Designing a two-stage tube circuit with compressive cathode follower ala merlinb

[PRR]

Jeremy "Tubegeek" Epstein March 2014

....I got a chance to try it yesterday .. Apologies to anyone who has been waiting three years for an update!

5 years since you started. Just-now facing lack of overdrive. Stop painting the house and PLAY!

[tubegeek]

P, your advice is spot on, as always, brother. How are your hummingbirds doing?

Here's what I have since I last posted (a mere 2 days ago!)

I added a pi filter after the rectifier, dropping a couple of volts but knocking down a ton of humbuzz. I made a q&d 12V power supply node out of a 100K resistor in series with a 12V Zener || electro cap. That's all fed by the 140V 138V B+.

I started messing with a breadboard and a 2N5457 to get the JFET to bias somewhat well off the 12V feed. Ended up with a 6K3 load and a 680R source resistor, just shuffling through parts using the Tillman JFET Booster preamp values as a starting point. Once I had a reasonable bias point I soldered that mess in between the input jack and the old input node. 1uF tantalum cap DC blocker.

Then I played a little!

So: this is still a "preamp" with gain <1. But the JFET got the input level up so I could hear a bit of nice smooth grunge - I'd say the design CONCEPT is still promising. I'm going to have to get a little more serious - bust out the scope - to get much further and get a sense of the signal levels in each stage.

The "BONE RAY" tone stack is a bit different from the typical, but I think I like it. I'll see what happens when I get a chance to mess with it more. I should probably try bypassing it completely too, to see what it sounds like with that signal drop neutralized, so that maybe needs a switch?

Oh hey, wait a second - I didn't bypass the source resistor with a cap - so I have some "free" gain for the cost and effort of soldering in one cap. I'll do that tonight for sure.

It's good to be back.

I'll finish with a cautionary tale: the reason I had to put this project – all my projects! – on hold for so long, was, I let my workspace get so disorganized I couldn't find anything nor find a clear space to work. A nightmare. It took a visit from some German cousins to get me to spend TWO WEEKS cleaning up and sorting through mess, both in the "guest room"/workshop and the "pit of hell"/basement.

I found all sorts of half-baked idea carcasses in various stages of undevelopment along the way. I started off easy, boxing up another Gus Smalley ASDF that I had already completed the circuit board for, but my real target was this project, which I had never stopped wondering about. Still a challenge but finally making some progress!

[tubegeek]

Bypassed the JFET source with a 47uF electro. On the one hand, there is now good, usable level. On the other hand, I now really can't tell where/how the clipping happens without a scope. Played around with it a bit more and got a feel for the Bone Ray tone stack - it's weird but I think I like it.

Here are some pics. No bottom on it yet.

What do you think of the name?

UNDERVIEW:


CIRCUIT BOARD:


BONE RAY TONE STACK AND POWER SUPPLY:


OVERVIEW:


ANOTHER OVERVIEW, WITHOUT THE UNUSED DRILL HOLES PHOTOSHOPPED OUT:

[PRR]

> I let my workspace get so disorganized I couldn't find anything nor find a clear space to work. A nightmare.

Indeed.

Gets worse when you have another project-shop 10 miles away.

[tubegeek]

Gets worse when you have another project-shop 10 miles away.

In my case, four stories away, but yeah. Definitely worse!

On the positive side, I've been finding SO MANY parts as I sort and organize. I do NOT need to buy resistors for a while.

[Ben N]

Why not just sub a 6n17b for V1? After all, nobody ever ran an actual Bassman with a 12au7 in the first or second position, AFAIK.

[PRR]

> nobody ever ran an actual Bassman with a 12au7 in the first or second position, AFAIK.

Bassman 5F6a specified 12AY7 in the front end.

12AY7 was introduced in late 1948 with much hoopla, but nobody seems to have stuck with it past 1960.

[Ben N]

Bassman 5F6a specified 12AY7 in the front end.

12AY7 was introduced in late 1948 with much hoopla, but nobody seems to have stuck with it past 1960.

I was aware of that (I actually thought about writing "12ay7, yes, with less than inspiring results, but 12au7, no"), but the the 12ay7, with a mu of 44, is still in the middle gain range, while the 12au7, with a mu of 20, is clearly nothing other than a current driver. The 6n16b is pretty close to that with a mu of 25, so it's really pretty hopeless to try to get any meaningful "gain" (in the imprecise way guitarists use the word), or to make up for the losses of a passive tone stack. Yeah, you can try to compensate somewhat for that by boosting the input, but why not just sub in the right tube for the job?

But tubegeek surely isn't the first to try: https://www.diystompboxes.com/smfforum/index.php?topic=83164.0

[tubegeek]

Why not just sub a 6n17b for V1?

Funny you should mention that - when I get back home from vacation there's a package with Cyrillic writing on it waiting for me

I tried with the 16's first because a handful of 16's thrown in with some other Russian parts were my first submini purchase and I wanted to try using them for something. It was more an attempt to work through merlinb's cathode follower design info than anything else. As it turns out it really wasn't worth building with 16's.

Now I'm either going to need to recalculate everything to try merlinb's grid-current idea with some level of precision - or just run two half-17 gain stages into two half-16 cathode followers and see what happens....

I'm trying to get together an order for the SMPS parts so they will be there when I get back too. I promised my wife a Nixie clock for her office and the next one of those I build will use a SMPS. Might as well build one for this project too!

[Ben N]

Yeah, well, I'm a big talker. I've got a few packages with Cyrillic writing on them hanging around here since...  But along with them, I also have a board from vigilante397 for a submini Alembic F2B clone (in my regular pedal-building queue, meaning there's a fair chance I'll get to it in a matter of months rather than years), so I'm watching your work here with interest, 'cause I've heard it said that a CF is the mysterious extra tube stage that was supposedly added to Gilmour's F2B. Then all I have to figure out is whether the power supply will handle the extra tube (Nathan thinks it will), and what to do with the other half of the tube. First world problems.

[tubegeek]

I'm trying to get together an order for the SMPS parts so they will be there when I get back too. I promised my wife a Nixie clock for her office and the next one of those I build will use a SMPS. Might as well build one for this project too!

I completed a nice SMPS board for this sucker today - haven't installed it yet but it tests quite well, a very cool addition and no more "whatever the transformers sag to" voltage selection. I may have to load down the 12V input a bit so it doesn't burn out the heaters - my 12V supply brick is running about 15 V unloaded, I'll have to put a resistor across it and see what it will actual put out with my heaters connected.

[vigilante397]

Then all I have to figure out is whether the power supply will handle the extra tube (Nathan thinks it will)

Totally will

Source: I have that same SMPS running three tubes simultaneously in a different design.

[tubegeek]

Tried to test with the SMPS last night - I must have a short somewhere in the audio circuit board - the heaters lit OK when only the 12V supply was connected but when I connected the B+ the IRF740 in the SMPS board got super hot and no B+ was measurable. I don't think I killed the power supply board but I'll have some troubleshooting to do, knifing the gaps etc. Serves me right for building a high voltage tube circuit on perfboard.

[Ben N]

Wait, what? There's something wrong with HV on perf?

[tubegeek]

Wait, what? There's something wrong with HV on perf?

Well, not a lot of room for using the insulating properties of air on perf, is all I meant.

[Ben N]

Yeah, but OTOH you get a nice wide track for all your little electrons to meander down, instead of having to squeeze by on an ultrathin copper trace like a salad with sprouts down a supermodels gullet.

[tubegeek]

2 steps forward, one step back...

I decided I needed a better power supply so I built the NE555-based SMPS design that I've seen in several places, and that works pretty well, within limitations.

Clearly a higher-gain arrangement was needed beyond 4 sections of 6N16b so I redesigned with 1 6N17b and 1 6N16b as follows: the two halves of 6N17b as input and recovery-from-tone-stack gain stages, each DC coupled to a 6N16b cathode follower. (In the drawing further up the thread, the 1st and 3rd triode sections have been substituted with 6N17b.)  Still trying to draw grid current into the cathode followers as per the original plan for smooth compression and distortion.

My first try at new parameters was a bust: the result was too much current drawn overall for the SMPS to supply (approx. 15 mA) - the voltage sagged, because the cathode followers drew too much current, it ended up overheating the SMPS and shutting it down. I calculated some new operating points based on lower current draw and a B+ voltage of 200 which seems to work if the overall current draw is much lower, about 7.5 mA.

The two 6N17b sections use -1.2V bias, 60K plate resistor, 1K6 cathode resistor self-bias, .75mA per triode. The cathode followers are designed to use -0.6 V bias and a cathode resistor of 50K and about 3.5 mA per triode. (I see I must've done something wrong here - that doesn't work to get the direct coupled operating point, with the cathode up around 150V which is what's needed.... back to the drawing board.)

With the values above the 6N17b sections are working fine, but the cathode followers are not biased properly - they seem to be at about VBias -22V (!!) I find this hard to credit but now that I've noticed that 3.5 mA across 50K is 175V not 150V that explains why they are so far off.

I connected the two units directly, omitting the tone stack (for extra gain and a quick setup) and tried it out. It sounded great! - very creamy, compressed and overdriven. Then I inadvertently left the SMPS running with no load on it for a while and it smoked. Back to work but VERY promising even with the weird operating points...got to review Merlin's book again!

[Ben N]

Excellent report. Do you think the first CF is contributing anything to that nice creamy tone? Functionally, one would think that without the stack, there is no need for a CF, but if CFs really add compression, then maybe it does make sense. How was noise?

[tubegeek]

Noise was not an issue so fas I could judge with an unboxed amp using only partially shielded signal cable. No HF whine from the SMPS or anything like that.

If I am getting grid current compression (? not with -22 Vbias, I'm not?) in the cathode followers then sure they are adding to the tone but if you're asking, is guitar signal through one triode stage enough to overdrive the CF, dunno, haven't scoped this thing yet. Doubt it....

My next move, I think, is going to be replacing the cathode R's on the followers with pots and seeing if I can dial in the bias better. It's just not doing what the plate curves say it's supposed to be doing.

One possibility, I guess, is that I'm getting an RF oscillation that's causing weird behavior - i have the buildout and stopper resistors as per Merlin's recommendation, though. It just now occurs to me that I have the 220K grid "stopper" R's at the wrong end of the jumpers between the 1 and 3 plates and the 2 and 4 grids - should be closest to the grids and they aren't.

I wish I could build this thing on a breadboard but I don't trust the breadboard connections. So a LOT of soldering/unsoldering going on when I want to make changes.

[tubegeek]

Status: with a working SMPS, this setup (at least with the tone stack bypassed) has PLENTY of gain - too much really. I'm having trouble measuring DC conditions properly, what I'm getting isn't making sense, maybe there is some oscillation going on when it's just on my workbench. Next steps are to box it up and then scope it.

Each pair of triode sections on its own sounds pretty good too, I may have to make the topology switchable, at least maybe bypass the 3rd triode section (run the tone stack into the final cathode follower with no gain stage ahead of it.)