My studio Mic compressor build

[Rob Strand]

Oh wow.  Where do we start! 

I'm not sure I trust any of those meters.

One thing though, when you did the measurements did you use the same multimeter leads?  Maybe the leads are broken?

The flickering on the red meter looks worse than leads.

The flickering on black meter could be the meter's auto ranging getting confused.   I've seen that happen.    In some cases  when the meter changes ranges the meter impedance changes then that changes the circuit voltage and the meter thinks it can switch down to the previous range.   It then gets stuck changing ranges and never settles on a measurement.

Here are all the pins:
Pin 1: Oscillates around 0.5V (less than 50 mV each way)
Pin 2: 1.4mV
Pin 3: 0V
Pin 4: 15.23V
Pin 5: 50mV
Pin 6: -0.637V
Pin 7: 4.21V
Pin 8: Oscillates around -0.9V
Pin 9: Oscillates around -0.35V
Pin 10: -11.85V
Pin 11: -14.89V
Pin 12: Oscillates around -0.35V
Pin 13: Oscillates around -0.43V
Pin 14: Oscillates around -0.37V

All sort of inconsistencies with these measurements.   If the meter is producing bad readings then that would explain a lot of the inconsistencies.
 
The only good thing I can extract from those measurement is pins 12, 13 and 14 now look like they are near zero volts - at least within the meter wobble.    It is still possible we have oscillation at IC5a and without the diodes the output at pin 14 is oscillating but averaging out to zero.   With the diodes in it's possible the oscillation was being rectified and that caused pin 14 to go negative.

Maybe leave the circuit with the diodes shorted then use your oscilloscope to probe the pins of IC5a and IC5d.  No signal input is fine.  What you are looking for is DC levels or very high frequency signals (perhaps 100kHz to 1MHz).    I hope your oscilloscope can test as high as 1MHz ?

As for the meters.  It's a big problem if you can't trust your meter.   Look into the leads being a problem first.    One test I can think of is get a 9V to 12V supply then connect it up to a 100k pot as a voltage divider (volume control).  Make sure you can dial up voltages 0, 0.5, 1.0, 1.5, 2.5 ... 12.0  on the pot and that the meter reading follows the pot position.   If the meter reads 4V when the pot is on "1" (pot = 0 to 10 range)  then clearly  the meter readings are jumping all over the place.  Next reverse the power connection and check voltages 0, -0.5, -1.0, 2.5, ... -12.0.

That will give you some idea of how much you can trust the meters!

[jfrabat]

Well, if you noticed, the red meter does not have ANY leads on it...  so it is certainly not the leads.  I just noticed that today, and I came to the same conclusion you did; I cannot trust those readings.  But the other 2 meters get stuck in auto range hell.

As for the scope, it only goes up to 2K (it is one of those $30 kit deals).  Now I wish I had a real one!

I will try the pot and power source thing tomorrow.  I should also have the second board ready to compare it to this board.

[11-90-an]

I made one of those LED voltage indicators that light up when given voltage... say,

4.5v, 2 LEDs light up
6v, 3 LEDs light up,
9v, 4 LEDs light up,
10v, 5 LEDs light up

But of course, it isn't accurate at all. Just for knowing if my battery needs recharging or not... 

[jfrabat]

One thing though, when you did the measurements did you use the same multimeter leads?  Maybe the leads are broken?

Yes, but these are the newest leads I got.

As for the meters.  It's a big problem if you can't trust your meter.   Look into the leads being a problem first.    One test I can think of is get a 9V to 12V supply then connect it up to a 100k pot as a voltage divider (volume control).  Make sure you can dial up voltages 0, 0.5, 1.0, 1.5, 2.5 ... 12.0  on the pot and that the meter reading follows the pot position.   If the meter reads 4V when the pot is on "1" (pot = 0 to 10 range)  then clearly  the meter readings are jumping all over the place.  Next reverse the power connection and check voltages 0, -0.5, -1.0, 2.5, ... -12.0.

That will give you some idea of how much you can trust the meters!

I used the same leads for all the tests.  I used a dual gang pot simply because it was the biggest one I had (easier to work with), but I only used one gang.  Notice I made a little mark on the pot to tell the orientation (I tried to repeat the same positions as much as possible), but I measured each MM and then repeated for the other, which means the pot position is not IDENTICAL.  I may repeat it with an A100K pot I have that has clicks to it (or change MM and keep the pot position).  This is what we have:

POSITION 1 (full counter clockwise):

OLD METER (Red and Black):


NEW METER (Black and Orange):


POSITION 2 (about 10:30):

OLD METER (Red and Black):


NEW METER (Black and Orange):


POSITION 3 (12 o'clock):

OLD METER (Red and Black):


NEW METER (Black and Orange):


POSITION 4 (1:30):

OLD METER (Red and Black):


NEW METER (Black and Orange):


POSITION 5 (3 o'clock):

OLD METER (Red and Black):


NEW METER (Black and Orange):


POSITION 6 (full clockwise):

OLD METER (Red and Black):


NEW METER (Black and Orange):


My (inexperienced) conclusion is that the red one seems more accurate, but neither is waaay off.  Since I did one MM at a time, the position of the pot is not EXACTLY the same for both MM, and considering how sensitive the pot is, I can see where the discrepancies come from.

[jfrabat]

Got the second board far enough to be able to measure it.



It is also showing -13V on Pin 14 IC5

[Rob Strand]

I used the same leads for all the tests.  I used a dual gang pot simply because it was the biggest one I had (easier to work with), but I only used one gang.  Notice I made a little mark on the pot to tell the orientation (I tried to repeat the same positions as much as possible), but I measured each MM and then repeated for the other, which means the pot position is not IDENTICAL.  I may repeat it with an A100K pot I have that has clicks to it (or change MM and keep the pot position). 

The main thing is both meters are in the same ball-park.  I would not make any conclusions about accuracy as using a pot like that *and* trying to get the pot in the same position is *very* inaccurate.   The initial measurement at 12V is probably the only one we can use and the meters agree within 1%, which is about all I would expect from those meters.   The red one agrees with the power supply meter but we don't know the accuracy of the power supply.

If I knew you had that bench power supply it would have much easier!    Just connect the meter to the bench power supply an vary the voltage over whole range.   Compare the power supply meter  with multimeter.      Those power supply voltage meters aren't that great but at least it's a "third opinion" that can be used to compare the meters.

More importantly though,  both meters didn't show or any crazy jittering like we see when measuring your circuit?

Also, that's bad news about the second build.

So there could be a problem with the layout or both circuits could be oscillating.    Problems with the layout can also cause oscillations.  It's more likely on a larger board like this.

Since your oscilloscope only goes up to 2kHz it's not possible to see oscillations.   You can build a little test circuit  to check for oscillations, it's only three or four parts.   Another idea is your black meter has a frequency range "Hz" you might be able to measure the output of pin 1 IC5a.   Try measuring frequency on the old board with the diodes shorted and also the new board with the diodes wired normally.    The frequency measurement isn't very convincing.

Another idea is to measure the voltages on IC6b, IC6c, D1 D3.    The circuit is similar to to the problem part of the circuit around IC5a & IC5d.    The aim of this test is to see if you can get sensible measurements in this part of the circuit with your multimeter.  The compressor circuit seems to be working so that would be a good test for the multimeters.

[jfrabat]

More importantly though,  both meters didn't show or any crazy jittering like we see when measuring your circuit?

No.  Red has jitters when unplugged, but with voltage, it is solid.  No movement at all (at most, 1mV).

Also, that's bad news about the second build.

So there could be a problem with the layout or both circuits could be oscillating.    Problems with the layout can also cause oscillations.  It's more likely on a larger board like this.

I figured something along those lines.  Odd that only this OpAmp is doing it, but none of the others (well, I only measured pin 14 on the new board, so I am not sure this one does not have OTHER issues!).  Anyway, having 2 compressors to daisy chain is never bad in my book, so at worse, I will have 2 in line compressors and no noise gate!  But I would really like to get them working properly!

Since your oscilloscope only goes up to 2kHz it's not possible to see oscillations.   You can build a little test circuit  to check for oscillations, it's only three or four parts.

Do you have a circuit?  If I got the parts, I can get working on that.

Another idea is your black meter has a frequency range "Hz" you might be able to measure the output of pin 1 IC5a.   Try measuring frequency on the old board with the diodes shorted and also the new board with the diodes wired normally.    The frequency measurement isn't very convincing.

So the above is a better idea?  I could always try this one, but I don't mind building a simple test circuit.

Another idea is to measure the voltages on IC6b, IC6c, D1 D3.    The circuit is similar to to the problem part of the circuit around IC5a & IC5d.    The aim of this test is to see if you can get sensible measurements in this part of the circuit with your multimeter.  The compressor circuit seems to be working so that would be a good test for the multimeters.

OLD BOARD (red MM):
IC6 Pin 5: -7mV
IC6 Pin 6: +4mV
IC6 Pin 7: +5.5mV
IC6 Pin 8: -91.4mV
IC6 Pin 9: +0.8mV
IC6 Pin 10: +0.1mV
D1c: +3.1mV
D1a: -77.5mV
D2c: -77.5mV
D2a: -40.4mV

OLD BOARD (black/orange MM):
IC6 Pin 5: -27.4mV
IC6 Pin 6: +3.4mV
IC6 Pin 7: +3.9mV
IC6 Pin 8: -76.7mV
IC6 Pin 9: +0.9mV
IC6 Pin 10: +0.1mV
D1c: +16.1mV
D1a: -45mV
D2c: -45mV
D2a: -28mV

NEW BOARD (black/orange MM):
IC6 Pin 5: -119mV
IC6 Pin 6: -79mV
IC6 Pin 7: -267mV
IC6 Pin 8: -83mV
IC6 Pin 9: +0.4mV
IC6 Pin 10: -0.8mV
D1c: +88mV
D1a: +77mV
D2c: +77mV
D2a: -99mV

EDIT: For reference, here is the new board's IC5 voltages (black and orange MM):
IC5 Pin 1: +2.22V
IC5 Pin 2: +0.4mV
IC5 Pin 3: -0.9mV
IC5 Pin 4: 15.35V
IC5 Pin 5: +5.8mV
IC5 Pin 6: -5.8mV
IC5 Pin 7: -5.11mV
IC5 Pin 8: 2.22V
IC5 Pin 9: -12.83V
IC5 Pin 10: -12V
IC5 Pin 11: -14.97V
IC5 Pin 12: -5.25V
IC5 Pin 13: -4.09V
IC5 Pin 14: -12.6V

D3c: +4.2V
D3a: +2.7V
D4c: +2.7V
D4a: -4.9V

[11-90-an]

if you guys up to the challenge, you can probably refer to the board layout Felipe posted here... assuming that's the exact same one?   

Maybe somebody may find a problem there.

[jfrabat]

if you guys up to the challenge, you can probably refer to the board layout Felipe posted here... assuming that's the exact same one?   

Maybe somebody may find a problem there.

Yup, that's the one...

[Rob Strand]

No.  Red has jitters when unplugged, but with voltage, it is solid.  No movement at all (at most, 1mV).

OK at least the measurements are stable.

Do you have a circuit?  If I got the parts, I can get working on that.

Here's *something*.   However,  we might have to reduce the caps in order to measure oscillation but reject mains hum.



Oh it's only good for connecting to opamp outputs like pin 1.  Elsewhere it could contribute to odd effects.

I should say how to use it:

- Connect the DMM where R2 is.
- Connect the test circuit ground to the circuit ground
- Connect the "probe" to the point you want to test.

If the DMM measures a DC voltage that means there's an AC voltage at the probe.

So the above is a better idea?  I could always try this one, but I don't mind building a simple test circuit.

Worthwhile doing both.   The more evidence the better.

I figured something along those lines.  Odd that only this OpAmp is doing it, but none of the others (well, I only measured pin 14 on the new board, so I am not sure this one does not have OTHER issues!).  Anyway, having 2 compressors to daisy chain is never bad in my book, so at worse, I will have 2 in line compressors and no noise gate!  But I would really like to get them working properly!

IC6 Pin 5: -7mV
IC6 Pin 6: +4mV

etc...

Yes it not looking good.   Two boards with the same problem.   The IC6  measurements show the somewhat different, and more like expected, behaviour on both boards.

It's all pointing to some local evil problem around IC5a, IC5d.

if you guys up to the challenge, you can probably refer to the board layout Felipe posted here... assuming that's the exact same one?   

Maybe somebody may find a problem there.

I found it hard to follow from those pics.     I'm lucky to follow what I think is there never mind finding what might be wrong.

I noticed a few things:

- I can see is the tracks are quite long and running close to unrelated parts of the circuit.   That could promote oscillations.

- I'm not 100% sure but it looks like C8 is wired to IC5 pin2 and not R64.   Like C8 and R64 have been swapped.  If that's the case the PCB doesn't match the schematic.  Very weird!

- Again, not 100% sure but the value of R64 looks like 4k7.   It could be the JPEG colors.   Red bands on blue resistors are often hard to read from photos.



EDITED:
Added explanation of test ckt.

[11-90-an]

I noticed a few things:
- I can see is the tracks are quite long and running close to unrelated parts of the circuit.   That could promote oscillations.
- I'm not 100% sure but it looks like C8 is wired to IC5 pin2 and not R64.   Like C8 and R64 have been swapped.  If that's the case the PCB doesn't match the schematic.  Very weird!
- Again, not 100% sure but the value of R64 looks like 4k7.   It could be the JPEG colors.   Red bands on blue resistors are often hard to read from photos.

Go for it, Felipe! there might be some hope! 

[Rob Strand]

I had a few thoughts which had promised but all that quickly dissappeared when I went over the stuff we have checked already.   Reply #56, where you checked all the connections and values, killed all my ideas 

In reply #56 it checks pin 2 IC5 going to R64, so I don't know why I though I saw pin 2 IC5 going to C8 (not that I was 100% sure).

With diode D3 and D4 shorted it makes IC5a operate independently of IC5d.   We are still seeing junk on pin 1 IC5a.   There's not much circuit around IC5a.

So after all that the only thing  left is to check the value of R64.   If R64 is 4.7k or shorted that might cause a lot of interference in IC5a.

It would be worthwhile to the continuity between IC5 pin 3 (at the top of the IC) and the negative terminal of C24.
 

[jfrabat]

I found it hard to follow from those pics.     I'm lucky to follow what I think is there never mind finding what might be wrong.

I noticed a few things:

- I can see is the tracks are quite long and running close to unrelated parts of the circuit.   That could promote oscillations.

They are in fact long.  I did not know this was an issue when designing the board, so I did not take it into consideration.

- I'm not 100% sure but it looks like C8 is wired to IC5 pin2 and not R64.   Like C8 and R64 have been swapped.  If that's the case the PCB doesn't match the schematic.  Very weird!

I just tested an empty board; YOU ARE 100% CORRECT!  The negative side of C8 HAS continuity to pin 2!  I have NO idea why this would happen (there is no possible connection in the schematics!).  Could this be the issue?

- Again, not 100% sure but the value of R64 looks like 4k7.   It could be the JPEG colors.   Red bands on blue resistors are often hard to read from photos.

I will take a leg out and re-measure this afternoon (getting ready for work now).

[11-90-an]

I just tested an empty board; YOU ARE 100% CORRECT!  The negative side of C8 HAS continuity to pin 2!  I have NO idea why this would happen (there is no possible connection in the schematics!).  Could this be the issue?

YES YES YES YES YES YES! This could be the issue! 

Probably just a miss-click when you were laying out the board?

[jfrabat]

So what is the best way to fix it considering the boards are already printed?

[kraal]

So what is the best way to fix it considering the boards are already printed?

I would do the following.

1. Open your pcb layout file, and run a DRC to check the exact errors.
2. Remove tracks that should not exist (write down each one of them)
3. identify missing tracks (write down each one of them)
4. based on 2 and 3, identify the components' pins that should be not connected to the PCB
5. identify the tracks that need to be recreated off-board
6. apply 4 and 5 (cut pins/desolder, resolder, wire)

[11-90-an]

So what is the best way to fix it considering the boards are already printed?

I would do the following.

1. Open your pcb layout file, and run a DRC to check the exact errors.
2. Remove tracks that should not exist (write down each one of them)
3. identify missing tracks (write down each one of them)
4. based on 2 and 3, identify the components' pins that should be not connected to the PCB
5. identify the tracks that need to be recreated off-board
6. apply 4 and 5 (cut pins/desolder, resolder, wire)

since your pcb file is probably lost, you can maybe just trace with your schem... will be really tedious, but it will work

[kraal]

since your pcb file is probably lost, you can maybe just trace with your schem... will be really tedious, but it will work

Maybe I missed something in this long thread, but why would the PCB layout file be lost ? Knowing the time one spends designing a layout one should always backup it.
I hope that Felipe kept his. Or at least that he kept the Gerber files to be able check visually and not have to do it with a DMM. If he didn't maybe he can recover the production files (you can do it with jlcpcb)

Edit: found the post where Felipe wrote that he deleted the file. Anyway, I would check if the production files are still available on the website where he ordered the PCB.

@Felipe: if you ordered them on JLCpcb, go to "order history", find  your order, click on "production completed", then click on "production file" in the popup window. They keep production files for a very long time (see screenshot, have retrieved some I have made in May 2019), then in the archive file, in one directory ("yg" in my case) you'll find all your gerber files. You can then open them with a gerber viewer.

[jfrabat]

since your pcb file is probably lost, you can maybe just trace with your schem... will be really tedious, but it will work

Maybe I missed something in this long thread, but why would the PCB layout file be lost ? Knowing the time one spends designing a layout one should always backup it.
I hope that Felipe kept his. Or at least that he kept the Gerber files to be able check visually and not have to do it with a DMM. If he didn't maybe he can recover the production files (you can do it with jlcpcb)

Edit: found the post where Felipe wrote that he deleted the file. Anyway, I would check if the production files are still available on the website where he ordered the PCB.

@Felipe: if you ordered them on JLCpcb, go to "order history", find  your order, click on "production completed", then click on "production file" in the popup window. They keep production files for a very long time (see screenshot, have retrieved some I have made in May 2019), then in the archive file, in one directory ("yg" in my case) you'll find all your gerber files. You can then open them with a gerber viewer.

I accidentally deleted the PCB file, so yes, it is lost.  I ordered from OSH Park, so I will check once I get home to see if they have the file (site is blocked form the office)

[kraal]

I ordered from OSH Park, so I will check once I get home to see if they have the file (site is blocked form the office)

If you uploaded a KiCad .pcb_layout or an Eagle .brd file instead of gerbers, you might even be able to recover your original files. Ask them (I cannot check as I never ordered at OSH park).