Debugging a Neutron

[taeagan]

A little more detail on the both-LED-only-one-FET scenario...

It gives more sound with U101 connected.  In bandpass, connecting U101 sounds like no FET connected except at the very ends, depending on how the Sweep Direction switch is positioned.  It's either no sound when you hit the strings really hard and then it sort of farts in, or it sounds OK, then slowly farts out.  

With U100 connected, bandpass and highpass are both just as bad.  Either no sound at all until the note is just ringing out at the end, or sound when you hit it real hard and it farts out real fast - depending on the Sweep Direction.

[taeagan]

With the meter, I re-checked all of the traces coming from the H11F3 pins and they're not shorting to any of their neighbors.  I also tried the NSL32's again - also with just one LDR side connected - and no difference.

What could I be missing?  RG, if the audio side is fine, what is causing the problem?  Where else can I check for shorts that I may have missed (even though it feels like I've been staring at that board and poking it with meter leads for days... oh, wait... I have been)?

[R.G.]

Sitting in a darkened, quiet room contemplating my too generous navel isn't helping. We need more data.

I need you to do some more invasive measurements.
1) pull out C4, and disconnect the wire that connects to C4 on the PCB and to the pole of one section of the direction switch. Disconnect it at the switch.
2) hook a pot and two resistors up with the resistors going from +9 to one pot outer lug, -9 to the other outer lug, and the wire that was connected to the pole of the direction switch that you removed. Make the resistors equal and both twice (or so) the value of the pot.  Now you can input a DC level into the envelope detector.

3) Hook power back up, and verify that you can get about +/- 3V or so at the wiper of the pot.
4) have a friend or significant other strum while you twiddle the pot and play with the switches. Does the problem recur? When?
5) measure DC voltage on the output of the two filter opamps, U2 pins 7 and 1. Do they vary with the pot twiddling? When the pot is still, do they 'recover' to 0V or stay off 0V?
6) what is the voltage across Rx while you twiddle the pot? How does that vary?

This is a PITA, but it may be the only way to fix it. We have to find what is varying with what, and that will lead us to the defect.

[taeagan]

So if I use a 100K pot and two 220K resistors, will that work?  

And connect the wiper of the pot to the wire that I disconnected from the direction switch?  

And do this with the H11F3's out, right?

[R.G.]

So if I use a 100K pot and two 220K resistors, will that work?

And connect the wiper of the pot to the wire that I disconnected from the direction switch?

Yes and yes.

And do this with the H11F3's out, right?

Nope, put them in. We've exhausted the forward looking "ahah, that voltage proves it was the butler in the drawing room with the steak knife" approach. Now we're using the "make it fail and see what's happening... slowly..." approach.

Making the defect happen with a guitar is hard to catch if you're working by typing things over the internet, not sitting there with a 'scope. So what I told you to do is to put in a pot that will let me fake any degree of "note loudness" by twisting that pot, and seeing how that affects the audio path. The supposition is that somehow the envelope is getting into the audio path, since removing the H11s removes the fartiness. I have asked you to disconnect the input signal from the envelope detector so we can apply our own "envelope".

[soggybag]

I notice in the PCB image in the PDF from GEO a few of the traces touch other traces they are not meant to touch. I'm not sure if these matter or if they are meant to make a connection. From the look of the traces it seems they were not meant to make a connection.

R7 touches the pad that leads to Hi Pass.
Pin 1 of U1 the very corner just touches the pad leading to the bypass switch

[taeagan]

1) pull out C4, and disconnect the wire that connects to C4 on the PCB and to the pole of one section of the direction switch. Disconnect it at the switch.
2) hook a pot and two resistors up with the resistors going from +9 to one pot outer lug, -9 to the other outer lug, and the wire that was connected to the pole of the direction switch that you removed. Make the resistors equal and both twice (or so) the value of the pot. Now you can input a DC level into the envelope detector.

- First, let me make sure that I did this right.  I de-soldered C4 and removed it completely.  I took a 100K pot and soldered a 220K resistor onto each outer lug.  I then connected one 220K resister to +9 and the other to -9.  I connected the middle lug to the wire that I removed from the switch and is connected to R10 and where C4 used to be on the board.

3) Hook power back up, and verify that you can get about +/- 3V or so at the wiper of the pot.

- This may be the first problem.  The wiper goes from .2V to -.7V as I rotate the pot.

4) have a friend or significant other strum while you twiddle the pot and play with the switches. Does the problem recur? When?

- It doesn't recur.  It also doesn't sound like an envelope filter.  It sounds like the H11's are out completely in lowpass (they are in).  Bandpass and Highpass, it sounds like the H11's are out until the pot gets about half way and then no sound.  As I turn the pot and it shuts off you hear a little bit of the filter effect.  Flipping the direction switch basically reverses the effect of the pot.

5) measure DC voltage on the output of the two filter opamps, U2 pins 7 and 1. Do they vary with the pot twiddling? When the pot is still, do they 'recover' to 0V or stay off 0V?

- Stays at 0V the whole time.

6) what is the voltage across Rx while you twiddle the pot? How does that vary?  

- When the effect is producing no sound, the voltage across Rx is about 3.5V.  When I rotate the pot and the effect comes back on, it drops to something like .2V.

Am I doing this right?

[R.G.]

1) pull out C4, and disconnect the wire that connects to C4 on the PCB and to the pole of one section of the direction switch. Disconnect it at the switch.
2) hook a pot and two resistors up with the resistors going from +9 to one pot outer lug, -9 to the other outer lug, and the wire that was connected to the pole of the direction switch that you removed. Make the resistors equal and both twice (or so) the value of the pot. Now you can input a DC level into the envelope detector.

- First, let me make sure that I did this right. I de-soldered C4 and removed it completely. I took a 100K pot and soldered a 220K resistor onto each outer lug. I then connected one 220K resister to +9 and the other to -9. I connected the middle lug to the wire that I removed from the switch and is connected to R10 and where C4 used to be on the board.

You did that correctly.

3) Hook power back up, and verify that you can get about +/- 3V or so at the wiper of the pot.

- This may be the first problem. The wiper goes from .2V to -.7V as I rotate the pot.

And I did that incorrectly. R10 and R11 are 22K and 12K, and a 220K to _9 will only pull that point up by .4 to .9V. Sorry - you really needed a 10K pot and 22K resistors because of the low resistance of R10 and R11. My fault. Can you change those?

- It doesn't recur. It also doesn't sound like an envelope filter. It sounds like the H11's are out completely in lowpass (they are in). Bandpass and Highpass, it sounds like the H11's are out until the pot gets about half way and then no sound. As I turn the pot and it shuts off you hear a little bit of the filter effect. Flipping the direction switch basically reverses the effect of the pot.

Curiouser and curiouser. That's suspicious, but we'll have to wait for the results from the 10K pot.

6) what is the voltage across Rx while you twiddle the pot? How does that vary?

- When the effect is producing no sound, the voltage across Rx is about 3.5V. When I rotate the pot and the effect comes back on, it drops to something like .2V.

This is suggestive of a problem as well, but again, we need the results of the lower impedance input to the envelope rectifier.

You're doing it right - I just told you wrong
:oops:

[taeagan]

OK, I tried it with the 10K pot and 22K resistors.  The voltage on the middle lug of the pot swings from -1.5V to .9V.

Symptoms are basically the same.  Doesn't sound like an envelope filter, just sounds like it did with the H11's out.  In highpass and bandpass it shuts off half way.  Now, in lowpass, at halfway on the pot, the tone changes.  All of a sudden there's a treble boost about half way through.  No voltage at pins 1 and 7 of U2.  And the voltage across Rx is 3.5V when there's sound, 0V when I rotate the pot and it shuts off.

[R.G.]

OK, now we take some data. I apologize, but it's necessary. It's what I would do if I could get my hands on the parts.

1) Set the pot to full negative (-1.5V)
2) measure and record in a table the voltages at:
- pot wiper
- junction of R13/R14
- junction of R14/R15/R16
- junction of R17/Rx
3) Note for this step - is it still sounding OK?
4) Is the pot turned full positive yet? If not, turn it up about 1/10th and go back to 2)
5) when the pot has been rotated fully and all the measurements taken, email it back.

It appears that something is funny in the rectifier/inverter/LED drive from what you've said, and we just have to run it down.

[taeagan]

I can't find your e-mail address or any way to send you an attachment, but I took measurements at each 1/10th of a rotation on the pot:

V wiper: -1.5
R13/14: 6.9
R14/15/16: 6.7
R17/X: 3.5
Sound? No

V wiper: -1.5
R13/14: 6.7
R14/15/16: 6.6
R17/X: 3.4
Sound? No

V wiper: -1.5
R13/14: 6.7
R14/15/16: 6.5
R17/X: 3.4
Sound? No

V wiper: -1.5
R13/14: 6.6
R14/15/16: 6.5
R17/X: 3.4
Sound? No

V wiper: -1.0
R13/14: 6.6
R14/15/16: 6.5
R17/X: 3.4
Sound? No

V wiper: -0.5
R13/14: 6.6
R14/15/16: 6.5
R17/X: 3.4
Sound? No

V wiper: -1.0
R13/14: 5.0
R14/15/16: 4.9
R17/X: 3.3
Sound? No

V wiper: 40mV
R13/14: -.8
R14/15/16: -.8
R17/X: 0
Sound? Yes

V wiper: 0.2V
R13/14: -.9
R14/15/16: -1.0
R17/X: 0
Sound? Yes

V wiper: 0.8V
R13/14: -1.0
R14/15/16: -1.0
R17/X: 0
Sound? Yes

[sean k]

This may or may not be of relevance but on my neutron I basically got it working fine except the horrid loud switching noise which is actually occuring when the LED's switch on and off.There is no gradual brightness between the on and off state(I can verify this as the down switch is replaced with a 100k pot between resistors and so I can tune in the LED's).So my thinking is the problem is the LED's need to be replaced with a set that turns on gradually and doesn't send shockwaves through the earth...?.I picked this thinking up in the thread about rolling your own vactrols and think it may be significant.

[taeagan]

I think I might be giving up on this for a while.  I tried it with the H11F3's as well as with a pair of NSL32's.  So different pair of LED's, same result.

One thing I did that seemed to improve things a bit was replace the 4148 signal diodes with germanium diodes (1N34A).  Definitely not as bad, but still a problem.  A couple of other things I noticed is that it behaves better as I turn the gain down, and also sounds much better with one of the LDR's unplugged.  I've also noticed a switching sound like what you described, but it's mostly with lower frequencies.  When I get it working OK, it works well on highs, but lows get blown out.

I've spent many hours on this and also tried breadboarding it  - re-building it from the ground up.  No idea what I'm doing wrong.  RG's been awesome about helping me, but we can't nail it down.  I even bought some new TL072's from Radio Shack today to rule out a bad batch of them from Mouser.  No dice.

[R.G.]

We've been working on Tom's Neutron by email.

Near as I can tell, his unit may be one of the few that suffers from a bad new component. Either that, or the recent H11F3's are a different pinout or are simply defective. His unit also displays a problem with NSL LED/LDRs. Very strange.

You say that there is a horrid loud noise when the LEDs switch on and off. That's not a characteristic of LED/LDRs in general. For there to be a loud noise, there has to be a sudden transient going into the circuit from the LED/LDR, and LED/LDRs simply don't work that way. LEDs are fast - very fast - and they turn on in microseconds. All of them do. But LDRs are quite slow, and even an instantaneousl LED will not make an LDR change faster. That's not the problem you have.

Can you tell me
- what is the origin of your board?
- what opto setup are you using?

[sean k]

the layout is my own that I etched up myself and the LED's are just plain old bright greens and the LDR's are VT935G's  with 10-50k at 10 lux (min-max) and 1M min dark resistance (these are specs as published) with a rise time of 35ms and a fall time of 5ms.I can try taking some photos of the board,top and bottom,and send them to you if you'd like to see it.
 I've also got some other LED's and another type of LDR to try out in the circuit.