The 6550 Experience

[bill bokey]

Thanks for listening to my soundclips Dave, glad you liked them. I would have loved to keep the 1987, it sounded really good !

Can you tell us more about class B ? Once I've tried the 6550s I'll give the EL34s a go as well. But I still don't quite understand how to bias in class B. Is it really class B or very cold class AB ? What was the plate dissipation of the 6550s you use back then ? I read the very first ones had 35W plate dissipation and the later ones 42W but which one did you use ? How did you come up with the 25-30mA range ? Did you use an oscilloscope to "see" when they were operating in class B or did you do it by ear only ?

So if I were to do the same with EL34s, how would I find the right bias range ?

[daveweyer]

You can use a scope to tell, but your ears will also hear that when you get near cutoff the amp will appear to have no sustain, the lower level guitar signals will just disappear and the amp will go silent, with a little grit when it cuts off.
Specifically, class B operation requires the opposing output tube to be in cutoff for a substantial portion of the wave, which means less than 5ma of current.
Generally, class B output stages also ran the output tubes with positive grid current at the peak of the cycle, which meant that you had to have a driver tube which could deliver power, but that doesn't happen with the audio power tubes designed for the Hi Fi market of the '50s; the point of their design was so that circuit designers didn't have to worry about power driver circuits, the tubes being able to draw their maximum cathode current with no grid 1 current.
So in Marshalls, Fenders and the like, all class B means is that the opposing tube goes into cutoff during its off portion of the push pull cycle. Obviously, when the tube is off, it draws no current and dissipates no power, and the efficiency of the circuit goes up. So either you get more power out, or you get more life out of the tubes, take your pick.
When the opposing tube cuts off there is a small switching transient that appears at the zero crossing of the waveform, notch distortion they call it, but it hardly matters when the amp is in full clipping. It just makes the distortion a little nastier.
Plate dissipation never really changed in the 6550 or KT88. It is a rating system which makes it look different. I have found that all these type tubes will get rid of about 40 watts of heat without turning the plates red. You can see for yourself by just turning one on in a dark room at 40 to 50 watts of power, and watching the plate color.
Adding a small fan increases the plate dissipation by removing conducted heat (not radiated heat) from the bulb.
Tubes with plate caps also can get rid of a bit more heat.

[daveweyer]

Gotta admit, "Castle made of sound" is a pretty catchy phrase.......

[C J H]

First off; Dave, sincere thanks to you for your contributions! Both technical and musical artifacts like that Kazani clip. Man, those divebombs at around 1:20 sound so sinister, give me the same kinda chills as horrormovies did when I was a kid.

I´ve been on the 6550-track for a while and have a quad laying around that I never made it to my amp (recently started playing again after a longer hiatus). Anyway, my JTM100 -build with a bass circuit is currently running el34´s and the PT puts 540 on the plates.. Stuck with examstudies atm but will open ´er up this weekend, piggyback the bias dropper to get it down to 15K and, out with the el34´s and in with the 6550´s, set bias at 25ma and see what it sounds like.

I have a PPIMV on this one, does this in anyway alter how the class B "method" should be approached?

[shakti]

Dave,

what about the 12AT7? Is that a useful minor mod in V3 if everyting else stays the same, or do you find it necessary to modify any component values in the PI to make it worthwhile?

[daveweyer]

Any circuit or output tubes will be fine as class B. There are some possible issues if you are using screen taps on the OT, but they won't hurt anything electrically.
The only thing you may find is that the phase inverter/driver doesn't have enough swing to turn the output tubes completely on during the positive cycle. If you lower the bias by 5 volts, the driver has to swing 5 volts more to get the same current in the output tubes.
That's where the 12AT7 comes in. It is a much higher transconductance tube, and will draw more current for the same input voltage. So if you run out of swing, just pop in a 12AT7 and see what happens. Then, if you like where it's headed, adjust the cathode resistor (the one with all the voltage across it, usually from 70 to 100 volts). Both Fender and Marshall used a 470 ohm resistor directly from the cathode to the junction of the grid bias resistors and that set the bias current in the tube. The 10K to 22K resistor from that junction down to the NFB injection point determines the ratio of positive to negative swing in the driver stage. Adjust for the maximum wave amplitude on each phase so they clip about the same time, in other words send an equal amount of signal to each output tube pair.
Marshall used a 5K pot from the 10K to ground, but sometimes bypassed that with another 4.7K. All this depends on the plate voltage available for the driver, the purpose being the maximum swing from the driver plates at various plate voltages. That lower value also determines how much range is in the presence control. The more negative feedback you have at that junction, the more of it is available for the capacitor to short out at maximum presence.
Fender abandoned the 12AX7 driver idea completely in favor of the 12AT7.
All this is nice, but you have to follow your ears and hands to get the sound and feel that really makes it for your style of playing.

[OnTheFritz]

Dave, I still don't understand the "class B" output, and how an A-B amp with a low bias achieves this. ?
Obviously I'm missing something, especially regarding under-bias, what I understand to be responsible for odd-order harmonics that are unpleasant.
Not trying to be a pill, just honestly curious.
Thanks in advance for your reply.

[daveweyer]

Onthefritz,
An amp in class AB is just an amp in which the output tubes live in near class A most of the time, in other words they are never in a complete cutoff condition. They are also never in a complete Class A condition where they draw the maximum plate current at idle. In class B they go into complete cutoff during the half cycle in which they are not contributing to the transfer of power from the supply through the output transformer.
That's really the key. In class AB the tubes will still draw quite a bit of current even when they are not contributing to output power. This is supposed to reduce distortion, and it does. It also limits the power output of the amp, and dissipates more power in the tube, power which is not producing any output.
Now here is an important concept; even harmonic distortion becomes odd harmonic distortion whenever there are two frequencies going through the amp which produce sum and difference frequencies that are odd multiples.
So if you have ANY distortion, you have both odd and even harmonics!
I know there is a whole world of even verses odd harmonic distortion debate, but if you can see the simple fact that any distortion produces both even and odd harmonics it will help to understand the significance of the debate.
Class B has some different harmonics than class A, but the measure of its value is really how all those sums and differences strike the ear.
One thing is important here; Class B stages, in which one of the tubes is in cutoff condition while the other one is in full conduction, offer the highest possible efficiency in an audio amp. And that means extended tube life.
I allow just a small amount of conduction at idle, just to keep the amp sounding smooth. That's the 25ma range.
So I don't get the highest possible efficiency, but about 5% away from it.
Anyway, if this still doesn't make sense, just ask again and I'll try to do better.

[OnTheFritz]

"Class B has some different harmonics than class A, but the measure of its value is really how all those sums and differences strike the ear.
One thing is important here; Class B stages, in which one of the tubes is in cutoff condition while the other one is in full conduction, offer the highest possible efficiency in an audio amp. And that means extended tube life."

But isn't that A-B, push-pull?

[C J H]

Ok, so I took a long needed studybreak last night and decided to give this a try. Ended up with this;

- 520ish platevoltage, I don´t know why this was lowered from the orginal 540.. I did change the cathode cap on v1, same value (330) but from a sprauge to a BC, same with bias caps, sprauges to BC. Only reason was that those sprauge caps came from an old build and looked a bit hacked (one without insulation and so on... ). I also changed V1´s plate resistor from cc to cf (didn´t have those on hand when building this amp). Changed cathode resistors on V2 to nos allen bradley CC´s (used to be metalfilm). All these changes I´ve been intending to do for a while since I was "replicating" a specific amp when first building this, had nothing to do with the 6550´s though.

- As I said before this is a JTM100-style build, build after one I´ve seen with all caps below, blackflagpanel but singlerectifier. 47uf mains, 23,5uf screens and PI, dual 32 on board (v1 and v2). It has 27k negative feedback (can´t remember on which tap though.. Super-Basscircuit but currently with no cap on the mixers (PA ?). All this adds up to a kinda "mellow" top end, it still cuts trough but sounded very "pleasant" for lack of a better word. Think Kossoff rhytm with a good Lester and creamy, smooth leads in spades.

- Piggybacked a 33k cf on top of the 27k in there for a total of 14.85ish K on the bias dropping resistor. Biased at 28 ma, couldn´t get lower and didn´t have appropiate resistor at home.

- The 6550´s i have in now are Tung-sol RI´s (russian).

- Fired it up last night but it was late so couldn´t really play. I did today though...

The verdict:

The 6550´s don´t sound stiff at all, which is was I was fearing; a too "clinical" sound. They do however stiffen up the lows and add attack in the highs. I would call the amp more percussive now than with el-34´s. Regarding the class-B or A/B push/pull I can´t really add to that discussion-- but it sounds darned good! There is, as Dave said, a different harmonic response, kind of like the musical overtones you get from a good set of pickups. When you listen to Jimitunes and you hear him making "octave" sounds through just the fuzz? Thats the harmonics I´m talking about.

When talking about effects. Fuzz sounds different (RM-axis fuzz), fatter and not as piercing as before. Octavios sound f-cking huge! Wah´s is about the same if maybe a bit more muffled, this might be due to how I set the other effects now though. I haven´t really figured out the vibe though (sweetsound mojo) which sounds huge when it has some speed but the slower settings aren´t shining through the way it used to. Again, probably a tuning issue so I´ll open it up when I get the chance..

To sum it up: I´m a believer, this is way "closer" than any other combos I´ve tried (most of them.. 10x series w/el34, split cath 12x w/el34, later superbass) and does get the jimisound indeed! And Xplorer, yes the BOG tone, according to my ears, is in here for sure. It´s no onetrickpony though, still sounds real nice with a Les Paul, on its own or with a touch of OD.

Dave, thank you again for your input. Next I´ll try the PI 12at7 mod.. But for now I really need to get back to studying.

[daveweyer]

Did anyone notice the reflections off the 6550s? Bright band on top, dark band below, then reflections off the shoulder of the bottle, which in the American tubes had flashing as well.

OTF, back for a moment to class AB and class B. There is another element to the story regarding the use of class B in Marshalls. I tried to explain class B as an operating system where the tubes are biased right next to cutoff or right at cutoff, and since I left 25ma of current in the tube at idle there is still a bit of class A current in the tube at idle. But it is only a couple of driving volts away from cutoff. I left this much current so that the amp wouldn't sound distorted at very low volumes, the guitar string sustain tail, for instance, can get cut off if you bias the tubes any lower.
So if we get very technical at the idle current situation, the amp is mostly class B with a tiny bit of class A at idle.
We could make a special name for it, class BA, where the amp is 90% running in class B and 10% in class A.
Sometimes this operation is called class AB2, but that designation also denotes positive grid current on the output tubes at full swing, and we have none of that in a Marshall--actually a 12AX7 won't even bring the output tubes to zero bias at full swing, it doesn't have enough current.

So you are right to want to include class A somewhere in the designation of the output form. The point is that it is such a very small percentage of class A that it can be put aside for the general discussion.

But here is the variable: when the amp is cranking, the peak value of the power supply voltage is reduced substantially, from 1.414 times the AC RMS transformer voltage, to closer to 1 times the AC RMS transformer voltage, depending on the regulation of the transformer, the size of the main filter caps, diode losses, etc.

When this happens, suddenly your output stage has only 470 volts on the plate and screen instead of 540. At this operating point, the bias voltage used to get 25ma at idle, now gives you no current in the tube at the zero crossing point, meaning the output tubes are operating in complete class B. So for all practical purposes, especially for crunch operation, you have a class B amplifier.

Is there more distortion in class B? Yes. Is it even or odd harmonic distortion? Mainly odd.
So let me rehash this for a minute if you can bear with me: One of the main purposes of using push pull operation was to cancel distortion in the output stage, it works the same in any push pull stage; the EVEN order harmonics are cancelled or substantially reduced, leaving only the odd order harmonics in the resultant signal.
So how does that square with the love of even order harmonics in tubes? It doesn't.
Let's take it one step further, if you have ever measured a square wave with a selective voltmeter, you will have noticed that all the harmonics are odd order.
So when you run your Marshall up into clipping, especially square wave clipping, you are generating a signal with nearly all odd order harmonics!
So should the generation of more odd order harmonics in class B be of concern for Marshall lovers? It doesn't seem to make sense.

Really, the only place to generate even order harmonics is in the class A stages which are not push pull-- but that operating point must be below where intermodulation distortion creates sum and difference frequencies of nearly equal magnitude to the generator frequencies, which of course brings in all the odd order harmonics.
Any time you hear a stage start to clip, you are forming a square wave top, and introducing all the odd order harmonics.
So I'm going to conclude that Marshall lovers also love odd order harmonics.

[Xplorer]

very cool !!! please, post some clips !

[bill bokey]

Great rig, C H J !!!
Thanks for sharing ! My 6550s should be here tomorrow... can't wait !!!!

Thanks Dave, I'm beginning to understand !

[daveweyer]

OnTheFritz,
I was just looking in the Radiotron Designers Handbook under Class B amplifiers to see if there were any passages to make the meaning of class B any clearer.
I'll quote from page 587, "A class B amplifier is an amplifier in which the grid bias is APPROXIMATELY equal to the cutoff value............. and so that the plate current in a specific valve flows for APPROXIMATELY one half of each cycle when an alternating signal voltage is applied." Note the specifier, "when an alternating signal voltage is applied'. Emphasis mine.
And on page 589, figure 13.47, "Composite characteristics of class B amplifiers, (A) biased to cutoff (B) biased to the point of minimum distortion."
What I did in Jimi Hendrix's amps and many other similar amps, was bias the class B stage for minimum distortion instead of absolute cutoff, as in graph (B), figure 13.47, page 589 in the Radiotron Handbook.

You can comfortably call it a Class B amp if the output tubes are in cutoff condition for over 2/3 of their off cycle, which they certainly were in all the Jimi Hendrix amps that I worked over.

Anyway, I hope this is more clear than not.

[OnTheFritz]

OnTheFritz,
I was just looking in the Radiotron Designers Handbook under Class B amplifiers to see if there were any passages to make the meaning of class B any clearer.
I'll quote from page 587, "A class B amplifier is an amplifier in which the grid bias is APPROXIMATELY equal to the cutoff value............. and so that the plate current in a specific valve flows for APPROXIMATELY one half of each cycle when an alternating signal voltage is applied." Note the specifier, "when an alternating signal voltage is applied'. Emphasis mine.
And on page 589, figure 13.47, "Composite characteristics of class B amplifiers, (A) biased to cutoff (B) biased to the point of minimum distortion."
What I did in Jimi Hendrix's amps and many other similar amps, was bias the class B stage for minimum distortion instead of absolute cutoff, as in graph (B), figure 13.47, page 589 in the Radiotron Handbook.

You can comfortably call it a Class B amp if the output tubes are in cutoff condition for over 2/3 of their off cycle, which they certainly were in all the Jimi Hendrix amps that I worked over.

Anyway, I hope this is more clear than not.

Thank you Dave. Appreciate the input and explanations. Just got home from work and need to also read your previous post regarding this.