Circuit Design Help: Solve Audio Driver Issue on PCB

[Firefox123]

Circuit Design help...

Hi all...

I had a question about a little audio circuit I am working on...its pretty simple it just consists of an audio amplifier (OPA134PA), an audio balanced line driver (DRV134), a audio differential receiver (INA137), and a few caps and resistors.

When I hook this circuit up on a bread board everything works fine...the op amp gives a gain of 8 and the line driver outputs two signals 180 degrees out of phase with virtually the same peak to peak voltage.

I am inputting a .5 volt peak to peak signal into the amplifier and getting 4 volts peak to peak out of the amplifier. Then the line driver provides me with two out of phase signals each 4 volts peak to peak.

But on my PCB something seems to be wrong...the audio line driver (DRV134) outputs are not equal...the positive output is around .30 volts higher than the negative input which is out of specs for the DRV134. I called Texas instruments and they said I might have a bad chip, but since it works on a breadboard I am convinced it is a issue with the final PCB circuit.

Any ideas what could cause the difference in the two outputs?

I have attached the ORCAD files I used for the circuit for reference...any help would be appreciated.

Regards.

 

[berkeman]

Is the amp through-hole or SMT? Any way to easily swap chips?

Is the power supply decoupling adequate on the PCB? Are you sure the IC is connected correctly? Maybe reflow the solder joints to be sure you don't have a soldering issue.

BTW, on the ground connection below C9 on the schematic, I'm seeing an extra wire stub. It still passes the DRC checks, though. I'm viewing it in Orcad Capture 9.0, so my conversion from your version may have introduced the wire stub if you don't see it.

 

[Firefox123]

Is the amp through-hole or SMT?

The amp and the line driver are both through hole...

Any way to easily swap chips?

Sure...I've already swapped line drivers and the ones on the breadboard work great...but the ones I solder onto the PCB have the same issue...

Is the power supply decoupling adequate on the PCB?

As far as I know, yes...

Are you sure the IC is connected correctly?

I have gone over the data sheet several times and compared it to the circuit so I believe it is correct...

Maybe reflow the solder joints to be sure you don't have a soldering issue.

Thats a thought...Ill do a continuity check on the solder joints as well to make sure I don't have some kind of a crazy short somewhere...

BTW, on the ground connection below C9 on the schematic, I'm seeing an extra wire stub. It still passes the DRC checks, though. I'm viewing it in Orcad Capture 9.0, so my conversion from your version may have introduced the wire stub if you don't see it.

Yeah...good catch. I see it also. That shouldn't be there...the ground connection should look like all of the other ones. I can't believe I didn't see that...I don't think that should impact the circuit operation...


So any thoughts on a possible cause? Is this likely a grounding issue or could it be EMI? EMI seems very unlikely since I am operating at audio frequencies...nothing above 20kHz...

 

[berkeman]

Since it's through-hole, put a socket in the PCB. That makes it easier to swap chips around to debug what's going on. How fast is your oscilloscope? You might have a high-frequency oscillation on you PCB circuit, and your 'scope is not seeing it. You may also have a loading problem on the line that is showing the smaller signal. Or did you say that one of the balanced outputs was 0.3V bigger than it was supposed to be?

 

[Firefox123]

Since it's through-hole, put a socket in the PCB. That makes it easier to swap chips around to debug what's going on.

Thats a good idea...

How fast is your oscilloscope? You might have a high-frequency oscillation on you PCB circuit, and your 'scope is not seeing it. You may also have a loading problem on the line that is showing the smaller signal.

I'm not sure what the fastest rate of the O-scope is...I'll have to check and get back to you.

What kind of a loading problem would create this issue?

Or did you say that one of the balanced outputs was 0.3V bigger than it was supposed to be?

Yes...that is exactly what is happening. The positive output is around 0.3V higher than the negative is. The spec sheet for the audio line driver has the range from .007 Volts to .07 Volts so 0.3V is definitely something outside of the manufacturers specification...

I am going to do some grounding continuity checks today...maybe something will turn up.

Thanks for the replies...if you think of anything please let me know...

 

[Firefox123]

The O-scope can do up to 10ns/division.

I checked the grounding and it looks good...all the ground points passed my continuity test...

 

[Alfi]

Don't focus solely on the chips. A leaking capacitor can pull down a voltage output.

 

[Firefox123]

Don't focus solely on the chips. A leaking capacitor can pull down a voltage output.

How should I test for that?

I have looked at the capacitance of the capacitors and they look good so far...the grounds on the caps look good also...

 

[Alfi]

Check the signals and voltage in and out with the scope.

Any chance you have a signal generator available?
Then you could inject a low freq and a high freq to test the capacitors band pass qualities.

 

[Firefox123]

Check the signals and voltage in and out with the scope.

Any chance you have a signal generator available?
Then you could inject a low freq and a high freq to test the capacitors band pass qualities.

No signal generator available but I do have an audio signal generator that I can use...

 

[Firefox123]

Check the signals and voltage in and out with the scope.

Any chance you have a signal generator available?
Then you could inject a low freq and a high freq to test the capacitors band pass qualities.

I don't think its a capacitor since the problem is occurring on 2 different PCBs, but if I make the circuit on a bread board (including the capacitors) then the line driver output is within specification...

Sweeping through various audio frequencies doesn't seem to affect the issue...

I'm at a loss as to what could cause the line driver to have a positive output that is .3 Vpp higher than the negative output.

 

[Alfi]

Ok
possible weird stuff to look at then.
( I am -not bad- at truoble shooting but not on line - sorry )

It's possible that some trace lengths are a nice little radiating antenna just by virtue of their length and the frequencies 20-20Khz involved ( I assume.)

If D.C. values are correct, then it points to a dynamic gain loss due to ... something.

If two circuit boards show the same result I would also look hard at the board layout. and a good trace ohm check on a blank board. I again assume it's not a multilayer board.
( I'm sorry, I can't open the attachment. I'm just running through what I would do with the facts given. )

I also once found that a tiny tiny crack really messes with VHF circuits.
Took us 4 days to find it and it was only by dragging a freq analyzer over every inch of the board do we find a perfect 3db loss crack in a damn trace.

This is fun. I hope you get the right clue to solve the mystery.

 

[Firefox123]

First of all I want to thank everyone for the replies...lots of good thoughts...

My investigation into the DRV134 (the line driver) seems to suggest that perhaps it is the NEGATIVE output that has a problem and not the positive output. I assumed that if I inputed a 4Vpp signal into the line driver that I would get 4Vpp out of both outputs.

But (on a breadboard setup) I actually get around 4.1 - 4.2 Vpp out of each output...so the driver does seem to output slightly higher than what you put into it.

This leads me to believe that perhaps the positive output of, say 4.3, is correct and the negative output of 4.0 is the one that is wrong.

Ok
possible weird stuff to look at then.
( I am -not bad- at truoble shooting but not on line - sorry )

I know...its tough when you can't physically look at the circuit...but I appreciate the replies...

It's possible that some trace lengths are a nice little radiating antenna just by virtue of their length and the frequencies 20-20Khz involved ( I assume.)

The interesting thing is that this problem occurs in both PCB versions (version 2 and 3) of the circuit, but does not occur on a breadboard. This makes me suspect that perhaps we do have some weird problem like you are describing.

If D.C. values are correct, then it points to a dynamic gain loss due to ... something.

I agree...that does appear to be the case...

If two circuit boards show the same result I would also look hard at the board layout. and a good trace ohm check on a blank board. I again assume it's not a multilayer board.
( I'm sorry, I can't open the attachment. I'm just running through what I would do with the facts given. )

Yeah its just a two layer board. You would need either the free or paid version of ORCAD to open it...

And yes this problem has occurred so far on 3 PCB boards...two are "version3" and one is "version2". The main differences between version 3 and version 2 are the molex connectors (these are plastic connectors mounted on the board with pin connections)...and the traces are slightly different since the different molex connectors are different sizes and mounted slightly differently (one has the pins facing up and one has the pins facing to the side).

I will try to get the schematic in a different format so you can open it...

I also once found that a tiny tiny crack really messes with VHF circuits.
Took us 4 days to find it and it was only by dragging a freq analyzer over every inch of the board do we find a perfect 3db loss crack in a damn trace.

Yikes...that must have been a real pain in the neck...

This is fun. I hope you get the right clue to solve the mystery.

I hope so too...Ive been working on this thing for a few days now and I can't figure it out...

 

[zeitghost]

If the worst comes to the worst, and it sometimes does with Cad & Orcad, you may find you have to compare your schematic with the traces on the pcb...

You can do this by increasing the size of the layers to fit something like an A3 printout & systematically comparing what you thought you drew on the schematic with what it thought you meant. I use coloured pens to do this.

I've been stung by Orcad a number of times when it did "interesting" things... such as shorting all the pins of a matrix keypad together because a line was just that little too close to a symbol on the schematic... that one required mods to a mere 100 pcbs...

 

[Firefox123]

If the worst comes to the worst, and it sometimes does with Cad & Orcad, you may find you have to compare your schematic with the traces on the pcb...

You can do this by increasing the size of the layers to fit something like an A3 printout & systematically comparing what you thought you drew on the schematic with what it thought you meant. I use coloured pens to do this.

I've been stung by Orcad a number of times when it did "interesting" things... such as shorting all the pins of a matrix keypad together because a line was just that little too close to a symbol on the schematic... that one required mods to a mere 100 pcbs...

Good idea...I will have to do that and see if anything seems out of the ordinary...

That suck that it shorted your keypad together...I'm sure that was slightly annoying...

 

[Firefox123]

circuit update

There have been some new developments since my last post...

I think I have discovered what causes the negative output of the audio line driver to have a .3 V peak to peak voltage drop although I do not fully understand it...

I had taken the output from the audio line driver (DRV134) and used this as the input for the audio line receiver (INA137). When this is connected we observe the voltage drop. I have connected the driver directly to the receiver without any capacitors or resistors, connected it with the resistors only, and connected it with both the capacitors and resistors present and each time the negative output has the .3 volt drop.

If I take the output of the line driver and I just connect each output to ground through a 2.2 K resistor then the line driver output does not have the .3 volt drop and both outputs look good.

So the receiver is doing something...but I am not sure what. This happens on both the PCB and the breadboard if the driver outputs are looped back as inputs to the receiver.

Any thoughts?

 

[Firefox123]

Update

The issue has been resolved. After talking to Texas Instruments a few times they ran a simulation and the cause was simply found to be the receiver inputs.

The receiver is basically a differential amplifier so the negative and positive inputs have different input impedances. That is what causes the .3volt drop on the negative terminal.

I incorrectly assumed that TI would have internally compensated for this difference and internally made both inputs equal in impedance. TI has assured me that this does not affect the receivers output, so perhaps there is something internal that eventually takes this into account (although looking at the circuit I don't see it).

Anyway...mystery solved and thanks everyone for responding.