Why some amps use two triode in parallel?

[soundmasterg]

America sure safe France and Europe by protecting them from Soviet. Without us, your sorry butt will be fried. We protect oil mainly for your benefit. We sure have our own oil. Americans should boycott Europe by not travel there, not buying their product.....which is very easy. Other than Mercedes and BMW etc, There not much worth wanting.

Alan,

I don't know where you live other than that you live in the US. I don't know anything about you other than that you are from Hong Kong originally and are of Chinese descent. Some of the stuff you have said like the above strike me as being from the typically reactionary Republican dominated southern areas of the US, but who knows if that is the case or not. It doesn't matter. The thing that we all have to remember is that we live in a world that interacts and reacts just like chemicals in a test tube, and there will be pros and cons to anything. I don't much like what Russia is doing right now in many areas but specifically in the Ukraine and Crimea. I also notice that the Europeans tend to not want to get involved in world affairs as much as the US, especially when it comes to contributing soldiers or money, yet they like to tell everyone what to do with legislation like the ROHS stuff. Perhaps that is the American in me expressing that opinion, but in the end it doesn't matter. America is not perfect. Americans are not perfect. Every country has pros and cons, and issues and problems...we have enough problems needing solutions here to not worry about others, and many of these problems are of our own making. Whether Obama is good or bad doesn't matter. Congress makes the laws and controls the money and they haven't done their job for 20 years. Failure to compromise is a large part of the problem.

NONE OF THIS MATTERS HERE!

This is an amp forum and we were talking about parallel triodes and why some amps use them or not. Merlin explained where the 3dB noise improvement comes from and the circumstances of that often repeated maxim, and if you don't care or disagree with it, that is your perogative, but it is taught worldwide in schools that it exists, and it is also taught that it is in an ideal situation, and reality will give you less of an improvement. It has been taught this way since the beginning of electronics, and for good reason....because it works. It is a general rule of thumb that you can apply to almost any amplifier and you will know that it is very close to reality. It saves time from having to do a calculation every time you have a theory about something. You've gotten hung up on all these semantics about the noise improvement yea or nay, but what matters in a guitar amp is the sound. Does it sound good or not? Does it give some benefit or not? I can guarantee you that Leo Fender didn't sit there with noise analyzers trying to decide if a 100k or 220k resistor gave more 2nd harmonic distortion or hiss or noise. He had ideas and tried them out and let musicians try them out whose opinions he trusted and we have some great amps as a result. You will never get the answers to your questions if you sit around and concentrate on the math all the time. Get out there and build some of these circuits and listen with your ears....let them decide what sounds good or not. Then after you have done this and come up with some conclusions, you can go back and try some of the math to figure out why you conclusions exist but since we are talking about an audio thing here, you have to hear it to decide no matter how much math you do.

Greg

[Alan0354]

I live in the liberal land of " People's republic of Kalifornia".!!!!That's the reason I feel so strongly about this. I sure hope more American born people feel proud of this great country instead of thinking we are always wrong, then take for granted freedom is free.

The disagreement is only about noise, that has nothing to do with the sound. I did worked with low noise designs for a while. I sure never run across situation that people use parallel devices. This is mainly because in semiconductor, they already designed wider channel FETs, low base spread resistance BJTs to do the job. I don't want to repeat this as it is written in the past posts here.

That's the reason I wrote post #39 as the two "ideal theory" started out to be true. It is the pasky real life situation that make it not useable or of very limited usage.

[Tooboob]

Don't take too much grief. I believe that this is your thread, and you're staying on subject.

Somebody asked you how it sounds.

I've designed some hairy stuff too. I studied tube amps a bit (O'Connor, Blencow, Keuhnel [?] and various other sources). I got to the point where I felt like I understood them. I know a few ways to do everything, and I know their advantages and drawbacks. Time to design an amp, right? Wrong.

Especially when it comes to a "lead" sound, I didn't know what the various distortions sounded like, and I'm still surprised sometimes. It's all about the sound, tone, timbre, expressiveness, etc.

You can trace most amps back to work at Fender and Marshall. Those guys were trying to solve a problem. Acoustic guitars weren't loud enough to keep up with multiple singers, much less a full band. Jazz guitarists used to bang out chords as loud as they could, the horns would drown them out, and solos were inaudible. So Fender and Marshall wanted to use those new tube things to make a loud guitar, and make it cheap enough that even a musician could afford it. Input sources varied in level, so they left lots of headroom on the input volume and followed it with a volume knob that went way too far with a strong source. They reasonably figured that if you turned it up and it distorted, you'd turn it back down. Their product met a need, and they made money.

Then some of those darned kids found out that if you turned it up all the way, it made a cool really loud sound. The sound was the sound of a design being misused. It became an iconic tone, with it's own emotional connotations. It's the sound of rock, and it's kind of an accident. It's in each of our heads, and we perceive it as a real sound, but there isn't really sound in our heads, so everybody's version is a little different, and our personal taste determines our favorite flavor.

There is no math in identifying "sounds good". It's all a matter of taste and experimentation, and you'll even find that that sound you dialed in perfectly one evening needs more tweaking in the morning.

Your background will serve you well in identifying problems, understanding overall architecture, and making changes in the desired direction, but the real challenge is in setting the sonic goal in your head, and achieving it in your ears.

More to the subject, I do lots of complex boards, and if you make a circuit larger to reduce noise, I can tell you that you need a pretty good reason.

I think we all concede that the noise advantage of paralleling triodes can easily be swamped by other noise sources. It's still good to compute the advantage, because intuitively, you might expect the noise to double.

I don't care much about your shot noise and stuff. An electric guitar and cable set a certain noise floor to begin with, and there's enough 60Hz EMI in there to dominate anyway. I'm glad that you can optimize the grid stop for noise, but which 12AX7 will you use? Are microphonics a greater concern? You're evaluating a system with guitar, cable, amp, speaker, cabinet, room - and don't forget the drummer, bass player, and singer.

Why are you spending 15 to 18 hours a week in retirement geeking out?

[Gingertube]

It is interesting to note that Paul Rivera used parallel triode input stages on nearly all the Fender Amps he designed.

By the time he got around to the Fender 75, however, he dropped the parallel triodes on the input and changed the grid stops from 68K down to 33K.

This tells me (well suggests strongly anyway) that his intent in using parallel triodes on the input was always about noise.

Cheers,
Ian

[Alan0354]

It is interesting to note that Paul Rivera used parallel triode input stages on nearly all the Fender Amps he designed.

By the time he got around to the Fender 75, however, he dropped the parallel triodes on the input and changed the grid stops from 68K down to 33K.

This tells me (well suggests strongly anyway) that his intent in using parallel triodes on the input was always about noise.

Cheers,
Ian

That's the reason I came out so strong in challenging this. Because a lot of people read only the first two lines and take it as the truth and do the design accordingly.

That is the reason I challenged it so hard making the point that it is useless to set a theory based on an unreal set of parameters that in real life you cannot fulfill. Anyone can come up with a "truth" if you put enough restrictions on it. This parallel triode thing sure does not seems to be better from my "LIMITED" calculation. GIVE ME THE 1/f DATA, then we complete the calculation. But so far, taking only the grid current, the gain and the input grid stop noise, The parallel triode is loosing BIG TIME. From my understanding, a lot of the 1/f noise( flicker noise) is GAIN INDEPENDENT. So you don't have the excuse of higher gain gives lower noise. And the most important fact is parallel triode DOES NOT DOUBLE THE GAIN.

I stand by everything I said in this thread, that you can make any theory you want with enough restrictions. It's whether it can be applied in real life that matters. It FAILED on one of the most important and money making type of amps.......the transimpedance amp.

[Leo_Gnardo]

It is interesting to note that Paul Rivera used parallel triode input stages on nearly all the Fender Amps he designed.

By the time he got around to the Fender 75, however, he dropped the parallel triodes on the input and changed the grid stops from 68K down to 33K.

This tells me (well suggests strongly anyway) that his intent in using parallel triodes on the input was always about noise.

Cheers,
Ian

That was then, and not long after, Paul Rivera started his own company and revived the parallel triode input stage, and uses it up to today. WHO will hang the bell on the cat? Who will tell Paul to wear the dunce cap. All those triodes, wasted. All that noise, not eliminated. Oh the tubeanity! The theories transgressed, the equations ignored... pass me a kleenex. I'm verklempt.

[Gingertube]

Do we have (anywhere on the www) a modern'ish Rivera schematic at which we can have a "squizz"?

Once we get the grid stops down in size to address their noise then parallel triodes at lower current than a single triode is the next sensible "low noise" mod.

Or you can use parallel triodes for tone purposes as per my post #4 from the top. This circuit was developed in semi ignorance along the lines of:
1) I got a spare triode, I should do something with it (as per a post or 2 above)
2) I might as well run the two triode at different operating points to get a harmonically richer sound
3) Since I'm going to have double the Miller Capcitance from 2 triodes (a bit of a dodgy arguement actually) then I will use half the value of grid stop (and therefore actually do something about the noise by accident).

For guitar Amps then this is covered by Merlin to some degree.

If you want to be "mathed to death" by boredom then it is covered better by Morgan Jones in "Valve Amplifiers".

It is probaby covered in some of the older (but excellent) texts such as "Vaccuum Tube Amplifiers" by Valley and Wallman (which is also on my shelf) but I was'nt sufficiently motivated to get off my butt and go and look.

Cheers,
Ian

[Alan0354]

Just by eliminating the grid stop lower the noise a lot, but not eliminating all the thermal noise and shot noise from grid current. The pickup can easily be 10Kohm. That is a thermal noise generator and shot noise caused by input grid current. You still have to go through the exact calculation like what I did in the past posts to determine which is the dominant factor.

Noise calculation is device independent, it's physics. You need the complete noise information..........flicker noise etc. After that, it does not matter whether it's a tube or transistor. You don't need a tube book to tell you.

There is a very big difference between getting the "Sound" and basic physics like noise, ground noise/loop/interference and stability. Those have NOTHING to do with the sound. For sound, it's all touchy and feely, no theory, just experiment and listen.

For physics, you don't just experiment and see, You can roll your eyes or whatever. These are fundamentals and you can work it out on paper. The three I mentioned are not intuitive at all, you just have to know the physics, you cannot look at these on the surface and see it is or it's not.

Noise calculation is well written and documented. You just have to go through it one by one and then sum up the root mean square and see.

[Leo_Gnardo]

Do we have (anywhere on the www) a modern'ish Rivera schematic at which we can have a "squizz"?

I dunno how modern but thanks to Doug Hofmann's site:

http://el34world.com/charts/Schemati...ic-diagram.gif

http://el34world.com/charts/Schemati...nucklehead.pdf

With or without schemos, I operate on Riveras & the parallel triode input is in just about all I see, and so far has never proved a problem.

Now off in search of a towel, so I can toss it in.

[Alan0354]

It is very easy to look at noise if you don't want to look at theoretical math. build a parallel stage using a 12AX7 as the first stage of the amplifier. Put a switch that you can disconnect the cathode and the grid of one of the triode. Set up the input circuit like a real amp with the guitar with grid stop resistor and resistance of the pickup. Turn on the amp and listen to the hiss.

You have to use a signal generator to calibrate as you have more gain when two triodes paralleled. Input a sine wave from a signal generator in single triode mode and read the output amplitude at the output of the amp. Then without touching the signal generator, switch to have two triode in parallel. You will see the output amplitude increase. Adjust and mark the volume knob on the amp to get the exact same amplitude when you have single triode.

Now you have the amp set up so you get the same gain with either single or parallel triode. The next step is simple, just listen to the hissing sound. Switch back and fore and adjust the volume pot to make sure you are listening at the SAME gain setting. You can tell right away which one has lower noise or the same. It is preferable to use the cascade gain channel as you get more gain and the hiss is more pronounced.

The only factor you cannot compensate is the input capacitance. You have more input capacitance when two triodes in parallel and this lower the noise bandwidth. So you will expect the hissing is slightly lower for parallel triodes. But if you eliminate the grid stop, you should have enough bandwidth that it does not matter much. But just keep that in mind.

This you don't need calculation, just observe. You can then play with the parameter like removing the grid stop, raising and lowering the plate current and compare. Now you can do the touchy feely way. Just make sure you adjust the gain so you have exactly the same gain through the amp in each configuration.

[Gingertube]

Leo - thanks for those.

Heaters spread between +/- 18V DC with the input triodes heater running at +18V to +6V for slightly elevated heaters.
That will help with noise too.

Cheers,
Ian

[Alan0354]

Just want to update. I just did an experiment with input shorted to ground on the high gain channel. I put a switch across the 20K grid stop resistor of the first stage and compare the noise looking at the scope. I switch the switch back and fore. Just the noise of the 20K resistor produce 3 Times the noise amplitude.. To confirm, I have the speaker on also and you can hear a BIG difference in the noise volume with and without the resistor. THAT IS THE DOMINANT NOISE. UNTIL YOU REMOVE THE RESISTOR, EVERYTHING ELSE IS SECONDARY.

I know you guys think it's crazy to calculate noise, that I am wasting your time keep at it. This is how noise work. You find the dominant one and Fix it one by one.

With the 10K pickup resistance of the guitar pickup, I am not sure any of the parallel triodes is even worth considering. Even you remove any grid stop at the input, reduce to 10K input resistance (from the pickup) only lower the noise by 3dB. I am seeing 10dB (3 times) min. difference of noise just from the 20K resistor. So you expect the noise is still double just from the 10K resistance of the pickup. There is NOTHING you can do about it.

Again, my calculation is DEAD ON. The flicker noise is not very dominant from this. You are limited by the resistance of the pickup no matter what. Don't believe me, it's easy to do this experiment!!!


Risking JMF being sarcastic about my long post, I do confirm that rectifying the filament lower the power supply noise. It became obvious in the noise test environment where the gain is all the way up and no input signal, I can hear the hum. Using simple rectifier lower it by a lot. Now, if the grid stop is lower from 20K, the 120Hz becomes dominant. The next step is to use regulator.

[Merlinb]

THAT IS THE DOMINANT NOISE. UNTIL YOU REMOVE THE RESISTOR, EVERYTHING ELSE IS SECONDARY.

Yes, see post #20.

[J M Fahey]

Don´t worry about me
Sometimes I write in a certain way to shake fighters back into the Real World when I see the discussion goes *real* far away from the original one (which sometimes is not bad) but the real trigger is when I see it turning into a cats and dogs fight while the poor original poster sits at the ringside in despair, seeing that *nobody* is paying attention to his original problem or provides any useful answer.
But now I see a civil discussion, so everything´s fine
Just in case it isn´t absolutely clear, nothing I write is ever personal

[silverfox]

Hey! Shut for a minute and take a listent to this

This looks like one of the threads I got involved in over at the Parkin Lot or Lobby.

Okay, noise schmoise. Here's what it sounds like. A couple of parallel tube examples. That's what it's really about right- THE SOUND, and that is part of what is desired in an amp of this design. Are the richer harmonics influencing the perceived noise threshold?

Take a listen:

Watkins Dominator - YouTube

Alright now cool down people.

Watkins Dominator "Golden Anniversary" reissue - YouTube

Marshall 18 Watt ReIssue

Marshall 18 Watt 1974X Combo - YouTube

Yes there is noise here and there. Is it acceptable?

That's what the Fox says...