• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Tube Amplifier Design and Rolling Tubes

I change sockets to go from 'ring' to split connectors. I have found that these are much easier to remove.

Whenever I remove tubes, I always gauge the connectors to make sure they are not sprung. See Socket Retension for 'gage pin' suggestions

Espresso is another of my passions and a Puly Caff Lifty or something similar on either side levers the tube up easily
 

Attachments

  • PulyLifty.jpg
    PulyLifty.jpg
    16.4 KB · Views: 27
Last edited:
Just my experience - not making claims for every amp / tube created. Don't take it too seriously.

In my 15 years of using the same EL34 PP tube amp every day I find that there is a difference between pentode types and beam-tetrode type tubes. (the difference is power output). EL34 (Pentode) vs Sylvania 6CA7 (beam-tetrode)

Between brands of pentodes, you can definitely hear a difference between something like a Chinese Ruby and a NOS Mullard or similar (bass is loose with Ruby)
But the difference in sound between a good Russian / Euro EL34 (Winged-C, New Sensor Mullard branded) and a NOS EL34 from the 50's or 60's is fairly minor - the main difference is reliability / longevity (it's close enough to double the life with NOS).

Where NOS or certain builds of new tubes becomes critical is in circuits where plate voltages become extreme - as apart from JJ's none of the modern EL34's seem to be able to cope as well with high plate voltages - which will obviously impact the sound of they don't meet the required spec.

As far as Phase inverters go (Nos for reliability but I can't detect any difference in sound, and I doubt measurement would show it either.
Input/preamp tubes again, I can't pick any difference between brands - they either work or they don't.
When they sound different it's usually because they are out of spec gain wise. If the gain factor is the same - very difficult / impossible to tell types apart.

Lots of marketing these days influencing peoples purchasing and opinions - I'm of the opinion if they meet spec, they are good to go (A lot of 12ax7 types these days do not meet spec - some of which are rather expensive current production types.

I wouldn't get too invested in Tube rolling at today's prices - what I was paying $20 for 15 years ago is now $500. The juice isn't worth the squeeze.

My amp needs a recap and a bit of TLC - I decided to pack it away and just buy a Yamaha SS amp instead. it's nice not having to break out a multimeter to play music.
 
Last edited:
In my 15 years of using the same EL34 PP tube amp every day I find that there is a difference between pentode types and beam-tetrode type tubes.
That will STRONGLY depend on the circuit/topology.
Between brands of pentodes, you can definitely hear a difference between something like a Chinese Ruby and a NOS Mullard or similar (bass is loose with Ruby)
But the difference in sound between a good Russian / Euro EL34 (Winged-C, New Sensor Mullard branded) and a NOS EL34 from the 50's or 60's is fairly minor - the main difference is reliability / longevity (it's close enough to double the life with NOS).
Any actual data to support this?
 
That will STRONGLY depend on the circuit/topology.

Any actual data to support this?

Yes. 35w with el34 in pentode and 40w with Kt88. Stereophile and others have tested the amp (Cayin A50T/A55T).
16w vs 20w in triode mode which I prefer if running more aggressive speakers which go below 4 ohms.
I don't think the Rubys meet spec. They also don't last long in the circuit. 500hrs vs 2000hrs


I had another set of data - but can't recall the site that did the testing.
 
Last edited:
Which is it?
what do you mean ?

6ca7 output is between el34 and kt88 in this amp. (I prefer Pentodes in this amp).

EDIT: A50T and A55T is the same amp other than a resistor change for bias values. (current builds of the amp have a switch).
 
what do you mean ?
You compared 6CA7 to EL34, then when I asked about that, you pivoted to EL34 vs KT88. So my original question remains. And the sonic claims remain unanswered as well.
 
You compared 6CA7 to EL34, then when I asked about that, you pivoted to EL34 vs KT88. So my original question remains. And the sonic claims remain unanswered as well.

did you mean this quote ?

In my 15 years of using the same EL34 PP tube amp every day I find that there is a difference between pentode types and beam-tetrode type tubes. (the difference is power output). EL34 (Pentode) vs Sylvania 6CA7 (beam-tetrode)

They don't sound the same. I think it's due to power - but it could be something else.

KT88 and 6CA7 sound more similar than the EL34 - I assumed it was due to power.

EL34 wise -Ruby vs Everything else I have tried. data is Rubys don't last as long - my guess is they don't meet spec or are of inferior quality.
 
Last edited:
did you mean this quote ?



They don't sound the same. I think it's due to power - but it could be something else.

KT88 and 6CA7 sound more similar than the EL34 - I assumed it was due to power.

EL34 wise -Ruby vs Everything else I have tried. data is Rubys don't last as long - my guess is they don't meet spec or are of inferior quality.
Re-asserting claims is not data.
 
Re-asserting claims is not data.

So it's just in my head ? I'm not disagreeing either. It could well be.
But how do we explain the significant differences in lifespan of the different types ?

I looked at the wiring and the 6CA7 appears different to the EL34 - (pin 1) is that unimportant?

I'm here to learn - so appreciate any wisdom you can pass on.
 
I have rolled tubes in my Sonic Frontiers Line 1 preamp quite extensively, both about 12 years ago and again a couple of years ago. The "rolling" I did was at the gain stage only. My subjective impression is that the sound is similar but not quite the same amongst the tubes.

If you decide to ignore naysayers' advice and have a little fun, DO roll a few different makes of tubes. If your pocketbook is like mine, you shouldn't start with exotic or expensive NOS ("new old stock") tubes, but new ,modern tubes of which there is a reasonable selection depending on the type of tube. (In case of my SF preamp, the tubes are of the '6922' variety for which there is a relatively huge choice.)

If objective proof of the sound differences you hear is what you care about, you certainly should arrange for some rigorous blind comparisons. If you don't care about "proof" but are content with you personal impressions, then don't bother.
 
Are spec sheets for tubes of any value as far as how a tube will impact a circuit?
Because it looks like the 6CA7 and EL34 are not the same spec or build wise - yet people selling them say they are fully compatible / interchangeable.


lead 1 looks different.
They look pretty much interchangeable to me. Pin 1 is either the grid or beam-forming plates, which are similarly connected.

Much like the KT66 and 6L6GC tubes, with the small advantage that the 6L6GC takes a lower heater current so the mains transformer runs a little cooler than with the KT66.

S.
 
I grab an octal-based tube by its socket, rather than the glass, when removing it.
 
Tube rolling within a type in my experience only makes a difference when you take them out of their linear region. As such, it matters much more in a guitar amp (where there is ample intentional distortion) than in hi-fi, where everything is meant to be low distortion.
I've fooled around with different preamp tubes in guitar amps used for amplified harmonica. A lower output tube than the normal 12AX7 will give distortion that's more fun and reduce feedback. If you go to low you may not get enough output power. In audio amps I'm not sure there's as much reward.
 
It's nice not having to break out a multimeter to play music.

It's my view that many audiophiles worry too much about the bias. With NEW tubes adjust initially, do 24 hour burn in, adjust.
Check again - not necessarily adjust - @ 50, 100, 200, 500, 1000, 2000 hours.

NOS are a whole 'nother kettle of fish as they sometimes take 24 or more hours to degas and many re-gas if not played quite frequently.

Input tube rolling REQUIRES level and channel balance adjustment to match what came out.

If one checks the cathode resistors, they may be 5% inductive parts with 300PPM TempCo.
I use precision wire wound non-inductive 1% 50PPM TempCo.

Line voltage variations and transformer temperatures may affect both bias and the output level, so do any rolling on fully warmed gear.
A small change in line voltage may affect B+ which affects bias, doubly so if the bias supply is unregulated.
I keep a cheep & cheerful digital voltmeter plugged into the system circuit.
If the voltage is off a little and the bias is off a little, all is well - provided all in the same direction.

The bias meters in my amps [avatar] are accurate to ±2.5mV - verified!
As long as they all track equally, fuhgedaboudit.

If your bias check system has a switch, exercise it a couple of times before checking as the part likely has built up a little contamination.

Today's precision meters are both a blessing -accuracy- and a curse -precision.
A curse because they encourage knob-dicking.

Single turn cermet pots have a relatively short adjustment count, around 100-200 cycles.
Some lower grade parts suffer early mortality which can be fatal if open.

Do yourself a favor and replace single turn bias pots with multi-turn. Precision and life are both increased. Bournes 3339 series.

Bournes3339.png
 
Tubes are horribly inconsistent even between a "matched" set from same manufacture if you graph distortion vs frequency at various levels of drive. And this is with even 20db of global feedback I can demonstrate this! Like 10 or 20db differences in 3rd harmonic in a quad of identical tubes that bias the same. Yeah it's sitting 40-50db down, more with nfb, but it's a shit show.

Zero feedback circuits (which are so popular) are a non-starter for me these days for this very reason.

The gains can also be way mismatched yet I rarely see designs that handle that elephant in the room. I'm done with folks bragging about superior imaging and information retrieval and all that when your amp has 2-5db of channel imbalance because you don't believe in feedback and tone/balance controls are also "evil" and avoided. Lots of B.S. out there and has been since the triode revival of the late 90s.

I still love tubes but I wish I knew 20 years ago what I know now. Lots of wasted time going the wrong direction with these things. Trust the measurements and science. Too many of these "designs" are dashed together by people whom don't even own a scope. No thanks.
 
Last edited:
I've done a fair bit of tube rolling, to see if there's any difference between tubes of the same type but from different manufacturers. My findings are that as long as the tubes are within spec, and I do have a tester, there's no measured difference. The circuits I've done this have had overall feedback, so the results might be different for no overall feedback designs, but I can't see why.

Statements I've seen along the lines that ABC has better bass, or XYZ has more detail, I think are rubbish and wishful thinking.

S
I respect your earlier post and think that they are well founded. But on this topic my experience is different , different makes of tubes can make a significant difference in my low/no feedback tube amps.both measurable and audible
Here is the difference between different tubes!in the same amp.

Just one many examples.. in another recent case of tube rolling my headphone amp went from 0.1% distortion to 2%


1734776351211.png

Just a comparison I had handy, Cayin 300B with the stereophile measurments and mine with two different driver tubes makes Tungsol resissue 6SN7 VS Shuguang . TungSol 6SN7 had a remarkable low distortion in this amp
1734776669382.png

The tube rolling effects on measured frequency response is minimal in my amps. What changes is gain and distortion.


Effect of different tubes label. New 6SN7 from Shuiguang had 2.8% distortion same tube from Russian new production Tungsol gave 0.6% . I think this is audible difference too
capture-jpg.583508

Effect of different input tube 6SL7 in same amp

capture-jpg.583510
 
Last edited:
Back
Top Bottom