Best way to combine two dissimilar channels into PI?

[dehughes]

Another thought..... you said 6g3.....

Your going thru an awful lot for two dissimilar channels.
Why not stick closer to the 6g3 scheme.... have half the 12ax7 as one pre
use the 6sn7 in various forms as the other...... and the other half of the 12ax7
as a common stage before the PI..... It will split the difference as for as gain
differences go, and you will still have a 6sn7 pre, the volumes become the
mixer and you will very much simplify the signal path.

OKAY!!! Now that's cool! (I'm responding to both posts here...)

First of all, your replies are so thick with ideas and information that I have to read them several times just to get it.

Second, the ideas are all great.

Third, 6SJ7? I didn't consider using that tube as well. I do have a few OLD stock used oddball tubes (very odd...) in a box upstairs, so I'll have to go check for a 6SJ7...that'd be fun to have a pentode in there instead...have to do some rewiring....but worth it.

Fourth, your idea about putting the 6SN7 BEFORE the other half of the 12AX7 is just brilliant. Get this: When I was installing the 220k mixer resistors in between the 12AX7 stages, I remember thinking to myself "This seems kinda silly, by whatever...I'll just follow the schematic." Frankly, I think that's the way I'm going to roll with this....maybe parallel the 6SN7 triodes and run them into the 2nd half of the 12AX7 and see how that goes. Brilliant! Different from what I was intending to do, but great nonetheless...especially as what I was intending to do really didn't end up to be what I was intending to do.

So, I'm going to go with the simplest solution, which is the 6G3 idea. I find the simplest ideas are are often the best....though now the question is how many different ways of wiring up the 6SN7 I'm going to try before I settle on one that works with the 12AX7 side.

One question: If I run the 6SN7 in parallel, with, say 300v of B+ available, would I just double up on the plate and half the cathode resistor values? As it is now, each plate has a 100k resistor, and each cathode has 2k7. So, would I just put a single 220k on both plates, and a 1k2 or so on the cathodes?

[Andy Le Blanc]

Thats the fun part. You have a bakers dozen, of different ways to dress a
socket and couple of gain stages. Start with the most primitive, zero bias.
put a .1 cap and 4 to 15m grid return on the grids ground the cathode and
put a 220k load on the plates. The thing about a 6sn7 is its current handling.
this will give the most gain, and give you chance to fart around generating
your own data when you change the plate load, the supply voltage will
change a bit and you can measure grid bias and if you can put 250 to 500ma
signal to it an a know freq. , you can get dynamic numbers for the circuit.

Look for the "Rightmark Audio Analyzer" online too, its a free down load you
can use to turn you computer in to a audio tester and get more info than
you could ever use for each configuration that you try, as a pre and with a dummy load, how each
effects the over all performance of the amp.

[dehughes]

Awesome. Thanks. I'll get to it tomorrow!

[dehughes]

Sooo, that's it! FIXED! YESSSSSSSSSSSSSSSS!

I've run the 6SN7 triodes in parallel with a 1k2 shared cathode resistor and a 220k shared plate resistor (yes...I know...wrong way...will change it to a 47k right now...), and ran that into the other half of the 12AX7 via a 220k mixer resistor, and voila! PERFECT! The 12AX7 side sounds MUUUUCH better, and the 6SN7 is nice and THICK and distinct. I'll go replace that 220k, but I'm pretty sure this is going to work wonderfully. I'll post back after I get a chance to really crank the amp up tomorrow afternoon...

THANKS to you all...especially Andy for thinking simple when I was headed off to convolution...

Okay...update...I put in a 55k Ra shared between both triodes on the 6SN7, so they're running about 112v each now (as opposed to 33v each with a 220k shared Ra). MY GOODNESS! FRICKING AWESOME....wow wow wow. I only hope this sounds as good when I crank it up tomorrow...

[dehughes]

Well, it sounds reallly good... The only thing I noticed was some fizzyness when it was really cranked up...but this was more of an "old school amp" fizzy (r.e., an amp that was not "meant" to be CRANKED fizzy) than a "yuck, that's lame" fizzy. I may try swapping out the 500pf and .01uf caps in the tone stack later, but for now I'm just going to enjoy it. Dark and rich is how I'd describe it....but not muddy. Fun...it's something completely different from what I'm accustomed to.

One thing is for sure...you can really hear the 5Y3. That is, the voltages are where they need to be for this build only because the PT is seeing a 5Y3 instead of a 5AR4. I much prefer the 5AR4, but that'll come after a PT swap...so for now, I'm good.

Thanks again!

[Andy Le Blanc]

Excellent .......
For the fizz, try putting a 100p to 500p cap parallel with the 82k plate resistor
smooth out the top, theres mention in the trainwreck pages, if you implement
it towards the back of the circuit it can be a subtle fix.

[dehughes]

Excellent .......
For the fizz, try putting a 100p to 500p cap parallel with the 82k plate resistor
smooth out the top, theres mention in the trainwreck pages, if you implement
it towards the back of the circuit it can be a subtle fix.

Right on. Thanks. So, what 82k plate resistor are you referencing? I don't have one in the amp....it would have had one on the PI, but I put in a more Voxy PI with a 47k tail and dual 100k resistors on the plates.

[Andy Le Blanc]

Dohhhh....... the plate resistor of the input stage of the PI. I was looking at the
scheme early in the post. Simple trick to try from the back of the amp to the front.
it'll be more apparent in the pre but maybe more effective on the PI.

[dehughes]

Dohhhh....... the plate resistor of the input stage of the PI. I was looking at the
scheme early in the post. Simple trick to try from the back of the amp to the front.
it'll be more apparent in the pre but maybe more effective on the PI.

Right on...that's what I thought. Very cool...does this differ from the small PF cap between the two plates themselves? I'm thinking yes, because the cap between the plates would keep high frequency oscillations down, whereas the cap across just one of the plate resistors, especially the 82k side, would mute the high end somewhat (depending on the cap value). Am I close?

Good deal about it being more effective in the PI but more apparent in the Preamp. I'll store that one away for future use, for sure!

BTW, I just finished really cranking this thing up, and wow...it's a little monster. The fizzy isn't as bad as I first ascertained, and I think that with a snip and a swap of caps, I'll be much closer to good (right now the 6SN7 side is a bit bass-heavy with the .002uf 6PS cap on the plates...I'll try a 500pf polystyrene and see what I think...). The 6G3 12AX7 side is just a snarly little gainy beast....yikes...

[Firestorm]

Is it a parasitic fizz ... or an odd order harmonic fizz? I wish I could find where I read it, but someone posted a spectrum analysis of composite/parallel tubes and demonstrated that when we do the conventional thing and halve the values of cathode and plate resistors in paralleled triodes, third-order harmonics become dominant. It's the plate load that's the problem apparently. Don't know if it applies in your case, tho.

[dehughes]

Is it a parasitic fizz ... or an odd order harmonic fizz? I wish I could find where I read it, but someone posted a spectrum analysis of composite/parallel tubes and demonstrated that when we do the conventional thing and halve the values of cathode and plate resistors in paralleled triodes, third-order harmonics become dominant. It's the plate load that's the problem apparently. Don't know if it applies in your case, tho.

Nah, not a parasitic fizz so much as a "70's cranked amp fizz", if that makes sense. It's like the amp is saying "wait, I'm not meant for this!" instead of "eww...something's wrong". I like the former.

As I'm really beating on this amp I'm noticing that the thing is mellowing out nicely and the fizz is waning, so I wager it's a combination of me breaking in the (relatively) new G12H30 and me slowly reducing coupling cap sizes to eliminate some of the (admittedly) excessive low end. I'll swap another cap tomorrow and see if that gets me where I need to be.

Interesting deal on the third order harmonic plate resistor value... What was the suggested "solution"?

[Firestorm]

I think the implication of the spectrum analysis on paralleled triodes (which I still wish I could relocate -- anyone?) is that the plate resistor should be what it was with a single triode with the cathode resistor adjusted accordingly. For example, if you used 100K/1K5 with a single triode, conventional wisdom dictates 47K/820 when you parallel the triodes. Instead, you would keep the 100K plate load and adjust the cathode resistor accordingly. Technically, you'd want a cathode resistor something like 1K35, so a stock value would be 1K2 or 1K5. (The resistors should probably be 1 watt for safety). If the cathode resistor is bypassed, changing its value won't affect gain very much, so you might have to adjust gain elsewhere if necessary.

I read the post I referred to when I was tweaking a 180-watt jazz amp I had built and it had a parallel triode reverb mixer. I would up using 82K on the plate and 1K on the cathode and it sweetened it up quite a bit.

In your case, though, you're probably right that as everything wears in it will settle down.

[Wayne]

...paralleled triodes (which I still wish I could relocate -- anyone?) ...

Was it this you were thinking of?

http://greygum.net/sbench/sbench102/dax7.html

W

[dehughes]

I think the implication of the spectrum analysis on paralleled triodes (which I still wish I could relocate -- anyone?) is that the plate resistor should be what it was with a single triode with the cathode resistor adjusted accordingly. For example, if you used 100K/1K5 with a single triode, conventional wisdom dictates 47K/820 when you parallel the triodes. Instead, you would keep the 100K plate load and adjust the cathode resistor accordingly. Technically, you'd want a cathode resistor something like 1K35, so a stock value would be 1K2 or 1K5. (The resistors should probably be 1 watt for safety). If the cathode resistor is bypassed, changing its value won't affect gain very much, so you might have to adjust gain elsewhere if necessary.

I read the post I referred to when I was tweaking a 180-watt jazz amp I had built and it had a parallel triode reverb mixer. I would up using 82K on the plate and 1K on the cathode and it sweetened it up quite a bit.

In your case, though, you're probably right that as everything wears in it will settle down.

Huh. Very cool. I'm running a 56k plate resistor with a 1k2 cathode resistor, and so far it's good. I didn't think about leaving a 100k on the plate...wouldn't that drop the overall output of the stages?

Ah...got it...that chart is super helpful. So, it seems that if one is to parallel triodes, it's best to leave them unbypassed...but either way you're sacrificing the 2nd order harmonics. Now, what's unfortunate is that it seems there is now way to reproduce the 2nd order harmonic level if triodes are paralleled...

Perhaps I'll try putting the cathode cap on a DPDT switch, to act as a boost of sorts...?

[Firestorm]

...paralleled triodes (which I still wish I could relocate -- anyone?) ...

Was it this you were thinking of?

http://greygum.net/sbench/sbench102/dax7.html

W

That's the one. Thanks. Even more interesting than I remembered.