Marty Marts Tornado built, but has a small problem

[MartyMart]

Did your noise go up at all with the gain increase? I have mainly hiss now proportionate to the gain level
it seems tolerable, but less noise is always good. If you get a few minutes could you make a maxed gain clip? 

I posted that answer a few steps back ! ( It's getting complex - I know )

I have no decernable change in hiss having gone back to "stock" with it.
There's some and a touch of earth hum, but I'm not going to get this HUGE version
in a box, it would be silly to try.
At some point, I'll remake a smaller one, though I also have a Thor/LA screamer/BBE/Q&D comp
and a few other things to get in enclosures first !!!
If this PCB is sorted, perhaps one of you could etch an extra one for me ? that would be V- cool.
This is good work and in the true DIY spirit chaps 

MM.

[Eb7+9]

My Dr. Boogie makes that same noise so I look forward to someone figuring it out(because I've failed too). I almost have to think that this is a downfall of JFet circuits that have many cascaded stages.

the rushing noise has nothing to do with layout I'm afraid ...

Marty, you've got cap bypassing on every gain stage, giving lots of gain but increasing noise gain everywhere at the same time - which is already poor in jFET's running at 9v ... then you throw away a bunch of signal between gain stages using resistive voltage dividers between each stage ...

Dude!   

the combination of both will give you poor S/N ... that poor S/N will sound proportionally hissy if you're using relatively small caps to bypass Sources along the signal path ...

compare this to the Thor ... no bypass caps, no signal throw away with dividers ...

you can always try shaving off some Treble/Hiss by paralleling the drain loads with 0.001uF to 0.003uF caps on every stage, not a perfect solution by any stretch ... still, I think you're asking for trouble here ...

it's easy fix to remove Source caps and dividers, but there's the obvious potential loss of voicing ... that's shitty !

maybe a good compromise is to remove all Source caps, and voltage dividers, then stick the one Source bypass cap that best defines the current voicing and re-bias Drain loads carefully ...

g'd luck ...

[MartyMart]

That's great information JC - Bless you !!
I see exactly what you're saying, I can probably compensate for the changed "voicing" by
adjusting the interstage caps from the fet drains and roll off caps/RC network etc.
I have to repeat that I DONT have a real hiss/hash problem with mine, perhaps I
just got lucky if it's not a layout issue.
I think it's time for a Tornado Version 2.0  !!!
My dilemma will be that if it sounds this good AND is quiet, I may not want to post it up !!

Good man !
MM.

[Eb7+9]

just got lucky if it's not a layout issue.
I think it's time for a Tornado Version 2.0  !!!
My dilemma will be that if it sounds this good AND is quiet, I may not want to post it up !!

... your call n'est-ce pas ?!

the first BSIAB I built sounded dark and muddy, to the point of being unusable ... everything biased right but it seems that channel capacitance must have been xtra high for some of the devices I was using then ... this is very possibly another one of those high-variance jFET issues ...

I thought about this and forgot to mention, yeah, could be that your devices are perfect for the circuit as you've arrived to it ... it's something that's almost to be expected with FETs ... I wouldn't say it's luck on your part, more so that you adjusted the circuit to work correctly with your starting ingredients ... like any competent chef would 

just going back to the cap thing for a minute - old school purists in high end hi-fi circuit design land avoid Source/Emitter/Cathode bypass caps like the plague (... well, almost) ... the idea that the more you can achieve your (gain) goals without it the more you stand a chance of maintaining a certain level of transient "accuracy" (for lack of a better word) ... "focus" perhaps ... in a distortion circuit that can matter even though the definition of fidelity goes out the window ...

some interesting set of design constraints this music stuff ...

[MartyMart]

Makes me wonder why almost all Tube amps have quite large cathode bypass caps  !!
.... probably they dont suffer the same noise issues from those and the resistor dividers ( answers own question )
Yup, fets can be a real "bugger" sometimes !

MM.

[Eb7+9]

triodes have about the same device noise as transistors but the signal swing in a typical HV app is way higher leading to better S/N

[John Lyons]

Nice insight JC!
The dividers were kind of nagging on my mind as well. The resistive dividers are throwing away signal to ground but removing them will cause the gain to increase correct? I know how they work I just was wondering about where to go once they are removed or lessened.

Removing source bypass caps will lesten the gain as well as some tone shapping as well. What is a better way to reduce the gain outside of the divider and removing source bypass caps? Rasing the the voltage to the drains?

Some Engle high gain amps have 10uf bypass caps on all the stages!

Marty
It does seem our builds are very different gain wise. Initially I thought it was more of a layout issue which is was to a point, but I have screaming gain to the point oscillation and my wiring isn't cluttered and the layout seems pretty close to ideal with the exception of grounding..but no hum on mine either... Plus I used shielded wiring on all the ins ans outs to the switch and back.
Getting tired of this i'm sure...

John

[MartyMart]

No, not tired I'm learning a LOT from this thread !!

I'm onto Version 2.0 as we speak
Schemo adjusted and I'm trying things out, most bypass caps have gone
and divider resistors adjusted down also more power filtering too !
May take me 'til next week - but this WILL be right

MM

[gaussmarkov]

john,

i think that's right.  i believe i read somewhere that raising the voltage supply to these circuits helps.  as a caution to others:  if you try this, be sure to respect the ratings on your caps and resistors.

we are moving on, but just for the record, here's a final version of the star-grounded tornado v1 beta:

[MartyMart]

OK, here's some "adjustments" to the Tornado, sorry I have not redrawn the
complete schem yet but this should make sense, schem comes v soon :
-----Tornado V 2.0-----

Q1 :
MPF102 bypassed with 2k2 and 680n cap, from the 22n drain cap output direct to 1M log pot.
470k r at gate to ground from Q2
Removed the 3m3/6n8 link
Q2 :
MPF102 bypassed with 3k3 and 470n cap, from the drain cap output ( now 10n), divider of 470k/470k ground
into the Muamp section.
Muamp :
2N5457's As before but with 680Ohm bypass r, from C11 (now 22n ) a 180k/180k divider to Q5's gate.
Q5 :
J201 Bypassed with 3k3 r no cap ! drain cap output now 22n then 470k/470k divider into Q6 gate
Q6 :
J201 Setup as a source/follower, 9v to drain source bypassed with a 10k which is the output to the 1uf
C16 to tone stack - as normel here.
Getting good tones from this with minimal noise, not quite as much "clean" range but this is about
gain here !!
Anything else not mentioned, such as gate to ground pf caps remain.
My layout is smaller and required a second ground and 9v strip, so I'm double filtering the 9v rail
one that feeds Q1/Q6 and one that feeds Q2/Q3&4/Q5 both use 100uf and 100n caps from 9v to ground.
The feedback loop was causing a touch of "howling" and also a wierd "compression" in this version
so was removed ( see above )
Not 100% there yet but getting closer ! it sounds a touch "thicker" so may well use more of the 10n and
smaller caps from drains between stages.
Thanks to all for the useful contributions BTW
MM

[MartyMart]

http://aronnelson.com/gallery/Martys-layouts-and-photos/Tornado_V_2_0?full=1

Schematic - bit ragged but readable !!  sorry Gauss 

MM

[gaussmarkov]

Marty,  C2 (120pF) to source of Q1 instead of ground?  The older schem went to ground and we changed that, right?  gm

[gaussmarkov]

... and the output of the mu amp is the source of Q4?  sorry, another blem on my first schem.

[gaussmarkov]

how's this?  i marked it up to show the separate +9v power rails, now VA and VB, adding nominal 100 ohm resistors to distinguish them.  the  components are relabelled to follow the schem.  C2 is a "miller-capacitance-sim cap" across the gate and source of Q1.  and the output of the mu-amp is the source of Q4.

[John Lyons]

Marty

You said "2N5457s as before" for the Mu amp. Is this correct? On the schamtic and all gauss's schems the Mu amp has been J201s.
A possible sourse of why my gain was way more?

I've been reading over the Fetzer valve material over at ROG. There is a lot of good material there.
Some points I've been thinking about and need to try.

Biasing and determining the operating points of the stages.
Getting a good amont of gain but not so much as to need to throw away gain with resistive dividers.
Adjusting the source resistors bassed on the gain of the FETs. ROG mentions much smaller values.

What is the difference between gate caps tro ground and gate to source caps? (simulating miller capacitance)

Source bypass caps and the idea that unbypassed source resistors run closer to tube triode curves.

Running at more than 9v to get more current strength.

Marty
Thanks for posting your reworked Tornado here.
A dual 1Mpot could be added easily here to get more clean range. After C7 in the v2 schem posted above.
Replacing R10 with the second half of the pot.

John

[MartyMart]

Chaps, thanks for jumping back in

So, John, my mistake the first version MU was indeed J201's !
I've gone with 5457's now- seems to help and may help you on the gain front Too !
The dual 1M or pair of 1M pot idea is sound, I'll implement that on monday.
Gauss, schem looks right, though I now have all the 120pf's gate to ground and not "millered"
Not sure how much change this is, could be loosing some high end, so the 2n2 at end with the
10k possibly not required at all.
I had a touch of your problem john, it came on intermitantly but was solved with a re-bias and
fresh 9v battery !
It's failry quiet now, about the same as V 1.0 
It's much smaller, perhaps half the size, with no "ill" effects, so I'm pleased.
I recommend a possible reduction of drain cap size, all 10n and off Q2 at 4n7 sounds real nice
too.
Ditch the bypass cap at Q2 also works fine and reduces gain some, could be on the edge of
"howl" at full tilt still !
Still some experimenting to do but it's 90% there I think.

Marty.

[gaussmarkov]

thanks, marty.  shall we leave our schems as they are for now, given the fluid nature of things just now?

also, i wonder whether everyone should measure Vgs for their transistors and report those, perhaps using R.G.'s JFET measuring circuit?  might be good to match the mu-amp transistors, too.  both would be helpful for keeping the supply under 9V, per John's comment.

i will be lower profile for the next coupla weeks.  pls don't interpret that as a lack of enthusiasm on my part.  i'm just slammed with other stuff. 

all the best, gm

[MartyMart]

thanks, marty.  shall we leave our schems as they are for now, given the fluid nature of things just now?

also, i wonder whether everyone should measure Vgs for their transistors and report those, perhaps using R.G.'s JFET measuring circuit?  might be good to match the mu-amp transistors, too.  both would be helpful for keeping the supply under 9V, per John's comment.

i will be lower profile for the next coupla weeks.  pls don't interpret that as a lack of enthusiasm on my part.  i'm just slammed with other stuff. 

all the best, gm

Gauss, no problem at all, I'm going to do some final tweaking during the next few days and post an updated
schem, V 2.1 !
I'll implement some changes including John's "dual/pair" of gain pots, for better overall range and some gain/tone
changes, as it's very close to GREAT now !!
The dual pots remain in my original and I like how that works for the cleaner settings.
I think this should end up as a Marty/John/Gauss design, given the number of suggestions and input !
This is all much appreciated BTW
More in a few day's people, so be patient !  I'll sort out a new post for it and include perhaps a vero layout
also, as mine is quite a decent size now, about 8cmX5cm it will go "sideways" in a 1590BB

MM.

[John Lyons]

Marty, wow; you really got that vero sized down! This new build has the same amount of parts pretty much, doesn't it?
I'm going to tackle Gauss's New PCB layout for it and see how that goes. I don't have any vero & haven't used it before.
What's the gain like now on yours marty? Still a fire breather with the lowered gain FET type?


Gauss
The Fet Matcher would be a good idea. I bit of a pain but once you get it up and running it would be simple.
Especially if you use FETs a lot.
I think there is one layed out in the Gallery if I'm not mistaken.

Maybe a new thread topic but I was thinking of trying Parallel FETs as in some Matchelss, marshall etc type circuits.
If the FETs were not matches how would this affect things such as bias? I'd have to average the bias setting of the trimmer but I wonder how the parallel FET thing would translate from a tube triode pair?

John

[MartyMart]

It's still a fire breather but a wee bit more tame than my V 1  !!
I could make a vero version for you John, just main parts and trimmers with
wires ready for pots and I/O if you like ?
Could be interesting to compare direct to a PCB version ?
I'd appreciate a PCB also ....

Parallel Fets !!!  would that work like BJT's ?
Lowering gain then right ?

MM