Siglent SDG1000 (aka LeCroy WaveStation) firmware updates

[nctnico]

The firmware from Lecroy's website says its version ..2.28

[Wuerstchenhund]

The firmware from Lecroy's website says its version ..2.28

It seems the LeCroy firmware is always +1 more than the comparable Siglent firmware (odd numbers like the current .27 for Siglent , even numbers like .28 for LeCroy firmware).

The next Siglent firmware will probably be .29 or .31, and the next LeCroy version should be .30 or .32.

[rf-loop]

Also, the DB level indication seems way low for the actual output.  An actual measured zero dbm (into a high-z input millivolt-meter) level is shown as something like -32dbm on the generator screen.  The maximum settable dbm level is -13 on channel 1, and I know the maximum open circuit voltage is 20 volts pk-pk.  Am I missing something? 

Next, numbers are only "calculated in head" for thinking purpose, not exact values.

0dBm means 1mW!   

Lets look littlebit.

0.225Vrms to 50ohm is around 1mW power, so we say its level is 0dBm in 50 ohm system.
but example 0.225Vrms to 100kohm what is dBm?  It is around  -32.9dBm
For 1Mohm it is -42.9dBm.

Also it can not 20Vpp to 50 ohm load. Ok lets take High-Z example 100000k
10Vpp sinewave is roughly 3.5Vrms (rounded)
it is -9dBm
But for 50ohm load what happend. V drops around half. If there is 1.75Vrms sinewave and load is 50ohm there is around 17.9dBm level.

Now I do not know what impedance SDG use for High-Z  dBm value calculation. Of course in SDG display value is calculated value from voltage set and impedance. And how this math accuracy.

Las time I have used SDG1025 with new FW it show around right dBm for sinewave when also output impedance was set for 50ohm and power was measured with HP power meter (of course 50ohm impedance).  I do not remember results but this I remember that result was "acceptable" for cheap instrument.

It is good also to note that SDG output is always 50ohm source impedance independent of output  setting High-Z or 50ohm. Setting affect only to displayed level setting number.

If load is other than High-Z (I do not know what value is High-Z) or 50ohm it need user calculate real level.

If you set output to 50ohm (utility menu) then look agen dBm values. Different?

"into a high-z input millivolt-meter" and you look what power SDG display.

If you use dBm with SDG you need also use 50ohm setting and 50ohm load. Then it show around right. Becouse High-Z... as before. How much it is. If you drive then 1k load or 10k load or 100ohm load.  SDG do not know your real load. Number is right if your load is same what SDG use for calculating displayed.
(basic things for this read also in User Manual)

[commongrounder]

Thank you everyone, and especially rf-loop for the excellent discussion. 

I do understand that dbm refers to measured power into a load.  I come from an audio background, so I am used to the reference impedance for dbm as being 600 ohms.  My millivolt meter is calibrated to measure 0dbm as .775 volts rms (1mw>600ohms).  Personally, I would prefer dbV output levels, which would be independent of loading (i.e. always into "hi-z"), and then be aware of the 50 ohm output impedance to calculate level drops.  Most of the gear I work with has a 10k ohm, or higher, input impedance, so there is relatively low circuit loading.  The fact is, it is "all in our heads" anyway, and I just need to adjust to how they execute the level setting in the generator.

The generator output number one is rated 20 volts pk-pk into an open circuit (so-called "hi-z").

Anyone have thoughts on the log sweep issue?  If the Lecroy firmware made a fix to this, I would change over.

[commongrounder]

Just checked the SDG1005 with my Fluke meter that has dbm with an adjustable reference impedance.  With both the meter and the generator set for 50 ohms, into a 50 ohm terminator, the output, as compared to the generator display, is within .01db of right on, so not too bad.  In "hi-Z" mode, -30dbm, as displayed on the generator, measures exactly 1 volt rms or 0dbV.  I can work with that, since I usually just need to make relative db jumps in level to calibrate equipment.

Back to work!!!

[commongrounder]

I want to draw attention back to the sweep function of this generator.  One of the often used functions for me is performing sweeps over the audio range and beyond.  My other analog function generators (Tektronix FG504 and B&K 3020) have always produced the expected so-called "log" sweep, which is around a decade of upward frequency change over a particular period of time.  I have also externally swept the generators at one octave per period of time.  The first time I set up this SDG1005 for "log" sweep operation, I could tell right away that the frequency was not increasing as I expected.  I put the generator on the scope and set up the graphing function for frequency over time.  It seems to me that the rise is exponential (I have not calculated the values, someone better at math could do that).  The sweep envelope does appear to match a couple of the built in arbitrary waveforms.  I have shown below the graph from the SDG1005 and, for comparison, the same sweep set up on my Tektronix FG504.  I would like to ask Siglent to change the sweep envelope for "log" to either octave/time or decade/time (or both).  If someone has a suggestion for language to help convey that (mathematically or otherwise), I would appreciate the input.

[Bored@Work]

First image looks like a logarithmic, second like an exponential function. The trick, however, to get a logarithmic display on an oscilloscope with a steady x (time) deflection is to drive the signal with the inverse function of the desired logarithm - which is an exponential function.

[commongrounder]

Okay.  Now the graph displayed on the scope screen shots has the x axis running in real time, i.e. linear seconds per division, so any instantaneous reading of the frequency at any particular time along that graph should fit the curve of the sweep I am desiring.  By the way, I put log in quotes, because I came to understand that the rising curve of the sweep envelope isn't really a "log" function.  I think this is where the firmware designers are missing the boat.  The FG504 sweep envelope is what I am looking for.  The scope is reproducing that envelope just as I expect.  I have in the past plotted out the frequencies for the FG504 to be sure they fit my response plots.  I am looking forward to being able to automate this process using the trigger output function of the generator.  I also want to be able to use the arbitrary function to sweep a burst waveform to check peak power output over a range of frequencies with some automation.  I can get an approximation of what I want by reversing the direction of the sweep, but it still is not conforming to either a decade or octave per time unit rise (or drop, in that case) in frequency.  If you look at the little graphic on the generator screen in log sweep mode, you can see they have the correct visual idea of the "log" sweep function, but not the actual sweep.  I am also wondering if they use one of the arbitrary waveforms as the envelope for the sweep.  I haven't tried extending the sweep to the maximum time to see if there are discreet steps that suggest the 16k record length.  I might try that next, just to get some additional insight.

[commongrounder]

I went ahead and sorted out the math needed to create the proper sweep envelope for a decade "log" function.  I am attempting to contact Siglent through their web site, but I am not at all certain that the messages are getting through, as there is no auto-acknowledgement email, or anything I have seen, so far.  I surely am not the only end user who has issues with this.   If anyone has a direct email address to customer service at Siglent they are willing to share, that would be great.
Here is the formula with an example of an audio sweep of 20hz to 20khz over 10 seconds.

f(t) = fstart * 10(t * c)

c = log10(ffinal / fstart) / S

 where

 f(t) = a function giving the frequency in hz at time t
 t = the current time in seconds since the start of the sweep
 S = how long the sweep lasts in seconds (e.g. 10 seconds)
 fstart = the frequency at t = 0 in hz (e.g. 20hz)
 ffinal = the frequency at t = S in hz (e.g. 20,000hz)
 c = the 'time constant', as defined above (e.g. 0.3 for the example numbers above)

[nixxon]

I went ahead and sorted out the math needed to create the proper sweep envelope for a decade "log" function.  I am attempting to contact Siglent through their web site, but I am not at all certain that the messages are getting through, as there is no auto-acknowledgement email, or anything I have seen, so far.  I surely am not the only end user who has issues with this.   If anyone has a direct email address to customer service at Siglent they are willing to share, that would be great.
Here is the formula with an example of an audio sweep of 20hz to 20khz over 10 seconds.

f(t) = fstart * 10(t * c)

c = log10(ffinal / fstart) / S

 where

 f(t) = a function giving the frequency in hz at time t
 t = the current time in seconds since the start of the sweep
 S = how long the sweep lasts in seconds (e.g. 10 seconds)
 fstart = the frequency at t = 0 in hz (e.g. 20hz)
 ffinal = the frequency at t = S in hz (e.g. 20,000hz)
 c = the 'time constant', as defined above (e.g. 0.3 for the example numbers above)

Where do you apply that function to integrate with the AWG. Do you make a waveform in MS Excel and export into the AWG?

[commongrounder]

Right there is the problem.  We, as end users, do not have access to the location where the log sweep function is stored.  It is set in firmware, and requires a change in an updated version.  I have sent an email outlining this to Siglent.  They are apparently on vacation until May 3 for "Labor Day".  I think it would be excellent to be able to create and store custom sweep curves to suit the end user's needs.

[nixxon]

I went ahead and sorted out the math needed to create the proper sweep envelope for a decade "log" function.  I am attempting to contact Siglent through their web site, but I am not at all certain that the messages are getting through, as there is no auto-acknowledgement email, or anything I have seen, so far.  I surely am not the only end user who has issues with this.   If anyone has a direct email address to customer service at Siglent they are willing to share, that would be great.
Here is the formula with an example of an audio sweep of 20hz to 20khz over 10 seconds.

f(t) = fstart * 10(t * c)

c = log10(ffinal / fstart) / S

 where

 f(t) = a function giving the frequency in hz at time t
 t = the current time in seconds since the start of the sweep
 S = how long the sweep lasts in seconds (e.g. 10 seconds)
 fstart = the frequency at t = 0 in hz (e.g. 20hz)
 ffinal = the frequency at t = S in hz (e.g. 20,000hz)
 c = the 'time constant', as defined above (e.g. 0.3 for the example numbers above)

I tried your function in Excel and all I got was a straight line that increases at a rate of 10*c:

      A   B   C
14  fstart   1   Hz
15  ffinal   1000   Hz
16  S   10   seconds
17  c   0,3   (=LOG(B15/B14)/B16)) ("LOG"="LOG10" option in excel)
      
      
RESULTS:      
X-axis   Y-axis   
t   f(t)   
0   0   <--- the function is: =$B$14*10*(A23*$B$17) for value 0 in box B23
0,1   0,3   
0,2   0,6   
0,3   0,9   
0,4   1,2   
0,5   1,5   
0,6   1,8   
0,7   2,1   
0,8   2,4   
0,9   2,7   
1   3   
1,1   3,3   
1,2   3,6   
1,3   3,9   
1,4   4,2   
1,5   4,5   
1,6   4,8   
1,7   5,1   
1,8   5,4   
1,9   5,7   
2   6   

Did I do something wrong here?

[Galaxyrise]

nixxon, I believe commongrounder meant:
f(t) = fstart * 10^(t * c)

which would be
=$B$14*POWER(10,(A23*$B$17))

in your spreadsheet

[Fennec]

Hi,

I am not the expert in Excel, so please can you attach the file here ? Thank you.

[commongrounder]

Yes, that is correct.  (t*c) was originally superscripted to indicate exponentiation. I'm guessing the superscript got lost somewhere, so it appeared like 10(t*c), which is linear. It should be 10 ^ (t*c).

[nixxon]

Thanks. That made a difference:

A   B   C
fstart   1   Hz
ffinal   1000   Hz
S   2   seconds
c   1,5   
      
      
      
      
t (seconds)   f(t)   
0   1   <-----------column B now reads this: =$B$14*10^(A23*$B$17)
0,1   1,412537545   
0,2   1,995262315   
0,3   2,818382931   
0,4   3,981071706   
0,5   5,623413252   
0,6   7,943282347   
0,7   11,22018454   
0,8   15,84893192   
0,9   22,38721139   
1   31,6227766   
1,1   44,66835922   
1,2   63,09573445   
1,3   89,12509381   
1,4   125,8925412   
1,5   177,827941   
1,6   251,1886432   
1,7   354,8133892   
1,8   501,1872336   
1,9   707,9457844   
2   1000   

[commongrounder]

Yup. That is the kind of envelope I expect from the "log" sweep of a function generator.  Let's hope that Siglent can incorporate that into the generator firmware, or better yet, let us create sweep envelopes ourselves, just like any other arbitrary waveform.  By the way, I did try using the frequency modulation function, but the frequency range (equivalent to start and stop frequencies) is too narrow.

[nixxon]

Hi,

I am not the expert in Excel, so please can you attach the file here ? Thank you.

Here you go. It's just a crude excel workbook.

[MasterOfNone]

You can create your own custom sweep curves using FM as follows:
1: Draw the sweep curve using the Easydraw app (Equation, Line or hand draw etc).
2: Download the curve to the SDG.
3: On the SDG select the “Arb” function and then load the Stored Waveform.
4: Next select the type of wave you want to sweep (Square, Sine etc)
5: Set the frequency of the wave to the midpoint of your sweep range, so if you want to sweep from 0Hz to 20KHz you would set 10KHz.
5: Set the amplitude etc.
6: Press the Mod button and select "Type" to "FM".
7: Set the “FM Freq” to the sweep duration. Freq = 1/t ,  so for a 10 second sweep set “FM Freq” to 100mHz.
8: Set the ‘FM Dev’ to Half your sweep range ( so for a 0Hz to 20KHz sweep set 10KHz again).
9: Set the “Shape” to Arb.
10: Enable the output.

Note: It should be possible import sweep curves from Excel or other apps as CSV files into Easywave and then download them to the SDG, you will probably need to add the header information which you can see by looking a saved wave. But I haven’t tried this.

[nixxon]

I added the sine curve data as well, simply by adding a sine function in the excel file. I threw in Vpp and DC offset as well. Enclosed is the new excel-file for anyone to have fun with. Even if you know very little about excel, it is OK to change the values in the green cells.

The graph below shows a 2 second log sweep from 1 to 100 Hz with 2Vpp and 1V offset. It is based on a table with 1 mS increments. (2000 values).

It should be easy to export the 2 relevant columns (A (timecode) & C (Voltage level)) as a csv file (with correct headings etc.) and then import this csv-file into the AWG.

[Fennec]

Here you go. It's just a crude excel workbook.

But it works with OOO and I can play with them. Thank you

[nixxon]

Here you go. It's just a crude excel workbook.

But it works with OOO and I can play with them. Thank you

Cool. What is "OOO"?

[Fennec]

I dont't use Windows. Im on Linux.
OOO means Open Office Org...   But the better one is LibreOffice. Both are free and both works fine for me. I don't need this M$ crap.  Works on Windows, Linux, MAC

http://www.libreoffice.org/

http://www.openoffice.org/

[commongrounder]

MasterOfNone  Thank you for the description of the workaround using FM.  It does have the advantage of allowing any "shape" (i.e. sweep curve) to be used, although I have to say it is clunky to implement and adjust on the fly.  Also there is the issue of output triggering at the beginning of each sweep pass, which is essential for my work plans. I would prefer that Siglent correct the improper curve of the built in log sweep.
nixxon  It would be interesting to see if you could create this decade sweep curve in a CSV file that fit the 16k point record length for the SDG1000 series arb files.  Haven't messed with Easywave to see what values are used for the levels (14 bit, plus/minus 8192 levels, or plus/minus 3 volts)?

[rf-loop]

Also there is the issue of output triggering at the beginning of each sweep pass, which is essential for my work plans.

Is it possible you explain what is this "issue"?
It do not work as you hope or is there some real error?