Please critique my tube amp schematics

[patricks]

Hi all,
I'm posting up three schematics that I've mocked up for versions of a tube amp, I'd love some feedback on them. The background behind is that I read an article in Premier Guitar on how tube amps work. The example they used was the 1960 Vox AC4. I thought "hey, that doesn't look too hard" so I read some build threads and then posted here about possible mods.
Some suggestions were made about swapping the EF86 in the original for a pentode such as the 6GH8, which is a pentode and triode in the same tube (it's also less microphonic, higher gain and more plentiful). Not wanting to waste a good triode, I cooked up some ideas and here they are. I used the Valve Wizard website
to calculate values for anode resistors, bias resistor, bypass caps etc., but this is my first stab at circuit design/modification and tube calculations (dropbox link to word doc with calculations below).

If it's not too much trouble, I'd love to know whether any of the ideas will/won't work, whether some component values are wrong, etc. This is my first time, though, so be gentle!
Images are thumbnails, full-size images will load in a new tab

1. Pentode and triode used in preamp, with the option to select one, the other or both in series


2. Pentode and triode in parallel, with an option to blend some triode signal into the pentode signal


3. Pentode preamp, triode used as a cathodyne inverter to power a push-pull power section


Calculations: https://www.dropbox.com/s/9m59v6800s8g1dc/AC4%20mods%20-%20load%20lines.docx

Cheers
Patrick

[Digital Larry]

I'm no expert in tube amp design, but just on general principals, in schematic #1 the right hand of C2 wants a large value bleeder resistor to ground to avoid a massive pop when you switch to position 1.

Also it looks like the triode is always driving the output tube, but the triode's input is wide open in position 1.  I'd try to either ground its input or disconnect its output in position 1 but you'd need another switch section (maybe).

[Digital Larry]

On the second schematic, I'm curious about the mixing strategy.  When the triode is turned all the way down it is going to dump its signal to ground through the cap.  And that might be OK but I'd feel a little better if the two preamp stages had some series resistance between the caps and the pot to do some passive voltage mixing and limit the current when the triode is turned down.

Again take my comments with a grain of salt - I've not done any tube amp designs.  Perhaps some fab tube amp has used your approach and I'm just unaware of it.

Another thought that just popped into my head is the idea of doing a "blend" pot where at the left it's all pentode, at the right it's all triode, and halfway it's 50/50.  For that you could just connect either end of the blend pot to the right side of each stage's output coupling cap.  Use a linear pot or possibly a stereo log/anti-log pot would give better results as in a pickup blend pot.

On your third schematic, the tone/volume control before the phase splitter wants to be referenced to ground, not the lower phase splitter output.

I would also review the power supply filtering approach and compare it to common push pull amp designs.  Normally the B+ feeding the output transformer center tap is closest to the power transformer and voltage taps for the plate circuits are taken from subsequent filter sections moving towards the input of the amp.  As it is you have a filter but everything is hanging off the end rather than being distributed for isolation between gain stages.

[tubegeek]

OK: really quick look so far, one thing that will cause a small problem. Easily fixed (especially at this point!)

Generally if you operate more than two stages off of the same B+ (tube amp shorthand for the main high voltage supply) you want to separate their high voltage feeds with an RC filter section. This has the added benefit of filtering the power supply more, for the high gain (preamp) stages that will amplify any ripple seen at this spot the most. Result: quieter (less 120 Hz humbuzz) amp, and more likely to be stable this way. The preamp stages don't draw very much current, so there is less of an issue with power wasted as heat in the filter resistor here.

You have this correct in two out of the three schematics, but the push-pull one has all the stages pulling B+ from the same spot.

(A good exercise: calculate voltage drop and power dissipation for each resistor in your amp. If you don't know the amount of current each tube is expected to draw, use the value given in the data sheet operating point as an approximation. Use a resistor rated for about double the calculated dissipation as a rule: there are always a few resistors in a tube amp that will need to be rated much higher than the typical 1/2 watt for a small resistor - almost always the plate loads and the power supply filter Rs, also sometimes cathode resistors as well. Grid resistors usually are fine with small ratings - they are not subject to much current draw if any.)

This is quite likely to cause a low frequency oscillation - called "motorboating" because of how it sounds - and so it's usually avoided. Fix is easy - just move the center tap of the output transformer to a point closer to the rectifier as you have shown the EL84 connection in the other schematics.

More later. Good circuit thoughts in general - the combination preamp is something I am considering with a little amp I have built recently, because I want to hear what a pentode input stage sounds like, I've never built one myself.

You'll want to follow Digital Larry's advice about the grounding resistor behind the switch (called a pull-down resistor) as well. Another possibility would be to have a pot to blend the preamps in any proportion you like, I can explain that when I get more time (maybe tonight if it's slow at work.)

In general I think that there is a whole world of tube choices besides the ubiquitous 12A?7 types copmmonly used that have much better potential for tube guitar amps. There is a tendency in guitar amps to copy classic circuits and the hi fi tube world has been more adventuresome in this regard. My preference is to use tubes with lower plate resistance/higher gm because they can drive the stages following them more easily while still keeping plenty of stage gain. (That is an oversimplification, but it's still true.)

There are dozens and dozens of cheap and excellent Russian tube types that are promising for guitar amps and the only thing holding back the DIY'er from using them is - sometimes - a lower place on the learning curve of design. Which is why getting feedback from a community like this as you work is very helpful. I started my education into electronics in a similar fashion in about 1989 - on an Earthlink BBS - and that's how I've learned almost everything I know about electronics.

Guess who were two of the most helpful people I "met" back then, who were generous with their time and advice?

Hints: "two letter first name" for one of them, and "last name = carpenter's tool" for the other. They're still here!

Using "oddball" tubes can be a tricky business when building a guitar amp because their data sheets are written to show operating points (meaning choices of B+, plate load R's, and bias voltages) which offer minimum distortion (or sometimes max. power with "acceptable" distortion) as a general rule. So sometimes you need to choose "dirtier" operating points for guitar, which involves experimentation.

Which is fun!

[JRB]

As the others have mentioned you should include pull down resistors at the caps that are connected to the switch to prevent pops when switching.

My knowledge about single end poweramps is very limited but in push pull amps a screen resistor is usually used on the power tube screens, you might want to see if you need them I think the valve wizard has a article about single end poweramps.

You also want to add some extra fuses to the design to protect your transformers again the valve wizard website has some articles about it.

If you are going to use relays for the switching you want to use DC relays and get enough filtering in the DC supply. In my own amp I didn't filter the DC supply enough and it caused massive hum in the amp.

And finally a first time start up guide,
http://www.paulrubyamps.com/info.html#FirstPowerUp

Building a amp isn't a big difference from building a pedal except that the power in it can actually hurt you pretty bad or kill you. (sorry if you already knew this but somebody had to write down the warning).

[patricks]

This is great, thank heaps guys!

So, I've added a bleeder resistor to each schematic. In the first, I've added another switch pole to ground the input; in the push-pull schematic, I've added grid stopper resistors to each EL84 and separated the B+ for the OT from the B+ for the EL84s.

Another thought that just popped into my head is the idea of doing a "blend" pot where at the left it's all pentode, at the right it's all triode, and halfway it's 50/50.

That's a sweet idea! Taking it one step further, how about this:

It uses a 3PDT switch to link the two parts so you can choose series or parallel. With the switch in "series" mode, the blend pot controls the amount of pentode signal feeding the triode. With the switch in "parallel" mode, the blend pot functions as you suggested (all pentode -> 50/50 -> all triode as the dial's turned clockwise). It solves the problem of leaving the input to the triode circuit open, too.
Would the "blend" pot provide enough resistance after the capacitors when the circuit's in parallel?

JRB, thanks for the link, I've bookmarked it for when I get to the end of the build. Thanks for the caution, too, it's much appreciated and definitely worthwhile, Out of the four stompboxes I built recently, two worked straight away. The other two didn't work immediately because they both had issues with components shorting to ground - can't afford to do that with a filter cap or a B+ lead...  

[Digital Larry]

I'm suspicious there's a problem with the switch in the series position?  That puts the pot between the input and output of the triode, which I don't think you want to do.

[patricks]

Thanks for catching that, I've fixed it - a 4PDT solves the problem

[Digital Larry]

Don't mean to be nit picky but you should probably ground the bottom end of the pot when you go to series mode.

[patricks]

No, nit-picking is exactly what I want - I want to make sure I catch mistakes as early as possible

Getting back to the pot between the input and output of the triode, I might need some more help on that. What I'd like to happen when the pentode and triode are in series is that when the pot's turned all the way anti-clockwise, only the signal from the pentode goes to the power amp. When the pot's fully clockwise, all the signal from the pentode goes through the triode. The only way I can think of making that happen is with a pot between the triode's input and output (pic below), but if that's a no-no then I'll have to scrap the idea.


If the idea's valid, then I'd need to make sure that the value of the pot was greater than the triode's input impedance, wouldn't I (so that the path through the triode was the one of least resistance)?

[Digital Larry]

My initial comment is that you can already get 100% pentode in your parallel configuration.  Ah but you are the customer, and you want what you want. 

The triode's input impedance is very high. 

I don't know that I'm going to be able to assist much more on this one because I don't want to steer you in the direction of thinking's something possible when I don't have the experience to be confident in my suggestions.  Interested to see what you come up with though.  I suspect that you may discover why pentodes are not very popular in guitar preamps.  Or you just may hit on the next big thing!

[patricks]

My initial comment is that you can already get 100% pentode in your parallel configuration.

True. I thought about that at first, but then I thought "maybe if I make the value of the pot large enough, it'll work the way I think it will". Shows how much of a noob I am at circuit design

I'll keep exploring different switching options - I could even go bananas and wire up a rotary switch to give me pentode/triode/pentode into triode/triode into pentode/pentode and triode in parallel. The wiring on that switch might be a bit hairy! .
Probably best if I just keep my first build simple, though...

I don't know that I'm going to be able to assist much more on this one because I don't want to steer you in the direction of thinking's something possible when I don't have the experience to be confident in my suggestionss.  Interested to see what you come up with though.  I suspect that you may discover why pentodes are not very popular in guitar preamps.  Or you just may hit on the next big thing!

I really appreciate you sticking with me this far, so thanks very much! Since my first post, I've found a couple of threads on other forums that talk about running pentodes and triodes in the same preamp in different configurations, so I think it's worth exploring, but I'll probably have to do a lot of tinkering with the values of plate, cathode and screen resistors.

Here's hoping it's the next big thing and it catapults me into rock god stardom! 

[amptramp]

A typical pentode like a 6AU6 may run 500K to 1.5 meg output impedance.  The typcal triode like a 6AV6 / 12AX7 may run 62.5 to 80 K.  With reasonable values of the pot between them, the triode is going to dominate the response regardless of the pot setting.  There is another way to get a transition between triode and pentode characteristics: replace the plate resistor for the pentode with a pot and feed it into the triode grid and use the triode cathode output as the screen supply.  Connect the triode plate to B+ so you have a voltage follower driving the screen between B+ and the output of the pentode.  This gives you a triode-connected pentode with the slider at the plate and a pure pentode with the slider at B+.  It also lets you go smoothly through an ultra linear connection at some value, probably around 40% of the way from B+ where the output of the stage is at its most linear.

[tubegeek]

Probably best if I just keep my first build simple, though...

I agree with your quoted statement because you will have a bitch of a time troubleshooting your multiswitched arrangement if it doesn't work out of the gate.

In support of that excellent idea:

Just choose a desired initial arrangement, while preserving your options by 1) making sure you leave some room on the front panel for future features/switches etc., and 2) when you wire the amp up, be sure to have easy access to the spots where you might break the circuit (via a terminal strip or suchlike) so that making revisions is painless and non-confusing.

amptramp's suggestion is a very cool way to achieve your aims - and the point about the pentode output impedance is one I had overlooked.
-j

[patricks]

you will have a bitch of a time troubleshooting your multiswitched arrangement if it doesn't work out of the gate.

As you can tell from the previous posts, I had a bitch of a time just trying to plan it out. Simple it is, then!
Thanks for the tips on leaving room for future mods, that'll be much easier to plan than doing all the switching first off.

amptramp, that sounds like a great idea. I've made a note of it
As far as the different output impedances go, would the triode signal still dominate if I set the circuit up as Digital Larry suggested in the third post - the outputs of the pentode and triode connected to lugs 1 and 3 of a pot, respectively, with the wiper (lug 2) going to the power amp section? It sounded like a cool idea to me, but if I've understood what you've said about the relative impedances, the pentode signal would be only be apparent over a small part of the pot's rotation, before getting swamped by the triode signal.

[amptramp]

Yes, you understand the output impedance problem.  Even if the pot is 1 meg, you don't have to have the slider very far from the pentode end before the results get swamped by the triode.  The triode output signal will be a bit lower than the pentode, so the pot may act a bit as a volume control and it will be difficult to compare two signals at different volume levels.

The concept of the pentode with the buffered connection to the screen variable from B+ to plate is not original with me - it was on the tubecad site many years ago amd I thought it was a cool idea.  If you go to the tubecad site, you will get an education from one of the best tube audio engineers on the planet.

[patricks]

Great, I've learned something now, thanks heaps!

I had a look on the TubeCAD site, I think I found the post you're referring to (image below). I've been looking for the TubeCAD software, too, but from the googling I've done it seems that the software is no longer available because it doesn't work on 64-bit systems. A pity, because it took me days to do the calculations for this circuit!

About the triode-connected pentode setup you described earlier, is it something like this (please excuse the cartoonish nature of the drawing, I whipped it up in about 2 minutes):


Or more similar to the version on the TubeCAD site (modified so that the pot replaces R1 and R2 as mentioned earlier on that page):


The ulta-linear preamp idea is certainly cool. I found plenty of references for ultralinear power amp sections, but none for preamps (in guitar amps, anyway). I'm guessing that the sound would be more "detailed" in pure pentode mode, "warmer" in triode-connected pentode mode and more "hi-fi" (i.e. more clean headroom) in ultralinear mode, but that's just a guess
After some more googling, I found the "spotted dog", that uses a 6GH8 in the preamp and I like the way it sounds, so at least I know I'm not tilting at windmills using this tube. The page doesn't say what he did with the triode part of the tube, so I'm guessing he either fed the signal from the pentode into the triode, or just left the triode unused.

[amptramp]

The ulta-linear line stage schematic lifted from tubecad is the right place to start.  Keep an eye on the heater-to-cathode voltage in the triode stage - some tubes have limitations and maybe a separate 6AU6 and 6C4 with separate heater supplies might be a good idea.  You do have room for a small filament transformer in a tube amp.  With the pot slider at the top, you have a simulated triode.  At the bottom, you are approximating a pentode.  Somewhere in between is ultra linear.

I learned a lot on that site.  Even our best gurus may want to peruse some of the designs and derivations.  The µ-amp/SRPP stage is well documented with derivations there.

[patricks]

The ulta-linear line stage schematic lifted from tubecad is the right place to start

Excellent, thanks. Maybe separate tubes would be the way to go, since I'm still not sure how to calculate expected voltages at different points around a circuit. I know the formulas, so I can calculate voltage, resistance, current or power dissipation if I have the numbers in front of me, but if I start with just a schematic I'm a bit lost.

As an aside, I did some more googling on pentode/triode/ultralinear options and found a few references to a "morph" control, which is apparently on page 95 of Merlin's tube preamp book. It's meant to be another way of choosing between pentode, triode and ultralinear operation, but just from a single pentode. This schematic has one after the EF86: http://music-electronics-forum.com/attachments/24041d1372353946-8031948687_debfcc9886_o.jpg, but I'm guessing there's an advantage to running it with a buffered connection to the screen as described on TubeCAD.

The TubeCAD site's going to provide plenty of good reading

[tubegeek]

Even our best gurus may want to peruse some of the designs and derivations.

I have no illusions that you were calling ME out with that one, amptramp but I do just want to chime in with an enthusiastic agreement. John Broskie (Mr. TubeCAD)  afforded me the privilege of meeting him on a trip to NYC some years back and it was a very enjoyable afternoon. Just about anyone interested in tubes - or hybrid circuits for that matter - should read everything on that site from A to Z and then some. John has an enormous talent for investigating standard circuits with fresh approaches and he has redesigned a huge number of basic blocks that most people take for granted. The only problem with his writing is that there are too many ideas flowing from him to know which ones to try!

I run an Aikido line stage as he designed it in my home system and it has served well for several years now. I had forgotten about the morphing pentode -> triode idea. It would surely be worth trying in this proposed guitar preamp.

Regarding the heater-to-cathode potential issue: we need to look at the relevant section of the 6GH8 data sheet.

http://tubedata.milbert.com/sheets/049/6/6GH8.pdf

The part we care about for this is at the bottom of Page 2. It says:

PEAK HEATER-CATHODE VOLTAGE:
(Heater negative with respect to cathode)
Triode Unit: 200 max. volts
Pentode Unit: 200 max. volts

(Heater positive with respect to cathode)
Triode Unit: 200* max. volts
Pentode Unit: 200* max. volts

*the DC component must not exceed 100 volts.

OK, so what this tells us is the absolute maximum difference we can establish (as a quiescent operating point, zero-signal conditions) between the triode section's cathode and the pentode section's cathode is 300 V DC. If this maximum is the choice, then the heater would have to sit 200V below the higher section so as not to exceed the first requirement. It would have to sit 100V above the lower section's cathode so as not to exceed the second requirement. So the heater would most likely be configured as 6.3VAC referenced to about 95VDC. This is usually done by connecting the AC heater's center tap to a voltage divider with a small current bleeding through it arranged so that it has the desired DC voltage. That voltage divider would likely be connected to the B+ to be able to provide that high a voltage after division, but it could also come from, say, a regulated screen supply for example.

In addition we would have to ensure that the peak swing on the lower cathode (with respect to the heater) could not be more than 100 V below its quiescent bias voltage.

None of these seem like deal-breakers for this design to me - bias is more likely to be in the ones or tens of volts than in the hundreds, so the lower cathode (the pentode's) will never swing 100V DOWN from where it starts out.

The G2 of the pentode - where the upper cathode (the triode's) will be referenced must be no more than 300V higher than that bias point and must not swing to any voltage more than 300V higher either. If the triode cuts off completely that means that the cathode will be at B+ so we shouldn't use any higher than 300V plus the bias voltage of the pentode. And the data sheet operating point shows typical conditions like -1V for bias and -8V for cutoff with a plate violtage of 125 V (so maybe a B+ of twice that as a guess.)

How much signal swing are we looking for from this stage? Maybe 25 Vpk or so? How much will that make the G2 swing? Depends on how this pentode + triode circuit works - I haven't re-read the article yet so I'm not too sure. In any case if we don't use a B+ larger than 300V (350 is the max allowed by the data sheet anyway) we can be pretty safe trying it out. That would mean sending the data sheet's 12 mA of current through a 14.6Kohm plate load resistor to drop 175V so as to end up with 125V at the plate of the pentode.

Seems quite doable with this tube without exceeding the heater-cathode potential limits. And the tube's ratings imply it's a pretty robust hunk of glass - both sections max out at 2.5 watts plate dissipation, so that's no wimpy tube. A 12AT7 has the same 2.5 W per section rating for max. dissipation, a 12AX7 has a 1 W per section rating, just for for comparison.

-j