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

Started by tubegeek, March 21, 2014, 09:29:50 PM

Previous topic - Next topic

tubegeek

Designing a Two-stage Tube Circuit With Compressive Cathode Follower.

Jeremy "Tubegeek" Epstein March 2014

This design process is directly inspired by Merlin Blencowe's analysis of the Fender Bassman-style two-stage combination in his terrific book "Designing Tube Preamps for Guitar and Bass" 2 ed. I am indebted to Merlin for his clear and thorough explanation of the design process as well as his insightful demonstration of the distortion which results from the typical Bassman two-stage circuit. All mistakes are mine; Merlin has cheerfully agreed to permit me to refer to his explanation and method.  Thanks Merlin!

Some design goals and parameters:

A) To design a two-stage tube circuit including a gain stage, direct coupled into a very compressive, soft clipping cathode follower, as described in Merlin's book. Essentially to replicate the Bassman-style circuit's tonal qualities, using different tubes suitable for a very small build.

B) To use subminiature dual triodes I have on hand: 6N16b.

C) To use a small 12V:120V power transformer on hand, fed by a 120V:12V wall wart. In other words, a simple linear power supply design. The 12V will feed the heaters and then the stepped-up 120V will be rectified, for a B+ voltage of approximately 160V DC after filtering.

Note that this transformer could also be configured for 240VAC output, I expect to use the lower B+ though.

Also note that the final unit will utilize four triode sections: I have 6V-heater tubes on hand and a 12V heater supply. My expectation is that I will use one two-stage circuit to drive a tone stack, and a second two-stage circuit to make up gain and drive the output. This is still TBD.

Design steps:

1. On the standard anode curves for the 6N16b, plot grid-anode curves (shown in red, these curves are nearly vertical for a cathode follower.)

2. Choose a B+ value, and add a reversed scale for the cathode voltage (red scale.) B+ chosen to be 160V (rectified 120V minus a few volts for filtering.)


Iteration I.

3. Plot a load line in the normal fashion (blue line, Rk = 10K.)

4. Mark 0 V end of load line on both voltage scales (left and right green crosses.)

5. Locate approximate center bias point on load line (center green cross at approximately Vgk = -2.5V, Vak = 109V, Vk= 51V.)

6. Point A: Vk = 95. Point C: Vk=0. Output V = 95 V pk-pk.

7. Point A: Vga = -75. Point C: Vga = -160. Input V = 85 V pk-pk.

8. Gain = 85/95 = .895

9. Operating point at B: Vgk = -2.5, Vk = 51. Vg = 48.5  Ia = 5 mA

Direct Coupling to previous stage:

Va must be Vg for the cathode follower = 48.5 V

10. Using Ra = 10K and B+ = 160V, plot load line.

11. 2.5 / 5 ma = 500 ohms Rk, 5 mA * 10K = 50V. (Not shown, no good.)

Evaluation:

One design goal is to try the grid-current mode for the cathode follower. With 2.5 VBias, we are far away from the grid current region.

Iteration II.

Using a different bias point, at B"' we will have about .8V Vbias which should get us some grid conduction and the following other results from steps 3 and 9-11. (We will skip steps 4-8 from here on out because we aren't really interested in linearity here and because a huge output swing also isn't needed.)

Vga = 78V
Vgk = -.8V
Vk = 79V

For the gain stage, using Va = 160-79 = 81V and Ra = 17K
1.5V/4.7 mA = 320 ohm Rk



Iteration III.

Iterating again with small Vbias (hotter bias on CF.)

20K load line (blue line)
A' at 0Vgk. B' at 1Vgk, Vga about -73. Vga = 92. Vgk = -1. Vg = 91.

Direct Coupling to previous stage:

Va must be Vg for the cathode follower = 91.

Using Ra = 20K and B+ = 160V, plot load line (same blue line.)

(At E) 2.2 / 3.4 ma = 650 ohms Rk, 3.4 mA * 20K = 68V.

This looks like one good candidate.


We will iterate again, using Merlin's suggestion of a lower value for Rk on the CF. We'll try a load line of 8K this time.

Iteration IV.

8K load line for CF
7 mA Vgk -2 (at B")
Vga 104
Vg 56
Vk 58

Va for gain stage = 56
load line = 160 / 2mA = 80K
VBias for Va = 56V = -1.5V
1.5/1.3 = 1.15K Rk

Iteration V.

8K load line for CF
(At B*)
Vgk for CF = -1
Vga = -90
Vg = 70
Vk = 71

For gain stage, Va = 70

80K load line
Ra = 80K
VBias = -2 V
-2V/1.1mA = 1.8K = Rk



Iteration VI.

The Bassman cathode follower has equal Rk and Ra on the two stages. It looks like a load line of somewhere around 15K might get that done - let's see how that works out. 16K makes the line easy to draw...

Rk = 16K
For Vgk of -.6V at X, Vga = -65, Vg = 95

For the gain stage:
Va = 95 VBias = 2.2V Ia = 4 mA, Rk = 550 ohms.



Using these values, here is most of the circuit. Merlin's suggested grid stoppers and build out resistors are shown here. Tone stack TBD.



Next step is to breadboard the circuit above. The 550 ohm Rk's will need to be tweaked to dial in the bias most likely, however, it's been my experience that direct-coupled circuits are nowhere near as tricky to set up as you'd think. They'll always settle in somewhere!

Comments welcome. Thanks again to Merlin Blencowe = merlinb !






"The first four times, we figured it was an isolated incident." - Angry Pete

"(Chassis is not a magic garbage dump.)" - PRR

jazbo8

Nicely done! What's the expected drive voltage into the gain stage?

tubegeek

Guitar levels I guess - consensus on that is less than 10V, so I suppose the input could be overdriven by high output humbuckers now that you mention it. Or do you mean the 2nd gain stage? That's dependent on the loss in the tone stack which I haven't really dealt with yet. I'd like to do Stephie Bench's inductor tone stack actually, that seems like an interesting one to try.
"The first four times, we figured it was an isolated incident." - Angry Pete

"(Chassis is not a magic garbage dump.)" - PRR

jazbo8

I meant to ask what is the signal going into the first tube on the schematic, because the "2-tube CF" stage is usually used well down the chain from the input jack, so its input signal is far from the guitar level, which is commonly only 100mV, so when you say 10V, I assume it is already been through a few gain stages, i.e., the CF before the TS is reaaa...lly over-driven. ;)

tubegeek

Well, the plan is for the schematic at the bottom of the first post to be in largish pedal format, and to accept guitar or another pedal as input.

I seem to remember seeing that high output HB's could give a volt or more out? Anyway, 2+ volts of bias is enough input headroom for every guitar amp I've ever looked at so that should be OK.

This design would certainly be different if it had to worry about REAL CF signal swing but I'm not really trying to do that - there's no requirement to scale the signal up to drive a power amp, certainly not with clean swing.

In Merlin's book, he explains that running the CF section in the grid-current region (his example shows .6 VBias on a 12AX7 a la Bassman and Marshall) leads to a very rounded-off distortion characteristic and a lot of signal compression. Loading the gain stage with the grid current flowing into the large cathode resistor of the CF is a medium-heavy load so you get a rounded-off clip instead of a flat-topped clip from this setup.

In the actual Bassman you don't really use it because the output stage already clips before this happens, but in a master volume setup like a Marshall (using the same two stage setup) it's an important source of sweet grunge, according to Mr. Blencowe.

His description of this distortion source is what's leading me to try and isolate that behavior and try it in a pedal.

So - just to be clear - those inputs and outputs on that draft schematic at the end of the first post are meant to represent in & out jacks and the designed circuit stands alone (just doubled up coming in and out of a tone stack.)

If I had an appropriate 6V transformer on hand I'd try just the two stage circuit with nothing else. I will set it up that way first for testing and I hope to show oscilloscope shots as soon as I get some time in my basement lair.

But since I'm going to be using a 12V transformer, I'll build up two complete stages to use two 6N16b's (this is a twin triode, so I have four triodes to work with.) And I figure I'll put a tone stack in between just because it'd be a good use of the first CF.

I guess I'd need an output trim on an actual build, I'll use Merlin's suggestions there too for driving an output with a CF, from the section on effects loops in his book.

Book Plug Time: "Designing Tube Preamps for Guitar and Bass" 2 ed. is an excellent book with a superb organizational style and outstanding explanations of tube audio circuits as well as hybrid arrangements where appropriate. If you are familiar with Kevin O'Connor's books, this is at that level and Merlin does a better job - to me - of explaining his design process thoroughly. I'll have to go back to "The Ultimate Tone" to see how O'C deals with this material but I did look through Valley & Wallman, and Reich, and even the Bible RDH4, and Merlin's explanation was the most useful by miles. Especially since he goes into the forbidden topic of this special distortion/compression resulting from grid current flow, which of course isn't even acceptable dinner-table conversation between Langford-Smith and his pals.
"The first four times, we figured it was an isolated incident." - Angry Pete

"(Chassis is not a magic garbage dump.)" - PRR

jazbo8

I see, the best way to tell is breadboard it to hear what it sounds like but I think the gain might be a bit low, here is my reasoning: without cathode bypass capacitor, stage one has a gain of ~10x, so the first CF stage's output is only 1V with an input of 100mV (the humbuckers could put out more but most tube amps are designed with input sensitivity of 100mV). Of course, this is moot if you have another pedal(s) before this one. Let us know how it turns out...

PRR

> gain might be a bit low

+1.

The 5F6a has a gain-stage and gain-control *before* this stage.

The stage you show can put out 10V-30V. Gain will be under 10. So input overload is 1V-3V. You will have to work hard to get any clipping. With additional gain=10 in front the input overload becomes 0.1V-0.3V, still hard work but feasible.

> I'd need an output trim

Your second stage also can output 10V-30V, you only want ~~1V into a guitar amp (or further pedals).

You propose gain of 10, loss of 10 (tonestack), gain of 10.

Your second stage level is no higher than your input stage level. You are doubling your hiss power.

Your input overload is awful high for guitar.

Your output level is awful high for guitar-amp input.

Re-order the stages. Gain stage. Gain pot. Gain stage. Now you have gain of 100, can boost guitar to 10V-30V and clip it. Now tone-stack, loss of 10, output level 1V-3V. Only minor trim needed to keep guitar amp input happy.

Working around 5mA means fat filter caps in B+. 1mA-1.5mA was good enough for Leo.

> acceptable dinner-table conversation

Try Valley & Wallman, Pulse-Shaping.
  • SUPPORTER

Brymus

FWIW I used Merlins "bootstrapped" cathode follower setup in my last build it really gave me nice clipping and compression over a standered CF.
I'm no EE or even a tech,just a monkey with a soldering iron that can read,and follow instructions. ;D
My now defunct band http://www.facebook.com/TheZedLeppelinExperience

tubegeek

I got a chance to start building this project today.

1st pic shows the terminal strips laid out with the 6H18b Russian subminiature tubes threaded into place. This is from the bottom. The ground busses are holding the terminal strips in position.


2nd pic shows the same thing from the top.


3rd pic shows the progress I made before turning in for the night.
"The first four times, we figured it was an isolated incident." - Angry Pete

"(Chassis is not a magic garbage dump.)" - PRR

jtn191

Cool ideas, I'm always interested in seeing the finished project and hearing a clip ;)

tca

Are you running  this at 160V?!?!?



You can get electrocuted easily!
"The future is here, it's just not evenly distributed yet." -- William Gibson

tubegeek

Quote from: tca on May 15, 2014, 02:21:09 PM
Are you running this at 160V?!?!?

That's the idea!

Quote
You can get electrocuted easily!

I'd better be careful!

(It's not my first rodeo. But thanks for the concern!)
"The first four times, we figured it was an isolated incident." - Angry Pete

"(Chassis is not a magic garbage dump.)" - PRR

tubegeek

Question for Merlin:

Re: Section 6.16, Bootstrapping for More Gain, Page 130 & 131.

Doesn't the signal from the cathode follower into the junction between R1 and R2 cause the junction to have MORE signal than the anode? There should be 1/2 the signal there already, the cathode follower is adding approximately 1x the anode signal with the same polarity, so would there be an additive effect that causes the top of the resistor to have MORE than the anode?

I'm not sure how the mixing behavior would work - I guess that the low resistance output of the cathode follower would approximate a virtual earth..... ???

Or am just I looking at the current described in the last paragraph of P. 130 from a different point of view, and is this the same current that gets dumped into the low output resistance of the CF? If I understand virtual earth correctly, there still would be a mixing of the signals there though, which would give the output of the CF some additional gain as a result.

I think?

Confused!

Thanks,

-j
"The first four times, we figured it was an isolated incident." - Angry Pete

"(Chassis is not a magic garbage dump.)" - PRR

tubegeek

I know you've all been just holding your breath waiting to hear what's been going on with this project.

Well, until today, nothing for quite a while.

I made a little progress today: I replaced the broken 6N16b, which wasn't nearly as difficult as I had foreseen (I had thought that re-lacing the eight wires inbetween all the other stuff that was connecting to the terminal strips was going to be a nightmare. Naaah. Not so bad!)

The second thing I did was a bit of a one-step-forward, one-step-back kind of thing.

I had been getting too-low voltage off the high-voltage winding of the second back-to-back transformer in the power supply. If you look at the older posts, you'll see I have been blaming under-rated, junk-box parts. I'm getting about 145VDC on the first filter cap instead of the 160 that the math says I am supposed to.

I don't think that's the problem. I tried somewhat less junky, properly-rated transformers and the problem is still there, despite the fact that the power supply is not yet loaded. I think I figured out what's going on, and it's something else - I think it's related to the back-to-back trick.

Small power transformers have so-so regulation. They deliver an output voltage that's a little high under no load and they sag down to their rated voltage under load. This means that their turns ratio is a little closer than it should be for the rated voltage: instead of a 120:12 transformer having a 10:1 turns ratio, it'll be more like 10: 1.1 or 1.2 and the "12V" winding will be more like 14 to 16 no-load.

Well, if you turn this transformer around, you don't get as much out the primary as you wish you did if you drive it with the rated voltage on the secondary. The step-up ratio is not as high as the ideal 10:1 in the example, it's a little short. To get the rated primary voltage OUT, you'd have to drive it with the NO-LOAD voltage onto the secondary. Damn.

What this means is I need to either:

a) run the unit off a somewhat higher input voltage, say 18VAC, and put some resistance in series with the tube heaters to bring the heater voltages down to 12V, while driving the second transformer with this higher voltage (which may or may not actually deliver a bigger output voltage in the real world under these conditions - experimentation needed.)

b) go to a SMPS running off of the 12V power supply, which I have been avoiding because I've never built one and I thought that adding a new wrinkle would give me additional development issues and I didn't need the extra complexity. Joke's on me - sticking with the tried-and-true linear supply gave me a head-scratcher that I just figured out after about a year on the back burner. Oh, well. I know that most people reading this will say that's what I should have done in the first place - in my defense, I don't have the parts for a SMPS on hand. But I may have to step up to 2016....

In any case, I *think* if I go with option a) I may be able to go back to the smaller-sized, more pedal-sized transformer - I don't think it was the transformer's fault. We'll see once I get the circuit further along and I can load down the power supply completely. But I think I was wrong about placing the blame for the low output voltage on an under-rated transformer. I think it was the ratio.
"The first four times, we figured it was an isolated incident." - Angry Pete

"(Chassis is not a magic garbage dump.)" - PRR

vigilante397

Quote from: tubegeek on May 15, 2014, 06:25:53 PM
I'd better be careful!

(It's not my first rodeo. But thanks for the concern!)

With a name like tubegeek I would be shocked (pun intended) if you didn't know what you were doing here ;D

Never saw this thread first time around, so it's cool to see it now. Looking forward to see how it turns out!
  • SUPPORTER
"Some people love music the way other people love chocolate. Some of us love music the way other people love oxygen."

www.sushiboxfx.com

PRR

> you've all been just holding your breath waiting

That's why I look like a blue Smurf!

> I'm getting about 145VDC on the first filter cap instead of the 160 that the math says I am supposed to.

Yeah, so?

IMHO, perhaps better for this application.

I'd sure never fret (much) about 20% variation in supply voltage on tubes.

If it works on 145V, but gives 1.1% IM distortion, where your contract calls for 1.0% IMD, try finding 160V.

If it doesn't work on 145V, it won't work on 160V.

My long-forgotten suspicion is that you want distortion and are unlikely to get a bunch. This suggests a LOWER supply voltage, not higher, to get some "strain" happening.
  • SUPPORTER

tubegeek

Quote from: PRR on February 18, 2016, 08:54:41 PM
My long-forgotten suspicion is that you want distortion and are unlikely to get a bunch.

If I could get just exactly what I want, I'd get gorgeous Bassman-style compression. If Merlin is to be believed (and I believe he is) the Bassman is a happy accident of grid current drawn from a voltage amp into a cathode follower that sounds the way it sounds.

This whole project is an experiment to see if I can do that, twice, inside a little box.

It's a specific kind of distortion I guess. Dynamic distortion instead of harmonic.
"The first four times, we figured it was an isolated incident." - Angry Pete

"(Chassis is not a magic garbage dump.)" - PRR

tubegeek

Quote from: vigilante397 on February 18, 2016, 08:20:21 PM
With a name like tubegeek I would be shocked (pun intended) if you didn't know what you were doing here ;D

Still I appreciate the concern for my safety, thank you. I always try my best to give the same high voltage warnings. Only takes one electrocution to bring everything to a halt. And I do get pinched every now and again - I am ashamed to admit how untidy I let my workspace get.

Quote from: vigilante397 on February 18, 2016, 08:20:21 PM
Never saw this thread first time around, so it's cool to see it now. Looking forward to see how it turns out!

You and me both. I made up a little testing power supply that puts a (nominal) 18V into the (nominal) 12.6V winding, we'll see what that all really means with the heaters and some compensating resistors drawing current from the low voltage side, and the rectifier feeding a simulated circuit load from the high voltage side, should be about 15 to 20 mA for the 4 tube sections. I anticipate I'll get close enough to spec to move forward with this stack of kludges in the next days. Might get a little warm, but so what?
My biggest issue now is I can't find the layout drawing I made, so I have to re-trace the schematic around the wiring I did (see the photos above) to see what connects where on the controls. I had put two jacks on already, so those are going to be good hints, I just have to figure out which is IN and which is OUT to get a good start!
"The first four times, we figured it was an isolated incident." - Angry Pete

"(Chassis is not a magic garbage dump.)" - PRR

tubegeek

Just parking this info here for myself for testing purposes:

To simulate the circuit current draw on the rectifier: 160V drawing 16 mA would require a 10K resistor for testing, 2.5 Watts dissipation on this resistor.

6N16b heaters draw 400mA, both in series will drop 12V, dropping another 4V at 400 mA for a total of 18V (which is what I expect from the power transformer under load) will require a 10 ohm resistor in series with the heaters, 1.6 Watts dissipation.
"The first four times, we figured it was an isolated incident." - Angry Pete

"(Chassis is not a magic garbage dump.)" - PRR

tubegeek

The project is built up to the extent that I got a chance to try it yesterday (power supply is still not right, waiting on parts to use SMPS instead.) Apologies to anyone who has been waiting three years for an update!

The comments about inadequate gain are spot on. Weighing two options: a JFET booster ahead of V1 (quick and dirty) or adding one more tube to the input, ahead of the circuit as drawn. Both just to get the levels where they belong to see whether the clipping is going to happen the way I want it to. Probably going to go with the JFET just for a quick trial.

More later. When later? Good question!
"The first four times, we figured it was an isolated incident." - Angry Pete

"(Chassis is not a magic garbage dump.)" - PRR