Hack my function generator to get at sweep and sweep trigger

[jancumps]

Inspired by the blog on bode drawing on an oscilloscope I started a hack on my function generator.
It doesn't have outputs for sweep signals. I'd like to use the sweep signal as trigger, and if possible also as X-axis in XY mode.

I had a look at the schematics to see where the sweep signals are generated. Didn't search too hard - figured that if I know approximately where these signals should be, I can poke around with my scope to find them.
I have some success: I found the sweep signal, both for log and linear sweep, and a nice negative pulse at the sweep start that can be usefull for triggering - and it is long enough to be used as Z signal to black out a scope trace  n XYZ mode (I couldn't test this as my scope doesn't have a Z).

I attached pictures of the signals and the generator's output in linear and logarithmic mode, schematic and nudies of the PCB.
Next step is to find out how to buffer the signals, remove the DC component and give them the right impedance so that they can be safely used as outputs.

[Lightages]

Your image of the schematic is too low resolution to read easily, and you have omitted too much to see where the best point of connection might be.

[jancumps]

Your image of the schematic is too low resolution to read easily, and you have omitted too much to see where the best point of connection might be.

Unfortunately, that is the resolution I have to work with. I can't find a better resolution schematic for my unit. I had restricted the view area in my post to limit bandwidth. I attached the complete schematic to this post, and a higher resolution picture of the signals in linear mode.

I had to use the same blurry schematic to repair this unit. The quality of the scan was more difficult than finding and fixing the issue itself (output TR1 and driver TR3 transistor shot, see last photo).

[jancumps]

The glass of whine you see in the first photo helps to get a sharper view of the schematics.

[bitwelder]

The glass of whine you see in the first photo helps to get a sharper view of the schematics

The glass of what? 

[SeanB]

The somewhat consumed glass of what I presume is a fine French red wine in the right hand side.

[jancumps]

The glass of what? 

Alcohol impacts my writing skills.

[w2aew]

Seems to me that you're basically there.  You've identified the signals, and been able to get them on the scope. That's all they'll have to drive, so there shouldn't be much more that you'll need to do.  You could add simple emitter follower or unity gain opamp buffers if you want, but since you have demonstrated what you want w/o adding them, it's probably not necessary.

In case you're interested, here's a video I did a few months ago showing the same basic sweep application that Dave recently showed:

[alm]

Seems to me that you're basically there.  You've identified the signals, and been able to get them on the scope. That's all they'll have to drive, so there shouldn't be much more that you'll need to do.

Unless he wants to use standard 50 ohm coaxial cables to connect the trigger output to the scope. These will present either a capacitive or 50 ohm resistive mode, both of which may upset the circuit. I glanced at the schematic earlier and my impression was that they would have a fairly high output impedance.

[jancumps]

Seems to me that you're basically there.  You've identified the signals, and been able to get them on the scope. That's all they'll have to drive, so there shouldn't be much more that you'll need to do.

Unless he wants to use standard 50 ohm coaxial cables to connect the trigger output to the scope. These will present either a capacitive or 50 ohm resistive mode, both of which may upset the circuit. I glanced at the schematic earlier and my impression was that they would have a fairly high output impedance.

I have seen that video. Think it was refered to in the blog here about Bode drawing.

I 'd like to buffer the signals for several reasons:
-the sweep ramp signal has a DC component that I want to get rid off
- getting it to same standard BNC impedance.
- protect my generator from shortcuts or stray signals on these outputs.

The frequency of these signals isn't that high that it would need hf components.
The sweep trigger signal is +5V, with a trigger signal going to 0 at sweep start.
The sweep ramp is 10V, with a -15V DC component.
Voltage rails available in the scope: +/- 21, 15 and 5 V.
Advice is welcome!

[SeanB]

741's run off the 21V rails will do, you can just use them as buffers to drive the coax via a 47 ohm resistor. If you want to remove the offset use 2, the first being an inverter ( virtual earth) with the sweep sent in via a resistor, and another resistor connected to a source between 0 and 21V to provide the DC offset and the opamp configured with a gain of -1. Another with a gain of -1 will invert it again. To buffer the sweep trigger just use a 741 as a buffer.

[jancumps]

I have been mulling over this for a while and tried to use an opamp adder to offset the -15V component in the sweep signal with a +15V signal.

Photo 1 shows the signal coming out of the generator: a -15V to -5V sweep.

Photo 2 shows the schematic that I tried: two opamp stages, both with gain=1:
- an opamp adder: It adds - and inverts -  the sweep signal and a +15V DC signal.
- an opamp inverter: inverts the output from the first opamp, resulting in the sweep signal from 0V to +10V.
As power I used +15V, 0, -15V symetric (this is also available in the generator).
I didn't use the scope sweep signal and power rails for testing (wanted to proove and verify the setup first), but simulated this signal with the normal generator outputting a 10V sawtooth with a -15V offset. You can see in the scope photo 4 that the generator does not go to -15V completely, it is more like -14.5V.
I have added a 50 Ohm resistor to the output (this is my naive attempt to get 50 Ohm impedance - need to do some additional reading to check if that is correct - it's been a while).

Photo 3 shows the circuit on a breadboard, using two 741s

Photo 4 shows input and output on the scope. The bottom signal is the input, between -15V and -5V, the top signal is the processed signal, from 0V to +10V.

This is what I wanted to get. I'll have to do some more measurements and stress testing.

[jancumps]

Parts have arrived, ready to go

• 2 BNC connectors and the right coax cable
• LM324N 4 x opamp (I need 3: one to buffer the gate signal, 2 to remove the negative dc component in the sweep. 1 unused, maybe in future that one I can use to suppress the sweep's falling slope with the gate signal
• Resistors.

I had a  TriPad board lying around.

[jancumps]

Circuit built and tested.
I'm going to prepare the mount points for the pcb, and drill the holes in the back of the unit for the two BNC connectors.
The BNC mounting hole is a bit odd:

I don't have a drill in that format   - so it's probably going to be a round hole. Hope the connector doesn't get loose when I start to use it.

If anyone here has experience making such a hole in an iron plate casing with simple tools (I have drill and a round file*), please chime in.


* looked this up in some translator. Dutch Vijl translates in English File. Hope it is right

[SeanB]

Drill the hole the diameter of the smallest dimension of the connector, around 9mm, then use the file to get it large enough for the connector leaving the flat to lock it. I have good success just drilling it to be slightly undersize then filing it larger until you can just start the connector in the hole and turn it in with a little force. Then place the lock washer and solder lug, and then the nut, then tighten it up and lock with clear nail varnish ( cheap, good for this and easily available, plus it is good for locking trimpots) and you are done and it will not come loose.

[w2aew]

Good job on the circuit board.  The only thing I'd recommend is putting local decoupling capacitors on the power supply lines on the board.  Good practice.

[PA4TIM]

I liked your video, very clear explanation and demonstration but I would like to add some comments.
Filter shapes are influenced by the impedance termination and connection of used clips and cables. Their capacitance and inductance makes them part of the filter.

In this case the inductor of the can gets ectra inductance from the two testclips and leads, those clips also have capacitance so you get a total L with a small cap parallel, and to taps  with an other cap parallel loaded by 50 Ohm and this gives a Phaseshift/delay so the tank will be on an other frequency as on the scope. ( if mounted back in a receiver). Also for the output.

If you couple the signal in through a small cap ( so there is minimal loading) somewhere before an amplifier before the filter and look with a 10x probe, high Z or other low loading probe after some amp the filter will stay terminated the right way. Just try it, tune the filter and then play with the connections and loading on both sides to see the effects. On 1 MHz they will be minor but on 40 MHz or 400 MHz they will be huge. I did some tests on 50 MHz while measuring a teslalens and the very light inductive coupling between the tanks ( all tuned excaxt at 50 MHz) caused coupled together a 3 MHz frequency shift.

Second " problem" is you have a linear vertical axis. If you add a diodedetector/RFhead  to transform the response into DC and then make a little log amp with an opamp you get a nice logarithmic picture from the outline of the filter. Just like a netwotk analyser. For real filters with 40 dB attenuation you need this. If the response is 1 V and you use 5 vertical devidions you see 200 mV /division. If you want to see bandpass ripple the details are lost. 40 dB (in power, so in a 50 Ohm environment) is 10000X so if you have a 10 dB passband ripple this will be 10X. So 10 dB will be 10 mV ( no calculator on hand so excuse me if I go wrong on the numbers)  10 mV on 200 mV/div is will look very nice , you probably will not see it at all if you do not use a log detector. So you will be pleased while the result is dramatic bad, because in real life it should be less then 1 dB. Also harmonics that are 30-40 dB down, you will not see. An analyser shows a dynamic range around 80 dB or more.

Just two tips to make the sweeping even more usefull.

[jancumps]

Everything is built in now.
I used the fuse holes on the back of the unit as mount holes for BNC connectors. I'll have to open the unit in the future if one of these fuses blows.
When not in use, the connections are covered by a plastic door.

[jancumps]

I've been testing the hack with an RC high pass filter.
For this picture, I used the gate signal as external trigger, put the sweep to log. Fine-tuned the time/div so that one full sweep fills one screen. Sweep from 600 Hz to 20 kHz.

The only thing I do not have at this moment is a log vertical scale.
I will have to search a bit to find a broadband/wide voltage range linear to log converter.