Where Is My Signal Revisited

[KazooMan]

Long post with an update on my U-Boat.  It should make a good mystery.

You may recall the thread about my problems with my build of Merlin's U-Boat Sub-octave pedal.  I am using the schematic and layout found at valvewizard.co.uk/uboat.pdf.  There is an earlier version that has several differences in component values and in the circuitry near the trim pot.

My build worked for a signal generator signal, but barely passed any guitar signal.  I received many good suggestions and couldn't identify the problem.  After a lot of messing with the board, I began to worry that I had created an intermittent short or open circuit, so I etched a new board to move the components over.

The last thing I did (fortunately) was measure the bias voltage coming from the voltage divider resistors R4 and R5.  It was only 1.7 V not the desired 4.5.  I thought that confirmed that I had screwed something up in all of my attempts at repairs.

I installed new sockets for the ICs and JFETS, new jumpers and moved all of the components over one at a time, checking the value of every component as I went.  Board looked great on both sides, but........  I still had the original problem.  I started measuring and found that the bias voltage was, once again, 1.7V.  OK, so I measured and I had a solid 9V coming in (Smallbear 200 ma regulated power supply).  As you might imagine, I had a voltage drop of 7.2V across R4 and 1.7V across R5.  I lifted one end of both resistors and they measure perfectly at 1M.  I removed all of the ICs, and the JFETS and the voltages remain the same.  Lifting R1 and R3 also no change.  

There has to be another path from the R4 R5 junction to ground with a resistance of about 300K.  It had to be in BOTH builds.  I can't seem to find it.  The older version of the schematic has R4 and R5 as only 10K.  So, I installed two 10K resistors.  That should overcome the parallel resistance, and it does.  The bias voltage is now 4.5V on the dot.  I haven't solved the phantom resistance issue, but I can check the pedal.

The pedal still works fine with the signal generator.  It passes the dry signal just fine and when engaged the mix pot sweeps from dry to octave down just fine.  

The pedal now passes the dry guitar signal just fine.  However, when engaged, there is a terrible ticking.  I can detect the octave down effect, but the ticking is louder than the signal.  It is interesting (perhaps telling???) that the frequency of the ticking changes with time.  It is the fastest when the string is plucked and then slows down as the volume decays.

Any thoughts on the source of the parallel resistance screwing up the voltage divider?  Remember that it is the same with all ICs and JFETS and R1 and R3 out of the circuit.  Removing the ICs really truncated most of the circuitry.  There is little still in play.  Also recall that I got the same voltage with two builds and I used all new sockets for build two.  I have swapped out different ICs and JFETS with no change.  I should point out that I am using different components that Merlin.  I didn't have any J112's so I used J201's as suggested by others.  I also do not have any LM833's and substituted an LM1458 (as well as others).  I plan to order the correct parts from Mouser.  That may solve the ticking, but the voltage problem remains.

Thoughts on where to look to solve the ticking?

[KazooMan]

After typing my message I got to looking at the schematic again.  I think I may have found the "phantom" 300K path from the bias voltage to ground.

Referring to the schematics in the pdf I cited.

With all of the ICs, JFETS, etc. out of the circuit.  There is a path from the 4.5V as follows:  R9, R8, link 1, R10, R11 to +signal.  Then from +signal on the second page through R24 to sub_oct.  Back to the first page, from sub_oct through P1 to ground.  My P1 is actually a 220K, not 200K.  It all adds up to a bit short of the predicted value, but it is there.  I could measure the resistance, but I would need to pull some of the other components to do so.  I traced It on the layout and it seems to be there. 

[Luke51411]

I am following this thread with interest as I too have a ticking uboat. Mine is built on a thcustoms pcb.

[merlinb]

Well spotted on the Vref resistors- I will change the BOM to 10k.

I also made a correction to the BOM by changing R28 to a resistor rather than a wire link; which do you have?

As for the ticking, you might try tacking in a really big capacitor across the power rails. Like 470uF or more. I'm beginning to worry this is a decoupling issue...

[KazooMan]

Hi Merlin,

I have a 10K resistor for R28.

Do you think 10K is optimal for R4 and R5?  I only chose 10K because that was the value in the older version.

I'll give the cap a try when I get a chance.  The ticking is actually interesting.  It reminds me of the sound of one of those big wheel of fortune things in a casino.  Starts out fast and the rate slows with time.  There may be another pedal in there!

[merlinb]

Do you think 10K is optimal for R4 and R5? 

Yes 10k is fine. It only consumes half a milliamps, which is much less than the rest of the pedal + LED.

[KazooMan]

Well, it turns out that I actually have a BIG 470 uF capacitor that I have been saving for a special occasion. 



Actually, I used a more reasonably sized 470 uF cap and it had no effect.  I tried to use my signal probe to see where the tick was coming from, but it is present at virtually every point that is not at ground, at varying levels.  I will fool around with the pedal a bit more and I plan to order the correct JFETs and Op Amp from Mouser.  I can't see how the substitutes I used could create the problem, but I will eliminate that from the equation. Mouser may take about a week, so I may not have anything else to add before then unless I get a flash of brilliance and fix the problem.

Oh, I should note that I mounted the mix pot on the component side of the board, rather than the back.  It fits the enclosure I had better that way.  It is held in place with double sided sticky foam.  I also had to use off-board wiring to connect the DPDT switch since the ones I had access to have a much larger footprint than the spacing on the board.

[merlinb]

According to THCustom, ticking is cured by reducing the value of the capacitor in the envelope follower (C11) to 680nF or 470nF, or so.

[Luke51411]

According to THCustom, ticking is cured by reducing the value of the capacitor in the envelope follower (C11) to 680nF or 470nF, or so.

I tried changing that cap which is C8 on the THcustoms board and I'm still getting quite a lot of ticking on my build. Hopefully it will work better on your build. Would lowering it any more be of any use?

[KazooMan]

I changed C11 to 470nF and the constant ticking was gone.  The pedal works fine (as it always had) with a signal generator input.  However, with the guitar input it works fine in bypass, but there is still a distorted signal with a "ticking" or however you would describe a periodic peak that decreases with the decrease in input level.  I am encouraged by this!  I have a shipment of the "correct" ICs and JFETs in transit from Mouser.  I will try to solve the final issue when they arrive.

Thanks to Merlin for pointing me to the parallel thread on another forum.  As I read through that I got the impression that there may well be a sensitiveity to the exact values of components.  Many good builds and several with issues.  Hopefully I can use the information gained to solve my problem and provide some help to others.

[Luke51411]

Thanks to Thomas of THcustoms I've got my build sorted on the madbean thread. Great product support!!! The culprit of the ticking on my build is distortion coming from 1n4148 in d4 and d5 that have a lower forward voltage. I replaced them with 1n4001 and no more ticking. Heres a link to that thread in case you are interested. http://www.madbeanpedals.com/forum/index.php?topic=16677.45;topicseen

[KazooMan]

Thanks, and congratulations on fixing your pedal.  I have read through that thread over at Madbean's site.  As I mentioned I did the initial cap swap and my constant ticking went away.  However, I still had a distorted signal.  I then tried the two additional fixes, the switch to 1N4001 diodes and adding another 470 pf cap.  Neither of these modifications changed anything.  I also switched to the original JFETS and IC when they arrived and this also had no effect.

I should note for anyone else who may come across this thread that they really need to compare all of the available schematics and layouts for this pedal.  I used the layout and schematic from the Valvewizard PDF I cited earlier.  This is different from the layout and schematic you can find in a link that was given in my original posting.  The THCustom design is a third variation.  The part numbers do not correspond.  D4 and D5 on that layout are not D4 and D5 on the layout I am using. 

I have not given up, I just need to set this aside for a while to think through where the problem may reside.  I also noticed the comments about the sensitivity to specific components.  Now that I have the constant ticking eliminated I should go back and swap out the rest of the ICs one at a time.  That's easy.  If the problem is a specific resistor or cap I will probably never find it.

[merlinb]

If the problem is a specific resistor or cap I will probably never find it.

You really need an ocsilloscope (and know how to use it), and a systematic approach to solve a problem like this.  

[KazooMan]

If the problem is a specific resistor or cap I will probably never find it.

You really need an ocsilloscope (and know how to use it), and a systematic approach to solve a problem like this.  

I have a Rigol DS1052E (and I know how to spell it  ). I know enough about using it to have it show me two simultaneous traces for the input and output when I input a nice sine wave from my Valve Wizard signal generator.  My first build gave a great octave down sine wave.  I mentioned that previously.  It was really remarkable how clean it looked on the scope.  My current build that now has multiple part substitutions isn't quite as nice, but it still shows what my ear hears from my audio probe.

BTW... I save space on my bench by using my build of Rick Holt's Vibratone amp for my audio probe. Two subminiature "valves" with tone, master gain, tremolo, and switchable bias all in a normal pedal enclosure.  There's a lot going on in there,

As far as using the scope to trace the problem, That would be easy with the signal generator input, but that works just fine.  It is the guitar input that is the issue.  When the effect is engaged with the mix pot on full dry signal there is NO distortion.  The wet signal is what another poster on the Madbean Board called "farty crackling".  I even tried the "Fake Guitar Oscillator" that I built following RG's design.  It emulates the attack and decay of a struck guitar string and it works just fine with the U-Boat.  I (perhaps naively) take his to indicate that there are no crazy impedance issues with the guitar input.  I should also note that the sine wave input works well over a wide range of input levels. 

There was some mention on the Madbean board about the actual sourcing of the 1N4148 diodes being an issue.  All of my components are from either Smallbear or Mouser.  Multiple swaps of all JFETS, ICs, etc., have no effect.

I can be pretty methodical about tracing a signal though an effect to see where things go off the rails, but I don't know how to do this for a strictly guitar signal related problem.  I will admit that my understanding of all of this is strictly that of an amateur hobbyist.  I actually enrolled as a freshman in college with a declared major in electrical engineering.  After one year I realized that I liked chemistry more and switched majors.  Many years later after a successful career in the pharmaceutical industry I am retired and building pedals for no other reason than I can.  Never had a class in oscilloscopes, but they did mention electricity when I was at Cal Tech and MIT.