• TalkBass has been independent since 1998. Add your voice.
    Create a free account to reply to discussions, view embedded media, and browse with fewer display ads.
    Join free Log in
    Want zero display ads or expanded classifieds tools? Compare plans.

SVT-CL: Mixing Power Tubes

#101
Great topic. This has got me digging and googling. Brain muscle is getting tired!

B-string said:
Cathode bias, not feedback. In cathode bias the tube self biases through a resistor and bypass cap. Lower power option with interesting distortion character, no feedback loop. Basic triode to a transistor; Plate-collector, grid (control)-base, cathode-emitter.
Click to expand...

I'm not sure if we're talking about the same thing, or not. Just for clarification, if the cathode R is unbypassed, then you automatically have cathode feedback (also known as cathode degeneration). The SVT schematic linked to by iiipopes (on page 2?) shows unbypassed cathode resistors. Thus, cathode feedback. (Don't know about the amp shown on page 4 -- its cathode resistors are off-page).

I've pretty much convinced myself that cathode feedback lessens the effect of transconductance mismatch upon gain (Check this link: http://www.angelfire.com/planet/funwithtransistors/Book_CHAP-4A.html . If you are masochistic, scroll down to section 4A.3 "small signal analysis" and take a look at the equation for gain without the cathode bypass cap (4A.10.10). First, keep in mind that gm (transconductance) is directly proportional to mu, so when we talk about gm variation, that is equivalent to talking about mu variation. Now, back to that equation... you'll see mu on top, and mu+1 on the bottom, so in the ideal case where the sum of Rbc+rp is less the mu*Rk, and mu is, say, 10 or more, mu (and thus transconductance) drops out of the gain equation and the gain becomes simply -Rbc/Rk (exactly analogous to a transistor amp with emitter degeneration). Gain is now independent of mu (and thus transconductance).

Now, I can't speak to how much less Rbc+rp actually is to mu*Rk (and this might be where my flaw is -- those who have studied this stuff, help!!!). But I will say that paralleling tubes has the effect of reducing rp by the number of tubes paralleled. Therefore, in an amp with 3 tubes on a side, rp on each side is reduced by a factor of 3.


beans-on-toast said:
There is something else going on. In part it has to do with how the tubes interact with the transformer and noise. Check out this link at the end, the paragraph starts with "Transconductance matching is important for AC (signal) balance in the output stage." There are other useful points made in this document.
Click to expand...

Thanks very much for the link -- very interesting!

Regarding the paragraph you pointed out, the next two sentences are also important. They state:

"A push-pull amplifier has inherent power supply hum rejection and common-mode input rejection due to the symmetry of the output stage. If one side has a different amplification factor than the other, this symmetry is lost, and the amplifier won't be able to reject power supply hum and noise as well."​

I agree with this, but I would like to dig down a bit deeper and clarify what they're taking about.

1. Most of what that section discusses is the necessity of matching the gains of the two "halves" of a push-pull amp (per iiipopes comments). They really aren't talking about matching transconductance of multiple tubes within a side (at least, not per my reading), and I completely agree with what it says, at far as it goes.

2. They also don't discriminate between amps with and without cathode degeneration. I believe there is a difference between the two. If the cathode resistors are not bypassed, they are a form of negative feedback. As such they also act to equalize the gains of the two side of the push-pull amp and thus lessen the effects of transconductance mismatch between the two sides. In other words, matching transconductance has just become less important.

So what this means to me: if the goal is to have the two sides of a push-pull amp have equal gains (to prevent even-order harmonics, etc.), then, in amps without cathode feedback (that is, amps in which there is no cathode resistor, or the cathode resistor is bypassed with a cap), you must match transconductance between the two sides of a push-pull amp in order to match gain.

But in amps in which there is cathode degeneration, matching transconductance between sides becomes less important, as long as mu*Rk >> Rbc+rp.

Anyway, maybe I'm completely wrong or I'm missing something. But this makes sense to me.
 
#102
The vintage SVTs had a hum balance for the heater circuit and a power tube balance. The SVT-VR has the power tube balance. They eliminated the heater balance in the new amps. They don't have the power tube balance on the SVT-CL for some unfortunate reason.
 
#104
beans-on-toast said:
The vintage SVTs had a hum balance for the heater circuit and a power tube balance. The SVT-VR has the power tube balance. They eliminated the heater balance in the new amps. They don't have the power tube balance on the SVT-CL for some unfortunate reason.
Click to expand...

Yeah, looked at my schematics and saw it was probably just the CL. You can do something similar with the CL after setting the bias. You very slightly adjust one bias control to null any hum and noise you hear with the gain up and no input. Very slightly means you should hit this in well under 1/8 turn on the bias pot, or else something else might need looking at. Both bias lights will remain green when this is done correctly. The better the match in at least triplets of tubes, the closer and deeper the null will be.
 
Last edited:
#105
B-string said:
jcanderson, you will need to pick a specific amplifier operation class before bypassed cathode, cathode resistor or grounded cathode operation can be fully addressed. Class A/B, A/B1 or A/B2, as an example are grounded cathode and any resistor is generally for test measurement purposes.
Click to expand...
Hi B-string,

Good point about the resistor -- I was going by the SVT schematic linked by iiipopes, and, although I could see it had unbypassed emitter resistors, I couldn't read their values (schematic was pretty illegible). So I was assuming it was cathode degeneration and that's the rabbit hole I went down.

Just looked up a schematic that's actually readable, and the cathode resistors are 1 ohms. To your point, this is a value you'd use for measuring cathode current, NOT for cathode feedback.

Ooops! My bad!

So with this new info in mind, I'd say: certainly match transconductance between sides so that their gains are equivalent and thus the output waveform is symmetrical (no even-order harmonics). And match currents within a side so that no one tube is dissipating more power compared to the others on that side. Won't hurt to match transconductance within a side, too, but I'm not sure what's gained, as long as the overall transconductance of a side matches the overall transconductance of the other side.
 
Last edited:
  • Like
Reactions: B-string
#106
...and I'll add...iiipopes technique of matching the gm of pairs of tubes and splitting the pair between the two sides of a push-pull amp should result in the two sides of the amp having equal transconductance (because overall transconductance of each "triplet" side is gm1+gm2+gm3), should matching transconductance be your goal.
 
#107
The only "fault" iiipope had was in considering the output tubes as running in "pairs". Yes you can run a like imbalance in the triplets of both halves of the waveform and the net should be symmetry. Or you can leave a triplet balanced within itself and adjust for the imbalance of the other half with bias adjustment.

Semantics in the wash as we are talking a M.I. amp and not a lab amplifier. Pure sine waves are hardly ever the end result.
 
  • Like
Reactions: FenderOli, okcrum, beans-on-toast and 1 other person
#108
jcanderson said:
I've pretty much convinced myself that cathode feedback lessens the effect of transconductance mismatch upon gain
Click to expand...

Thanks for the link. Based on you comments, you might be interested in buying this book. It goes into great detail in the analysis and operation of the vintage Fender Bassman. They perform a network analysis of the circuit and compare the estimated plots to measured ones. Not an easy read but very interesting. There is also two books on pre-amp and power amp design that are full of examples. I've read all three books. The web site touches on some of the topics covered as well as an analysis of a number of classic circuits. Worth checking out.

One thing to keep in mind, cathode bias is typically only used in low power amps. Most often amps over around 25W are fixed (grid) bias. Of course there are exceptions. The amp we are discussing is a fixed bias, not cathode biased. The cathodes have a small 10 ohm cathode resistor that is used to monitor the cathode current for setting the bias. In this case, the resistance is too small to be doing any cathode regeneration that matters. The voltage across the resistor is only a fraction of a volt. In the case of a 6L6 push-pull pair in a cathode biased amp, the resistor is typically 250 volts, the capacitor is 50uF and the voltage across them is around 25V. With the SVT-CL, the fixed bias voltage that is applied to each grid is around -44V. When looking as the equations it is good to look at the real numbers to get a better handle on what is going on.

The schematic link to on page two is not for the SVT-CL that we are talking about. It is for a vintage version of the SVT. But it's close enough and isn't an issue. The vintage amp has the cathodes in each trio tied together through a single 1 ohm cathode resistor. Again the resistance is too small a value to affect the bias.

The three tubes in each trio are connected in parallel and their cathode currents are summing. They are effectively working as a single unit. Matching is performed to a tolerance so the tubes aren't perfectly matched. The cathode currents may be a little higher or a little lower, but they balance out. The same applies to the other trio. You can do an interesting experiment if you have an IR thermometer. If all the power tubes are the same model, the bulb temperature can be measured and it's proportional to the heat being dissipated as a result of the current flowing through the tube. The temperature can be used to estimate how closely each tube is matched. It is an interesting technique to get a ball park view of what's going on.

Concerning parallel tubes. The general design rule of thumb is transconductance and inter-electrode capacitance is multiplied by the number of tubes connected in parallel; the plate resistance is divided by the number of parallel tubes.

Concerning cathode degeneration. It is hard to get into the formulas here. But cathode degeneration does lower the transconductance. How much so depends on the impedance of the resistor and parallel capacitor if there is one. We are back to noting that the real life value of the resistor in the circuit is important when considering what is going on. The input impedance of the next stage (transformer in this case ) affects things as well. The transconductance of the tube will vary as the voltages and currents swing. Tubes with different transconductance will not affect things in a proportional way. There is less of an effect because of the negative feedback than an amp with a bypass capacitor. So in effect, the effect of transconductance differences in diminished as you said. Transconductance differences are less important when cathode degeneration is in play. There are a number of advantages to cathode regeneration. The drawback is that it reduces the voltage gain.

In the case of the SVT-CL, the amp is fixed bias push-pull design and cathode degeneration does not play a significant role so transconductance and cathode current matching is desirable. When buying a matched sextet, you want the tubes to be matched as closely as possible. Some sellers do not match transconductance, only cathode current. It's easier to group tubes together when you are only looking at a single parameter so it is more cost effective for them. I'd want to deal with someone who has the attention to detail to do both. Ideally you want the matching done at the plate voltage that the tube operates at. A tube that is matched while being pushed produces more realistic matching results.
 
Last edited:
  • Like
Reactions: jcanderson, coreyfyfe, FenderOli and 1 other person
#109
beans-on-toast said:
Thanks for the link. Based on you comments, you might be interested in buying this book. It goes into great detail in the analysis and operation of the vintage Fender Bassman. They perform a network analysis of the circuit and compare the estimated plots to measured ones. Not an easy read but very interesting. There is also two books on pre-amp and power amp design that are full of examples. I've read all three books. The web site touches on some of the topics covered as well as an analysis of a number of classic circuits. Worth checking out.
Click to expand...

bean-on-toast, thank you very much for the link to Keuhnel's book! Looks like exactly what I want, and I've just ordered it from Amazon, along with his two other books.

And thanks, too, for your explanation regarding aspects of tube amplifier operation and tube selection -- you've described pretty much what I would expect, now that I've dug into the details a bit more.
 
  • Like
Reactions: beans-on-toast
#110
jcanderson said:
bean-on-toast, thank you very much for the link to Keuhnel's book! Looks like exactly what I want, and I've just ordered it from Amazon, along with his two other books.

And thanks, too, for your explanation regarding aspects of tube amplifier operation and tube selection -- you've described pretty much what I would expect, now that I've dug into the details a bit more.
Click to expand...

I hope that you enjoy them. I've been meaning to order the ohm's law book. I have some of his original papers on the subject and they are fascinating.

A point that I should have mentioned earlier about tube matching is that how tubes are matched also depends on the amp that they are going into. A tube that's going into a fixed bias amp can be dialed in by adjusting the bias voltage to provide the required current flow. It helps if each power tube has it's own bias adjustment pot. That same tube might not work as well in a cathode biased amp where the bias is defined by the circuit. Cathode biased amps are more what you would call plug and play because they are self biasing and manual adjustments are not made.

Also, some people like the added harmonics and distortion that comes with unmatched tubes. They use this to help define their tone. Clearly not acceptable in a hi-fi application but musical instrument amps are in a class of their own.
 
Last edited:
  • Like
Reactions: okcrum and jcanderson
#111
OK, now that you guys have taught everyone else about all the considerations of AB1 tube output stages, how many tubes at a time (if less than six) do you replace in an SVT-CL? You are required to use only the LEDs for setting bias, simulating an actual user.

B-string, I'm already pretty sure you'd replace 3. ;)
 
Last edited:
#112
If the tubes were old, I would replace all six in a CL because you want them to be balanced. This helps minimize THD. If you just replaced three, it could be noisier. But that noise may not be an issue in a powerful amp that is played so loud. So replacing three is not so bad. In a smaller amp, perhaps used for recording, it would be a different story.

If a single tube blows, I would replace it with one that matches the others as close as possible. A perfect match is good but you don't have to be nuts about it.

That's why, when you change your tubes, keep the old ones. You never know when you'll need one to pop in.
 
Last edited:
  • Like
Reactions: okcrum
#114
Pretty much agree with Beans. Just a bit more clinical on my views the socket that the tube failed I would place a sticker with the date. If the tubes have light use, one with a matching number from the same "house". If the tubes have lots of hours on them the whole set of six (socket still labeled). This all for new production tubes. I have been doing this enough decades (like many others here) that taking two (or even one) operational tube(s) out of service just because, in a musical instrument amp is a hard case to justify.
 
  • Like
Reactions: beans-on-toast and okcrum
#115
JimmyM said:
Sounds like a good plan on the surface, except for one small detail...power tubes in SVT's are tied together in triplet sets, not pairs ;) But hey, doesn't matter to you now.
Click to expand...
hi jimmy i have a question.i have an svt and i need tubes because the one tube is out.the tubes are 6 or 7 years old .here in greece i cant find 6 matched tubes.can i take two sets of four and put them in each triplet and have spares the other two?is it good?thanks
 

Latest posts

Top Bottom
Back