• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Output tube interchangeability - using different types of tubes in an amplifier???

Xulonn

Major Contributor
Forum Donor
Joined
Jun 27, 2018
Messages
1,701
Likes
5,473
Location
Boquete, Chiriqui, Panama
Output Tubes.jpg

Although there has been some discussion here at ASR about tube-rolling and it's usefulness or worthiness, none of those discussions focused on output tubes. The big, power-hungry tubes that feed output transformers (or not, in the case of OTL amplifiers like Atmasphere) have been mentioned, but not discussed in detail. I am aware that some amp designers and manufacturers push them to the limit for more power (which reduces life-span), some design for multiple tube type compatibility (which I assume limits the optimization for performance), and some design for excellent performance combined with lowered stress and long tube life (as with the classic McIntosh amplifiers).

It is fairly common knowledge that power transformers are a primary limiting factor, because the more powerful tubes - like the KT120 and KT150 - draw lot more filament current than an EL34 or KT88. Indeed, if I put KT150s in an KT88 amplifier that was designed with enough current capability for KT150's, will them amp become as powerful as one designed specifically for KT150s?

Some of the parameters for multiple output tube-type compatibility, including the fact that EL34, 6CA7 and KT77s are very close in specs, and seem to be very interchangeable. However, I don't know enough about the performance curves of those related tubes to understand optimization options and rules. Surely there are some parameters for optimal performance and life span that include running them conservatively.

I am awaiting the arrival of a quad of Electro-Harmonix 6CA7s next week to replace the no-name Chinese output tubes in my Yarland Class-A PP EL34 amplifier. Like the original 1960s Sylvania "fat bottle" 6CA7s (which, were designed specifically to provide a domestic American option to the British EL34), the Russian-made Electro-Harmonix 6CA7s, like the 6L6 tubes they are said to emulate, are "beam tetrodes" rather than pentodes. OTOH, most other modern 6CA7s, including JJ versions, seem to be pentodes. What's the practical difference between the two types of tubes with their different number of grids? How can a single tube type (6CA7) have a different number of grids and be either a tetrode or a pentode, depending on the brand?

Plate voltage and current ranges for EL34 KT88, KT90, KT120, KT150 are quite different. How can putting EL34s in and amplifier optimized for KT88, or KT150s in an amplifier optimized for lower-power KT88's be an intelligent option? And of course, I can imagine that putting KT88s in an amplifier optimized for KT150s would not be good - or safe!

In audio, even with vacuum tubes, most knowledgeable music lovers want flat frequency response and low noise and distortion, and many of us buy tube amplifiers for reasons other than purported "magical sound enhancement". However, guitar players are looking for other differences, and talk like this:

On a sonic scale, the 6CA7 tube is a hybrid between EL34 and 6L6 tube. Like the 6L6 and 6550, 6CA7 tube has a more pronounced low end. In a guitar amp, this makes for a tone that is more substantial. People often describe 6CA7 sound as heavier and darker than the EL34 tube.
and...
Yes, tetrodes are more glassy and HiFi than pentodes. Both technologies have their advantages.

As it happens there are no true kinkless or beam tetrodes - if the number of electrodes matters. The difference between a tetrode and a pentode is the 5th electrode which in the pentode is a grid and in the tetrode are two beam forming plates. That's why some tetrodes still are referred to as pentodes. As the 4th and 5th electrodes, which are essentially performance boosting, aren't used in triode connection tetrodes and pentodes operate identically from an electrical POV.
 
Last edited:

atmasphere

Member
Manufacturer
Joined
May 26, 2021
Messages
75
Likes
179
However, guitar players are looking for other differences, and talk like this:
They don't know what they are talking about. What's happening is how well the tube works with the output transformer, and the difference in feedback caused by the change in loop gain. This in turn affects output impedance.
Surely there are some parameters for optimal performance and life span that include running them conservatively.
There is. Get a copy of the RCA Receiving tube manual. 'Design Center' specs are for conservative operation. With regards to maximum plate voltage as opposed to maximum plate current, they should not occur at the same time. IOW if you stay within the parameters laid out in the manual you should be good.

IMO being able to roll different pentode types like the KT150 as opposed to a KT88 is a pretty good bet that the amplifier is always leaving performance on the table- IOW there's a compromise required to make that happen.
 

DVDdoug

Major Contributor
Joined
May 27, 2021
Messages
1,009
Likes
1,227
I barely remember tubes from when I was a kid... ;)

There were equivalents but I believe that was for general use, not for high-fidelity (or optimum fidelity).

It's probably best to check with the amplifier manufacturer. I wouldn't trust 3rd-party recommendations unless they've done measurements with the particular tub & amp combination. ...I'll give guitar players an exception since they are usually looking for "good distortion" and that's not really measurable.

Some circuits are more sensitive to tube changes than others. And, in some cases you might have to adjust the bias when you change tubes (even the same part number) or you might even have to tweak the bias as the tube ages to maintain the same performance.

Other circuits are more immune to part variations. IMO - That's how an good amplifier should be designed. You don't want the performance to change as the tube ages or if you have to replace it. Almost all solid state audio circuits are designed to allow-for part variations... Every MOSFET or transistor is going to have different gain but usually there are no pots (or tweaked/selected parts) and the circuit as a whole is designed to work properly as long as the parts are within spec. And as a bonus, solid state devices don't "age"... Sometimes they "die randomly" but they don't deteriorate over time.
 

dfuller

Major Contributor
Joined
Apr 26, 2020
Messages
2,026
Likes
2,801
They don't know what they are talking about. What's happening is how well the tube works with the output transformer, and the difference in feedback caused by the change in loop gain. This in turn affects output impedance.
They also clip a bit differently - beam tetrodes have a noticeably different behavior to pentodes re: screen current.
What's the practical difference between the two types of tubes with their different number of grids?
The beam tetrode was invented to sidestep the Philips patent on pentodes. Time to get a bit in depth...

The original amplifying tube was the triode. 3 elements, plate, cathode, and a grid between. It works well, but has limited voltage gain (you'll rarely see a maximum gain above 100). Plus, it has a relatively high grid to plate capacitance, making it unsuited for RF duties.

Next was the tetrode. It has an extra grid between the grid and cathode, called the screen grid. It's held at a positive voltage, but usually not nearly as positive as the plate. Its goal is to screen the control grid from the plate, thereby increasing the gain and improving its behavior at high frequency. The problem is, you can develop a negative resistance kink in the transfer function, which makes the tubes unsuitable for amplification (they make just dandy oscillators, though). This is caused by secondary emission from the plate - the electrons from the cathode are accelerated to such a velocity that they knock more electrons off the plate. The positively charged screen grid can attract them, and this causes a drop in plate current despite an increased voltage.

So, to solve this, some genius at Philips (the huge Dutch conglomerate - that one, yes) named Bernard D.H. Tellegen figured out that if you just stick a grid at cathode potential between the screen grid and plate, you could have all the benefits of the tetrode (namely very low inter-electrode capacitance and super high gain) with none of the bizarre negative resistance kink instability. The problem was solved, and pentodes became ubiquitous for most things (triodes stuck around too, they're much more linear sans feedback so good for small signal amplification).

The problem was, Philips owned the patent for the suppressor grid, and RCA (and others, namely EMI/GEC) didn't want to pay that royalty. So what they did instead was invent something really novel - the beam tetrode. Instead of using a 3rd grid, they instead used some clever design tricks (namely adding beam confining plates at cathode potential and aligning the control and screen grids) to form the electrons emitted from the cathode and pulled to the plate into beams, which formed a virtual suppressor grid. They have the secondary advantage of being more efficient too, because the screen grid draws less current by virtue of being aligned with the control grid (and in its "shadow", so to speak).
OTOH, most other modern 6CA7s, including JJ versions, seem to be pentodes.
This is a misprint on JJ's part - I've broken them open. They're beam tetrodes as are their KT77s. Their EL34s are true pentodes though.
 
Last edited:
OP
Xulonn

Xulonn

Major Contributor
Forum Donor
Joined
Jun 27, 2018
Messages
1,701
Likes
5,473
Location
Boquete, Chiriqui, Panama
A remarkable and thorough response, @dfuller . Again, the breadth and depth of knowledge among ASR members amazes me.
This is a misprint on JJ's part - I've broken them open. They're beam tetrodes as are their KT77s. Their EL34s are true pentodes though.

And the above quoted line tells me that there is still at least a couple of options for me if I want replacements in the future and the Russian EH6CA7s are not available.

One more question. I have no rational reason for wanting the 6CA7 beam tetrodes other than curiosity - and a desire to replace the supposedly unreliable no-name Chinese EL34s. Are the 6CA7s likely to change anything in the amplifier's measured performance? Distortion? Noise? Efficiency? Frequency response? Rolloff? (I'm not part of the veil-lifting crowd - but I think that is a cool thing tor traditional weddings.)
 

dfuller

Major Contributor
Joined
Apr 26, 2020
Messages
2,026
Likes
2,801
A remarkable and thorough response, @dfuller . Again, the breadth and depth of knowledge among ASR members amazes me.


And the above quoted line tells me that there is still at least a couple of options for me if I want replacements in the future and the Russian EH6CA7s are not available.

One more question. I have no rational reason for wanting the 6CA7 beam tetrodes other than curiosity - and a desire to replace the supposedly unreliable no-name Chinese EL34s. Are the 6CA7s likely to change anything in the amplifier's measured performance? Distortion? Noise? Efficiency? Frequency response? Rolloff? (I'm not part of the veil-lifting crowd - but I think that is a cool thing tor traditional weddings.)
I can't speak for hi-fi, but in guitar amps they have slightly more headroom before the onset of clipping, presumably due to the lower screen current causing less of a nonlinear behavior at extreme ends of their operation.
 
F

freemansteve

Guest
Having just changed valves/tubes in my guitar amp, from eletro-harmonix to JJs (de-Russianing you might say, and I needed spares to be available immediately), I can't say it sounds any different. Given that the thing is designed to add crunchy distortion, and that one is always playing slightly differently, through several pedals, and that a guitar has ~8KHz bandwidth, I can't see why anyone would care, or really be able to make statements like "glassy" etc - it's plain daft.
 

raindance

Addicted to Fun and Learning
Joined
Sep 25, 2019
Messages
747
Likes
657
Chinese EL34 tubes are pretty decent and reliable provided you don't run them too near their limits. In fact I'd rate the common Shuguang EL34-B above any of the JJ variants, for example, if you don't like fireworks in your music room. I see way to much derogatory stuff online about Chinese power tubes - I repaired tube gear for many years in South Africa, the UK and the USA. Their small signal tubes have issues with consistency, but the power tubes are OK.

The Shuguang 6550 tubes are also quite robust, but aren't substitutes in many amps.

Just adding my 2c.
 

dfuller

Major Contributor
Joined
Apr 26, 2020
Messages
2,026
Likes
2,801
Chinese EL34 tubes are pretty decent and reliable provided you don't run them too near their limits. In fact I'd rate the common Shuguang EL34-B above any of the JJ variants, for example, if you don't like fireworks in your music room. I see way to much derogatory stuff online about Chinese power tubes - I repaired tube gear for many years in South Africa, the UK and the USA. Their small signal tubes have issues with consistency, but the power tubes are OK.

The Shuguang 6550 tubes are also quite robust, but aren't substitutes in many amps.

Just adding my 2c.
Alternatively, in my experience with guitar amps (which generally run power tubes much harder than hi fi equipment), their small signal tubes are fine - the distortion is a little brighter, perhaps - but the power tubes are often suspect. Chinese EL34s have a reputation as firecrackers in guitar amps. JJs generally don't particularly care and are super robust, as are the EHX EL34s... which is unfortunate, given that you can't really get Russian tubes at the moment.
 

raindance

Addicted to Fun and Learning
Joined
Sep 25, 2019
Messages
747
Likes
657
Alternatively, in my experience with guitar amps (which generally run power tubes much harder than hi fi equipment), their small signal tubes are fine - the distortion is a little brighter, perhaps - but the power tubes are often suspect. Chinese EL34s have a reputation as firecrackers in guitar amps. JJs generally don't particularly care and are super robust, as are the EHX EL34s... which is unfortunate, given that you can't really get Russian tubes at the moment.
Yeah, the qualifier about running the power tubes hard is true - they won't work near their max ratings. In guitar amps, noise isn't as much of an issue, and what I have found with the small signal tubes is that they seem to be consistently more noisy, more microphonic and, in the case of dual triodes, less well matched in the same envelope. I used to use JJ for all small signal applications because, with the exception of the "long plate" versions of their tubes, they were quieter.
 

dfuller

Major Contributor
Joined
Apr 26, 2020
Messages
2,026
Likes
2,801
Yeah, the qualifier about running the power tubes hard is true - they won't work near their max ratings. In guitar amps, noise isn't as much of an issue, and what I have found with the small signal tubes is that they seem to be consistently more noisy, more microphonic and, in the case of dual triodes, less well matched in the same envelope. I used to use JJ for all small signal applications because, with the exception of the "long plate" versions of their tubes, they were quieter.
JJ's frame grid E83CC is non-microphonic almost to the point of comedy. You can whack that thing with a stick while it's running and it won't do anything. And yeah, the Chinese tubes seem to trend toward microphony, for some reason.
 

atmasphere

Member
Manufacturer
Joined
May 26, 2021
Messages
75
Likes
179
I have found with the small signal tubes is that they seem to be consistently more noisy, more microphonic and, in the case of dual triodes, less well matched in the same envelope.
Emphasis added
I don't find this to be the case at all! We use Chinese 12AT7s which are consistently lower noise and lower microphonics than any others we've tried and the sections well matched (our circuits are differential). Out of a box of 100 we reject about 10%. With JJs, about 40%. With Russian made, about 90%. We've had better luck with the 12AU7s; the Chinese is still #1 but the JJ version is also quite good and we're happy to use either.

The Shugang 6SN7 was also better than the Russian 6SN7s. The JJ tube was really pretty good, but was more prone to cathode/filament leakage and outright filament failures. Since the Shugang factory burned down JJ is really the only game in town and now they are at least 6 months back ordered.
 

raindance

Addicted to Fun and Learning
Joined
Sep 25, 2019
Messages
747
Likes
657
Emphasis added
I don't find this to be the case at all! We use Chinese 12AT7s which are consistently lower noise and lower microphonics than any others we've tried and the sections well matched (our circuits are differential). Out of a box of 100 we reject about 10%. With JJs, about 40%. With Russian made, about 90%. We've had better luck with the 12AU7s; the Chinese is still #1 but the JJ version is also quite good and we're happy to use either.

The Shugang 6SN7 was also better than the Russian 6SN7s. The JJ tube was really pretty good, but was more prone to cathode/filament leakage and outright filament failures. Since the Shugang factory burned down JJ is really the only game in town and now they are at least 6 months back ordered.
Interesting. Perhaps the issue was that the OEM small signal tubes used in China-produced amps are lower quality? Typically these are what I have been dealing with. I dealt mostly with 12AX7, 12AT7 and 12AU7. I had the most issues with the 12AX7, so perhaps I generalize too much... :)
 

ta240

Addicted to Fun and Learning
Joined
Nov 7, 2019
Messages
790
Likes
1,266
It is fairly common knowledge that power transformers are a primary limiting factor, because the more powerful tubes - like the KT120 and KT150 - draw lot more filament current than an EL34 or KT88. Indeed, if I put KT150s in an KT88 amplifier that was designed with enough current capability for KT150's, will them amp become as powerful as one designed specifically for KT150s?
Aside from the heater current draw there are other issues. Even if the power transformer and output transformers were rated to handle the current of the KT150 you would want a different plate voltage and need the bias on the tube to be adjusted to draw much more current through the tube to get the power out of the amp.

Without changing the plate voltage and bias the KT150 would be running well under its specs, which will increase its life but will also put you in a weird place on the load line curve where you could end up with not very good sound too.

I often see it referred to like underpowering them wouldn't be a bad thing, but as is shown here with a 6V6 you can impact the tubes performance quite a bit. The main tube amp I use is touted as one you can roll all sorts of tubes through but running a KT88 at the same specs as a 6L6GC doesn't sound good to me. At least they are okay with similar plate voltages and heater current so a simple switch to change the bias resistance can fix it in that situation. But something as powerful as a KT150 would be really hard to make interchangeable with a KT88 or lower power.
 

captainbeefheart

Active Member
Joined
Apr 18, 2022
Messages
276
Likes
357
The swapping of output tubes between completely different types can lead to different performance certainly. There are amps out there for "tube rollers" and they list just about every bottle possible as usable but without a question the output transformer is at a fixed impedance ratio, so there will always be a compromise between tubes, with a specific load there is most likely an optimal tube for that load which should outperform all others.

But are we discussing open loop? What happens with a split load design like the McIntosh circuit or quad II, the latter isn't an even split between plate and cathode but there is still feedback in the cathode winding. With the McIntosh circuit we are near 100% feedback, would other tube types perform as well? Again the loading might be slightly off but there is enough feedback in the entire circuit that I believe once bias is adjusted an EL34 may be close enough to the 6L6, even though the latter has higher plate impedance loading between the two isn't grossly different. I'd wager not a huge difference in output performance.

One thing I don't see discussed very often in amps that are pushed for "tube rolling" is I would think the amount of feedback would need to be reduced in order for stability purposes with each tube type. A well designed amplifier with feedback will have very specific compensation networks, the more feedback used or the greater the phase shift the more critical these compensation networks become. So for an optimized amplifier with one tube type you can easily use -20-30db of feedback and get very low distortion but changing tube types is not recommended for stability, if you relax the feedback to say -10db or less then the compensation networks aren't critical or even needed due to the lower amount of phase shift. With much less feedback you are either going to suffer performance or possibly include nested feedback loops, localized feedback to help reduce the overall global feedback needed. Myself I don't use high levels of global feedback, -20db is often plenty with the help of localized nested loops but I can't imagine using different power tubes for the sake of stability as mentioned, it takes me quite some time to work the compensation networks out so I don't want some end user to throw in some random power tubes and ruin the stability of my design. I'm old school, I don't know how other builders do it but I use Lissajou patterns to find the breakpoints in my amplifiers. Basically I hook my scope up in x y mode with probes at input and output of the amplifier, I'm looking for a very specific ellipse shape on my scope screen (45° ellipse), I increase frequency until I see that exact shape and mark down the frequency, this allows me to draw my Bode plot with the amplifiers break points. I find this to be far more accurate than using gain vs frequency measurements. After I have my break point frequencies I can calculate my compensation networks and install them. I still go through an empirical process where I adjust the values up and down to try and dial in but often the calculated values are just about perfect and lead to very nice square wave response.

Crap I just let out my secret method ;)
 

egellings

Major Contributor
Joined
Feb 6, 2020
Messages
1,801
Likes
1,325
Different plate characteristics, bias points, etc.--not directly interchangeable if optimum performance is sought in any one setting.
 
Top Bottom