Tubes for amps- do they sound the same?

[SineWave]

I got my PrimaLuna integrated tube amp about 20 years ago. Since then, I have figured out, from personal experience, that inexpensive (say > $250ish )amps and CDPs sound the same as more expensive ones. I was pretty sure this was the case even in my teen years, in 1985, when I only owned a set of $200 Dali IV speakers. This got me thinking, do different tubes sound different?

The PrimaLuna uses four EL34 tubes and 2 no longer glow. So I replaced them with the Electro-Harmonix brand (I think this is the brand the amp originally came with but there is no brand labelled on them). Do all brand tubes sound the same? JJ, Tung-Sol, Mullard are some of the other brands I saw available as EL34. These are in the $17-$50 range, but I saw some go up to $300. I paid $39 each. If they all sound the same, next time I'll just get the lowest cost replacement tubes. Thanks in advance!

 

[MattHooper]

I wish ;-)

 

[DVDdoug]

It depends. I think someone here on the forum did some experiments.

Tube characteristics do very from tube-to-tube, not only brand-to-brand. But a good amplifier should be designed to work well and stay in-spec as long as the tubes are in-spec. Transistors & MOSFETs vary too but they don't age and eventually die like tubes. Again, a good design handles these normal variations and a high-fidelity amplifier is supposed to a "straight wire with gain".

Some amps my require a bias adjustment when the tubes are changed but that should be the exception and it should say that in the owner's manual.

Some tube amplifiers are designed to have a "tube sound" (distortion?) and the sound may change with different tubes or as the tubes age.

Guitar amplifiers aren't supposed to be high fidelity. They all sound different so most guitar players have their favorite amp as well as their favorite amp. Tube guitar amps are popular because tubes tend to soft-clip when overdriven for a "pleasing distortion" (but you wouldn't want the whole band and the singer to have that distortion). A guitar amp's distortion characteristics my, or may not, change when the tubes age or are replaced.

P.S.
Tube Rolling - Does it make a difference? (In this particular experiment, it did not.)

 

[levimax]

I got my PrimaLuna integrated tube amp about 20 years ago. Since then, I have figured out, from personal experience, that inexpensive (say > $250ish )amps and CDPs sound the same as more expensive ones. I was pretty sure this was the case even in my teen years, in 1985, when I only owned a set of $200 Dali IV speakers. This got me thinking, do different tubes sound different?

The PrimaLuna uses four EL34 tubes and 2 no longer glow. So I replaced them with the Electro-Harmonix brand (I think this is the brand the amp originally came with but there is no brand labelled on them). Do all brand tubes sound the same? JJ, Tung-Sol, Mullard are some of the other brands I saw available as EL34. These are in the $17-$50 range, but I saw some go up to $300. I paid $39 each. If they all sound the same, next time I'll just get the lowest cost replacement tubes. Thanks in advance!

It is a little hard to generalize as it depends a lot on circuit the tube is in. For many "standard" circuits with negative feedback the difference in sound between tubes is going to be minimal. In some case though if there is not much of any feedback in the circuit some tubes can sound different. I believe the PrimaLuna is a pretty standard design so the sound of different tubes should be minimal.

While the sound of tubes is usually not a big deal the quality of the tube i.e. how long it lasts and how resistant they are to catastrophic cascading failures is another story. Almost all of the "modern" tubes are made in Eastern Europe or Russia and have little if anything to do with the brand names they are sold under (the brand names were bought and in some cases along with some of the original equipment). In addition to the equipment being 50+ years old some of the materials used in vintage tubes are no longer available due to environmental concerns. The end result is that modern tubes in general do not last as long as original NOS tubes. An extreme example are vintage NOS Mullard tubes. They cost a fortune these days but are much more durable than current production and can withstand much more abuse and more challenging operating points. In my experience a vintage Mullar rectifier tube will outlast a cheap modern rectifier tube by a factor of at least 10 and more like 100 times. NOS Mullard power tubes will outlast modern equivalents by are large margin especially if pushed hard. Whether or not they are worth the price depends on a lot of factors but for rectifier tubes at least, the failure of which can take out the output tubes, I find them well worth the cost. Output tubes not so much.

 

[Keith_W]

Of course tubes have the potential to sound different, and even different tube manufacturers might have a different sound.

I had a conversation with an audio engineer who worked on tubes. He told me that what nobody ever discusses are manufacturing variability between tubes and between manufacturers. He has never found a pair of so-called "matched tubes" to measure exactly the same.

Why do tubes from the same manufacturer sound different? Because the effect of the multiple tolerances required is cumulative. Are the plates separated by the same distance? Is the purity of material consistent across batches? Are the plates warmed to the same temperature? Is the vacuum in every tube the same? How much deviation from tolerance does the manufacturer's QC permit before junking the tube?

Why do tubes from different manufacturers sound different? Because they all have different interpretations of compliance to the published specification, might use different plate voltages, heater temperatures, etc. He gave the example of the 300B tube, one of the most well known tubes in audio, and then he showed me voltages and current draw he measured from various 300B's from his tube tester. He said that if you were to put them on the same graph, you would see a normal deviation from a manufacturer representing in-house variability, with the standard deviation representing manufacturing quality and QC, and multiple normal deviations from each different manufacturer. This is the same with any electronic component, or everything that has been manufactured, except that the high manufacturing cost of tubes and the nature of tubes means that the manufacturers accept a higher tolerance to reduce the reject rate.

I asked him why would somebody design a push-pull Class A amplifier with no global negative feedback (i.e. my amplifier) knowing that the positive and negative parts of the AC current would vary due to variances between tubes? (Don't judge me, I didn't know all this when I bought my amplifier). He said that it is because of a marketing imperative. There are people who want a push-pull Class A amplifier with sufficient power and current delivery for modern speakers, so they do the best that they can within their limitations, which is the same as what any engineer who is asked to design any product would do.

This means that for any given amplifier, there is an optimal "ideal" tube for which the amplifier was designed in mind. Even then, tolerances still add up from the other electronic components in the amp meaning that different amps from the same manufacturer might not measure the same, especially when you add tube variations in the mix. Deviations too far from the specification laid by the amplifier's designer will result in reduced performance or even damage to the tube or amp. A tube that works well in your friend's amplifier from a different brand might not work so well in your amp.

Of course, measures the same and sounds the same are different things, but that is a different topic. (EDIT) Let me rephrase that: "Whether something that measures differently might be audible" is a topic for a different debate is what I meant to say.

 

[MattHooper]

Of course tubes have the potential to sound different, and even different tube manufacturers might have a different sound.

I had a conversation with an audio engineer who worked on tubes. He told me that what nobody ever discusses are manufacturing variability between tubes and between manufacturers. He has never found a pair of so-called "matched tubes" to measure exactly the same.

Why do tubes from the same manufacturer sound different? Because the effect of the multiple tolerances required is cumulative. Are the plates separated by the same distance? Is the purity of material consistent across batches? Are the plates warmed to the same temperature? Is the vacuum in every tube the same? How much deviation from tolerance does the manufacturer's QC permit before junking the tube?

Why do tubes from different manufacturers sound different? Because they all have different interpretations of compliance to the published specification, might use different plate voltages, heater temperatures, etc. He gave the example of the 300B tube, one of the most well known tubes in audio, and then he showed me voltages and current draw he measured from various 300B's from his tube tester. He said that if you were to put them on the same graph, you would see a normal deviation from a manufacturer representing in-house variability, with the standard deviation representing manufacturing quality and QC, and multiple normal deviations from each different manufacturer. This is the same with any electronic component, or everything that has been manufactured, except that the high manufacturing cost of tubes and the nature of tubes means that the manufacturers accept a higher tolerance to reduce the reject rate.

I asked him why would somebody design a push-pull Class A amplifier with no global negative feedback (i.e. my amplifier) knowing that the positive and negative parts of the AC current would vary due to variances between tubes? (Don't judge me, I didn't know all this when I bought my amplifier). He said that it is because of a marketing imperative. There are people who want a push-pull Class A amplifier with sufficient power and current delivery for modern speakers, so they do the best that they can within their limitations, which is the same as what any engineer who is asked to design any product would do.

This means that for any given amplifier, there is an optimal "ideal" tube for which the amplifier was designed in mind. Even then, tolerances still add up from the other electronic components in the amp meaning that different amps from the same manufacturer might not measure the same, especially when you add tube variations in the mix. Deviations too far from the specification laid by the amplifier's designer will result in reduced performance or even damage to the tube or amp. A tube that works well in your friend's amplifier from a different brand might not work so well in your amp.

Of course, measures the same and sounds the same are different things, but that is a different topic.

I've been tube rollin' a whole lot this last year, both the power tubes and the small tubes and it's been fascinating. It seems I can get more thick and lush with some, more clean and precise with others, or mix 'n match. The biggest surprise was for the first time swapping out the 6550 tubes for KT120s. The impression was like my speakers got "bigger," deeper sounding bass, sonic images seemed to have more heft and scale, soundstage seemed to expand. I've gone back and forth tons of times and every time it seems to have the same effect.

(Sorry...no measurements...hence not put forth as anything someone wanting objective data has to take seriously.)

 

[levimax]

Of course tubes have the potential to sound different, and even different tube manufacturers might have a different sound.

I had a conversation with an audio engineer who worked on tubes. He told me that what nobody ever discusses are manufacturing variability between tubes and between manufacturers. He has never found a pair of so-called "matched tubes" to measure exactly the same.

Why do tubes from the same manufacturer sound different? Because the effect of the multiple tolerances required is cumulative. Are the plates separated by the same distance? Is the purity of material consistent across batches? Are the plates warmed to the same temperature? Is the vacuum in every tube the same? How much deviation from tolerance does the manufacturer's QC permit before junking the tube?

Why do tubes from different manufacturers sound different? Because they all have different interpretations of compliance to the published specification, might use different plate voltages, heater temperatures, etc. He gave the example of the 300B tube, one of the most well known tubes in audio, and then he showed me voltages and current draw he measured from various 300B's from his tube tester. He said that if you were to put them on the same graph, you would see a normal deviation from a manufacturer representing in-house variability, with the standard deviation representing manufacturing quality and QC, and multiple normal deviations from each different manufacturer. This is the same with any electronic component, or everything that has been manufactured, except that the high manufacturing cost of tubes and the nature of tubes means that the manufacturers accept a higher tolerance to reduce the reject rate.

I asked him why would somebody design a push-pull Class A amplifier with no global negative feedback (i.e. my amplifier) knowing that the positive and negative parts of the AC current would vary due to variances between tubes? (Don't judge me, I didn't know all this when I bought my amplifier). He said that it is because of a marketing imperative. There are people who want a push-pull Class A amplifier with sufficient power and current delivery for modern speakers, so they do the best that they can within their limitations, which is the same as what any engineer who is asked to design any product would do.

This means that for any given amplifier, there is an optimal "ideal" tube for which the amplifier was designed in mind. Even then, tolerances still add up from the other electronic components in the amp meaning that different amps from the same manufacturer might not measure the same, especially when you add tube variations in the mix. Deviations too far from the specification laid by the amplifier's designer will result in reduced performance or even damage to the tube or amp. A tube that works well in your friend's amplifier from a different brand might not work so well in your amp.

Of course, measures the same and sounds the same are different things, but that is a different topic.

It is sad that this is true for many modern tube amps. Well designed amps from the past ( and a few from the present) put performance first and any decent design can use different tubes without issue.

 

[fubarnow]

My Primaluna with stock tubes el34's

 

[Waxx]

I also got a Prima Luna amp (a Prologue 4), and all Prima Luna amps have an own autobias system, where the amp measures the tubes and adjust the B+ voltage and bias to make them work like it should. That is one of the big differences compared to similar tube amps and makes them sound so good. Other amps also sometimes have autobias, but none is so extensive and exact as the ones from Prima Luna i (and most) think. And the rest of the amp is just build like it should. Off course it's very colouring and not low distortion, that is general with tube amps, and that is why we love it (the harmonic distortion) and why we pay so much for it...

Prima Luna used selected and tested Shuguang tubes (before the factory burned down a few years ago) and now use PSvane (A Shuguang spinoff) equivalents, of the higher quality series (Valve Art EL34B) that are rebranded to Prima Luna. That is what the owner (Herman van den Dungen) said to me, and it looks like he is not lying. The same tube is also rebranded as the Black Treasure EL34B and the TAD EL34B-STR. You can order the tubes direct from Prima Luna (EU/UK) or Upscale Audo (US/Canada) if you wish, but only when you have a registered ownership of a Prima Luna amp. But if you don't want to deal with them (especially Upscale audio is a bit to much a snake oil seller for many), use the Psvane or Shuguang (if you can still find them) EL34B or EL34C models if you want to stay close to the original sound. Many others, also other types also work like said in the documentation as the autobias adjust the amp to it.

And yes, tubes sound different, from different type and brands. It's not neutral sounding and each has it's own colouration profile, depending on the spec that are variating a little. It's almost impossible with this tech to make them exact to spec.

The picture of my amp is from 4 years ago, in my former house. The amp is now in my office at home driving a pair of bookshelf speakers...

 

[fubarnow]

I prefer buying my tubes from Upscale Audio.
I know they've been measured.
No snake oil from Kevin, He's Fun to watch.
Audio should be fun.
Too many people are too bitchy about audio
and they need to lighten up.
You have a very nice amp.

I have discovered that the 42 Watts
my PL puts out is plenty for what I need.
I've run various speakers with this amp
and never ran out of juice.
I think the design of PL is first rate.

 

[Waxx]

I prefer buying my tubes from Upscale Audio.
I know they've been measured.
No snake oil from Kevin, He's Fun to watch.
Audio should be fun.
Too many people are too bitchy about audio
and they need to lighten up.
You have a very nice amp.

Yes, he measures and matches, but the snake oil part is that he pushes very expensive NOS tubes as better than modern production, which is very often not the case. He earns more on that i guess than selling high quality tubes made today (or at least recently). But his measuring and matching is just like Prima Luna themselves, on point (and with values written on the box), that is already a lot better than a lot of other vendors that sell "matched" tubes.

 

[fpitas]

Being a thrifty engineer, I'd mostly try varying the output quiescent bias to see if improvements could be made. I have to wonder: if the bias is re-adjusted to factory spec after any replacement (as it should be), will most of the differences disappear?

 

[MattHooper]

The value of measurements was shown again in this month's Stereophile. Measurements pulled the pants down of more than one piece of gear.
The worst was I think the Zesto tube amp which measured waaaay lower than it's rated power. We've seen that a number of times in the stereophile amp measurements (though I think it tends to happen more often with tube amps?). I'd be kinda pissed if I bought an amp thinking it put out a certain power rating and it wasn't providing anything like that power.

 

[fpitas]

The value of measurements was shown again in this month's Stereophile. Measurements pulled the pants down of more than one piece of gear.
The worst was I think the Zesto tube amp which measured waaaay lower than it's rated power. We've seen that a number of times in the stereophile amp measurements (though I think it tends to happen more often with tube amps?). I'd be kinda pissed if I bought an amp thinking it put out a certain power rating and it wasn't providing anything like that power.

Good transformers are expensive and heavy.

 

[sergeauckland]

A few years ago I did try to measure what difference there was between different manufacturers' valves.

I used a Leak TL10 amplifier http://44bx.com/leak/TL10.html

and the valves originally present, a mixture of RCA and GEC, and a few unbranded no-name valves I happened to have about.

This is clearly a very small 'study' more for my own information and to see if there are the 'night and day' differences so often mentioned, so not done with academic rigour.

I measured noise, frequency response and distortion for the original, then changed one valve at a time and repeated the measurements.

There was next to no difference between the measurements, all had essentially the same figures for noise, distortion and frequency response.

I therefore concluded that valve rolling is a mug's game, and any amplifier should work to spec with any valve that in turn meets its spec.

Furthermore, the idea that, say, RCAs sound like this, and GECs sound like that for the same valve type is nonsense. If they meet the specification for that valve, they will work correctly. Even matched output pairs shouldn't be necessary with good design and adjustable AC & DC balancing.

If an amplifier design is fussy about valves, it's a crap design and should be eliminated from serious consideration. Audio is very simple, even for valves, and for power amps it's ALL in the output transformer.

S

 

[fpitas]

A few years ago I did try to measure what difference there was between different manufacturers' valves.

I used a Leak TL10 amplifier http://44bx.com/leak/TL10.html

and the valves originally present, a mixture of RCA and GEC, and a few unbranded no-name valves I happened to have about.

This is clearly a very small 'study' more for my own information and to see if there are the 'night and day' differences so often mentioned, so not done with academic rigour.

I measured noise, frequency response and distortion for the original, then changed one valve at a time and repeated the measurements.

There was next to no difference between the measurements, all had essentially the same figures for noise, distortion and frequency response.

I therefore concluded that valve rolling is a mug's game, and any amplifier should work to spec with any valve that in turn meets its spec.

Furthermore, the idea that, say, RCAs sound like this, and GECs sound like that for the same valve type is nonsense. If they meet the specification for that valve, they will work correctly. Even matched output pairs shouldn't be necessary with good design and adjustable AC & DC balancing.

If an amplifier design is fussy about valves, it's a crap design and should be eliminated from serious consideration. Audio is very simple, even for valves, and for power amps it's ALL in the output transformer.

S

I tend to agree. As I said above, I think the output bias varying after replacement *may* account for sonic differences. Of course, this simply means the right procedure was not followed per setting bias.

 

[MattHooper]

A few years ago I did try to measure what difference there was between different manufacturers' valves.

I used a Leak TL10 amplifier http://44bx.com/leak/TL10.html

and the valves originally present, a mixture of RCA and GEC, and a few unbranded no-name valves I happened to have about.

This is clearly a very small 'study' more for my own information and to see if there are the 'night and day' differences so often mentioned, so not done with academic rigour.

I measured noise, frequency response and distortion for the original, then changed one valve at a time and repeated the measurements.

There was next to no difference between the measurements, all had essentially the same figures for noise, distortion and frequency response.

I therefore concluded that valve rolling is a mug's game, and any amplifier should work to spec with any valve that in turn meets its spec.

Furthermore, the idea that, say, RCAs sound like this, and GECs sound like that for the same valve type is nonsense. If they meet the specification for that valve, they will work correctly. Even matched output pairs shouldn't be necessary with good design and adjustable AC & DC balancing.

If an amplifier design is fussy about valves, it's a crap design and should be eliminated from serious consideration. Audio is very simple, even for valves, and for power amps it's ALL in the output transformer.

S

I believe you've mentioned that before, and I think this kind of contribution is always very welcome. Even if it was just for your own benefit, it can serve as a "data point" for others to consider. Same, I think, when people post about blind testing results. It may not be a strictly scientific exchange, but "Well, if he/she could have been fooled...so could I."

It also re-enforces the power of having personal experiences in forming our beliefs. It's one thing to consider the arguments and evidence for one side or another; it's another to have a personal experience. In my case for instance my blind testing of AC cables, Video cables and other things have led me to mostly conclude those are a "mug's game" too. It's hard to shake the impact, even if strictly speaking we can't over-extrapolate just from our own test.

As I've also mentioned before, I do feel I'm hearing differences in tube rolling and actually had something of an opportunity to do a blind test because I had two pairs of the same tube amps for a while (two sets of CJ Premier 12 monoblocks). But I couldn't pull off the blind test for various reasons before selling one pair, which is a bummer.

At this point I still think I'm hearing differences between different sets of tubes in my system. It's tentative because of course I could be wrong. But the impression is so strong, and so consistent it's hard to shake. When I switch between KT120 and 6550 power tubes for my amps, the change in sound seems unmistakable - changes in bass quality, impact (e.g. horns will sound softer, drums less forward etc), soundstaging, imaging. I also have certain NOS sets of the small input tubes - one set reliably produces a super clean, precise, tight, forward sound, the other set produces a thicker, more lush and laid back sound.

I'd be confident enough to put some money down I could tell these apart in blind tests; about as confident as I was before doing my tube pre-amp/ss pre-amp blind test (where I could identify the tube preamp). So, I'm continuing to tube roll between these tubes "hearing" these differences in my system. Of course, again, I'm not for a moment trying to convince anyone else as my anecdotal experience is useless in settling such a question. (And even if it turned out to be a bias effect, it is so strong and reliable I don't mind taking advantage of it).