Debugging Devi Ever Hyperion Fuzz build

[JCGH]

Hi all,

I'm relatively new to the pedal building world and it's got me well and truly hooked! So I apologise in advance for any newbie questions.

Inevitably I've come up against a couple of problems where I've had to start learning how to debug the circuit, all the pages on here have been really helpful but unfortunately I haven't been able to solve the problem.

My current problem is with the Devi Ever Hyperion fuzz circuit and seems to be one that has come up when others have tried to build it but I couldn't see any resolutions to the problem on the forum. (I was hoping it would be a nice simple circuit!)

The circuit I have built is on stripboard and follows this layout:

http://tagboardeffects.blogspot.co.uk/2012/03/devi-ever-hyperion.html?m=1

Key things to note:
I am using 2n5089 transistors instead of the MPSA18 transistors. I am also using carbon film 2M2 resistors, all others are MF. (Not sure if using different types matter.) Other than that the circuit is the same.

Problem: I have the "blatty" or gated sound as described on the debug page here:
http://www.geofex.com/fxdebug/bias_prob.htm

So, I'm pretty sure it's a bias problem.

This was backed up with my voltage readings, which seem to show a high reading at the collector of Q1. The below schematic shows my voltage readings. (Q3 is actually a 2N2907A, not a 2N907A as my notes suggest...it was late.) If the usual format of R1 X Y, R2 X Y etc. is more useful then let me know.



I decided to breadboard the circuit with the same parts and see if I could get it working, which I did and it did and the readings are as follows:



So I'm assuming it is the first 10K or 2M2 resistors that are the problem? Based on what I've read. Or am I wrong?

I've checked all joints and resoldered them, checked for solder bridges, replaced the 10K resistor and nothing has changed.

I'm hoping I've missed something somewhere or that I have just got the wrong end of the stick.

Here's a picture of the board, in case it is useful.





Any help would be appreciated! Thanks.

[slacker]

Nice work with the debugging, looks like you've almost got it figured out, something is wrong with the biasing of the first transistor. Looking at your vero layout is the right hand side 2M2 resistor connected correctly? I can see that the bottom of it is connected to the base of the transistor (the middle pin) on the third row up but what is the top of it connected to, it should go to the collector of the first transistor (the top pin of the transistor) I can't see that connection.

Welcome aboard

[JCGH]

Hi, slacker, thanks for the response and taking the time to look. I've had a look at that resistor and there is a link two columns to the right that goes back down to the collector of the transistor, you can just see it poking out from the top of the third capacitor.

My current thinking is that there is some resistance causing the voltage drop from 9.31 to 8.90, so the 10K is doing something. But it's not enough drop, I've measured the resistance and it's correct, so there is probably some voltage that is bypassing the 10K resistor straight to the collector thereby causing only a slight drop in voltage.

Is this the correct thinking? Should I just be looking for solder bridges? I can't see any but I can have another look.

[PRR]

Guess: One end of the first 2.2Meg is not really connected.

Observation: some of that solder is blobby. IMHO you should be able to see the wires thinly coated with solder. In particular you should be able to see if the solder has "wet" the wire. It should cling like ice-cream on a hot sidewalk. Not bead-up like raindrops on a well-waxed hood.

My instinct is to heat the joints and beat the board on the bench. Get the excess solder off. If the joints are good, they will still be joined by well-wetted solder in the crack. However hot-beating also flings "threads" of solder all over, giving 99 shorts to pick-off (and perhaps a serious eye-burn!!). Maybe a bright light and a solder-sucker on selected joints would be a better way to go.

[JCGH]

Great! I'll try it out as soon as I have the chance as get back to you. Cheers.

[Mike2E]

After checking the circuit, This is just a bad design. As I look at the circuit it makes my hackles rise. The bias design having no resistor from base to ground and no emitter resistor is very bad, because it relies on the gain of the transistor to control everything.
Firstly, add a small resistor emitter to ground, then add a base ground resistor.
Measure the collector voltage, it needs to be midway to work properly
I have a step resistor that I would put in and adjust it until I get the result.
The difference you get in the two circuits is more likely to be transistor differences if your solder joints and circuit is good.
Look here at a real circuit http://www.electrosmash.com/big-muff-pi-analysis and note how the input buffer transistor is biased.

[PRR]

> collector voltage, it needs to be midway

It "will" be midway, once the fault is found.

This is Collector-Base negative feedback. It is pretty stable.

As a hasty analysis: assume Vbe is "small" compared to supply voltage.

Current in the 2.2Meg is multiplied by transistor hFE to pass the Collector-Emitter current.

If hFE is tiny, Vce is high; if hFE is huge, Vce is low.

We may quick-figure the Vce by dividing 2.2Meg (2,200K) by assumed hFE, and pretending the transistor "acts-like" that resistance. Then simple resistor-divider math.

hFE
30 *** 2200K/30= 73K *** 9V*73K/(10K+73K)= 8V  {Gv=38 Rin=6.9K}
100 *** 2200K/100= 22K *** 9V*22K/(10K+22K)= 6V  {Gv=115, Rin=6K}
300 *** 2200K/300= 7K *** 9V*7K/(10K+7K)= 3.7V  {Gv=204, Rin=6.3K}
900 *** 2200K/900= 2.4K *** 9V*2.4K/(10K+2.4K)= 1.7V  {Gv=280, Rin=6K}
2700 *** 2200K/2700= 0.81K *** 9V*0.81K/(10K+0.81K)= 0.67V  {Gv=310, Rin=6.5K}

All of these conditions will work, and most pass signal plenty large enough to hammer the next stage.

Note that a 9:1 change of hFE, 100-300, causes just 2.4:1 change of gain, and the input impedance hardly changes at all. (I suspect the 15% wobble above is rounding-errors from my round thumbs.)

The part-number specified has maybe 3:1 range of hFE so we expect only 1.4:1 change of gain.

The reverse connection of Q3 ("can't work!") is more interesting. However Devi was exploring "new approaches" in musical electronics. This product pleased some people. I suspect it depends on some un-documented device parameters; it may have worked with the devices Devi had and not with devices from other lots or other factories. I try not to think about it.

However JCGH's numbers suggests that Q1 has hFE near 10. That's not right. It is *possible* that Q1 has been inserted with C and E swapped. Reverse-hFE is highly variable but (on any modern device) very much lower than the forward-hFE (for '5089, 400-1200 IIRC).

[JCGH]

Hi guys, thanks for the attention.

I'll work through all your suggestions/numbers as soon as I get the chance (I'm in the middle of exams which is inconveniently drawing me away from the more pressing matters of pedal debugging!)

However, as a quick update, PRR, I did have a look over the board and removed/resoldered areas that looked blobby, not sure if I did a great job though. I also replaced the 2M2 resistor. No changes, still getting the 8+V at collector of Q1.

Some further information: I have realised that this is becoming a common problem with many of my builds - which suggests a degree of incompetence on my part. The problem being the 'blatty' sound of an unbiased transistor.

The circuits I have built with the problem are all the Emerson EM-Drive, two on strip board and one on a Fuzzdog Pedalparts PCB. The readings at collector appear to highlight the problem as with the Devi Ever fuzz. I think I've read somewhere that this circuit is also very sensitive to many different factors? But it is another one that I have breadboarded and it all seems fine.

Anyway, the current plan is to rebuild either of the above circuits again. I'll breadboard the circuits and check that they work and then aim to clean everything thoroughly (legs and stripboard) so that I can try and have the cleanest joints possible (with unblobby solder!). I'll see where that gets me.

Thanks again for the help.

[PRR]

If they all work good on push-in breadboard, but crap-out on solder-board, that may be a clue.

Rip-up an old VCR or somesuch. Look at the factory solder joints. Take a part out, clean legs and pads, solder the part back in. Try to get it as "right" as the factory did. (If the VCR worked, it should still work, though working VCRs are getting scarce, and rip-up re-rig and test is tedious.)

[Mike2E]

I agree with PRR's circuit analysis. Especially the last point about the transistor placing incorrect. I checked your photos and then the pin out of what you used and think you should recheck.
PRR, I think I have spent to long working on high accuracy audio and instrument repairs.

[PRR]

> I think I have spent to long working on high accuracy audio and instrument repairs.

I've been there too. Designing so the operating point is *nailed* and is obvious by inspection, without knowing the active device specs.

The ultimate became throwing in a 19 cent chip and letting the chip-maker sweat the details. Leaving discrete design to funky fields like guitar pedals.

But this cheezy scheme works a lot better than you might guess at a glance.

Perhaps the main flaw is a lowish input impedance. (But a high stability base bias network for a 4-R design will also have low input impedance.)

What stunned me (I never worked it out this far) is that if you had a need for Input Impedance = 6.3K +/-5%, this job will do it with ANY transistor the cat drags home and without trim. Swings and roundabouts. One factor of Zin drops and another rises as hFE varies.

Don't think I ever needed a +/-5% Zin. If I did I'd probably do the obvious and make the amp ~~100K Zin, swamp it with a 6.8K.

And the change of voltage gain is fairly large, especially for hFE<200. (I was taught to "assume hFE>50", and with a slight shift this plan would be semi-stable; in the meantime hFE>>50 has become cheap, and >300 is a standard for the small parts.)

[Mike2E]

Hi guys, thanks for the attention.

I'll work through all your suggestions/numbers as soon as I get the chance (I'm in the middle of exams which is inconveniently drawing me away from the more pressing matters of pedal debugging!)

However, as a quick update, PRR, I did have a look over the board and removed/resoldered areas that looked blobby, not sure if I did a great job though. I also replaced the 2M2 resistor. No changes, still getting the 8+V at collector of Q1.

Some further information: I have realised that this is becoming a common problem with many of my builds - which suggests a degree of incompetence on my part. The problem being the 'blatty' sound of an unbiased transistor.

The circuits I have built with the problem are all the Emerson EM-Drive, two on strip board and one on a Fuzzdog Pedalparts PCB. The readings at collector appear to highlight the problem as with the Devi Ever fuzz. I think I've read somewhere that this circuit is also very sensitive to many different factors? But it is another one that I have breadboarded and it all seems fine.

Anyway, the current plan is to rebuild either of the above circuits again. I'll breadboard the circuits and check that they work and then aim to clean everything thoroughly (legs and stripboard) so that I can try and have the cleanest joints possible (with unblobby solder!). I'll see where that gets me.

Thanks again for the help.

After another look at your pictures of the pcb I am convinced you have the Collector and emitter of the 2N5089's swapped over. The collectors appear to be connected to the second from bottom copper strip, which has the no breaks in the track and connected to emitter of the metal can transistor.
I would not rebuild it, just get sucker and then soderwick to remove the excess. The joints are not bad, just too much solder. When I get circuits like this to fix I heat and hold the board in my hand, and quickly flick off excess, but then you need to clean up the splatter.

[Mike2E]

> I think I have spent to long working on high accuracy audio and instrument repairs.

I've been there too. Designing so the operating point is *nailed* and is obvious by inspection, without knowing the active device specs.

Perhaps the main flaw is a lowish input impedance. (But a high stability base bias network for a 4-R design will also have low input impedance.)

I don't think the low input Z is a huge problem. Most pickups are about 6k to 10k, a bit of HF loss (maybe), but it will feed a valve amp with transformers reducing high end response. If it is after another pedal, even less of an issue. I have out of interest performed simulated response tests using test equipment, and changed loads on pickups.

[JCGH]

I think that might be it! It's so simple... I think must have all of them the wrong way round in that case, I think I've confused myself when mirroring the layout on the other side of the stripboard ready for soldering.

I did find recently that if I bread boarded a circuit and reversed the transistor that the voltage reading were similar to those I was getting on this circuit (i.e. 7ish volts at collector - which was actually base...) so switching it round was my next plan and I think it was mentioned as a potential problem earlier. Your diagram has helped confirm this I think.

Next stop, success. Hopefully!

Thanks for the continued attention, I'm in the middle of some major exams so I haven't picked up the iron in over two weeks!

[JCGH]

It works! (I've managed to fit in an hour of soldering!) Not only that but I've fixed all my other builds (those being two other EM-Drives and a ZVEX SHO) as they all had the same problem! I can't quite believe I made such a silly mistake! It all happened when moving from the strip board diagram to the actual board... At least I know now for future builds.

Thanks once again!