Marty Marts Tornado built, but has a small problem

[MartyMart]

The only PCB layouts that I've built are things like tonepad etc, where they
are proven "workers" built by many dozens of people, not knocking the layout for
the sake of it, but with such high gains available, there's bound to be some changes
and revisions required.
Doing your own, or perfing/veroing your own will be very worthwhile.
I'd try to space things out a little on this one if I were you.
Mine is just too huge and I need to get a smaller one done, I'm scared that I'll
also run into similar problems !!

MM.

[gaussmarkov]

The only PCB layouts that I've built are things like tonepad etc, where they
are proven "workers" built by many dozens of people, not knocking the layout for
the sake of it, but with such high gains available, there's bound to be some changes
and revisions required.
Doing your own, or perfing/veroing your own will be very worthwhile.
I'd try to space things out a little on this one if I were you.
Mine is just too huge and I need to get a smaller one done, I'm scared that I'll
also run into similar problems !!

MM.

john (basicaudio) thoughtfully pm'd me about this thread.  i have corrected the C2 connection to the source of Q1 and revised the original layout.  i would also be happy to space things out a bit.  are there some particular suggestions as well?

gm

[gaussmarkov]

i have also added some comments to gaussmarkov.net and the project file to alert people that my layout may be noisy.  i do not want anyone to mistake a problem with my layout as a problem with Marty's tornado circuit.

again, suggestions for improvement are welcome.

all the best, gm

[John Lyons]

I worked on another layout today. Its as linear and straight forward as I can get it.
The feedback is a little messy with a jumper but better than fooling with running across
all those stages on the PCB with a high gain signal.
Anyone see anything I should change here?



C2 (120p)in gauss's schematic was corrected and now goes from gate to source on Q1
The .001 cap was added to Q5 to cut a little fizz out
The output of Q3/4 now comes from the source of Q4 as on marty's build ( not shown on either schematic)


I hope this does it. The circuit sounds great...just a little background noise to cut out and we're set!

I'm afraid to see how big this board is now. Haven't even checked.

Thanks so much to Marty For making this one up, and to Gaussmarkov for making the initial layout.


John

[MartyMart]

I'm crap looking at PCB layouts !! however after a few minutes of going "square eyed"
it seems to be fine, dont see any obvious errors in there John.
Spacing seems better, lets cross some fingers !!

MM.

[gaussmarkov]

i'll reproduce john's layout in eagle and that will verify it.

i feel like a broken record on this point, but it may be relevant so i will repeat it:  maybe the "noiseless biasing" approach described by R.G. is helpful here?  this would involve reducing the magnitude of the 1M resistors in the voltage divider for the mu amp to 10K and adding an extra resistor (470K or 510K) coming out of the divider.  edit: one also adds a cap to the divider junction to shunt thermal noise.  admittedly R.G.'s example is a pair of 10M resistors where the thermal noise problem is probably greater.  but if the noise persists, maybe it's worth a try.

cheers, gm

[MartyMart]

Gauss, you may be right, after all the ROG "Thor's" muamp is set up like that,
1m from the 10k's and an additional 100uf to ground.
Whilst I "dont" have a noise problem, I'll try this tomorrow to see if I can get an
even quieter build !

MM.

[gaussmarkov]

i just checked john's layout with eagle, after adding the 1nF cap to the trimmer on Q5 and changing the output of the mu amp to be the source of Q4.  i laid it out on a .1" grid so it can be perfed as well as etched.  the width is 35 squares, so the board will measure something over 3.5" built this way.  i did squeeze things together a little to get this dimension.  i just thought it would give an idea of the size of this layout.  thanks, john, for putting this together.  i hope it brings the noise down.  i am interested to know.

there are two items that may need fixing.  first, the 47nF cap (C11 on my schem) does not connect to a 200K resistor that goes to ground in the schematic. it is between a 1K resistor and a 200K resistor that connects to another 200K resistor going to ground.  second, R19 & R20 (both 470K) have C14 (33nF) connected to their junction whereas I think it should be C15 (150pF) that is connected there.  the third difference with the schematic does not matter:  the order of R6 and C6 has been reversed.

i called this v 1.1 beta, following Marty's initial label. 

here is a version that passes eagle's design rule check, so that it is consistent with the schematic:



here is the schematic that matches this layout:



all the best, gm

[John Lyons]

Ahh! Good catches there Gaussmarkov! I missed those goofs...

Here is what I have now.



I added an extra cap space for the Noiseless biasing setup 1M/1M, 100uf and 1M to Q4's gate .
If need be we can change the values from 1M to 100K, 10K etc. Not sure the cap is necessary but it doesn't hurt and can be taken out if need be.

Marty, Using a lower value for the biasing will give more current won't it? I Know that with, say a distortion + the high value Vref will starve the circuit a good bit. Not sure if this is the case here.

Gaussmarkov, Thanks for working on this with me! Your layout is "tighter" than mine. I snugged it up a bit some we'll see how it goes.

John

[gaussmarkov]

Ahh! Good catches there Gaussmarkov! I missed those goofs...

Marty, Using a lower value for the biasing will give more current won't it? I Know that with, say a distortion + the high value Vref will starve the circuit a good bit. Not sure if this is the case here.

Gaussmarkov, Thanks for working on this with me! Your layout is "tighter" than mine. I snugged it up a bit some we'll see how it goes.

nice john! 

i think the idea with the noiseless biasing is that you can keep the current the same.  the lower values in the voltage divider increase the current but you compensate with a high biasing resistor coming out of the divider into the gate of the transistor.  plus that resistor lets you put in the cap to sink thermal noise to ground.  unlike the distortion +, i think you want low divider resistor values, like 10K, not 1M.

gm

[John Lyons]

Sorry...I should have explained a little better. There was a thread about the Differences isn the Dist+ and DOD od250 schematics and the "better" sounding unit (OD 250)  has 10K/10 and 1M as the Vref. (no cap in the Vref through...) Mark Hammer brough up the value differences and suggsted lowring them as not to starve the circuit. I think the OD250 uses 22k/22k 470K.
I was thinking that 10K would be better or at least have more current with 10K and thn into a 1M and cap...

I'll have a look over the PCB one more time when I'm fresh and then try out the new layout (and hopefully the last one!!!!!!)
Three times is the charm with any luck...

John

[gaussmarkov]

yes, i wondered if you weren't thinking of that thread because i saw it also and thought it was pretty interesting.  there doesn't seem to be anything with such high resistors for both the voltage divider and the biasing resistors.

i hope this "linear" layout strategy does the trick.  i'd like to know where proximity matters and where it does not.  we still have lots of components right next to each other.     i just traced the signal path through my original layout and yours and i see the qualitative differences.  is that where you focused your efforts?

cheers, gm

[MartyMart]

GREAT work on the PCB's chaps
OK, John on very close inspection at LOUD volume, I do have a touch of your "high mid hiss"
However, it seems to be a least 5 X quieter than your example and only with gain/tone maxed out.
This is absolutely fine and I have that with a BSIAB and several other high gainers too.
You must just have a "compounded" version of that for it to be so noticable.
I'll try the noiseless bias system and report back in the next 1/2 hr  OK ?

MM.

OK, it's very hard to tell after 7 minutes of rewiring but it seemed to make little or
no difference to use 10k pair/1M to gate and a 100uf from the pair to ground, just
made a mess of my board !!
There is and will be some hiss when these things are at "max gain" but to put it into
perspective, the hiss is around 1/3 the volume of a single coil "hum"  when  switch
on the gtr !!!

[MartyMart]

There's one other thing that could be done here, set-remove and measure the 50k
pots and use fixed resistors to the drains !
Fiddly for sure, but I think that would stablize the noise a bit more.

MM.

[gaussmarkov]

here's a schem of what we are currently working with.

[John Lyons]

Marty

The Noise I've been getting isn't the usual hiss. If I crank the tone and or gain I get a little hiss, which is fine.
The noise I get is a mutted sound as standing next to a huge waterfall and covering your ears. It has some mid and low mid content thyat sets it aparts from the usuall hiss fro high gain stuff. Next to the Dr boogie and BSIAB2 the noise is significant even with the gain and tone turned all the way off. It may be a side effect of having the Feedback section run across the board and near the previous gain stages...?
How is your feedback line run on your vero build? (the 6.8n/3.3M from the MuAmp to the first stage output) did you run it on the baord or float it over with a wire jumper?

Well I'll go ahead and build this new layout and cross my fingers. If you can hear me yell Marty from your place there...you'll know there is still a problem.

Thanks gauss for the new revised schematic! Nice to have something for reference.
Seems it's just you me and marty here....

John

[mydementia]

Guys - just thought I'd chime in here...
I built my Tornado to GM's first layout (sounds like I have a cap in the wrong place after reading here a bit) but it works great.  I didn't make ANY subs and didn't take undue care in routing my wires (have a look at the picture below) and mine seems to work like Marty's...typical hiss at high gain/high treb but quiet otherwise.  I just made a couple more of GM's Beta1.0 boards but am going to wait until this issue gets sorted before I commit the parts...  Has anyone else built this circuit?  Now's the time to share!!

[gaussmarkov]

Thanks gauss for the new revised schematic! Nice to have something for reference.
Seems it's just you me and marty here....

John

that's enough for me! 

i've been going over your layout and the rog thor layout trying to infer good practice for these high gain circuits.  i see that the signal path (ignoring feedback loops for the moment) does not meander much.  but there are always a few places where it parallels itself for a couple of components.  like an RC pair in series placed side-by-side.  so it's confusing.  does anyone have any rules of thumb on what things to avoid?

Guys - just thought I'd chime in here...
I built my Tornado to GM's first layout (sounds like I have a cap in the wrong place after reading here a bit) but it works great.  I didn't make ANY subs and didn't take undue care in routing my wires (have a look at the picture below) and mine seems to work like Marty's...typical hiss at high gain/high treb but quiet otherwise.  I just made a couple more of GM's Beta1.0 boards but am going to wait until this issue gets sorted before I commit the parts...

this is nice to hear.  but also confusing.  thanks for chiming in!  nice perf also.  that warms the %^&*les of my heart, because i layout for perf and then convert to pcb.

cheers, gm

[MartyMart]

I know next to nothing about the "do's n dont's" of PCB layouts ...BUT on 95%
of the perf/vero that I use, I keep ground/power/VB and signal as far from each
other as possible.
On vero, this means power top, ground bottom VB where required and signal
throughout the middle !!
Seems to work for me and usually that means at least 1cm between any of the above
for runs other than a tiny 9v link to pin 8 of an IC - for example.
Perhaps the ground plane on the PCB need to be larger ??

MM.

[gaussmarkov]

I know next to nothing about the "do's n dont's" of PCB layouts ...BUT on 95%
of the perf/vero that I use, I keep ground/power/VB and signal as far from each
other as possible.
On vero, this means power top, ground bottom VB where required and signal
throughout the middle !!
Seems to work for me and usually that means at least 1cm between any of the above
for runs other than a tiny 9v link to pin 8 of an IC - for example.
Perhaps the ground plane on the PCB need to be larger ??

marty, thanks for helping with my question.  thinking along the same lines, in my last revision i put in some ground fill/pour (as i understand it, a "ground plane" is specific to multilayer (often 4 layer) boards in which there is a layer that is nothing but copper and that copper is grounded. hence a grounded "plane.")

i have been interested in this particular issue, keeping ground and +9v apart, and maybe you are interested in what i have read.  separation does not seem to be a universal principle.  i have also found online recommendations to keep these traces close together!    for example, see page 12 of this document:   PCBDesignTutorialRevA.pdf

in my tornado layout, i followed that recommendation.  i wonder if the reasoning is similar to the advantage of having a ground plane:   the changing current in a trace is accompanied by an equal and opposite image current in the ground plane so that there is less induced voltage on neighboring traces than without the ground plane.  by the way, the reference above also says to avoid unconnected copper fills; ground them or remove them.

and marty, i was wondering whether you could sketch your build for us so that we have another example of what works.  john asked about the connections for the feedback connection.  i would love to know roughly how you laid out the whole circuit.

cheers, gm