Marty Marts Tornado built, but has a small problem

[snap]

keep the outputs away from the inputz!

groundplane everything you don`t need, to save your etchant and the environment,
but don`t come too close...

[gaussmarkov]

keep the outputs away from the inputz!

groundplane everything you don`t need, to save your etchant and the environment,
but don`t come too close...

hey snap, thanks!  that reference i mentioned above also says to use ground fill liberally, but just for electrical reasons.  the environmental ones are good too.     i see that a lot of folks like to use extra wide traces for +9v, too.  my current guess is that if you are up in the 40mil neighborhood then you have met the "extra wide" criterion.

hmm.  i now realize that this could be a thread hi jack.  we should keep this focused on the tornado.  my bad.  i will start a new thread.

[John Lyons]

Well shit! I spent the day ignoring my "work" and built the new layout I posted above...and guess what! It's the got the same noise problem!!!!!!!!!
Not sure where to go from here.

Marty, if you could post or Email me a rough sketch of your layout or a pic that would be great.

John

[gaussmarkov]

dang, i'm sorry to hear that john. 

[gaussmarkov]

thinking about what i have read in other threads, i wonder if we could create a simple test for the presence of parasitic capacitive coupling in one of john's layouts using a copper clad board.  ground planes are used to fight unintended capacitive coupling, allowing the coupling to occur with the ground plane and, so, less with neighboring traces.  if john grounded a fresh board and placed it against the solder side of one of his builds, insulated by something very thin (but not a magnetic shield) could we get the ground plane effect?  if the noise reduced, wouldn't that confirm that some traces were capacitively coupled?

maybe there is some version of this idea that works even better?  if it works, i wonder if a more refined version couldn't be used to find the location of capactive coupling in a bad layout?

i'm afraid this sounds kind of crazy, gm 

[MartyMart]

John, that "sucks" !!
Sorry man, I feel responsible for your "mental torture" now 
I dont have the hours free for a total draw up, so I'll take a couple of snaps
and mark it up quick in photoshop in the next few hours.
You'll get a picture of how BIG this circuit is  !!!

MM.

Board photo and details :
http://aronnelson.com/gallery/Martys-layouts-and-photos/Tornado_vero?full=1

It's HUGE isn't it !
You can see just how far apart the stages are and the pots too, there's quite a distance between the
Jfets  and quite a lot of copper too !!
I need to rebuild a smaller one and see where that gets me 

[MartyMart]

John, Here's an mp3 of my build :
Gain/Tone full up recorded through Behringer DI/cab sim as usual
Silence then chords then pup switching and more silence
The bit of earth hum and pup switching is louder than the hiss by far.
This was recorded VERY loud so watch those speakers !!
I also have an end of circuit 2n2 cap "lifted" so have more high end going on !!

MM
http://aronnelson.com/gallery/albums/Martys-layouts-and-photos/Tornado_noise.mp3

How does this compare with your signal/noise ratio ?

[gaussmarkov]

hey,

i just got a nice note, through gaussmarkov.net, from a fellow named phil in nambour, australia who apparently does not use this forum. 

his opening comment about my layout is "The circuit layout looks very cluttered and with high impeadence/ high gain floating so close you are just lucky to even get it to work."  i was tempted to leave that out, but i am including it for assurance to john that he is not being singled out by the gods for mind twisting. 

he also comments "I am completly lost as to the need of a feedback loop like that one, I would suspect that to be problematic.
It does pull the mid down a bit.  [-4 Db @ 1kHz] you are probably better without it.  The combined loading effect of the tone with no amplification at the end also puts the potential output at a great disadvantage."

he made a number of suggestions that sound helpful to me:

1) the trace on my layout from R16 to R17 under R19 alone is a big issue.  i have realized that this is one of my worst offenses in having a nonlinear signal path layout.  john's version fixed this.

2) don't have c10 and c11 side by side.  both john and i have this in our layouts.  it's relatively easy to change in mine, with no traces running under either capacitor, so john could experiment with that immediately.  he also says that this junction needs to be surrounded by ground.  also, one can see whether these components are a problem by checking whether they are microphonic!  gently tap on them and see.  (first time i have heard that one a stompbox circuit.)  marty, where are your c10 and c11 relative to each other?

3) correct the output of the mu-amp (john already noticed this but i have not updated gaussmarkov.net yet--i'd better get on it)

4) scale down the mu-amp voltage divider to 10k/10k with a 4M7 bias resistor (which we've discussed, but not with such a high bias resistor).  john, what did you use in your second build?  marty, did you try this?

5) lowering the value of R13 will decrease gain and help reduce oscillation problems

i think phil clearly knows a thing or two, given that he has spotted independently a number of the issues we have identified ourselves.  so i am passing on the rest.  and now i am going to go back and thank him!    and if i have time, make another layout.

cheers, gm

[John Lyons]

Alright! Looks like we have somefeedback (literally and figuratively)  to work with. I need to check out Gauss's post in detail but I have this clip for comic relief. Marty, seems my built has way more gain than yours. This is with the voltage at 6 volts across the board. I can get singing feedback with out plucking notes...

www.mrdwab.com/john/tornadonoise.mp3 

This is the newest board with the parts spaced out.

Gain full up, humbucker bridge, after I stop playing I turn down the gain all the way and that's the noise I get.The static varies. The underlying rushing noise is consistant with all the 3 builds.

EDIT: Almost forgot that I disconnected the Feedback when the clip was recorded. I couldn't hear much difference and wanted to see it it affected the noise at hand...

I need to go over the posts above with the original layouts. I just wanted to post this so you know i'm not crazy and to let you hear what's going on.
Thanks for the pics marty!. I need to study those now as well. Your clip is much more tame than mine. If I didn't know better I'd say it's not the same circuit!

Marty. No hard feelings obviously. Just strange that we have such different behaving builds. It does sound great. Just need to tame this beast.

John

[mydementia]

Wow John - now I hear the waterfall you described earler. 
Maybe you could post some pix of your builds too... like I said in my post in this thread - my Tornado seems to work like Marty's... minor 'high gain circuit hiss' with gain maxed.  Any particular reason you biased all your FETs to 6V?  I biased mine to 4.5V per Marty's recommendation (original hand drawn schem in gallery). 
Are you reusing any parts in your builds?  Maybe you have a rogue JFET honking up the works?  I just went through a huge troubleshooting exercise with Brian over at ROG on my multiple misbehaving Thor builds - come to find out, the on-off-on switch I used (instead of single SPSTs for tone selection) was causing the problem (different switch in each build... but same wire proximity).
What an interesting problem...

GM - thanks for passing on the information on 'good' layouts... this thread will make all of us better at it!
Mike

[MartyMart]

Great info chaps, the feedback path came up in my original, posting someone mentioned
"problems" and "why?"
Well all I can say is without it, it just sounded like any "high gainer" and with it in place
it souded and responded - in a much more "amp&speaker" way , "speaker spank" is the
term that I used.
I tried the 2x10k silent bias with 1m to the fet - it's still there, but made little change for
me, no more and no less noise !
You seem to have a LOT more gain than me, perhaps my 2m2 pot and one or two tweaks
has made a HUGE difference, that I have not spotted.
Look at the build though - it's VERY close to the original schem !
I need to remake one as per the original schem, are you using the recommended Jfets in
all positions John ??
Thanks for all the input guys, frustrating but appreciated, I just wish that I knew more
"electronicly" about what's going on here !

[John Lyons]

I used all new parts on all three builds so far. I've socketed all the FETs and have changed all of them at least twice.
I biased them all over the place. Mainlt at 4.5v and for the clip I was having a bit of oscillation so I biased them at 6v to get some more headroom/clean gain.

Here are two of the boards I've built. The larger one is over 4 inches across!



Marty
My builds are as the original schematic except for:
Taking the output from the source of Q4.
.001uf across Q5 trimmer
Tried 10K/10K 1M and 100uf cap at the Muamp but didn't notice a difference.
All the FETs are as called for in the original schematic. Q1-6 MPF102, 2N5457, 201, 201, 201, 201
I did try a 1M at the place you use a 2M2 (with the stock 200K/200K) and it acted like a volume/tone control cut almost all the tone as well as volume
I could barely notice the connection or disconection of the feedback loop. Slightly less mids possibly...

John



John

[puretube]

have you done swapped the trimpots for fixed resistors,
to check whether the noise origines from them?

[MartyMart]

Gauss posted comments from the Australian guy, who stated that with this
much gain and a "crowded" board we were lucky that it worked at all !
In a way I agree, this is like building a "Train wreck" amp and NOT looking
after the lead dress.
Slight differences aside, my build is WAY more spread out, it's 12.5cm X 4 cm  !!
That puts gain pot and I/O connections more than 16cm away from each other.
Ton has just mentioned the trimpots, which I offered as a suggestion a page or two
ago, well worth trying a "new" type, or measuring and using fixed values.
I'm going to remove the 2m2 "xtra" gain control to get original values back in that
area to see what happens.
I'll get back tomorrow on that one.

MM.

[MartyMart]

OK, I very quickly went back to the original, no 2m2 pot and the 200k pair
after the Muamp section back in as per the original.
Quite a bit more gain, much like yours now John but NO extra hiss/hash going on 

With TC not maxed, it's quite quiet, earth hum is a slight problem which a box will solve.
Sorry ..... it looks like a layout / circuit size issue for sure.
I did'nt even use screened I/O cable yet either !

MM.

[gaussmarkov]

Great info chaps, the feedback path came up in my original, posting someone mentioned
"problems" and "why?"
Well all I can say is without it, it just sounded like any "high gainer" and with it in place
it souded and responded - in a much more "amp&speaker" way , "speaker spank" is the
term that I used.

marty, i hope you did not get the impression that i criticizing the circuit.  i recalled that you had already addressed this but felt that i should not edit what phil had to say.  so i was just passing it along.

Gauss posted comments from the Australian guy, who stated that with this much gain and a "crowded" board we were lucky that it  worked at all !  In a way I agree, this is like building a "Train wreck" amp and NOT looking after the lead dress.

again, passing on the comments regardless of how much i like or dislike them.    my original layout  has taken quite a beating.  the way i see it, this has been a great project from the learning perspective.

all the best, gm

[MartyMart]

Gauss, no problems at all, I'm VERY grateful that there's even interest in this
I'm just a bit frustrated for John and want to resolve this asap !

MM

[John Lyons]

Yeas, Marty... I'm frustrated about the noise but I just want to say that the circuit is huge sounding and I applaud your getting it designed built. Thanks! No way am I blaming you...

We're almost there and some new light hasw been shed on it which will help. It's a learning proces and if we can tame this then were learning something...
I'll try out using a fixed resistor for the trimmers. Not sure that will do anything but at this point who knows. I do get some DC scratch when turning them but that's expected I would think. No?

John

[puretube]

John: the signal`s ground from your Q1 has to "travel" all the way around the board, to reach the groundpoint of your Q2...
(here the signal strength is weakest);

on it`s way from Q1 to Q2, it passes through all the meshes of the higher order FETs` groundpoints, where larger signals are operating
and larger currents are flowing to and fro, with the battery and big cap somewhere in the middle...

this is not exactly a "star-ground", but rather a "string-ground";

now usually this is a danger for hum- or motorboating problems,
and it may not be the (a) reason for the hiss,
but worth a try testing to break up the ground trace somewhere right of the safety diode
and wiring a jumper from gnd Q1 to gnd Q2.

If I interprete Marty`s pics well, he got the power supply connections (battery and big cap) right at the input, with shortest possible wires...

[puretube]

I do get some DC scratch when turning them but that's expected I would think. No?

that`s "crackle OK"  (term created by Zachary Vex  ), and can be regarded as "normal"...
(I asked about fixed resistors, coz I had read Marty mentioning it, but a look at the build pix showed the trimpots,
so I just wanted to repeat that issue, to make sure it has been tried.

(I have the carbon-pot t-shirt, even with DC-decoupled circuits... depends largely on manufacturer...)