DIY guitar preamp in stompbox makes oscillate as hell

[chilidawg]

Not sure if this applies to your design, but I do know of voltage inverters/charge pumps in other guitar pedal designs (MAX1044 and others) that naturally oscillate at 10KHz unless pin 1 & 8 are connected to each other. Worth checking the data sheet for anything that might allude to this.

Doesn't apply to his DC-DC converter TRACO TEC 3-0923.
Data sheet doesn't mention anything like that.

 

[Chandra]

Does OP series opamps need minimum gain? Why not try replacing with TLO series?
Regards.

 

[chilidawg]

No electrolytic cap was reversed. And the last test was without electrolytic caps.

So. I Assembled only few components and tested. Power supply section and the first stage.

input signal goes through OP2_A and pin 1 connected to c4 - output. To turn off other section of opamp, I tied opamp's pin 6, 7 and connected to GND.

In this time oscillation still occurs.

You TIED opamp2_B pin 5, 6, 7 to GND?
Proper terminating an unused op-amp is already drawn in your schematic (opamp7_B)

Btw, I compared your schematic and board layout with the Fender's. It seems everything is drawn correctly. I did notice one thing though: you HAVE TOO MANY electrolytics for each supply rails. You probably don't even need one imo, because the 0.1uF ceramic cap is more than sufficient for bypass.

I have designed a much larger board (16" long x 4" wide) for a commercial mass-produced talkback mic amp + 4 channel headphone amp distribution unit. Also used a DC-DC converter in it, a buck/boost IC switching at 150kHz and a non-isolated design. Aside from the medium size output caps (470uF/35V) that are part of the converter design, I didn't place any small electrolytics everywhere else on the supply rails. Just 0.1uF X7R ceramic caps next to the op-amps supply pins. When I scoped for any unwanted noise in the audio signal band, I found nothing. It was very clean, even when powered from a cheap 5V phone charger or a USB computer port that everyone knows to be very noisy.

 

[abbey road d enfer]

Thank you for your reply!

You mean supply board's power with batteries? Or supply opamp directly through batteries?
So. I Assembled only few components and tested. Power supply section and the first stage.

input signal goes through OP2_A and pin 1 connected to c4 - output. To turn off other section of opamp, I tied opamp's pin 6, 7 and connected to GND.

Do you really mean you connected pins 6&7 to ground?

 

[Newmarket]

Thank you for your reply!

You mean supply board's power with batteries? Or supply opamp directly through batteries?

Okay. Thought this already done. But I meant simply use 2x9V batteries to supply +/-9V rails. As others have said and I think you're waiting to do now ?
To demonstrate if problem is due to the PSU or the circuit itself.

 

[freebird92]

You TIED opamp2_B pin 5, 6, 7 to GND?
Proper terminating an unused op-amp is already drawn in your schematic (opamp7_B)

Btw, I compared your schematic and board layout with the Fender's. It seems everything is drawn correctly. I did notice one thing though: you HAVE TOO MANY electrolytics for each supply rails. You probably don't even need one imo, because the 0.1uF ceramic cap is more than sufficient for bypass.

I have designed a much larger board (16" long x 4" wide) for a commercial mass-produced talkback mic amp + 4 channel headphone amp distribution unit. Also used a DC-DC converter in it, a buck/boost IC switching at 150kHz and a non-isolated design. Aside from the medium size output caps (470uF/35V) that are part of the converter design, I didn't place any small electrolytics everywhere else on the supply rails. Just 0.1uF X7R ceramic caps next to the op-amps supply pins. When I scoped for any unwanted noise in the audio signal band, I found nothing. It was very clean, even when powered from a cheap 5V phone charger or a USB computer port that everyone knows to be very noisy.

Do you really mean you connected pins 6&7 to ground?

I meant 5 and 6! 6, 7 was a typing error.
I didn't know that was wrong way to terminate opamp. I'll try proper termination to OPAMP2 B and report here.
And this time, I learn that too many caps could be and error.
I will use less electrolytic caps for board and use only 0.1uF mlcc for opamp's supply pins
Thank you!

Does OP series opamps need minimum gain? Why not try replacing with TLO series?
Regards.

I used TL072, OP meant just OP-AMP!

 

[freebird92]

Okay. Thought this already done. But I meant simply use 2x9V batteries to supply +/-9V rails. As others have said and I think you're waiting to do now ?
To demonstrate if problem is due to the PSU or the circuit itself.

Yes I am waiting. It is weekend so it will take some days. Thank you!

 

[abbey road d enfer]

I meant 5 and 6! 6, 7 was a typing error.

On a TL072, pins 5 & 6 are the non-inverting and inverting inputs respectively. Tying them both to ground is not correct. Only pin 5 should be grounded, and 6 & 7 tied together.

 

[freebird92]

Hi.

I guess there are some battery snaps in my stomp boxes.

So, I opened up stomp box and found them. I cut 2 battery snap from my stomp box !
I removed dc-dc converter and connect each batteries to positive and negative supplies.

It works. So, the DC- DC power supply section was the suspect.

So I try another board that has less components like below picture.
This time I tie OP-AMP's pin 6,7 together and pin 5 connect to the ground.
there were no voltage drop. But when I plug the board into the amp, it starts to make noise.
When I tried with another DC-DC converters the result was same.





It looks certain that the board's power supply section made the oscillation.
But I still wonder that why the oscillation occurs even when I removed all of electrolytic caps.

It made me frustrated. But I think I learned something from this experience and from all of you. Thank you!

Now, I want to draw new board.
Is there any guidelines to make proper design a board that has dc-dc converters?

 

[jokeramik]

I think the main problem is the noise on the rails from the converter: Ripple and Noise - 20 MHz Bandwidth 75 mVp-p typ.
That´s a lot. It´s important to decouple the rails of every OPA or make special stabilzation for the amp so no noise comes to the unit.
Switching frequency of the converter is around 100kHz.

Best regards!
jokeramik

 

[abbey road d enfer]

It works.

Can you measue the voltages on the opamps pins, in particular around IC6 (reverb tank driver).

So, the DC- DC power supply section was the suspect.

Don't jump to conclusions. It's the combination of this particular converter with this particular circuit. These converters usually work; when they don't it's because something is inadequate in the circuit they are connected to. You have to find what.
Batteries are much more tolerant of load than converters. Measure the current draw with a multimeter.

 

[abbey road d enfer]

I think the main problem is the noise on the rails from the converter: Ripple and Noise - 20 MHz Bandwidth 75 mVp-p typ. Switching frequency of the converter is around 100kHz.

The problem here is not at 100kHz. It's clearly an oscillation at a medium frequency. Does it come from the converter itself (which works unloaded) or from an interaction between the converter and the circuit? It seems to be the latter. But in absence of oscilloscope clues, it's very hard to determine.

It´s important to decouple the rails of every OPA or make special stabilzation for the amp so no noise comes to the unit.

The oscillation happens even with 180uF decoupling capacitance.

 

[freebird92]

Can you measue the voltages on the opamps pins, in particular around IC6 (reverb tank driver).

Thank you! I'll try it! I'm just worried that I'm bothering you with too many questions.
Then measure voltages with dc-dc converter or batteries or both??!

I think the main problem is the noise on the rails from the converter: Ripple and Noise - 20 MHz Bandwidth 75 mVp-p typ.
That´s a lot. It´s important to decouple the rails of every OPA or make special stabilzation for the amp so no noise comes to the unit.
Switching frequency of the converter is around 100kHz.

Best regards!
jokeramik

Thank you for your reply.
I made some boards with TEC3 series converters.(The converter I used in this time.)
To reduce ripple, I used some LC filters and decoupling caps. These boards worked right and hard to listen any noise.
But this time it doesn't work and I don't know what is matter.

 

[abbey road d enfer]

Then measure voltages with dc-dc converter or batteries or both??!

With the batteries.

But this time it doesn't work and I don't know what is matter.

That's what we're trying to find out here.

 

[chilidawg]

It works. So, the DC- DC power supply section was the suspect.

Sorry, did you mean with the battery connected to a full populated board?

To reduce ripple, I used some LC filters and decoupling caps.

These converters are usually pretty quiet without any additional LC filters. I know there are some people who sneered at them and quickly judged that they're noisy stuff simply because the switching freq is not in the MHz band.

Now, I want to draw new board.
Is there any guidelines to make proper design a board that has dc-dc converters?

I always follow a simple rule that a friend taught me when I design a 2 layers board. Signal on the top layer, ground plane at the bottom layer, guard sensitive signal traces with a thick ground trace, and keep the amount of vias to a minimum (always try to find a way to route directly from one point to another point without breaking the trace or the need to switch between layers, you can do this by moving components around or arranging them, then you will see a path opened up), and most of my vias are direct connections to ground plane.

This is the board that I mentioned in my previous post where I used a non-isolated DC-DC converter switching at 150kHz, and yes, I put it next to my stereo line input jack connectors. You can see the audio traces from the connectors running on the left edge of the PCB. Was that a bad idea? Maybe, but there's no unwanted noise whatsoever in the audio signal, so who cares...

 

[Bo Deadly]

I removed dc-dc converter and connect each batteries to positive and negative supplies.

It works. So, the DC- DC power supply section was the suspect.
...
Now, I want to draw new board.
Is there any guidelines to make proper design a board that has dc-dc converters?

That circuit can be made to work. SMPS are just a little fickle.

Try the following:

1) Add 1k resistors across each output of the SMPS so that each 15V supply is loaded with 15mA (or probably a little less since the supply voltage will drop a bit). SMPS work better under load. Even at 100%. You can also try more load like 500 ohms to load with what would probably be 26-28mA or so (but in this case you would need to parallel two 1k if you only have 1/4W resistors).

2) Replace the inductors with 47 ohm resistors. You will loose a volt or so but that's not really a problem for this circuit.

3) Remove all of the supply electrolytics except C27 and C29.

I would bet a guitar pick that fixes it.

For a future board, you should use capacitance multipliers after the SMPS. CMs nicely isolate the SMPS from the load and completely filter away the high frequency switching noise.

 

[freebird92]

With the batteries.

That's what we're trying to find out here.

Here is my report!

OP6 (reverb driver )Voltage measure // with 9V batteries supply

Pin8 (vcc+) 8.9 V
Pin4 (vcc-) -8.68 V

no input jack plugged
I removed all bypass capacitors(include 0.1uF MLCCs).

volume full
1 685mVAC
2 26mVAC
3 29mVAC

5 100mVAC
6 48mVAC
7 6.8VAC

Volume 0
1 -6mVAC
2 -6mVAC
3 -6mVAC
5 0V
6 20mVAC
7 0V

I recorded a demo.
0:00 ~ 0:10 reverb 0
0:11 ~ 0:18 reverb about 2 or 3
0:18~ reverb full
Volume knob is about 1(7:30 O'clock). When I increase volume over 9 O'clock I make IR loader clip.

audio demo(my blog link) 퇴역군인 : 네이버 블로그


Thank you for your kind advice!

Sorry, did you mean with the battery connected to a full populated board?

These converters are usually pretty quiet without any additional LC filters. I know there are some people who sneered at them and quickly judged that they're noisy stuff simply because the switching freq is not in the MHz band.

I always follow a simple rule that a friend taught me when I design a 2 layers board. Signal on the top layer, ground plane at the bottom layer, guard sensitive signal traces with a thick ground trace, and keep the amount of vias to a minimum (always try to find a way to route directly from one point to another point without breaking the trace or the need to switch between layers, you can do this by moving components around or arranging them, then you will see a path opened up), and most of my vias are direct connections to ground plane.

This is the board that I mentioned in my previous post where I used a non-isolated DC-DC converter switching at 150kHz, and yes, I put it next to my stereo line input jack connectors. You can see the audio traces from the connectors running on the left edge of the PCB. Was that a bad idea? Maybe, but there's no unwanted noise whatsoever in the audio signal, so who cares...

Yes, battery connected to the full populated board and it works.
Thank you for showing me the specific picture along with the advice. It will be a lot of reference.
I will keep your advice in mind when I make a board next time.
Thank you for your kind advice!

That circuit can be made to work. SMPS are just a little fickle.

Try the following:

1) Add 1k resistors across each output of the SMPS so that each 15V supply is loaded with 15mA (or probably a little less since the supply voltage will drop a bit). SMPS work better under load. Even at 100%. You can also try more load like 500 ohms to load with what would probably be 26-28mA or so (but in this case you would need to parallel two 1k if you only have 1/4W resistors).

2) Replace the inductors with 47 ohm resistors. You will loose a volt or so but that's not really a problem for this circuit.

3) Remove all of the supply electrolytics except C27 and C29.

I would bet a guitar pick that fixes it.

For a future board, you should use capacitance multipliers after the SMPS. CMs nicely isolate the SMPS from the load and completely filter away the high frequency switching noise.

Okay! I'll try it according to your advice.
Does "1K resistors across output of SMPS" mean
1.
+15V - 1K - GND
-15V -1K - GND

or

2.
+15V - 1K - -15K

When it's done, I'll report in this thread!

Anyway, The CMs cause some voltage drop. But would it not matter if it was designed like this?

Thank you for your kind advice!

 

[abbey road d enfer]

volume full

Is it the pot labelled "normalvol" (50kC)?

1 685mVAC
2 26mVAC
3 29mVAC

5 100mVAC
6 48mVAC
7 6.8VAC

Volume 0
1 -6mVAC
2 -6mVAC
3 -6mVAC
5 0V
6 20mVAC
7 0V

I asked you to measure the DC voltages.

 

[freebird92]

+ I fitted DC-DC converter and plugged DC jack of 9V 300mA adapter.


Now the board can deliver guitar sound to next stage.

It make some noise like boiling kettle. (like file below)
But it doesn't work right.
the adapter supply 8.93DC
but DC-DC converter make
+5.946 VDC 225mVAC
-5.963 VDC 237mVAC

demo file (file from my blog)

+ I take 12V 2A adapter.

DC converter makes +- 9V DC
When I turn Volume pot(C50K) output voltage and frequency varies.

When I turn to 0(full ccw) 0VAC

When I turn to CW it goes to 0.13VAC 1kHz

 

[freebird92]

Is it the pot labelled "normalvol" (50kC)?


I asked you to measure the DC voltages.

Yes. It is

DC voltage is (with batteries)

1 -0.006V
2 - 0.006v
3 -0.004V
4 -8.72V
5 0V
6 0.001V
7 0.006V
8 8.46V