12AU7 concertina/split load PI

[jeff]

I'm in the middle of(finally) building a new amp. I've decided to go with a split load PI just to try something different than every other amp I've built. I've come across a lot of sites suggesting to use a 12AU7 instead of a 12AX7, or a 12DX7edit 12DW7 with the 12AU7 half as the PI. I have a bunch of 12AU7s so I'll give it a try. 

Should the 56K plate and cathode resistors be changed to a different value to better suit a 12AU7 over a 12AX7?
Does anyone have a schematic or know of an amp that uses a 12AU7 splitload PI?

[sluckey]

Many of the Magnatones use a 12AU7 split load PI. Here's one...

http://home.comcast.net/~seluckey/amps/magnatone/Magnatone_M10A.pdf

EDIT... And some maggies use the low gain side of a 12DW7 as the split load PI. M15 for example.

[jjasilli]

I note that the Maggie 12au7 is driving 7189 power tubes with only 3.3V on cathode pins 7, a pretty low "bias voltage".  So the low voltage gain of the driver stage of the split load PI can probably still slam these particular power tubes pretty hard.  The result may be different with more signal hungry power tubes.  Any thoughts?

[echuta13]

The 12DW7 might be an interesting tube to try as well.

[sluckey]

I note that the Maggie 12au7 is driving 7189 power tubes with only 3.3V on cathode pins 7, a pretty low "bias voltage". 

Pin 3 is cathode. But take another look. You'll see that there is about -15.3V on the grids, for a total bias of -18.6v (grid to cathode). This particular amp uses a combination of fixed and cathode bias.

Very similar voltages to it's little brother, EL84.

[jeff]

I'm going with the 6V6. Think it's OK.

The amp I'm building's like a 5E3 deluxe with a different preamp. Can you just stick in a 12AU7 in a deluxe or can the 56Ks be optimized for using 12AU7s?

The 12DW7 might be an interesting tube to try as well.

(Maybe that was the tube suggested instead of 12DX7)

[tubeswell]

re: the 12DW7, that is JJ's ECC823. JJ also do an ECC832, which is the same, but with the triodes switched around (so all you need to change are the cathodyne load resistors for the 'AU' side when putting it in a BFPR)

[jeff]

That's what I need to know. When using the 12AU7(or AU side of that tube) as the PI, what do you change the resistors to?

[sluckey]

That's what I need to know. When using the 12AU7(or AU side of that tube) as the PI, what do you change the resistors to?

Did you look at the schematic I posted?

[jeff]

I note that the Maggie 12au7 is driving 7189 power tubes with only 3.3V on cathode pins 7, a pretty low "bias voltage". 

Pin 3 is cathode. But take another look. You'll see that there is about -15.3V on the grids, for a total bias of -18.6v (grid to cathode). This particular amp uses a combination of fixed and cathode bias.

Very similar voltages to it's little brother, EL84.

Can I use that inverter with 6V6s instead of the 7189s
Are the 100Ks OK when using 6V6's? For some reason I thought the resistors would be lower going from 12AX7 to 12AU7.

After more research, they suggest replacing the 56Ks with 22Ks does that sound right? What sould the cathode resistor be?

[jjasilli]

I note that the Maggie 12au7 is driving 7189 power tubes with only 3.3V on cathode pins 7, a pretty low "bias voltage". 

Pin 3 is cathode. But take another look. You'll see that there is about -15.3V on the grids, for a total bias of -18.6v (grid to cathode). This particular amp uses a combination of fixed and cathode bias.

Very similar voltages to it's little brother, EL84.

Oh -- I see now that I missed the -16V @ point F!

[tubeswell]

That's what I need to know. When using the 12AU7(or AU side of that tube) as the PI, what do you change the resistors to?

About 22k (2W) would be good, try a 1k-3k3 (2W) bias resistor. Also (to avoid blocking distortion) you could easily lower the coupling cap to the cathodyne stage to 500pF or 1000pF, because of the wider bandwidth

[Fresh_Start]

That's what I need to know. When using the 12AU7(or AU side of that tube) as the PI, what do you change the resistors to?

About 22k (2W) would be good, try a 1k-3k3 (2W) bias resistor. Also (to avoid blocking distortion) you could easily lower the coupling cap to the cathodyne stage to 500pF or 1000pF, because of the wider bandwidth

Not trying to be a bonehead, but sluckey showed you a schematic with 100K plate & cathode resistors and a 15K bias resistor of the 12AU7-type concertina splitter stage.  Why not start there?

Respectfully,

Chip

[tubeswell]

Effect of different DC load lines 10k Ra/Rk, 22k Ra/Rk,  and 100k Ra/Rk

20k load gives the 'highest' current drive

44k load would provide the current/voltage tradeoff.

200k load is running the 12AU7 on a very low current high voltage swing.

Depends what you want

[jeff]

Not trying to be a bonehead, but sluckey showed you a schematic with 100K plate & cathode resistors and a 15K bias resistor of the 12AU7-type concertina splitter stage.  Why not start there?

Respectfully,

Chip

I asked only because that schematic uses different power tubes. Wasn't sure if that setup would be good with 6V6s
Ok, my turn to be a bonehead(and hopefully learn something).

 Why do we want to increase the current? I always understood tube to be voltage controlled. The voltage on the grid either allows more or less current to flow through the tube which allows more or less current to flow through the load which changes the voltage at the plate due to the IR drop across the load. How does having the PI supply more current effect the power tubes?

For example if you had two different PI tubes set up to give the same output voltage but one could supply more current?

Also I'm gonna wind up with one unused triode section. Would there be any advantage to halving the load resistors and paralleling the two sections?

(Sorry if this is a stupid question. By asking a stupid question you run the risk of looking stupid, on the other hand by not asking you run the risk of staying stupid)

[RicharD]

If you look at the typical operating characteristics for a pair of 6V6's, you'll see it calls for around 38V peak audio grid to grid.  Counting on thumbs call that 40 volts and since this is a geetar amp you want your PI to be able to make double that.  Of course if your B+ to your 6V6's is higher, you're gonna wanna push things a little higher, but that's trivial.  Tubeswell provided you with 3 jolly good load lines and even went so far as to calculate your bias resistor should you choose you bootstrap.  Any of those will swing plenty of voltage.  It doesn't take a whole lot of current to drive a pair of 6V6's so you're good there too.  Enie meanie miney mo will work in this situation.  This 2x10k will give you the most linear performance but sticking a 12AX7 in there after the fact is gonna suck.  The 2x100k is not making the best use of the AU but will be a little more forgiving if you try to swap to a higher gain tube.  Personally, I'd run with the 2x22k scenario. 

Now..... a Concertina has a gain of less than 1, and we're looking for 80ish volts out when rockin.  This means the preceding stage needs to be making..... call it 85ish volts.  This is seldom a problem unless you have a high loss tone stack right in front of the PI, or the preamp circuitry simply isn't making enough gain.  I noticed you said you have the other 1/2 of the AU7 free.  Why not make it a gain stage directly coupled to the Concertina?  Now you're using the whole tube, 1 triode adds a gain of about 8 to 14 (depending on cathode bypass) and sets the bias of the splitter triode.  Realistically in this situation you can put about 10 volts in and get about 90 volts out.

[jeff]

Cool, Thanks for all the help guys, Sorry if I'm asking a bunch of dumb questions, I like the fact that you guys not only help me with "what do I do?" but also "why do I do that?"

I think "why do I do that?" is more important than "what do I do?"(give a man a fish.....)

Thanks for your time.
         Jeff

[RicharD]

"The only dumb question is the one that doesn't get asked."  I tell this to my guys in the field all the time.  With electrical work, this can prevent a situation from becoming dangerous.  It also kinda seems to be a motto of this forum.  We all love the tube thang and we all love to help. 

When designing a guitar amp, you need to pull together all the requirements first ie, output power, tone stack, over drive, reverb, and other bells-n-whistles.  Then I typically design from the speaker backwards to the input.  I know this thread is PI topical, but it's sorta hard to start at the PI.  This is why folks are throwing several different ideas at you. 

[tubeswell]

Why do we want to increase the current? I always understood tube to be voltage controlled.

Depends what you want. More voltage = more signal swing. More current = more bandwidth (better fidelity)

[PRR]

> Should the 56K plate and cathode resistors be changed to a different value to better suit a 12AU7 over a 12AX7

Back up. What are you driving??

Most self-biased output tubes may use a 500K grid resistor.

Most fix-biased output tubes should use a 100K grid resistor (we often cheat).

To the cathodyne, the two output stage grid resistors are in series for audio signals. Therefore a cathodyne "sees" 1Meg for self-bias, 200K (or a cheat higher) for fixed-bias.

We also know the maximum drive voltage required. Generally a Peak per-grid signal equal to grid-cathode bias is needed for "full" power. And for the usual power tubes it is related to the power tube screen voltage. We may need 25V peak at lower voltage (Princeton) or 50V at higher voltage (Bassman). Again, the two loads are in series, so we need twice this: 50V or 100V.

As Buttery says, in guitar we want "more!'. However we probably do not want "twice too much". that tends to give grid-block. Anyway for 6V6 6L6 6550 EL34 and friends it is hard to make "twice too much" (they were designed that way).

So "significantly more than" 50V-100V peak.

Those are peak voltages. We will want the "RMS" voltage. Multiply by 0.7. 35Vrms to 70Vrms.

In the last century, pocket-protected engineers compiled data for "best" operation for various tubes. Their jobs depended on tubes.

I've collected the 12AU7 and 12AX7 amplifier tables.


First pick a supply voltage. In most gitar amps you can have at least 300V, so use the "Ebb=300V" column. We already know the load, their "Rs", is either 1Meg or 200K. Look for Rs values nearest these numbers and follow over to the Ebb=300V column.

12AU7 driving 2*500K. 6th line down shows Rs=1.0Meg.

Look at the 300V "Eo" column. This is the RMS voltage possible. 54Vrms. This exceeds the 35Vrms needed for lower-volt output stages, OK.

Going back to the left, using Rs=1.0Meg, "Rp" is given as 0.51Meg or 510K. For "Split-Load" Cathodyne, we split this: two 255K resistors. (This is just ballpark: use two 220K or two 270K.)

The table gives "Rk" cathode _bias_ resistor, it should be 23,000 or 23K to get these results.

The "Gain" is not directly useful... Cathodyne voltage gain is always a little less than "1". Actually if amplifier gain is 12 then Cathodyne gain is about 11/12 or 0.9.

Do it again with 12AX7. 1Meg total load, best is Eo=46; however for "0.51meg" load there is a condition where Eo is 47. This may be round-off slop, or the "0.51Meg" condition may be a teensy bit better. Let's use it. Rp should be 0.24Meg or 240K for one load, so two 120K for split-load. Cathode bias should be 1800. Gain is "64", but really 63/64 or 0.98.

The 12AU7 can make 54Vrms, the 12AX7 can make a bit more than 47Vrms. The AU seems "better", but if we need "significantly more than 35Vrms" then either one is fine. If all our other tubes are AX, it may be best to stick with it.

We have design for 47Vrms-54Vrms in 1Meg, but hi-voltage power stages may need as much as 70Vrms. How can we get there?

The chart shows Ebb 90V, 190V, and 300V. Scanning across, "Eo" increases sorta in proportion to Ebb: 90V gives 19V and 300V gives 54V. We may assume that a higher Ebb will give higher Eo in rough proportion. Outputs that need 70Vrms total grid drive will usually be eating 450V. With a little filtering, we can ask for 400V to feed the cathodyne. Assume Eo rises like 400/300 or a factor of 1.33. 50Vrms at 300V would be 66Vrms with 400V supply. Close. Probably works. Must be bench-raced to know.

All of that is for two 500K loads such as used on self-bias power tubes.

Now try two 100K grid loads such as for fix-biased power tubes.

Total grid-grid load is 200K. We look for "Rs=200K" and take Rs=0.24Meg as the nearest suggestion. 12AU7 does best with Rp=100K at 41Vrms; for split-load this is two 50K resistors.  12AX7 does best with Rp=100K at 40Vrms; also two 50K resistors. While you should use different cathode-bias (12AU7=4K4, 12AX7=900), otherwise they are equivalent.

The 40-41Vrms covers lower-volted power stages but falls shy for the higher-volt power stages, even bumped to 400V. Since driver supply voltage can usually be near power stage voltage, it may seem we should be able to hit our mark. However HIGH-voltage power stages must usually run high negative bias to keep from frying at idle. The driver must swing the full bias voltage. So the high-voltage power stages can be very tough to slam with a cathodyne. We often have to re-think as we approach 5F6A turf.

[Tone Junkie]

Thanks PRR you just got another spot on my wall of essential knowledge. you just explained that in such a way I even understood it, not so easy to do.  .
Bill

[jeff]

Thanks for the explination.

Seems like either one would work well.

I think maybe I missed the point of using the 12AU7(12DW7) in the first place. If you want to change the way the PI operates you can change the tube or change the circuit. If you use a 12AU7 in place of a 12AX7 and optimize the circuit to use the 12AU7 it seems there is really not much of a benifit. Maybe the whole point of using the 12AU7(or 12DW7) was to use it in the circuit setup for a 12AX7. It's easier than rewiring the amp.

I misinterperted it as being an advantage of using a 12AU7 over a 12AX7.
Just two ways to skin a cat.

[Fresh_Start]

jeff - I'm not sure about that conclusion.  PRR's post indicates that the 12AU7 can produce 15% greater RMS voltage.  If you read the Valve Wizard's discussion of what he calls a cathodyne PI, he talks about the stage distorting.  Also, there's a lot of chatter about ways to improve the linearity of a Princeton Reverb's PI (for example).
http://www.freewebs.com/valvewizard/cathodyne.html

Cheers

Chip

[jeff]

OK thank you for bearing with me, I really am trying to understand this.

 I think the part that confused me is that if the splitload PI has about unity gain how can one have more gain than the other? Are what you saying is that even though it's unity gain one can produce a 15% greater voltage with a 15% greater input signal? In other words even though the voltage in is about the same as the voltage out one will stay cleaner while the other starts to distort?

"The 12AU7 can make 54Vrms, the 12AX7 can make a bit more than 47Vrms. The AU seems "better", but if we need "significantly more than 35Vrms" then either one is fine. If all our other tubes are AX, it may be best to stick with it."
Also I think my first read through I thought OK both are capable of at least 35V. I missed the fact that we want "significantly more".

"Back up. What are you driving??"
Sorry, you're right. The PI does not exist in a vacuum it's driving something.

Right now I'm using a 275-0-275 PT 5Y3 rect two 6V6 PP. Here's the schematic of what I have built so far:
I'm thinking of using jacks(as in schematic) and building an outboard unit with different PI I can switch to taste test.

I've been looking around through some schematics and came across a Gibson. It seems to have about the same voltages as what I'm expecting with two 6V6 PP and a 12AU7 PI.

[tubeswell]

Because the gain of a balanced cathodyne stage is unity, then the signal strength going in is 'equal' to the signal strength going out.  A 12AU7 grid is typically 6-8V lower than the cathode, so it will take a bigger signal without clipping. Because it is a higher current tube, it will achieve a greater voltage swing with a lower load than a 12AX7 (current x resistance = voltage). However the voltage difference between the grid and the plate is lower than in a 12AX7 so the amplification factor of a 12AU7 is lower.

[jeff]

I'm thinking of this as the outboard A/B using the values PPR suggested. Actually do I need to change them because I'm using 220K resistors in stead of 500K for the power tubes?(18K instead of 23K for cathode resistor):
A 12DW7 would be ideal for this but I'm going with what I got.

[tubeswell]

I'm thinking of this as the outboard A/B using the values PPR suggested. Actually do I need to change them because I'm using 220K resistors in stead of 500K for the power tubes?(18K instead of 23K for cathode resistor):
A 12DW7 would be ideal for this but I'm going with what I got.

Should be okay. Use 2W resistors for the 12AU7 side

[PRR]

> an advantage of using a 12AU7 over a 12AX7.

In the main example I outlined, and most places Fender used a cathodyne, the job is "easy", either tube can do the job, and cathodyne operation masks most of the difference.

Now look at older Ampeg SVT. Three 6550 per side. Maximum 50K gridleak per tube, 17K per side. As cathodyne you want the single-output "34K" load conditions. The short answer is that you don't do a six-pack of fix-bias 6550 with a cathodyne. The long answer is that a 12AU7 will do a whole lot better than a 12AX7. (The Ampeg answer is to use separate big-12AU7 {12BH7} cathode followers after a push-pull gain stage; the cathodyne is 2 stages left of the 6550s.)

> confused me is that if the splitload PI has about unity gain how can one have more gain than the other?

Do not confuse gain with maximum output. A 9V fuzz-pedal may have gain of 200 but max-out of only 1V. Your 6V6 has gain near 10 but max-out of nearly 200V at the OT primary. (Taken to OT secondary, depending on tap, gain may be a bit under unity and max-out near 12V RMS.)

Cathodyne's "about unity" might be said better as "almost unity". It is never more than unity. It is always less, It should never be much less than unity. One variant can be closer to unity than the other. It may make "no difference". For the conditions I outlined: 12AU7 0.92, 12AX 0.98. The difference is 0.98/0.92 or 1.06. All your other stages vary 10% or more, this isn't even significant.

> one can produce a 15% greater voltage with a 15% greater input signal?

Yes. Rather, 12AX7 needs 47/0.98= 48V in to make 47V out, 12AU7 needs 54/0.92= 59V in to make 54V out.

The stage before must be able to swing this 48V or 59V. But the cathodyne has VERY high input impedance. If you look at the "Rs=1Meg" rows, 12AX7 voltage-amp can make 46V RMS or 65V peak, so can be designed to slam the cathodyne to its limits.

> Here's the schematic of what I have built so far

Self-biased 6V6 with 220K grid resistors. That's easy. Any 3/4-assed driver can slam the 6V6es to their limits. You'd have to beat seriously hard, into over-over-drive, to hear any difference between different driver schemes.

High-volt fix-bias output stages need more care in their driver stages. Compare Fender Bassman 5B6, 5E6, and 5F6(a). 5B6 is self-bias and easy. (Also its paraphase is strong drive.) 5E6 is more "modern", and does the job, at least for bass or lightly overdriven guitar. 5F6(a) has the shared-tail driver which has excess balanced drive, and (with 5F6A NFB) has been popular for very large overdrive.

> Valve Wizard's discussion of what he calls a cathodyne PI, he talks about the stage distorting

That's an advanced study. Very pertinent, but tricky.

The naked cathodyne has funny overload action, true. Not horrible though.

What he is showing is cathodyne -inside- the feedback loop of the whole power amp. When the output stage overloads, the NFB forces the driver to try to overcome the clipping. That's what NFB does. You can get some VERY wild waveforms. If you do not know if your power is limited in power stage or driver stage, these waves are very confusing. In diagnosing a "weak" hi-fi amp, you would disconnect NFB to see how the naked amp overloads.

NOTE: many-many guitar amps which use cathodyne do not use NFB (or not-much) around power and driver stages, You won't get these extreme dips and humps.

Also note that most guitar amps capacitor couple driver to cathodyne grid. When cathodyne overloads, the cap-coupling will bias-shift, and stay clean a little longer, then go into cut-off distortion.

Don't re-invent wheels. Find a respectable amp with similar power-stage grid resistors and voltages. Steal the whole darn thing.

[jeff]

"Don't re-invent wheels."
AMEN.
Thanks for talking it out with me guys. Few more pieces of the puzzle.