Hacking up an old Garnet Amp

[spacelabstudio]

So I've had a late model Garnet Revolution I amp for a while.  It sounds fine, but it has never really excited me very much.  Perfectly functional, but not particularly inspiring.  These old Garnet amps seem to have dismal resale value, so I figure this can maybe be a decent platform for some experimentation.  (If one of you loves these old Garnet amps, you can buy it from me now before I kill again--I'll use the money to buy parts for a scratch build.)

At this point I've done a lot more lurking, reading and studying than I have real practical experience with modifying amps.  I'm likely to just cargo cult circuits from amps that I think sound good and see what happens, not necessarily understanding what I would expect the impact on tone to be.  The blackface Princeton is a good starting point.  I've also been digging a Gibson Scout I've been playing with lately. 

Here is the current schematic:

http://spacelabstudio.com/garnet_g45tr_late_version.pdf

In a lot of ways the basic circuit here is not that different from the blackface Princeton, just a lot of the values are different.  With the T filter (where the mid control is) bypassed (by using the pull switch on the master volume) the tone stack is basically a bf tone stack with some different values that give less of a mid scoop. 

The measured voltages at the tap points are A: 448V, B: 313V, C: 257V, D: 225V

I applied a 1V RMS sine wave to the secondary of the OT and got 15V on the primary, so I'm calculating an impedance ratio of 225:1 or an 1.8K primary assuming an 8 ohm speaker on the secondary.

I've been reading Merlin (http://www.freewebs.com/valvewizard1/pp.html) and am guessing, without actually having drawn the load lines, that the lower than other amps primary impedance and the lower than other amps screen voltage are related--it probably wouldn't be advisable to try to raise the screen voltage to be closer to plate voltage, like in the Princeton, unless I also raised the primary impedance.  (I guess I could get a new output transformer, or use a 16 ohm speaker if I wanted to go that route.)  I'm not sure what the tonal impact is of the different output design.

I can experiment with removing the master volume.  It's likely to go back in if I don't here a tonal difference, because having a MV is, you know, useful.

I can switch out the 12au7 3rd gain stage/PI and rebuild that part using Princeton values.

I could try cathode bias for the power tubes.

I could try getting rid of the T filter and building a BF or Marshall TMB stack.

I guess I'm just fishing for ideas/suggestions.  What would you guys try?

Chris

[PRR]

> 15V on the primary

Half-primary or whole primary??

Load on a pair of 6V6 should be 5K-10K, and Gar knew that. I would bet you have measured half-primary, the nominal impedance is four times higher, 7K.

> experiment with removing the master volume

Experiment? Turned full-up is the same, near-enuff, as not having it.

> has never really excited me

What speaker?

[RicharD]

At a glance, the tone stack looks lossy.  Play with that.  Play with the preamp cathode bypass caps.  +1 on the speaker. 

Crappy amp + good speaker = ok sound

Good amp + crappy speaker = crappy sound

[spacelabstudio]

> 15V on the primary

Half-primary or whole primary??

From anode to anode.  Maybe there's something throwing the measurement off, but that's what I got.  Amp schems rarely spec the transformers, for some reason.

> experiment with removing the master volume

Experiment? Turned full-up is the same, near-enuff, as not having it.

That's my instinct, but people seem to hate on master volumes for some reason, so I figure I should at least try it.  Easy enough to untry.  If nothing else, I'd get rid of another coupling cap.  I'm not expecting a big difference, but my practical experience is pretty lacking.

> has never really excited me

What speaker?

First the original Marsland and then an 80s era Celestion.  In the last couple of days I've plugged it into a Weber Sig ceramic 12b and an eminence gb128, both speakers I like a lot more.  Of course, what's the fun in convincing myself that it's fine as is?  ;)

Chris

[spacelabstudio]

> 15V on the primary

Half-primary or whole primary??

From anode to anode.  Maybe there's something throwing the measurement off, but that's what I got.  Amp schems rarely spec the transformers, for some reason.

Just measured again.  I think my cheap radio shack mm plays tricks on me.  This time I measured 27.5 / 0.92.  Assuming some slop in the measurement (29.89:1), I'll just assume they were shooting for 30:1, which works out to 7.2K primary.

Given, that, I probably have room to bring the screen voltage up.  Any word on what the audible effect of a change like that might be?  In the spirit of just trying things, I guess I should just try it and see.  Again, though, I'm more or less in cargo cult mode, just making things like other things, not necessarily understanding the why or wherefore.

Chris

[HotBluePlates]

I'd probably try to talk you out of giving it a facelift. The circuit looks quirky enough to appeal to me.

[The_Gaz]

Yeah, I don't think I'd hack up a Garnet either, maybe a boring old Traynor!  If I did, I'd probably look at some of his other amps because they all use a similar topology and James style tonestack. Perhaps there's a 'better'  model like a BTO or Sessionman, for example, that has different tonestack values, etc. that you could try.

[loogie]

I've undone a few hack jobs recently.  Just put a copy of the original schematic in there somewhere, plus some notes on the damage you did so the guy who ends up putting it right some day will have an easier time of it.

[spacelabstudio]

I'd probably try to talk you out of giving it a facelift. The circuit looks quirky enough to appeal to me.

Yeah.  I've enjoyed studying it.  I think the calculus here, though, is that I can't get any money for it and I don't use it as is, so this is a good way to save it from just sitting there collecting dust.  It ought to make a decent laboratory for seeing what does what.

Chris

[bluesbear]

It's your amp and, as you say, it isn't worth anything... yet. I say do what you want but keep in mind that things change. I've hacked a few that within a couple years suddenly WERE worth money. I have a Super Champ. All that's left is the chassis. Crappy amp; wasn't worth a nickle, now they bring hundreds. On the other hand, I love the amp it is now and if I had an original Super Champ to sell, it would just pay enough to build what it from scratch... and it would still be a crappy amp.
Dave

[jjasilli]

Ditto.  I'm in the middle of continuing the hacking of a Traynor Bassmate (EL84 version) that I picked up on eBay for the going rate of under $200, then. Now, if untouched, it's worth about $600!   I also have the 6V6 version of the amp, which I'll leave alone.  I just didn't like the tone of the EL84 version, but never expected the value to skyrocket.

[DummyLoad]

stick a 12AX7 in the last gain stage/cathodyne - i'm assuming they are same envelope. bypass mid controls. try other tubes for first stage - more gain use 12BZ7 - less gain use 5751 or 12AY7. don't bother hacking the MV just turn it up all the way. if you want to burn off some gain, parallel the MV with a 1M. simple mods that can be undone would be where i'd start. if you're looking for JCM800 or dumble ODS tone, you likely won't find it in there.   

[spacelabstudio]

I'm busy learning the hard way.  I decided to swap R1 and R2 in the power supply to raise the screen voltage.  Of course, I forgot take into account that the tube will draw more current this way and needs more negative bias voltage.  Cooked the power tubes.  Nice USA made Sylvanias.  Actually one still works.

Anyway, before I put another pair of power tubes in there I figure I better do something about the bias supply.  I figure I'll probably make it adjustable, but before doing that I'll just increase the magnitude of the bias voltage so I have something I can work with.  As far as I can tell (and I may very well be missing something, as we'll soon see) the bias voltage is basically set by the voltage divider of 120K and 33K resistors before hitting the rectifier diode.  The 100K and 82K resistors are just part of the RC filter and shouldn't really have much to do with determining the output voltage.  The output voltage is -34V.  The input to the rectifier is 56Vrms.  I found a 47K/2W resistor so I substitute that for the 33K, which as expected raises the input to the rectifier to 70Vrms.  Astoundingly, the output is still -34V. 

What am I missing?

http://www.el34world.com/charts/Schematics/files/garnet/garnet_g45tr_late_version.pdf

Chris

[spacelabstudio]

My birthday present for myself is a break from client work for a day to tinker around.  So, in the bias supply, I added a 100k trimmer pot in series with the rectifier diode and the 100k resistor to ground.  The 100k resistor and tremolo depth pot that form the output of the bias circuit are now attached to the wiper of the trimmer pot instead of directly to the rectifier diode.  I was expecting this would only give me the ability to reduce the output I was already getting from the circuit by up to half.  Surprisingly, the max output increased up to about -44V and I could adjust that down to -22V.  Not sure I understand why the max voltage increased, but I have to say I'm happy with the result.  (If the explanation of what I did doesn't make sense, I'm happy to draw it out and scan it.  I don't get along super great with schematic software for some reason.)

With a not quite new, but fairly low hours pair of EH 6V6s in there, it then starts to sound pretty great, biased to about 20mA per tube at the cathode.  Of course, enough has happened in the meantime, that I don't really get to evaluate whether raising the screen voltage had any positive effect on the tone.  I could see any difference that made being easily swamped by the effect of newer, peppier output tubes.  Almost sounds like a 6L6 amp to me this way.  Plenty of clean output for my taste.  I don't play super loud, usually. 

Of course, now it's more or less in the same tonal territory as my DRRI.  And my goal isn't really to make a backup or substitute for that amp.  I'm going for something a little different.  Not sure what, exactly.

So, intrigued by something Merlin wrote about mixing cathode and fixed bias, I decide to experiment with that a little.  Replaced the 1R I had tied to the cathodes with a 100R.  Shooting for 40mA total, again, I used the bias adjust to get 6V at the cathodes.  Interesting.  Not a huge difference in sound, though.   Dug around for other hi-watt resistors I might have lying around to try and found a 470R.  Installed that and adjusted for 18.8V at the cathode.  Now this made a big difference.  A lot closer to what I was looking for.  Which I guess, was just cathode bias, or at least a little more than a smidgen of it.  Not near as much clean headroom, but a more interesting sound, to me.  Bypassed the cathode resistor with a 150uF/25V electrolytic, which seemed to clean it up a hair. 

Starting to really warm up to this amp, now.  My hope has been to use it in a band where I'm currently using a Vibrochamp.  Our drummer is really nice to us and doesn't hit very hard at all, so we can get away with that.  So obviously, I'm not looking for tons of clean headroom, but I wouldn't mind being able to get a little more stage volume without being too clean.  This is starting to feel like it might work for that.  The fact that it's in a really light, compact cabinet doesn't hurt either for it's chances of getting used.

Anyway, I'm really digging the mixed fixed/cathode bias trick right now.  If you have enough resistors to play with you can dial in the amount of cathode versus fixed sound that you want.  Very cool.

Chris

[PRR]

> Not sure I understand why the max voltage increased

Where does the bias supply come from?

300V AC through 120K loaded with 33K, to rectifier loaded with 82K (100K?), and through the trem-pot and grid-leaks (no DC loss) to grids.

You changed the 82K/100K to 100K+100K or 200K. Less load on the rectifier. Since the rectifier is fed through the VERY large 120K, it sags, and sags less with a lighter load.

It's good.

[spacelabstudio]

> Not sure I understand why the max voltage increased

Where does the bias supply come from?

300V AC through 120K loaded with 33K, to rectifier loaded with 82K (100K?), and through the trem-pot and grid-leaks (no DC loss) to grids.

You changed the 82K/100K to 100K+100K or 200K. Less load on the rectifier. Since the rectifier is fed through the VERY large 120K, it sags, and sags less with a lighter load.

It's good.

Ah, that makes sense.  Thanks!