Slow Loris debugging assistance; inquire within

[vigilante397]

It's been a while since I've asked for help debugging a build, but I know I'll always have problems.

So I just finished throwing together Madbean's Slow Loris project, a modded RAT clone, about which I'm remarkably excited. I used a purchased PCB, the updated 2014 version. Unfortunately I'm getting ridiculously (like barely audible) low volume on the output. I found a thread on the Madbean forum wherein the poster had very similar symptoms to mine (distorted sound but very low volume, similar voltages, audio probe leads to everything normal throughout except sound dies on one side of C12), and his issue turned out to be a bad JFET, but I just socketed the JFET and tried every 2N5457 I have right now (fresh batch of 10) and they all gave the same result.

My voltages are as follows:

LM308

1 - 8.496      8 - 4.674
2 - 4.594      7 - 9.145
3 - 4.160      6 - 4.595
4 - 0            5 - 0

2N5457

1 - 8.394
2 - 8.380
3 - 9.146

I'm no genius, but I know those JFET voltages look bad, but as I mentioned every JFET I tried has the same issue.

So does it still just look like a whole bad batch of JFETs? Or is there something else I should be looking for?

Original build doc including layout and schematic: www.madbeanpedals.com/projects/SlowLoris/docs/SlowLoris2014.pdf

[bluebunny]

Do you have a bad ground connection somewhere around that FET?  What are the readings without the FET installed?

[vigilante397]

Without the FET in place I'm getting

1 - 0
2 - 0
3 - 9.202

[duck_arse]

yr bias volts are missing. check the 100k dividers for Vb. there was an errata on the gate resistor, has your board been "fixed"?

[bluebunny]

1 - 0
2 - 0
3 - 9.202

BTW, does "1, 2, 3" = "D, G, S"?

[vigilante397]

1 - 0
2 - 0
3 - 9.202

BTW, does "1, 2, 3" = "D, G, S"?

I should have clarified that. Do not debug pedals after midnight

D - 9.202
G - 0
S - 0

These are voltages without the JFET in place.

[jatalahd]

Your connections might be ok, but it just seems to me that Q1 is saturated (fully open) due to wrong operating point (when V_GS = 0, then I_D = I_DSS). This can possibly be explained due to deviance in JFET characteristics. Typically there is not that much deviance within the same batch, so all your JFETs might have the same deviation compared to the one the designer of the circuit had.

Since Q1 is isolated between capacitors, you could just build that part (R8, R9, Q1) independently to a breadboard and test different values for the source resistor. If my suspicions are correct you should increase the value of the 10k source resistor to maybe 33k, 47k or even more. See what you get..., measure V_GS, V_D and V_S for each source resistor.

When (hopefully) you start getting V_GS around -1V and a clear voltage drop across the drain and the source V_DS (due to JFET channel closing from saturation state), then you should just settle to the value you like best. It's just a buffer anyway.

[slacker]

I'd check for a short between source and gate, not sure how else you're getting 8 volts on the gate, unless what Jarmo said could cause that.

[vigilante397]

UPDATE: Sorry for going quiet on this for a while, I had almost given up on the Slow Loris and was about to go back to Tonepad, but it was still at the back of my mind. The RAT is my favorite distortion ever and I have successfully built 4 of them prior to this (3 tonepad, 1 vero) and I couldn't figure out why this was giving me so much trouble until I looked up the Tonepad layout and noticed one difference (aside from the usual mods): MadBean's Slow Loris calls for a 2N5457 whereas the Tonepad La Rata calls for a 2N5458! I looked up the datasheet and noticed that there are some significant differences, so I threw a 2N5458 into the Slow Loris and it fired right up!  

bool lessonLearned(bool lesson)
{
  if (lesson == true)
  {
     std::cout << "Learned";
   }
   else
      bangAgainstWall(head);
   return lesson;
}

[bluebunny]

thread.result = GOOD;
satisfaction++;

[duck_arse]

2n5457 > /dev/null ? I wouldn't have expected that. [I can't even remember any tcl.]

[vigilante397]

thread.result = GOOD;
satisfaction++;

Indeed

2n5457 > /dev/null ? I wouldn't have expected that.

I was a bit surprised myself. I know people have reported successful builds with a 2N5457, and I know Madbean knows his stuff so I have no reason to doubt it, I just thought it was odd that in my case it didn't work out.