

I also picked up one of these from Saelig during Cyberweek. Unfortunately, when it arrived, it looked like UPS played warehouse floor hockey with it!  The replacement arrived today.  Haven't had a chance to dig in yet but I'm following your posts. Good job so far.

[TomC]

Hi whipman,

Glad to have someone to compare notes with!

UPS wasn't too kind to my package either,  but it only got it on one corner and it was well padded inside so no damage.

I plan to post some sinewave tests later today.

[TomC]

Standard Waveforms - Sine
-------------------------


On this post I'm trying to illustrate the performance of the standard sinewave signal. The following AG1012F settings are the same for all the tests:

   Channel = CH1

   Output Load = 50 ohms

   Amplitude = 1Vpp

   Offset = 0mV

The AWG is connected to my SDS7102 DSO via a 20" RG58/U BNC cable terminated at the DSO end with a 50 ohm feedthrough terminator. If you try these tests on your unit, keep in mind that using longer cables or some RG58 A/U cables (stranded center conductor) will probably cause lower amplitude readings due to cable losses, specially at the higher frequencies.

The SDS7102 settings can be read from the images, they are located just below the scope's graticule.

For all the Sine tests Ch1 is set to 200mV/Div and a 0.00 divisions offset. The timebase setting is below the Y axis and is labeled (M:). The number on the bottom right of the graticule is the trigger frequency which in this case should be the same as the signal frequency. There are two signal measures on the bottom left of the screen, Vp which means Vpp, and Vk which means Vrms. Don't ask why they labeled it this way, it's a mystery to me too!

For the FFT tests Ch1 is set to 20dB/Div and a 0.00 divisions offset. The number below the Y axis is the frequency span per division. The number on the bottom right of the graticule is the trigger frequency which in this case should be the same as the signal frequency. The two signal measures on the bottom left of the screen are not used in FFT.

Note that for the FFT tests I used the "rectangular" window (no filtering) to get a better frequency resolution. Normally this is not a good idea because the harmonics that are normally generated when the number of sinewave cycles within the data window is not an integral number are not filtered. But in this case I could get around that by slightly decreasing the AG1012F output frequency until the spurious harmonics disappeared, an indication that the number of cycles was an integral number at that point. So that's the reason the trigger frequency on the FFT images is slightly lower than in the Sine images.

The following gives the specifics for each test:

Test #1, AG1012F Frequency = 1kHz.
Test #2, AG1012F Frequency = slightly below 1kHz.

These test images show the sinewave output of the AG1012F and its corresponding FFT at 1kHz 1Vpp for comparison to the published specs. As far as I can tell the levels are within the specified range.

Test #3, AG1012F Frequency = 10MHz.
Test #4, AG1012F Frequency = slightly below 10MHz.

These test images show the sinewave output of the AG1012F and its corresponding FFT at 10MHz 1Vpp for comparison to the published specs. As far as I can tell the levels are within the specified range.

The remaining tests show the results of using a 3 cycle sinewave ARB I created. As I described in my first post I wanted to see if this could be used to extend the sinewave frequency range of the AG1012F to 25MHz. The output is not as clean as the native signals, but I think it's usable if a better source is not available. This ARB was easy to create using the SDS7102 cut waveform feature, but it may be very time consuming to create if you don't have an Owon scope. So I attached a copy of the ARB file (3cy_1kHz.txt, please rename to .bin) to this post for anyone that may want to use it.

Test #5, AG1012F Frequency = 1MHz.
Test #6, AG1012F Frequency = slightly below 1MHz.

Test #7, AG1012F Frequency = 8.333333MHz.
Test #8, AG1012F Frequency = slightly below 8.333333MHz.

During these tests the AG1012F was set to output the 3 cycle sinewave ARB file I created (3cy_1kHz.bin). With the AG1012F frequency set to 1MHZ the frequency displayed on the SDS7102 is 3MHz as expected. To get a 25MHz output the AG1012F frequency was set to 8.333333MHz. The FFT images for these tests are not as clean as the others and more prominent harmonics peaks are evident. At the 25MHz output some signs of aliasing start to show up in the sine image. I'm not sure if this would also be the case with the AG1022F, since a 125MSa/s sampling rate can only sample a 25MHz output signal 5 times per cycle. When I pushed the output frequency to 30MHz (AG1012F set to 10MHz) the aliasing got worse.

[TomC]

Standard Waveforms - Square
-------------------------


On this post I'm trying to illustrate the performance of the standard square wave signal. The following AG1012F settings are the same for all the tests:

   Channel = CH1

   Output Load = 50 ohms

   Amplitude = 1Vpp

   Offset = 0mV

The AWG is connected to my SDS7102 DSO via a 20" RG58/U BNC cable terminated at the DSO end with a 50 ohm feedthrough terminator. If you try these tests on your unit, keep in mind that using longer cables or some RG58 A/U cables (stranded center conductor) will probably cause lower amplitude readings due to cable losses, specially at the higher frequencies.

The SDS7102 settings can be read from the images, they are located just below the scope's graticule.

For all the tests Ch1 is set to 200mV/Div and a 0.00 divisions offset. The timebase setting is below the Y axis and is labeled (M:). The number on the bottom right of the graticule is the trigger frequency which in this case should be the same as the signal frequency. There are various signal measures on the bottom left of the screen.

The following gives the specifics for each test:

Test #1, AG1012F Frequency = 1kHz, Duty Cycle = 50%.
Test #2, AG1012F Frequency = 5MHz, Duty Cycle = 50%.

These test images show the rise time (RT) of the square wave for comparison to the published specs. As far as I can tell it's within the specified range.

Test #3, AG1012F Frequency = 1kHz, Duty Cycle = 50%.
Test #4, AG1012F Frequency = 5MHz, Duty Cycle = 50%.

These test images show the fall time of the square wave for comparison to the published specs. As far as I can tell it's within the specified range.

Test #5, AG1012F Frequency = 1kHz, Duty Cycle = 50%.
Test #6, AG1012F Frequency = 5MHz, Duty Cycle = 50%.

These test images show the jitter at the leading edge of the square wave cycle immediatly following the square wave cycle where the scope was triggered. The number above the graticule indicates the time period from the trigger to the y axis. The scope trace was set to infinite persistance to capture the variations during a couple of minutes. As far as I can tell it's within the specified range. Note that 30ppm of 1ms is 30ns.

Test #7, AG1012F Frequency = 1kHz, Duty Cycle = 50%.
Test #8, AG1012F Frequency = 5MHz, Duty Cycle = 50%.

These test images show the overshoot at the leading edge of the square wave cycle.  As far as I can tell it's within the specified range.

Test #9, AG1012F Frequency = 999.999999kHz, Duty Cycle = 20%.
Test #10, AG1012F Frequency = 999.999999kHz, Duty Cycle = 80%.

These test images show the range of the duty cycle at the highest frequency that is not limited to 50% (just below 1MHz). As far as I can tell it operates as specified.

[TomC]

Standard Waveforms - Square & Pulse - Duty Cycle & Pulse Width
--------------------------------------------------------------


On this post I'm trying to illustrate the performance of the standard square & pulse signals throughout the range of duty cycle and pulse width settings offered by the instrument. The following AG1012F settings are the same for all the tests:

   Channel = CH1

   Output Load = 50 ohms

   Amplitude = 1Vpp

   Offset = 0mV

The AWG is connected to my SDS7102 DSO via a 20" RG58/U BNC cable terminated at the DSO end with a 50 ohm feedthrough terminator. If you try these tests on your unit, keep in mind that using longer cables or some RG58 A/U cables (stranded center conductor) will probably cause lower amplitude readings due to cable losses, specially at the higher frequencies.

The SDS7102 settings can be read from the images, they are located just below the scope's graticule.

For all the tests Ch1 is set to 200mV/Div and a 0.00 divisions offset. The timebase setting is below the Y axis and is labeled (M:). The number on the bottom right of the graticule is the trigger frequency which in this case should be the same as the signal frequency. There are various signal measures on the bottom left of the screen.

As implemented in the AG1012F, there are many similarities between the standard square & pulse signals. The main difference is that for pulse the duty cycle or pulse width is adjustable throughout the whole 5MHz range. In contrast, the square signal only offers duty cycle adjustments. In addition, for frequencies of 1MHz and above the square signal duty cycle is fixed at 50%. However, when the frequency is below 1MHz, the instrument's behavior seems to be identical for both square and pulse signals when the duty cycle setting is changed.

The behavior of the square signal at some particular settings was explored on the previous post. On this post I'd like to concentrate on the behavior of the pulse signal throughout its whole range. However, keep in mind that this behavior equally applies to square waves at frequencies below 1MHz.


The following gives the specifics for each test:

Test #1, Image of the AG1012F set to Standard square, Frequency = 5MHz, Duty Cycle fixed at 50%.
Test #2, Image of the SDS7102 while viewing the output of the AG1012F during Test #1.

These tests show that in this case the image displayed by the AG1012F accurately describes its output waveform including the Duty Cycle. This is typical of all square wave settings from 1MHz to 5MHz. In this range the Duty Cycle can not be changed and is fixed at 50%.

Test #3, Image of the AG1012F set to Standard pulse, Frequency = 5MHz, Duty Cycle set to 50%.
Test #4, Image of the AG1012F set to Standard pulse, Frequency = 5MHz, Pulse Width set to 100 ns.
Test #5, Image of the SDS7102 while viewing the output of the AG1012F during Tests #3 & #4.

These tests show that in this case the image displayed by the AG1012F does not accurately describe its output waveform's Duty Cycle or Pulse Width. In fact, although at this frequency the AG1012F can be set to duty cycles ranging from 20% to 80% in 0.1% increments, or a Pulse Width ranging from 40 ns to 80 ns in 1 ns increments, there are only four possible distinct Duty Cycles or Pulse Widths outputs available: 20% (40 ns), 40% (80 ns), 60% (120 ns) & 80% (160 ns).

The following tests illustrate what happens when you set the Duty Cycle anywhere between 20% and 39.9% or the Pulse Width anywhere between 40 ns and 79 ns. As shown by Test #10, the AG1012F output stays at 20% (40 ns) throughout this range.

Test #6, Image of the AG1012F set to Standard pulse, Frequency = 5MHz, Duty Cycle set to 20%.
Test #7, Image of the AG1012F set to Standard pulse, Frequency = 5MHz, Pulse Width set to 40 ns.
Test #8, Image of the AG1012F set to Standard pulse, Frequency = 5MHz, Duty Cycle set to 39.9%.
Test #9, Image of the AG1012F set to Standard pulse, Frequency = 5MHz, Pulse Width set to 79 ns.
Test #10, Image of the SDS7102 while viewing the output of the AG1012F during Tests #6 through #9.

When the Duty Cycle is increased to 40% or the Pulse Width is increased to 80 ns, as illustrated by Test #13, the AG1012F output finally changes to the next possible Duty Cycle/Pulse Width available.

Test #11, Image of the AG1012F set to Standard pulse, Frequency = 5MHz, Duty Cycle set to 40%.
Test #12, Image of the AG1012F set to Standard pulse, Frequency = 5MHz, Pulse Width set to 80 ns.
Test #13, Image of the SDS7102 while viewing the output of the AG1012F during Tests #11 & #12.

The above discrepancies between the AG1012F screen display and the output waveform are more evident at the higher frequencies. For example, if you lower the frequency to 25 kHz, Which has a 40000 ns period, there is a change in the pulse width for each 0.1% Duty Cycle increment. So as far as the Duty Cycle is concerned, the AG1012F screen display matches the output waveform in this case.

However, the Pulse Width at the output of the AG1012F never changes in 1 ns increments, so the screen display only matches the output waveform periodically as far as the Pulse Width is concerned. Owon has set this interval or segment to a minimum of 40 ns, the segment length can reach up to about 48 ns depending on the waveform's period. So you can expect the screen display to actually match the output waveform about every forty to forty eight 1 ns increments.

There is one other pitfall in this area that can cause trouble for a user that relies on the screen display as an indication of the output waveform. At some frequencies the lowest Duty Cycle or Pulse Width setting does not produce an output waveform at all. The following tests illustrate what happens at one of these frequencies.

Test #14, Image of the AG1012F set to Standard pulse, Frequency = 4.9MHz, Duty Cycle set to 19.6%.
Test #15, Image of the AG1012F set to Standard pulse, Frequency = 4.9MHz, Pulse Width set to 40 ns.
Test #16, Image of the SDS7102 while viewing the output of the AG1012F during Tests #14 & #15.
Test #17, Image of the AG1012F set to Standard pulse, Frequency = 4.9MHz, Duty Cycle set to 20%.
Test #18, Image of the AG1012F set to Standard pulse, Frequency = 4.9MHz, Pulse Width set to 41 ns.
Test #19, Image of the SDS7102 while viewing the output of the AG1012F during Tests #17 & #18.

The obvious conclusion is that the AG1012F screen display is not a dependable indication of the output waveform as far as the Duty Cycle and Pulse Width are concerned. One workaround is for the user to monitor the AWG with a scope as during these tests. I'm also working on an spreadsheet that simulates the AG1012F behavior and predicts within 0.1% of the Duty Cycle or within 3 ns of the Pulse Width the segment boundaries for a particular frequency. The last attachment shows the spreadsheet results when you enter a frequency of 4.9MHz. If there is any interest I will make the spreadsheet available in one of my next posts.

[TomC]

AG1012F Pulse Simulation Spreadsheet
------------------------------------


In this post I want to describe a spreadsheet I created in an attempt to alleviate the usage impact caused by the discrepancies between the AG1012F screen display and its output waveform as described in my previous post. The main objective or motivation for creating this spreadsheet was to provide an alternative to monitoring the AG1012F's output with a scope in order to obtain accurate Duty Cycle or Pulse Width settings.

I decided to design the spreadsheet aiming for a minimum of user interaction to obtain the desired information. So by default, I chose to unlock just the cell where the user can enter the frequency in MHz. From that action, I wanted the spreadsheet to respond by providing the following information:


1. The number of distinct Duty Cycles or Pulse Widths outputs available at that particular frequency.

2. The segment of duty cycle settings in % that would produce each distinct Duty Cycle output.

3. The Duty Cycle in % and the Pulse Width in ns associated with each segment of Duty Cycle settings.

4. The segment of Pulse Width settings that would produce each distinct Pulse Width output.

5. The Pulse Width in ns associated with each segment of Pulse Width settings.

6. Other relevant information associated with the input frequency.


After working on this for over a week, I now finally believe that the spreadsheet fulfills all of the above.


Attachment #1:

This image shows the spreadsheet's response after entering 0.6 into cell A10. Note that by default this is the only cell that can be selected or modified. The main response window is comprised of columns B through I, these contain all the information necessary to accurately set the AG1012F's Duty Cycle or Pulse Width.

Columns A & J mostly contain other information associated with the AG series and the input frequency needed to perform spreadsheet calculations. Although this may be of interest to some, it's not necessary to understand or modify any of this to use the information in columns B through I.

The top row of the main response window displays the input frequency and its period. The second row divides the window into a four column area for information associated with the Duty Cycle Segment and a four column area for information associated with the Pulse Width Segment. The third row is used to identify the information available on each column. Since these three rows act as column headers, they are frozen by default so that they remain visible as the user scrolls the information within the window. Depending on the input frequency up to 2000 rows below the header may contain information.

Looking at the information within the window, we can see that for this frequency there are segments numbered from 0 - 40. Segment 0 will always be displayed on the spreadsheet regardless of the input frequency, but may or may not be accessible at the AG1012F. In this case, it is accessible on my unit, but as indicated, it does not produce an output signal. This row has a light gray background to try to alert the user that settings in this row, if available, will not produce a signal.

Now let's say that we want to set the Duty Cycle as close to 75% as possible. If we just set it to 75.0% the AG1012F will produce an output signal, but is it as close to 75% as we can get? Looking at the spreadsheet window we can see that the closest settings are at segments 30 & 31. A 75.0% Duty Cycle falls within segment 30, but the Duty Cycle for that segment is the Start %, a 73.2% Duty Cycle. In this case setting the AG1012F to segment 31 would be closer to our desired output, a setting within segment 31 would produce an output signal with a 75.7% Duty Cycle.

There is one other thing that you should be careful about when selecting settings, the AG1012F is not always consistent as far as using exactly the same segment boundaries. I've observed variations of 0.1% at times in my unit. Also, comparing actual readings to spreadsheet predictions I often observed discrepancies of 0.1%. So I would recommend using settings in between the boundaries identified by the spreadsheet. In this case my unit switches to 75.7% exactly when I set it to 75.7%, but that may not be the case in every instance, so to be safe, I would recommend a setting like 76 or 77%.

If you are looking for a particular Pulse Width instead of a Duty Cycle, use the information on columns F through I in a similar manner. For example, if you want a pulse width as close to 850 ns as possible you should chose a setting within segment 21. Again, you need to be careful about segment boundary instability and discrepancies between actual values and spreadsheet predictions. I've observed discrepancies of up to 3ns, so to be safe, I would suggest a setting of around 870 ns in this case.


Attachment #2:

This image shows a portion of sheet 2 of the spreadsheet. This information is not needed to use the spreadsheet as described above. They are actual Duty Cycle and Pulse Width values I manually obtained from my unit and recorded here in tabulated form. These were used to help me develop and verify the spreadsheet. If you want to know more about what the different colors mean continue reading the paragraph below.

If you look at the three columns labeled 600 kHz, you can see that the values rendered in a black font match the spreadsheet predictions exactly. I devised a way of getting predictions to more closely match actual values by incorporating Correction Factors. Without correction, the prediction values tend to run slightly lower than the actual values in some cases, this is indicated by the values rendered in a red font. I tried 3 levels of correction, level one causes the predictions to also exactly match the actual values in light green background without causing overshoot of actual values elsewhere. This is the level that is enabled by default on the spreadsheet. Level two causes the predictions to also exactly match the actual values in light yellow background, but cause overshoot of actual values elsewhere. For this level the overshoot is indicated with light magenta background. Level three causes the predictions to also exactly match the actual values in dark yellow background, but causes overshoot of additional actual values elsewhere. In this case the overshoot is indicated with a dark magenta background.


Spreadsheet Attachments:

The attachment "AG1012F Pulse Simulation.ods.txt" is the spreadsheet in the native format I used to develop it. It runs on Open Office and you'll have to delete the ".txt" before using it.

The attachment "AG1012F Pulse Simulation.xls" is the spreadsheet as ported by Open Office to the Microsoft .xls format. I don't have Microsoft Excel, but I tried it on the online version and it seems to work correctly after slightly modifying the code for cell A9. Apparently I used features in that cell that are not available or don't port correctly to .xls.

The difference has to do with the background and text color displayed in this cell when the user inputs an out of range frequency. In the .ods format it changes to a red background and yellow text color while displaying "Input Out of Range". On the .xls version the message displays correctly but the background and text color doesn't change.

Both versions of the spreadsheet are locked by default except for cell A10 to prevent accidental overwrite of other cells. However, they are not password protected, so anyone is free to unlock them and do whatever.

If you unlock the spreadsheet, it's possible, for example, to change cell A25 to 250. That's the sampling rate of the higher members of the AG series (AG2052F & AG2062F). The spreadsheet contains code to modify its operation according to what I surmise these units will do. But I don't own or have access to these units to verify that the spreadsheet accurately predicts their behavior.

[TomC]

Standard Waveforms - Ramp
-------------------------


On this post I'm trying to illustrate the performance of the standard ramp waveform. The following AG1012F settings are the same for all the tests:

   Channel = CH1

   Output Load = 50 ohms

   Amplitude = 1Vpp

   Offset = 0mV

The AWG is connected to my SDS7102 DSO via a 20" RG58/U BNC cable terminated at the DSO end with a 50 ohm feedthrough terminator. If you try these tests on your unit, keep in mind that using longer cables or some RG58 A/U cables (stranded center conductor) will probably cause lower amplitude readings due to cable losses, specially at the higher frequencies.

The SDS7102 settings can be read from the images, they are located just below the scope's graticule.

For all the tests Ch1 is set to 200mV/Div and a 0.00 divisions offset. The timebase setting is below the Y axis and is labeled (M:). The number on the bottom right of the graticule is the trigger frequency which in this case should be the same as the signal frequency. There are various signal measures on the bottom left of the screen.

The following gives the specifics for each test:

Test #1, Image of the AG1012F set to Standard ramp, Frequency = 1MHz, Symmetry 0.0%.
Test #2, Image of the SDS7102 while viewing the output of the AG1012F during Test #1.
Test #3, Image of the AG1012F set to Standard ramp, Frequency = 1MHz, Symmetry 20.0%.
Test #4, Image of the SDS7102 while viewing the output of the AG1012F during Test #1.
Test #5, Image of the AG1012F set to Standard ramp, Frequency = 1MHz, Symmetry 50.0%.
Test #6, Image of the SDS7102 while viewing the output of the AG1012F during Test #1.
Test #7, Image of the AG1012F set to Standard ramp, Frequency = 1MHz, Symmetry 100.0%.
Test #8, Image of the SDS7102 while viewing the output of the AG1012F during Test #1.


These tests show that the image displayed by the AG1012F accurately describes its output waveform including the Symmetry. This is typical of all symmetry settings that I've tried. Note that in the SDS7102 images one cycle of the signal exactly matches the 10 center graticule divisions. This translates to 10% of the signal per division and allows the user to easily verify the symmetry accuracy.

This post concludes the series on the Standard Waveforms, which basically comprise the AG1012F Function Generator capabilities. On the next post I plan to cover the built-in waveforms.

                         A Recap of What I'm Thinking so far
                    -------------------------------------------

At this point my main complaint is the discrepancies between the screen graph and the actual output regarding Duty Cycle & Pulse Width for the square & pulse waveforms as described in the last two posts. I don't know if other AWGs in this price range have similar problems, so far I haven't seen any reviews that mention it, so if anyone reading this thread can shed some light on this I would appreciate it.

For now I think I'll just keep on gathering information while I continue to test the unit in detail. Once I get done I'll be contacting Owon and Saelig support about this Duty Cycle/Pulse Width problem and anything else that I may uncover.

[DJ]

It's kind of interesting to note how banner ads track subjects.

In the case of this thread, the TI AWG reference design ad appeared, and the appnote for it might prove useful for understanding and evaluating both this Owon unit and higher end models:

http://www.ti.com/lit/ug/tidu129/tidu129.pdf


You can even purchase their reference design, if so inclined:




Wide-Bandwidth and High-Voltage Arbitrary Waveform Generator Front End

http://www.ti.com/tool/TIDA-00075

[TomC]

It's kind of interesting to note how banner ads track subjects.

In the case of this thread, the TI AWG reference design ad appeared, and the appnote for it might prove useful for understanding and evaluating both this Owon unit and higher end models:

http://www.ti.com/lit/ug/tidu129/tidu129.pdf


You can even purchase their reference design, if so inclined:




Wide-Bandwidth and High-Voltage Arbitrary Waveform Generator Front End

http://www.ti.com/tool/TIDA-00075

Hi DJ,

That's kind of eerie!

Anyway, thanks for the tip. I'll add the app. note to my collection.

[TomC]

Built-in Waveforms
------------------


On this post I'd like to cover the Built-in waveforms. The basic settings and connections are the same as in previous posts. Other settings can be read from the AG1012F & SDS7102 images.

From my perspective, I find that the built-in waveforms perform as expected. There are some quirks with some waveforms, for example, more jitter at certain frequencies. Another example are disappearing waveform features or disappearing short pulses at the highest frequencies. However, as I understand it, these anomalies are not unique to the AG1012F and are characteristic of the DDS technology when a fixed frequency oscillator is used by the instrument.

Other than the above, I found the interface to the built-in waveforms somewhat clunky. There is no simple way to just browse the waveforms to see what they look like. Maybe this is not a problem for someone that is familiar with all the waveform names and uses. In my case, I think I'll need to see what they look like before I can decide if they meet my needs.

Attachment #0.

This illustrates the type of menu that is used to select a built-in waveform. You have to press two keys to get here, first the key that selects the built-in waveforms, and then the key that selects the category, in this case "Common". From here you turn the knob until the desired waveform is highlighted and then press select. At that point you see the image of the waveform on the screen and the waveform is output. If that waveform is not suitable you have to go through this process again to try a different one. There are 45 different waveforms and it takes quite a while to go through all of them.

Attachments #1, #2, #7, #8, #9, & #10.

These form a catalog of all the different built-in waveforms which I created by capturing the AG1012F's screen image while each waveform was selected. The motivation was to get around my complaint about the clunky interface. The waveforms are arranged in the same format used by the AG1012F built-in waveforms menus. For example, the waveform illustrations on attachment #1 correspond to the menu screen illustrated on attachment #0.

Attachments #3, #4, #5.

These is the PPulse waveform from page 2 of the Common waveforms as displayed by my SDS7102 at different output frequencies. This is a very short pulse, on attachment #3, where it's output at 1MHz, it's under 10 ns as measured by the SDS7102. This pulse is one that disappears if you try to output it at 5MHz or 10MHz. However, miraculously, it is present at 6, 7, 8, or 9MHz.

Notice what happens on attachment #4, where the pulse is output at 900kHz. What appears as two superimposed pulses, one about 10 ns and the other around 16 ns, can be viewed as a type of jitter. This is one of the quirks I mentioned before, and it persists at 800kHz & 700kHz. Once you get down to 600kHz, attachment #5, the "jitter" disappears and all you have left is the 16 ns pulse.

Attachment #6

This is the NegRamp waveform also from page 2 of the Common waveforms. I thought it was interesting that this ramp waveform can be used at frequencies well above 1MHz, since the standard ramp is limited to a maximum of 1MHz. No variable Symmetry on this one though, but if what you need is 100% symmetry this could be an alternative.

Attachments #11 & #12.

This is the DC waveform from page 1 of 1 of the Others waveforms. This provides an interesting feature that allows you to use the AG1012F as a relatively accurate voltage source. On attachment #11 you can see that I set the offset to 3.5V. Attachment #12 shows the voltage detected by my SDS7102.

[TomC]

Hardware
------------

You'd think that by now I'd have outgrown my desire to take new toys apart just to see what's inside! It seems it's still as strong as when I was 5!
The attachments show what I found.

[TomC]

Hardware (continued)
------------------------


A couple more images:

[TomC]

Hardware (continued)
------------------------

I had the unit apart for a couple of days to take pictures and read the markings on all of the ICs. A text file with all the markings I found is attached. From that I was able to find datasheets for most of the components. There were two or three items I couldn't crack, possibly Chinese made and only popular in China. The unit is back together now and as far as I can tell I was lucky enough not to break anything!

[Jamieson]

Thanks Tom, this thread has been informative.  I'm thinking about getting a similar Owon unit.

jamieson

[TomC]

Hi Jamieson,

So far I think is an OK unit for the price I paid. Hopefully I'll go back on sale again soon!

[whipman]

Great job Tom! For the cyberweek price, money well spent. Decent unit for the price! I really like mine.

[TomC]

Frequency Counter
------------------

On this post I want to explore the frequency counter.

For these tests a signal generator output is connected to a 3ft 50 ohm coax. The counter-in BNC has been equipped with a BNC T, one end of the T connects to the coax from the RF generator, the other end is equipped with a probe tip adapter for the scope probe. From here the signal is connected to the SDS7102 DSO through a passive X10 100MHz probe. By connecting the scope at the counter's end rather than the generator's end, the displayed signal should more closely match the signal that the frequency counter is examining.

Test #1 Image of the SDS7102 while viewing an 80mVpp 112.9MHZ signal from an RF generator
Test #2 Image of the AG1012F during Test #1
Test #3 Image of the AG1012F during Test #1 retaining the last valid display although the trigger is out of range
Test #4 Image of the AG1012F during Test #1 when the counter utility is entered while the trigger is out of range

Tests #1 & #2 show that the counter can operate with signal levels well below the specified range (250mVpp - 5Vpp for signals up to 100MHz and 450mVpp - 3Vpp for signals above 100MHz) However, 80mVpp seems to be the low limit for my unit at 113MHz. At this level, the trigger setting is very finicky, it only produces an accurate reading when set between -39mV to -35mV.

When adjusting the trigger is easy to go off range without noticing because the displayed frequency retains the last valid value when the counter is no longer detecting the signal, an example of this is Test #3. One clue that this may have happened is that you no longer see the least significant digits changing. However, if you are looking at a very stable source, a steady frequency display may be present all the time. The only fail proof test that what you are seeing is an active display is to exit the Counter utility function by pressing Back twice, then re-enter the Counter utility. Since the last counter settings are retained, if you still see the same frequency displayed then it is actively being detected. Otherwise you'll see the screen of Test #4.

Test #5 Image of the SDS7102 while viewing a 1Vpp 10KHz signal from a scope calibrator
Test #6 Image of the AG1012F during Test #5 displaying the frequency
Test #7 Image of the AG1012F during Test #5 displaying the pulse width
Test #8 Image of the AG1012F during Test #5 displaying the duty cycle

Tests #5 through #8 show additional measures available for signals up to 10MHz. In this case I'm using a scope calibrator as the signal source. At 1Vpp the Counter's trigger operates as I would expect and an accurate reading is produced with trigger settings in the 0mV to 900mV range. Note that with a stable source as this I had to exit and enter the counter utility to verify the trigger range, looking for changing digits doesn't help. The frequency, pulse width, and duty cycle readings are similar to those obtained with the scope's measure function, so I'm happy with those results.

In my case, I'll probably will seldom use the counter utility. For one thing, I can get the same functionality from my DSO, and the other reason is the inconvenient location of the counter's BNC. Otherwise, it seems that the counter functions as advertised and is capable of operating at signal levels well below the specs.

[TomC]

Modulation - AM
---------------

On this post I'd like to explore AM modulation.

First, let's review the Owon specs and some of Owon's term explanations verbatim from the manual:


Modulated Waveform - AM (specs)
---------------------------------------------------------------------------
Carrier Waveforms       Sine
Source             Internal/ External
Internal Modulating Waveforms    Sine, Square, Ramp, White Noise, Arbitrary
Internal AM Frequency       2 mHz - 20 kHz
Depth             0.0% - 100.0%

External Modulation Input (specs)
---------------------------------------------------------------------------
Input Frequency Range       DC-20 kHz
Input Voltage Range       ± 5 Vpk
Input Impedance       10 k ohms (typical)

Term Explanation
---------------------------------------------------------------------------
AM Frequency:
The frequency of modulating waveform.

Mod Depth:
The Amplitude Range of modulating waveform. In the 0% Modulation, the output amplitude is the half of the set one. In the 100% Modulation, the output amplitude is the same with the set one. For an external source, the depth of AM is controlled by the voltage level of the signal connected to the Ext Mod In connector in the rear panel. +5V corresponds to the currently set depth 100%.

Other relevant information from the manual
---------------------------------------------------------------------------

Modulation function is only used for CH1.

AM (Amplitude Modulation)

The modulated waveform consists of two parts: the Carrier Waveform and the Modulating Waveform. The Carrier Waveform can only be Sine. In AM, the amplitude of the Carrier Waveform varies with the instantaneous voltage of the modulating waveform.



             ------------------------------------------------------------------------

With the above in mind I'd like to start with some examples using the internal modulation. The basic settings and connections are the same as in previous posts. Other settings can be read from the AG1012F & SDS7102 images.


Attachment #1    - Shows the setup for CH2 of the AG1012F. During the internal modulation tests this output is connected to Ch2 of the SDS7102 and is used as the DSO's trigger and as a convenient reference to identify the beginning and end of two modulation periods. It is set to output a 100ns pulse at 10kHz, while the AM frequency is set to 20kHz. The AG1012F output waveforms during AM modulation are periodic, and since both outputs are synchronized to the same clock this arrangement provides a stable trigger.


Attachment #2    - Is the AG1012F Manual's figure that explains how to read the display during AM modulation.


Attachments #3 - #8   - These examples use a sine modulation shape at an AM frequency of 20kHz.

#3 - #5 Show the AG1012F setup at a 0%, 50%, & 100% modulation depth. Notice that as the modulation depth increases, the graph shows an increase in carrier amplitude. This agrees with Owon's "term explanation" where it states that at 0% the output amplitude is half of the amplitude setting. So in this case, since the amplitude is set to 1Vpp, the output amplitude should be 0.5Vpp and the graph reflects that. Conversely, at 100% the graph reflects a 1Vpp output amplitude as stated by Owon.

#6 - #8 Are images of the SDS7102 while viewing the AG1012F outputs during #3 - #5. Everything seems to look as expected, except the carrier amplitude. Regardless of the modulation depth setting, the amplitude remains the same, close to 1Vpp. Frankly, I think the way that the output is implemented makes more sense than the way Owon explains it and depicts it on the AG1012F screen graph. Nevertheless, as far as I can tell, this is another bug where the graph display doesn't quite reflect the instruments output.

The square, ramp, arb, & noise modulation shapes operate in a similar way, except that the noise shape produces a non-periodic waveform and the trigger arrangement I'm using of course can't produce a stable trigger. The discrepancy between the graph and the output waveform described above is also present in all of these four modes.

The arbitrary waveform modulation shape is an interesting feature that I think deserves at least one more example.


Attachments #9 - #10   - These examples use an arb modulation shape at an AM frequency of 20kHz. To use this modulation shape you set up the AG1012F to output an arb waveform first. For this example I'm using the Besselj built-in arbitrary waveform from screen 2 of Maths. Once this is done, when you press Mod and select ModShape Arb, the AG1012F uses that arb's shape to modulate the carrier. If you want to change the arb waveform used for modulation shape you have to first press the Mod button to exit Mod mode. The way the firmware is implemented it doesn't let you access the Arb waveforms menus while you are still in Mod mode. I don't particularly like that roadblock, but is workable.

Note that the Besselj built-in waveform contains 3 cycles, so, although the AM frequency is set to 20kHz, the carrier is actually being modulated at 20 x 3 = 60kHz. This type of effect is true anytime you use an arbitrary waveform that contains more than one cycle.


Attachments #11 - #14   - These examples use an Audio generator set to 2kHz as an external modulation source.

The audio generator output is connected to a 3ft 50 ohm coax. The Ext Mod-in BNC has been equipped with a BNC T, one end of the T connects to the coax from the Audio generator, the other end is equipped with a probe tip adapter for the scope probe. From here the signal is connected to CH2 of the SDS7102 DSO through a passive X10 100MHz probe. By connecting the scope at the AG1012F's end rather than the generator's end, the displayed signal should more closely match the signal that the Ext Mod-in BNC is seeing.

#11 - Shows the AG1012F setup. Notice that the ModShape, AM Freq, & Mod Depth functions are blanked out. However, on the left of the screen it's still indicating Arb shape, and the graph is still showing the shape of the Besselj built-in arbitrary waveform I used on the previous example. Seems the firmware didn't update these screen areas, I went back and tried other settings, it doesn't matter, it just shows whatever your last setup was. The rest of the screen shows the carrier frequency and amplitude, which should be accurate.

#12 - This is the SDS7102 image while viewing #11 after adjusting the amplitude of the generator's output until the carrier appears to have a 100% modulation depth. Notice that this happens when the generator's output is around 90mVpp. Also notice, that the carrier's amplitude, which is supposed to be 1Vpp, is only about 90mV. However, according to what Owon seems to state, 100% mod depth happens at 5V. So let's see what happens if I increase the generator's output amplitude to 1Vpp.

#13 - This is the SDS7102 image while viewing #11 after increasing the amplitude of the generator's output to 1Vpp. This looks to me like pretty bad over modulation, but the carrier amplitude is now 1Vpp, like it's supposed to be. So I'm wondering what will happen if I increase the generator's output even further, after all, the specs indicate that the range is ± 5 Vpk.

#14 - This is the SDS7102 image while viewing #11 after increasing the amplitude of the generator's output to 5Vpp. I'm not sure what to call this, it looks way over modulated and clipped, but the carrier output is still at 1Vpp like it's supposed to be.



Overall, I'm pretty happy with the way this feature works when using the internal modulation source. There are some quirks, but I can work around them without a lot of effort. The performance when using the external modulation source is a different story, I think that part needs a lot of improvement and it was released to market way before it was ready for prime time.

[TomC]

Modulation - FM
---------------

On this post I'd like to explore FM modulation.

First, let's review the Owon specs and some of Owon's Notes verbatim from the manual:


Modulated Waveform - FM (specs)
---------------------------------------------------------------------------
Carrier Waveforms       Sine
Source             Internal/ External
Internal Modulating Waveforms    Sine, Square, Ramp, White Noise, Arbitrary
Internal Modulating Frequency    2 mHz - 20 kHz
Frequency Deviation       2 mHz - 20 MHz

External Modulation Input (specs)
---------------------------------------------------------------------------
Input Frequency Range       DC-20 kHz
Input Voltage Range       ± 5 Vpk
Input Impedance       10 k ohms (typical)

Note:
---------------------------------------------------------------------------
The Sum of the Deviation and the Carrier Frequency should be equal to or less than maximum frequency of the selected function plus 1 kHz.

For an External Source, the Deviation is controlled by the voltage Level of the signal connected to the Ext Mod In connector in the rear panel. +5 V corresponds to the selected Deviation and -5 V to the negative selected Deviation.

Other relevant information from the manual
---------------------------------------------------------------------------

Modulation function is only used for CH1.

FM (Frequency Modulation)

The modulated waveform consists of two parts: the Carrier Waveform and the Modulating Waveform. The Carrier Waveform can only be Sine. In FM, the frequency of the Carrier Waveform varies with the instantaneous voltage of the modulating waveform.



             ------------------------------------------------------------------------

With the above in mind I'd like to start with some examples using the external modulation. The basic settings and connections are the same as in the previous post. Other settings can be read from the AG1012F & SDS7102 images.


Attachment #0    - Is the AG1012F Manual's figure that explains how to read the display during FM modulation.


Attachments #1 - #3   - These examples use an Audio generator set to 2kHz as an external modulation source.

The audio generator & DSO are connected as explained in the previous post.

#1 - Shows the AG1012F setup. Notice that the ModShape, & Mod Freq functions are blanked out. However, on the left of the screen it's still indicating Arb shape, and the graph is still showing the shape of the Besselj built-in arbitrary waveform I used on the previous post. This is the same quirky behavior that we experienced with external AM modulation.

The FM Dev function, which is set to 20kHz, is not blanked out, since it's required regardless of the modulation source. The note from Owon seems to state that when the modulation waveform is at +5 the deviation is what's been selected, so we should expect the frequency at that point to be 50kHz + 20kHz = 70kHz. Conversely, according to the note, when the modulation waveform is at -5 we should expect 50kHz - 20kHz = 30kHz, or at least, that's how I interpret it.

#2 - This is the SDS7102 image while viewing #1 after adjusting the amplitude of the generator's output until the carrier appears to have full frequency deviation at the peaks of the modulating waveform. Notice that this happens when the generator's output is around 340mVpp. Also notice, that the carrier's amplitude is about 1Vpp, which is what it's supposed to be. However, according to what Owon seems to state, full frequency deviation would not happen unless the modulation signal peaks were +5 & -5. So let's see what happens if I increase the generator's output amplitude to 1Vpp.

Note: The FM modulation output waveform is not periodic, so the scheme used in the previous post to obtain a stable trigger wasn't useful. In this case, to get a fairly stable trigger, I'm triggering on the carrier based on pulse width. Since the low end of the carrier frequency due to deviation is 30kHz, a period of about 33 microseconds, the width of the positive half of the cycle should be around 33/2 = 16.5 microseconds. So using normal trigger I started with the setting >16.5 microseconds and kept on lowering it until I got a trigger. As can be seen on the image the DSO is triggered at >15.4 microseconds, and the pulse that is triggering it is at the center of the graticule. This pulse more or less coincides with the negative peak of the modulating waveform, and as can be seen, the complete cycle yields a 30.3kHz frequency as indicated by the cursor measurement.

#3 - This is the SDS7102 image while viewing #1 after increasing the amplitude of the generator's output to 1Vpp. This looks to me like pretty bad over modulation, notice that the trigger pulse in this case is only >15.0 microseconds and it more or less coincides with the positive peak of the modulating waveform. At that point I'd expect to see a higher frequency than the carrier, not lower. Also notice lower frequency pulses at other apparently random locations. I think it's obvious that the Owon's note is again incorrect as far as the way the hardware actually operates.


Attachments #4 - #6   - These examples use a sine modulation shape at an internal modulating frequency of 2kHz. The FM deviation is set to 20kHZ. Since there is no modulating waveform amplitude setting, the AG1012F presumably automatically sets the appropriate modulating waveform amplitude to obtain full deviation. So with these settings, the output waveform should be very similar to what we got before when using the external source.

In addition to the above CH1 settings, I decided to set CH2 to output a 2kHz 1Vpp sine wave. This is not required and not used in any way to produce the FM modulation done by CH1, I just wanted to display this second signal on the DSO as a reference to help identify the areas of the FM signal where frequency changes should occur. As it turns out, I had to invert this second signal at the DSO to get the proper phase.

#4 - This shows the CH2 settings.

#5 - This shows the CH1 settings.

#6 - This is the SDS7102 image while viewing #5 and #4. I'm using the same DSO trigger scheme I used for #2 above. As can be seen, the carrier waveform is almost identical to what we got on #2 above, and the cursor measure yields pretty much the same frequency. Again, remember, that the waveform on CH2 is there just for reference, It's not the actual signal that is modulating the carrier.


Again, I have one more example using one of the built-in arbitrary waveforms for the modulation shape. In this case I decided to use the RoundHalf from page 2 of the Common waveforms. This waveform is about 90% positive, most of the round half, and about 10% negative. As a result, the carrier's frequency deeps below its nominal value for just a brief period during each cycle. As in the previous example, I decided to set CH2 to output the same waveform so I can display it on the DSO as a reference. In this case I had to alter the inter-channel phase at the AG1012F to get the proper phase on the DSO display.


Attachments #7 - #10  -  These examples use a Arb modulation shape at an internal modulating frequency of 2kHz. The FM deviation is set to 20kHZ.

#7 - This shows the CH2 settings.

#8 - This shows the CH1 settings.

#9 - This is the SDS7102 image while viewing #8 and #7. I'm using the same DSO trigger scheme I used before. As can be seen, the carrier waveform only drops below its nominal frequency for a brief period around the center of the screen, the incursion doesn't last for even one whole cycle. As a result, the cursor measure yields a frequency slightly above 30kHz. Again, remember, that the waveform on CH2 is there just for reference, It's not the actual signal that is modulating the carrier.

#10 - This is an expanded view of the center area of #9 above.



I think that overall, this feature performs better than the AM modulation feature discussed on the last post. In my opinion, there are fewer quirks, and although the external modulation doesn't quite match Owon's description, at least it doesn't mess up the carrier's amplitude.

[rf-loop]

Thank you about these all testings and information.

About AM modulation depth, it looks like they have messed with peak level and "nominal" level.

It is very common practice that if we set AM mod 0% and level example 1V then if we rise modulation depth to 100% minimum level is 0V and maximum level is 2V and nominal value stay fixed 1V.


Example like this:
http://www.radio-electronics.com/info/rf-technology-design/am-amplitude-modulation/modulation-index-depth.php

[TomC]

Thank you about these all testings and information.

About AM modulation depth, it looks like they have messed with peak level and "nominal" level.

It is very common practice that if we set AM mod 0% and level example 1V then if we rise modulation depth to 100% minimum level is 0V and maximum level is 2V and nominal value stay fixed 1V.


Example like this:
http://www.radio-electronics.com/info/rf-technology-design/am-amplitude-modulation/modulation-index-depth.php

Hi rf-loop,

Thanks for the link!

Very nice and complete review of AM modulation. It's a pity that Owon didn't get this right, but at least the graph reflects what should happen! I wonder if they can get this fixed with a firmware update!

[TomC]

Modulation - PM
---------------

On this post I'd like to explore PM modulation.

First, let's review the Owon specs (there are NO Owon notes on this) verbatim from the manual:


Modulated Waveform - PM (specs)
---------------------------------------------------------------------------
Carrier Waveforms       Sine
Source             Internal/ External
Internal Modulating Waveforms    Sine, Square, Ramp, White Noise, Arbitrary
Internal PM Frequency       2 mHz - 20 kHz
Phase Deviation       0° - 180°

External Modulation Input (specs)
---------------------------------------------------------------------------
Input Frequency Range       DC-20 kHz
Input Voltage Range       ± 5 Vpk
Input Impedance       10 k ohms (typical)

Note:
---------------------------------------------------------------------------
There are NO notes

Other relevant information from the manual
---------------------------------------------------------------------------

Modulation function is only used for CH1.

PM (Frequency Modulation)

The modulated waveform consists of two parts: the Carrier Waveform and the Modulating Waveform. The Carrier Waveform can only be Sine. In PM, the phase of the Carrier Waveform varies with the instantaneous voltage level of the modulating waveform.



             ------------------------------------------------------------------------

With the above in mind I'd like to start with some examples using the internal modulation. The basic settings and connections are the same as in the previous post. Other settings can be read from the AG1012F & SDS7102 images.


Attachment #0    - Is the AG1012F Manual's figure that explains how to read the display during PM modulation.


Attachments #1 - #3   - These examples use a sine modulation shape at an internal modulating frequency of 20kHz. The PM deviation is set to 90 degrees. Since there is no modulating waveform amplitude setting, the AG1012F presumably automatically sets the appropriate modulating waveform amplitude to obtain full deviation.

In addition to the above CH1 settings, I decided to set CH2 to output a 20kHz 1Vpp sine wave. This is not required and not used in any way to produce the PM modulation done by CH1, I just wanted to display this second signal on the DSO as a reference to help identify the areas of the PM signal where frequency changes should occur.

#1 - This shows the CH2 settings.

#2 - This shows the CH1 settings.

#3 - This is the SDS7102 image while viewing #1 and #2. The 90 degree phase deviation, 50kHz carrier frequency, and 20kHz PM frequency, were chosen to cause highly visible phase changes while viewing just a few cycles of the output waveform. Notice that in this case the phase changes appear to have been smoothed out by filtering. Also note the cursor measure of the signal raise period (15.4 micro seconds) center left of the graticule. As explained in the note below, this is being used to trigger the DSO. Again, remember, that the waveform on CH2 is there just for reference, It's not the actual signal that is modulating the carrier.

Note: Although the PM output waveform is usually periodic, I decided that a trigger other than a pulse at a multiple of the modulating frequency might allow a more accurate comparison of the output waveforms obtained using internal and external modulation. After trying a few different schemes, I decided to use a positive slope. As can be seen the thresholds are set close to the peaks, the period between thresholds was adjusted to the longest possible time that still provided a stable trigger. The idea was to use the same setting when using an external source for the modulating waveform. At that time, by slowly increasing the output amplitude of the audio generator, I hope that the output waveform when the DSO starts to trigger will be a close match to the waveform obtained using the internal source. Hopefully this will reveal the audio generator's output amplitude required to obtain full deviation.


Again, I have one more example using one of the built-in arbitrary waveforms for the modulation shape. In this case I decided to use the RoundHalf from page 2 of the Common waveforms. This waveform is about 90% positive, most of the round half, and about 10% negative. As a result, a sudden phase change is required when this brief deep below 0V occurs during each cycle. As in the previous example, I decided to set CH2 to output the same waveform so I can display it on the DSO as a reference. In this case I had to alter the inter-channel phase at the AG1012F to get the proper phase on the DSO display.


Attachments #4 - #6  -  These examples use a Arb modulation shape at an internal modulating frequency of 20kHz. The PM deviation is set to 90 degrees.

#4 - This shows the CH2 settings.

#5 - This shows the CH1 settings.

#6 - This is the SDS7102 image while viewing #4 and #5. I'm using the same DSO trigger scheme I used before. As can be seen, the carrier waveform exhibits a sudden phase change around the positive peak at the center of the screen as well as during a couple negative peaks visible on this image. Notice that some spikes are present that didn't get smoothed out, my opinion is that in this case this is normal. Again, remember, that the waveform on CH2 is there just for reference, It's not the actual signal that is modulating the carrier.


Attachments #7 - #9   - These examples use an Audio generator set to 20kHz as an external modulation source.

The audio generator & DSO are connected as explained in the previous post.

#7 - Shows the AG1012F setup. Notice that the ModShape, & PM Freq functions are blanked out. However, on the left of the screen it's still indicating Arb shape, and the graph is still showing the shape of RounHalf built-in arbitrary waveform I used previously. This is the same quirky behavior that we experienced with external AM and FM modulation.

The Phase Dev function, which is set to 90 degrees, is not blanked out, since it's required regardless of the modulation source. I couldn't find any explanation in the manual as to how the deviation is affected by the amplitude of the external source waveform. Maybe this can be considered an improvement, given the way the accuracy of the previous explanations turned out.

#8 - This is the SDS7102 image while viewing #7 after adjusting the amplitude of the generator's output until the DSO started to trigger. Notice that this happened when the generator's output was around 368mVpp. Also notice, that the carrier's amplitude is about 1Vpp, which is what it's supposed to be. However, at first look the waveform bears little resemblance to the one obtained while using the internal source. It appears to me that in this case, for some reason, Owon decided that the phase changes didn't need to be filtered. With this in mind, try to imagine what the waveform would look like if it was filtered, for me, a close resemblance becomes apparent.


#9 - This is the SDS7102 image while viewing #7 after increasing the amplitude of the generator's output to 1Vpp. This looks to me like pretty bad over modulation. Had to change the trigger slightly for it to be stable, and the waveform looks more random and possibly no longer periodic.



I think that when using the internal source, this feature performs as well as the FM modulation feature discussed on the last post. However, other than as a novelty, I don't see external modulation as being a useful tool as presently implemented. In my mind, it rates lower than external AM modulation.

[TomC]

Modulation - FSK
---------------

On this post I'd like to explore FSK modulation.

First, let's review the Owon specs and some of Owon's term explanations verbatim from the manual:


Modulated Waveform - FSK (specs)
---------------------------------------------------------------------------
Carrier Waveforms       Sine
Source             Internal/ External
Internal Modulating Waveforms    50% duty cycle square
FSK Rate          2 mHz - 100 kHz

External Trigger Input (these are the only specs related to the external source I could find in the Owon manual)
---------------------------------------------------------------------------
Level             TTL-compatible
Slope             Rising or falling (selectable)
Pulse Width          >100 ns
Trigger Delay          0.0 ns - 60 s

Term Explanation
---------------------------------------------------------------------------
FSK Rate:
The frequency at which the output frequency shifts between the carrier frequency and the Hop frequency (Internal Modulation only).

Other relevant information from the manual
---------------------------------------------------------------------------

Modulation function is only used for CH1.

FSK (Frequency Shift Keying)

The FSK Modulation is a modulation method, the output frequency of which switches between two the pre-set frequencies (Carrier Waveform Frequency and the Hop Frequency). The Frequency of the Output Frequency switch between the carrier waveform frequency and the Hop frequency is called the FSK rate. The frequency by which the output frequency switch from each other is determined by the Internal Frequency generator or the Signal Voltage Level offered by the Ext Trig/Burst/Fsk In connector in the rear panel. The Carrier Waveform can only be Sine.



             ------------------------------------------------------------------------

With the above in mind I'd like to start with some examples using the internal frequency generator. The basic settings and connections are the same as in the previous post. Other settings can be read from the AG1012F & SDS7102 images.


Attachment #0    - Is the AG1012F Manual's figure that explains how to read the display during FSK modulation.

Notice that there is no ModShape function. As a result, when using the internal source, the only modulation shape available is square wave with a 50% duty cycle. Since FSK only has two states, it makes sense that the modulation will be limited to digital waveforms, but only allowing square waves is taking it a bit too far in my opinion. I think it would have been nice if arbitrary waveforms were also included. It's likely that users interested in digital circuits will create two state ARBs for different applications. So why not let these waveforms be used to modulate FSK? Well, after all, I think there may be a way around this, as far as I can tell this limitation doesn't apply when the modulating waveform comes in via the external source input (Ext Trig/Burst/Fsk In connector)! As a matter of fact, the one spec available for this connector states that it's TTL compatible, so I plan to try it with a digital looking built-in ARB before I complete this post.


Attachments #1 - #3   - These examples use the only internal modulation shape available, a 50% duty cycle square wave. The frequency of this wave is the FSK rate which is set to 10kHz.Since the carrier is set to 100kHz and the Hop frequency to 50kHz, there should only be a few cycles of carrier and hop waveform during each half cycle of the square wave. As in FM & PM modulation, there is no modulating waveform amplitude setting, so the AG1012F presumably automatically sets the appropriate square wave amplitude to obtain stable carrier frequency to hop frequency transitions.

In addition to the above CH1 settings, I decided to set CH2 to output a 10kHz 2V 0 to peak square wave. This is not required and not used in any way to produce the FSK modulation done by CH1, I just wanted to display this second signal on the DSO as a reference to help identify the areas of the FSK signal where frequency changes should occur.

#1 - This shows the CH2 settings.

#2 - This shows the CH1 settings.

#3 - This is the SDS7102 image while viewing #1 and #2. The settings were chosen to cause highly visible frequency changes while viewing a limited number of output waveform cycles. In this case the FSK output signal is periodic every other cycle, so since the signal on CH2 is synchronized by the same clock, I was able to trigger on CH2 using the signal edge and a holdoff of a little over a 100 microseconds. This causes the DSO to trigger every other cycle. Again, remember, that the waveform on CH2 is there just for reference, It's not the actual signal that is modulating the carrier.


While using the internal source, I can't use an ARB waveform for FSK modulation as I was able to do for the other modulation modes at this point. So I'm going to try that at the end of this post while using the external source.


Attachments #4 - #5   - These examples use an Audio generator set to 10kHz square wave as an external modulation source.

The audio generator & DSO are connected as explained in the previous post.

#4 - Shows the AG1012F setup. Notice that the FSK Rate function is blanked out. To my delight, the left of the screen and the graph are correct in this case, no more quirky behavior like we experienced with external AM, FM, and PM modulation on this screen. I was so pleasantly surprised that I tried a couple times to get it to malfunction by setting ARBs, etc. before coming back to this screen, but no dice, Owon got it right this time!

The Hop Freq function, which is set to 50kHz, is not blanked out, since it's required regardless of the modulation source. The only hint in the manual as to what voltage levels are required to get the output to transition from carrier frequency to hop frequency is that the input is TTL compatible. So I'm assuming that a 0V to 2V peak square wave should do the trick.

#5 - This is the SDS7102 image while viewing #4. The carrier to hop frequency transitions start working once the amplitude of the generator's output is a little above 1V and continue to work normally as a I increased the amplitude to about 4V peak. So I left the amplitude set to 2V for this image. As for the DSO trigger, the scheme I used for #3 didn't work here, because the AWG clock is not in sync with the Audio Generator. So I triggered on CH1 using a combination of slope and holdoff to get the DSO to trigger on every other set of carrier/hop frequency transitions.


Attachments #6 - #9  -  For these examples I substituted my Audio Generator with CH2 of the AG1012F. This way I could setup CH2 to output a built-in Arb waveform and use that to modulate the FSK. So I have the CH2 output of the AWG connected to the external FSK in AWG connector instead of my Audio Generator.

#6 - This shows the CH2 settings. In this case I decided to use the CPulse (coded pulse) from page 2 of the Common waveforms. Notice that the frequency is set to 2.5kHz, this is equivalent to the 10kHz we were using before since this ARB contains 4 cycles.

#7 - This shows the CH1 settings. This should be exactly the same as #4 above.

#8 - This is the SDS7102 image while viewing #6 and #7. I tried to use the same DSO trigger scheme I used on #3 before but with a longer holdoff so that the DSO would trigger every second set of pulses. This makes the pulses stable, but the CH1 signal is still a little shaky because it's not exactly periodic in this case. So I stopped the scope to get a better capture. So, as can be seen, this way around the internal source limitation where only square waves are allowed seems to work very well. Again, remember, that the waveform on CH2 in this case is not just for reference, It's the actual signal that is modulating the carrier.

#9 - This is an expanded version of #8 for a better view of the FSK frequency transitions.



Well, in this case I have to admit That I barely have any complaints! Everything seems to work just as advertised and without quirks. I would still like to have the option of using ARBs with internal modulation, but since there is a workable way around that, it's not a major disappointment.