parallel triode question

[pinkmarkos]

hey everybody- looking for a little amp guru wisdom on the benefits/drawbacks of using a parallel triode method on v1 (ala matchless). From what i've been reading, this method is used to reduce noise and increase gain originating from v1, but at the expense of desirable 2nd harmonics. Is there any other reason to wire v1 this way? Would it be better to use two separate gain stages here? Is it better for vox style amps vs marshall style?
I have a trinity amps TC15 I built with the parallel triode in v1 and it sounds great. now i am messing around with a one channel (bright channel) 50W plexi style build, which leaves me with an unused triode on v1. i'm looking to voice it somewhere between plexi and jcm territory. any insight would be much appreciated- thanks

[Ken Moon]

Whether or not a parallel triode stage has more gain than a single triode stage depends on the rest of the circuit.

And don't worry about losing any 2nd harmonics - the signal just doesn't get big enough on an input stage to see that effect.

If you have a typical input stage, want to keep the same frequency response and voltage swing on the output, and just get the advantage of lower noise, you'd need to:

1) Cut value of grid stopper in half (since input capacitance doubles with dual triodes, halving resistance keeps same freq response).
2) Cut values of plate resistor and cathode resistor in half.
3) Double value of cathode bypass cap.

So a center-biased 12AX7 parallel input stage (in typical range of preamp B+ values) might use a 10k grid stopper, 51k plate resistor and 750R cathode resistor, and would be fully bypassed with a 44uF cathode bypass cap.

The lower noise (higher SNR) comes from using smaller resistor values in the circuit.

This also has the benefit of producing lower output impedance, which will mean less loss if a tone stack immediately follows the parallel triode stage.

[paulster]

And don't worry about losing any 2nd harmonics - the signal just doesn't get big enough on an input stage to see that effect.

This isn't strictly true. You don't have to overdrive a triode to generate harmonic content. You certainly get a lot more as soon as you are into overdriven territory but there exists harmonic content even with clean signals. It's one of the factors that makes tubes sound 'warmer' than solid-state equivalents.

This is also why a number of very high end audiophile tube amps are single-ended using high power tubes as push-pull also helps to cancel even-order harmonics, and it is the second-order harmonics that are very sonically pleasing. These amps certainly don't get near to overdriven levels as that's the last thing they want for this application.

To the OP, you could consider using different Ra and Rk values, to deliberately mismatch the triodes very slightly, which could certainly add to the complexity of the sound. Then you could switch one of the triode grids between signal and ground to give you a moderate gain boost switch.

[Ken Moon]

Good point.

I've experimented with parallel triodes quite a bit, including on the Express and Plexi-style preamps, and I don't notice any drop in harmonic content, but it could be there, and just be very subtle (after 53 years, my ears aren't as good as they once were).

If this 2nd-order harmonic reduction effect was more pronounced, it might be worth considering doubling up the third stage in the Express, since some feel this stage generates "too many" 2nd-order harmonics (the heretics!).

[pinkmarkos]

thanks for the input. i didn't realize that the grid stopper, Rp, Rk, and bypass cap values would need to be changed as well to keep the same response. however, i noticed that the matchless uses the parallel triode while keeping the same AC30 values for plate, cathode, and bypass cap. Is this just to increase gain?

I've experimented with parallel triodes quite a bit, including on the Express and Plexi-style preamps

Ken, how did you like the results? I am curious to know, for both the express and plexi.

Also, i like paulster's idea about the switchable 2nd grid to ground as a "boost". cool idea, i think i might try that.

[Ken Moon]

To tell you the truth, it's not a huge, head-turning difference.

I think the parallel triode input stage is more helpful on an Express, where you're usually doing everything you can to keep the noise floor low because of the high sensitivity of the amp.

With a plexi, just paralleling the input stage triodes is much less noticeable. If you're looking for an extra gain boost in a single-channel jmp-type preamp, I'd rather use the extra triode for another gain stage (a 2k7/100k unbypassed stage up front is popular - some even call this a "tubescreamer" stage, because the effect is similar to using a TS9 as a booster in front of the amp. (btw I think a 50w master-volume jmp amp, like the 2204, with a TS9 in front, is pretty darn nice sounding).

If you try the parallel input stage, you can try all sorts of stuff by using a shared plate resistor, but different cathode circuits. Try one hotter and one colder, one with led bias - the fun is endless

[pinkmarkos]

thanks for the feedback ken. already got the soldering iron warming up, . i agree with you about the 2204 sound- always seem to come back to that for some reason. anyways, guess i've got lots of circuit tweaks to try, i'll repost to let you know how it turned out.

[Darkbluemurder]

I also experimented with parallel triode stages.

The Matchless input stage with the 220k/1.5k combination obviously attempted to extract the most gain possible. While this is OK in a one stage preamp like the Spitfire it is not really necessary in a two stage preamp (AC30, Lightning, Plexi et al.).

In my Club Deluxe which is very similar to the Lightning I started out with 220k as the plate resistor but changed to 100k later for a more open sound but that is a matter of taste.

Cheers Stephan

[Andy Le Blanc]

you need to test the paralleled tube sections for dc balance and scale Rp, Rk.
mismatched paralleled sections will be unpredictable and what your looking
for is a change in harmonic products out to the sixth, a reduction.
distortion will be the "average" and slightly lower when parallel.

your also going to want to minimize mismatch issues with a common bias scheme.

the parallel causes simple addition of the currents.

the output impedance is divided...... (that's what your hearing)

the power is multiplied, both effect bandwidth.

[pinkmarkos]

you need to test the paralleled tube sections for dc balance and scale Rp, Rk.

could you use a dual triode with matched sides for this? seems like if the same Rp was tied to both plates, the only difference would be the inherent current draw of each side.

the parallel causes simple addition of the currents.
the output impedance is divided...... (that's what your hearing)
the power is multiplied, both effect bandwidth.

still trying to wrap my head around this... by bandwidth do you mean voltage swing of the stage, or frequency response? how am i hearing the lower OP impedance? because more signal is making it to the next stage?

thanks again for the input, you guys are awesome

[tubeswell]

you need to test the paralleled tube sections for dc balance and scale Rp, Rk.

could you use a dual triode with matched sides for this? seems like if the same Rp was tied to both plates, the only difference would be the inherent current draw of each side.

It all goes through the same plate resistor and cathode resistor anyway (when you have parallel triodes). If you want, you could put separate cathode resistors on each triode and then have a bypass cap from each cathode going to one side each of a 5-10k pot, with the wiper going to ground: then you can dial-in your perfect mojo. But that is probably unnecessary just for rock'n'roll (unless you are going for different rk/ck combinations - but then we aren't really talking about parallel triodes any more)

the parallel causes simple addition of the currents.
the output impedance is divided...... (that's what your hearing)
the power is multiplied, both effect bandwidth.

still trying to wrap my head around this... by bandwidth do you mean voltage swing of the stage, or frequency response?

Bandwidth = frequency response

how am i hearing the lower OP impedance? because more signal is making it to the next stage?

In a nutshell - yup. More bandwidth is making it into the next stage. The rule of thumb for good impedance bridging is to have output impedance of the source (stage before) at 1/5th (or smaller proportion) of the input impedance of the next stage.

[Andy Le Blanc]

A good parable for bandwidth....

how far back away do you have stand to see the whole elephant.....

say with a 12ax7 you've got the gain but your not seeing the whole critter.

a 12au7 has a good bandwidth, but the circuit needs to be critical to see
the elephants entire shadow.

Another factor I ran across, the capacitance that is seen by what ever is
driving the pair (or multiples of pairs) goes up, the capacitive loading.

Have you ever tried giving a quad of power tubes each their own coupling cap,
it does that, Hiwatt did that I think , good for clean/bass.

[PCollen]

I know I'm late for the party with this reply, but:

I had a Dr. Z MAZ-18 opened up once, and noted that it had a parallel 12AX7 at the input, and used a common 68K plate resistor and a common 1.2K cathode resistor bypassed by a (AIR) 22uF cap.

The ideal 'center biased' plate load for a 12AX7 is about 2x the plate resistance which, in this case would be 2x 35K = 70K. So I guess the good Doctor did his homework and chose the 68K accordingly for the clean tone and low noise he was designing for.

[David Root]

Andy, I tried separate 0.1 uFs on a #124 as I had individual bias pots. Mo'bass for sure!
But if you're going to crank it change the 220K bias to 100K otherwise the time constant is a bit too slow.

[renshen1957]

you need to test the paralleled tube sections for dc balance and scale Rp, Rk.
mismatched paralleled sections will be unpredictable and what your looking
for is a change in harmonic products out to the sixth, a reduction.
distortion will be the "average" and slightly lower when parallel.

your also going to want to minimize mismatch issues with a common bias scheme.

the parallel causes simple addition of the currents.

the output impedance is divided...... (that's what your hearing)

the power is multiplied, both effect bandwidth.

To add to the above,

As to parallel triodes:

The composite tube will have this relationship to the individual triodes it is made from: the same voltage capability; twice the current capability: twice the power rating: half the plate resistance/half the internal resistance; twice the capacitance; twice the transconductance; the same mu; and twice the plate dissapation

The parallel tube will have about 70% of the noise of a single section, the voltage gains is the same as a single triode (the measured gain is often higher circa 20-30%); resistance plate is so much lower, ergo there is less loss working against the plate load).

Or in other words, sonically the tube sounds more like a medium mu stage, smoother and mellower.

I used the parallel triode in a tweed build to simulate a 12AY7 with 12AX7 tubes in the first and second preamp positions at the time when 12AY7s were out of production (prior the Russians starting new production of the tube).

Best Regards,

Steve