parallel preamp tube - shared plate resistor/seperated cathode resistors

[Quatro]

if you run both sides of a 12ax7 in parallel and would like to bias each side differently (say 820 on one cathode and 2.7k on the other) and you would like to use a shared plates resistor, how do you pick the value? How does it compare to the same bias set up with separate plates resistors? Are there any sonic or other advantages or disadvantages to a shared plate resistor?

Thanks

[phsyconoodler]

If the tube is paralled you can't bias the cathodes separately.They are PRARALLED,meaning they are connected together.Same with the plates.You parallel the working components of each triode and they become one big tube.
  Fatter sounding but not a huge gain boost.I like 18 watt amps with paralled triodes.They sound yummy.

[Fresh_Start]

If the tube is paralled you can't bias the cathodes separately.They are PRARALLED,meaning they are connected together.Same with the plates.You parallel the working components of each triode and they become one big tube.
  Fatter sounding but not a huge gain boost.I like 18 watt amps with paralled triodes.They sound yummy.

Respectfully, I think that you can use separate cathode resistor/cap networks.  O'Connor writes about a "Tweed mixer" where two triodes have separate grids and cathodes but shared plate resistor.  Also, the AX84 Blues Preamp has parallel triodes, shared grids and plates, with separate and different LEDs on the two cathodes.

People almost always halve the value of a shared cathode resistor.  In theory the plate resistor should be halved as well AFAICT. However, it doesn't seem to be done that way typically.  I guess it's to get a hotter bias out of the paralleled triodes.

Cheers,

Chip

[tubeswell]

If the tube is paralled you can't bias the cathodes separately.They are PRARALLED,meaning they are connected together.Same with the plates.You parallel the working components of each triode and they become one big tube.
  Fatter sounding but not a huge gain boost.I like 18 watt amps with paralled triodes.They sound yummy.

Respectfully, I think that you can use separate cathode resistor/cap networks.  O'Connor writes about a "Tweed mixer" where two triodes have separate grids and cathodes but shared plate resistor. 

And Merlin Blencowe covers this in his 1st book (in the chapter about parallel stages)

[seedlings]

Given 12AX7, 280V and Ra=100k
if you use an 820r on one triode, that will draw about 1.2mA
if you use an 2k7 on one triode, that will draw about 0.7mA

total of 1.9mA and a drop of 190V across the 100k

If you went both 820R, the drop is 240V
Both 2k7, drop is 140V

If you went standard shard cathode of 820R, each tube would draw about .95mA for a drop of 190V (exactly the same as if you bias one with 820r and the other with 2k7)

CHAD

*edit: after you've calculated all of that, be sure to halve the Ra value you used for designing!

[phsyconoodler]

So I guess the question I should have asked is "why?"

[seedlings]

So I guess the question I should have asked is "why?"

Well, the way I look at it is more of a 'big picture' approach based on hunches and presumption (almost kidding).  I *think* that each triode's load line should be double the Ra resistor value (even though there might be another triode attached).  However the voltage dropped will be about double compared to one Ra per triode.

So, now that I said this out loud, my previous post should assume an Ra of 47k instead of 100k.  In other words, whaver Ra value you use to draw the load line and bias, solder in half that value since you're paralleling triodes.  Understand that since each section is biased differently, the load line will not be exactly what you planned, which is part of the beauty of this hobby.

Try it and report back!  It won't blow anything up :)

CHAD

*edit: It's like when you parallel power tubes, you halve the value of the OT primary.

[Fresh_Start]

So I guess the question I should have asked is "why?"

Why not?  Seriously though, I've thought about the tone shaping possibilities of parallel input triodes with separate cathode R/C networks.  You could create two frequency steps by using identical cathode resistors but different bypass caps.  You could switch bypass caps in/out on one or both triodes.  You could bias one triode relatively hot with a small cathode resistor and the other cold with a big cathode resistor.  Or you could do some combination of these things.

Why?  I don't know.  Tonal complexity?  I do remember some of the design discussion for the AX84 Blues preamp though.  They just experimented with different LEDs until they found a combination that appealed to the designers' ears.  Since each LED yields a different voltage at the cathode, the parallel triodes are biased differently even though they share a single plate resistor.  IRRC they were shooting for a warmer tone overall.

With respect,

Chip

[HotBluePlates]

if you run both sides of a 12ax7 in parallel and would like to bias each side differently (say 820 on one cathode and 2.7k on the other) and you would like to use a shared plates resistor, how do you pick the value?

If you know you what them to be different, you'll have multiple iterations of calculating... or you'll guess close enough, then you a pot wired as a variable resistor to dial it in until you like what you're hearing.

If you would like a demonstration of the calculation approach, we'd need to know which stage of your preamp will be the paralleled stage, what the supply voltage is, how big the input signal will be and what follows the paralleled stage. We'd need to know how big the signal must be at the output of the paralleled stage. We'd also need to know the desired final effect. If you knew something about how you'd like the stage voiced, that would help, too.

How does it compare to the same bias set up with separate plates resistors?

Parallel stages with different plate resistors, and output signals going to the same place, seems a little pointless. However, I suppose you could have one plate load very different than the other to try to induce some form of distortion in one of the triodes.

That said, I think the best use of this different-bias approach is to have a clean triode/distorted triode, or two different-distorted triodes. That's probably easier to manage with changes of cathode resistor value than plate resistor value.

Are there any sonic or other advantages or disadvantages to a shared plate resistor?

Not directly. Any advantage or disadvantage is probably seen in light of comparing parallel stages versus cascaded stages. Those advantages/disadvantages very likely depend on your needs and which circuit fills your needs the best.

[jbefumo]

Sorry about posting to an old discussion, but I think this fits here.


Reading Merlin's analysis of paralleled triodes really piqued my interest, and I've been messing around with the idea for a while now. For me, the channels on traditional Fenders and Marshalls always seemed like kind of a waste, unless you were jumpering them, ala Marshall, but that just seems like a rather funky solution.  To save panel space, I've built several prototypes using an NKK DP3T ON-ON-ON switch to give me 'Normal/Bright/Both' with just one (or two) inputs. (See pic.)


I ran simulations of three different configurations (shared cathodes, like early Marshalls, separate cathodes, like later Marshalls, and separate cathodes with shared plate resistors.)  I ran each one to show the effect when using the normal input only, the bright input only, or both (jumpered).  To keep the illustrations in order, I've put them into a single PDF (attached).


My electronic intuition is not all that keen, so I find that a graphic comparison helps me grok what's going on.

[silverfox]

Edited: Wow, misread the original post. Bad analogy for numerous reasons. I only leave it here as it otherwise will confuse the response to this post. Thanks for clearing up the Hetrodyne principle.

I've considered this in the past and obviously so have amp manufactures' long before I did. At one point I announced a design based on combining two voicings that would produce a metalicious blast of tone. It was shorty after that, I began to comprehend, a little, the Plexi design.

The concept of combining two frequencies to obtain a resulting mixed or composite frequency, was employed prior to audio amps in the field of radio. The Super Heterodyne receiver being one example. The Super Het combines the incoming signal with a locally produced oscillator signal to generate a resultant output signal that is then decoded as the desired station you tuned the radio to, (there is a little more to it than that but for simplification purposes).

With any combination of signals there arises harmonics as well as varying degrees of wave cancellations. This is what will make the development and fine tuning of the preamp circuit either a success or horribly sounding disaster. In my own contemplations, since I haven't built anything based on this yet, in my own contemplations I decided if I did this I'd add the single tube mixer circuit shown in Merlins tube preamp book. It seems also to me that changing the cathode resistors will also create a difference in gain between the two tubes and with a mixing tube you would be able to adjust for that.

Another analogy that may enlighten here, is taken from the colour spectrum. With four colours nearly any colour may be produce and in the area of monitors they only use three colours. So the concept has potential, but for the making it work.

Okay, I've raised some comparisons here taken from RF and these analogies are based on my conclusions so if they are faulty please, feel free to correct them.

silverfox.

[eleventeen]

"The concept of combining two frequencies to obtain a resulting mixed or composite frequency, was employed prior to audio amps in the field of radio."


But the same signal (one guitar) is being fed to both triodes. The result of this exercise should produce different gains; to what effect I would not venture to guess. But there is no heterodyning effect because there is only one frequency. Perhaps there would be a very slight difference in phase output but I doubt it would be detectable. You're right that different frequencies will produce a sum and difference type of output structure, but that is not what's being contemplated here, as I understand it.

[Willabe]

......because there is only one frequency.

What if you play more than 1 string at once, now you have more than 1 frequency?

[shooter]

play more than 1 string at once

I think eleventeens argument still holds since each tube *see's* the same thing.  To make it work, I'd think you'd need 2 guitars playing 2 different songs, or maybe same song just *delay* one guitar by some amount of time

[eleventeen]

Those sum and difference frequencies are indeed generated but in the case of consonant intervals (just talking about two notes) our ears have been conditioned (by years of listening to Western music including the equally tempered scale) to accept the "mix" of those tones. Additionally, the levels of those sum & diff freqs are way way below the notes that are intended to be heard. We've all had the experience of tuning a guitar by comparing two notes of supposedly the same pitch and "tuning out" the "beat frequency" produced when those 2 notes are out of tune. Even badly out of tune, most people have to listen pretty carefully to catch the "beats".


Anyway, anyway, anyway.....one guitar, playing 1 note or six notes, feeding the signal to two different triodes, should [edit] >not< produce heterodyning.

[shooter]

playing 1 note or six notes

I'll buy six notes, HBP answered that question for me in another thread, but 1 note is 1 frequency? and my understanding of heterodyning is you need 2 different freqs beat against each other?  since most players use lots of notes I guess it's kinda an academic thing more than a real world thing.

[HotBluePlates]

The concept of combining two frequencies to obtain a resulting mixed or composite frequency, was employed prior to audio amps in the field of radio. The Super Heterodyne receiver being one example.

......because there is only one frequency.

What if you play more than 1 string at once, now you have more than 1 frequency?

You can't apply heterodyning of RF to audio frequency amplification.

Heterodyning is equivalent to intermodulation distortion (2 frequencies in, 4 frequencies out), but it also has a tuned circuit at its output which only allows a single frequency (the "intermediate frequency" in an A.M radio) to pass on the the rest of the circuit. It also happens that the subharmonic which is generated by the Inter-mod-like process is the one frequency which is passed on (usually 455kHz).

However, as Eleventeen is saying, you have 2 of the same guitar signal applied to your 2 different gain stages. Doesn't matter if the signal is 1 tone or 10,000 tones, essentially the same thing is being applied to each. And all the 2 gain stages do is either EQ each a bit differently or distort one more than the other (or both).

The value of parallel gain stages with different cathode components is that you have access to 2 different gain levels, EQ curves and/or operating points. You could have both in action at the same time, or only one at a time for a selectable amp response.

playing 1 note or six notes

I'll buy six notes, HBP answered that question for me in another thread, but 1 note is 1 frequency? ...  since most players use lots of notes ...

To keep the concept to a scope we can wrap our brains around, pretend the amp gets only a single undistorted sine wave. That is, a pure tone with no harmonics which represent additional tones to complicate the picture. A real guitar signal (or almost any instrument sound) is actually a very complex wave with many tones, even for a single note.

That said, an actual example of heterodyning in audio would be a ring modulator. I doubt you'd argue that any guitar amp with shared plate/separate cathodes produces anything that sounds like a ring modulator.

[silverfox]

  Hetrodyne analogy stricken from the record. See above.



silverfox.

[jbefumo]

Not exactly on topic, but seems close -- I'm building a modified Tweed ODS amp with reverb, and since I have an extra tube slot, I figured I'd use some kind of dual-triode mixing circuit.  From the clips I've heard of this design, I really don't want to mess with the tone too much, and see no need for additional gain.  The circuit I've seen most online is the common-plate mixer, but that will add some degree of gain, depending on the choice of components.  It seems to me that a similar application could be done using a common-cathode design, but I've never seen that done anywhere, so I'm wondering if there's some good reason for that.  I'm attaching a pic of the two topologies I have in mind (values are just place keepers for the moment -- haven't bothered with the calculations yet.


Thanks!


Joe

[PRR]

Both forms of "mixer" involve substantial loss of potential voltage gain and maximum output.

The cathode mixer is especially bad.

"Passive" resistor mix with make-up gain as needed is usually a better plan.

[jbefumo]

Hmm -- now I'm confused (not an uncommon state) -- the cathode mixer I understand, but how is the common plate mixer losing gain? Seems like it would be >1 with any reasonable choice of resistors.
Joe

Both forms of "mixer" involve substantial loss of potential voltage gain and maximum output.

The cathode mixer is especially bad.

"Passive" resistor mix with make-up gain as needed is usually a better plan.

[jbefumo]

Here's a link that got me thinking about the 12AU7 in this application:


http://archive.ampage.org/threads/4/gadc/255984/12au7_in_Tweed_Mixer_-1.html


Particularly: "

"The isolation is perfect with none of the "tone sucking" that I got from mixing resistors. "  [/t] Hmm -- now I'm confused (not an uncommon state) -- the cathode mixer I understand, but how is the common plate mixer losing gain? Seems like it would be >1 with any reasonable choice of resistors. Joe Both forms of "mixer" involve substantial loss of potential voltage gain and maximum output. The cathode mixer is especially bad. "Passive" resistor mix with make-up gain as needed is usually a better plan.

[PRR]

> how is the common plate mixer losing gain? Seems like it would be >1 with...

"Potential gain".

You add one triode, you expect gain of 15-50. With two triodes in the pot, 225-2,500.

The plate mixer will have gain of less than half what each tube will do as a plain stage. Also maximum output tends to be much less.

You will in general have "better" (engineering) result with resistive mix then tube gain, or with two tube gain and a passive resistor mix after that (plates not parallel). The resistors can usually be figured to isolate any reasonable amount.

Try it. It may be fine.

If you don't want any added gain, an active feedback mixer does that with improved isolation.

[jbefumo]

The more I learn, the more I realize how little I know ....

> how is the common plate mixer losing gain? Seems like it would be >1 with...

"Potential gain".

You add one triode, you expect gain of 15-50. With two triodes in the pot, 225-2,500.

The plate mixer will have gain of less than half what each tube will do as a plain stage. Also maximum output tends to be much less.

You will in general have "better" (engineering) result with resistive mix then tube gain, or with two tube gain and a passive resistor mix after that (plates not parallel). The resistors can usually be figured to isolate any reasonable amount.

Try it. It may be fine.

If you don't want any added gain, an active feedback mixer does that with improved isolation.