Short answer:it is NOT a true 7591, even less "Tungsol"
but some Soviet designed tube which is "close enough" to be roughly used as an equivalent.Tube **resellers** can do nothing about that, they can´t publish real datasheet nor original name (if anything because it´s all written in Cyrillic) and specially because it kills sales.
Mind you, it´s probably better than the original one, military specs, etc. but that doesn´t cut ice with buyers, who want nostalgia, Mojo and a "famous name" , even when anybody thinking 30 seconds about it must realize there´s no connection whatsoever between old and new Tungsol.
And prefer inaccurate but magical datasheets than real ones.
That said, honest resellers at least hint at the difference and suggest proper biasing.
"If you want to measure the total dissipation of the valve it is neccessary to measure the Anode AND Screen grid current."
That's what the cathode current is - the sum if the screen and plate current (and grid current if there is any). And that's what I wrote about.
With all the explanation you gave - how would you measure dissipation in a triode?? They don't have screens.
I'm sure that in general what you are saying is worthwhile but it has little or nothing to do with establishing the safe operating area for the tube in question.
That's all I intend to say on this topic.
That's what the cathode current is - the sum if the screen and plate current (and grid current if there is any). And that's what I wrote about.
With all the explanation you gave - how would you measure dissipation in a triode?? They don't have screens.
I'm sure that in general what you are saying is worthwhile but it has little or nothing to do with establishing the safe operating area for the tube in question.
That's all I intend to say on this topic.
Short answer:it is NOT a true 7591, even less "Tungsol"
Do you have evidence or a reliable source to support that?
The reason I ask is that there was a tube a number of years ago called the 7591XYZ that was produced as a sub for the real 7591A when the old stock supply dried up and there were not yet any new 7591s in production. The "XYZ" was a repinned 6P3S-E Russian tube and it came with a cathode resistor that had to be added if you wanted to use the 7591XYZ in a real 7591 socket since electrically the 6P3S-E is much closer to a 6L6 than a 7591.
Ever since them there has been a steady drumbeat of people who insist that ALL Russian 7591 tubes are just repinned 6L6s. I do not believe - based on my own measurements and on the people I've spoken to from New Sensor - that is the case.
So if you're certain that the Tung-Sol branded tube is a rebranded/relabeled tube could you share the source of that information ? And could you share what the number is of the Russian tube that's being sold as a 7591? I'd appreciate it very much.
It might be all you want to say, but the easy fix is not messing with the control grid voltage, (bias) as I have pointed out over and over again.
The easiest is to put a simple zener in series with the screen grid supply to drop the supply voltage.
No doubt selecting a few random values from 6V to say 18V, will get the thing under control.
How come we have to go round and round to get to this?
As for the following comment:-
This statement is inaccurate, I will correct because I am often there.
Most all these valves from ex-USSR were based in the old 6L6 copy the 6P3S.
This was basically not a Soviet design, but as I made clear, a US design made in the USSR.
That was modernised to be the 6P3S-E with "coin" base and is really suprisingly small.
Hey suprise suprise, the last 7591 had coin bases too...and is small.
Now being as I happen to have the magic "cyrillic" data for the 6P3S-E (yes it's almost impossible to find, the REAL bias and current) , it's clear it differed quite a bit from the 6L6, AND they suffered from wide variations in quality and gassiness, a guy called "Wavebourn", an expert in USSR valves will confirm...
You only need to ask him.
The 6L6 really has a different more compact anode shape than the Russian one, I don't know about the Czech and Serbian ones.
It appears this 6P3S-E was altered or reverse engineered in various ways, the grid spacings alterred and hey presto it's a 7591.
What is not fake is the fact it is being produced by Reflektor Saratov, which has basically rebranded everything with fake names on the instructions of a certain person...ahum..and even invaded Svetlana's St Petersburg's space so badly they got so almighty p..ssd off with it, they stopped production of all consumer audio devices...
In my opinion SVET, would be the only company that could make the 7591 properly, being as I deal with another high quality company there.
Melz no longer make valves - specialising in military optics, Novossibirsk is dead, Foton Tashkent Uzbek is dead, I can't think of many more.
The easiest is to put a simple zener in series with the screen grid supply to drop the supply voltage.
No doubt selecting a few random values from 6V to say 18V, will get the thing under control.
How come we have to go round and round to get to this?
As for the following comment:-
This statement is inaccurate, I will correct because I am often there.
Most all these valves from ex-USSR were based in the old 6L6 copy the 6P3S.
This was basically not a Soviet design, but as I made clear, a US design made in the USSR.
That was modernised to be the 6P3S-E with "coin" base and is really suprisingly small.
Hey suprise suprise, the last 7591 had coin bases too...and is small.
Now being as I happen to have the magic "cyrillic" data for the 6P3S-E (yes it's almost impossible to find, the REAL bias and current) , it's clear it differed quite a bit from the 6L6, AND they suffered from wide variations in quality and gassiness, a guy called "Wavebourn", an expert in USSR valves will confirm...
You only need to ask him.
The 6L6 really has a different more compact anode shape than the Russian one, I don't know about the Czech and Serbian ones.
It appears this 6P3S-E was altered or reverse engineered in various ways, the grid spacings alterred and hey presto it's a 7591.
So yes that is "fake news".
What is not fake is the fact it is being produced by Reflektor Saratov, which has basically rebranded everything with fake names on the instructions of a certain person...ahum..and even invaded Svetlana's St Petersburg's space so badly they got so almighty p..ssd off with it, they stopped production of all consumer audio devices...
In my opinion SVET, would be the only company that could make the 7591 properly, being as I deal with another high quality company there.
Melz no longer make valves - specialising in military optics, Novossibirsk is dead, Foton Tashkent Uzbek is dead, I can't think of many more.
Last edited:
The grids are how one gets the current set up, sure.
Icathode = k (Vg1 + Vg2/Mu2 + Vp/Mup) ^1.5
Once the current is established however, power dissipation is Pdiss = Vp x Ip
and G2diss = Vg2 x Ig2 with Ip and Ig2 summing to Icathode
Since Vg2 is roughly similar to Vp (or B+ actually) for the usual audio tubes, to a good approx. Tube diss. = Vp x Icath.
Icathode = k (Vg1 + Vg2/Mu2 + Vp/Mup) ^1.5
Once the current is established however, power dissipation is Pdiss = Vp x Ip
and G2diss = Vg2 x Ig2 with Ip and Ig2 summing to Icathode
Since Vg2 is roughly similar to Vp (or B+ actually) for the usual audio tubes, to a good approx. Tube diss. = Vp x Icath.
Last edited:
What makes you say that?
Most of the normal examples Vg2 is half that of Vp...
The grid 2 current is typically only 5% to 10% of plate current at idle. It's just an approximation to simplify the formula to use Vg2 at B+ too (a small inaccuracy), it -will- be B+ for ultralinear mode.
If one is specifically calculating the grid 2 dissipation to check against tube grid 2 max spec, then of course use the real Vg2.
If one is specifically calculating the grid 2 dissipation to check against tube grid 2 max spec, then of course use the real Vg2.
Last edited:
I'm wondering if the OP's best solution might be conversion to cathode bias. It soaks up some of the excess anode and G2 voltages from modern power lines, proves a small cushion for G1 current runaway, requires no chassis holes, and generally is friendlier to use with modern valves. Four resistors of - what? - maybe 750 or 820 Ohms, four big 50 Volt 105C electrolytics. Another possibility.
All good fortune,
Chris
All good fortune,
Chris
There are well understood downsides, but total heat for a given bias is constant. I wouldn't ordinarily recommend this significant a modification, but this isn't an ordinary case. Horses for courses.
Restoring vintage electronics in 2019 requires some adjustments from ideological purity. There won't ever be true type 7591's again, or 115 VAC lines (although easy to make oneself). But life goes on.
All good fortune,
Chris
Restoring vintage electronics in 2019 requires some adjustments from ideological purity. There won't ever be true type 7591's again, or 115 VAC lines (although easy to make oneself). But life goes on.
All good fortune,
Chris
Hi Chris, I'm planning on running it on a Variac permanently so at least 115 will be assured.. I'm not too sure that there is any room for additional caps etc in the box. It's absolutely crammed full. Currently I'm waiting on parts to restuff the cans. Once I get them I'll update with cathode currents and go from there. I'm considering going ahead and drilling the case for bias pots. We'll see how I feel about it or I get there.
To everyone who has contributed to this thread, a big big thank you. There has been some amazing knowledge shared here, a lot of it is currently over my head. Nonetheless, it is valuable that it can be shared with green guys like myself. Cheers
As I said, the easiest method to lower the idle current of the valve is to drop the screen voltage anything from 20-40V less will drop it back to "normal values".
The anode voltage is not important. It can be easily 50V over the max without any harm at all.
Think of the screen like a 2nd control grid, and you are right in there.
That's easily proved by smoking amp's screen drive, where he uses mosfets to drive the screen instead of the g1.
All screen grids are sensitive to noise and voltage variation in high gain valves.
This isn't hearsay it's fact.
It's actually why ultralinear connections work the way they do, adding negative feedback to the screen grid.
The anode voltage is not important. It can be easily 50V over the max without any harm at all.
Think of the screen like a 2nd control grid, and you are right in there.
That's easily proved by smoking amp's screen drive, where he uses mosfets to drive the screen instead of the g1.
All screen grids are sensitive to noise and voltage variation in high gain valves.
This isn't hearsay it's fact.
It's actually why ultralinear connections work the way they do, adding negative feedback to the screen grid.
I did mention briefly in passing, - these high gain valves need low impedance drives or they self bias and run away. (Quicksilver did exactly that and constantly melted valves).
Sherwood appears to have done what quite a lot of 60s amps did, use a 12AX7 as an output driver/ phase splitter.
For a low gain valve it works fine (eg. the beautiful Bogen DB110), but 10W was no longer enough in 1963.
They seem to have carried over the circuit from their EL84 which also was a tube killer, chasing high output on the edge of what is sensible.
For a high gain higher performance valve it can rapidly become suicidal.
With a little bit of drift a 150k resistor can become 180k or higher.
I know RCA quote their maximum g1 circuit as 0.3Mohm, but that's at an anode/screen voltage of 300V aa -10V bias.
here we have over 400 and -18>-20 (double), and a repro valve.
so,-
Otherwise, off we go into self bias which rapidly becomes a vicious circle as it drags the entire secondaries down with the extra current.. (I note they derive their bias from an extra secondary winding), but the silicon diode will allow as much current as you like to dump into a dying valve inc a DEAD SHORT direct through the output tranformer primary.
What you would get is the equivalent of running a big end bearing in an engine, max damage and a rod sticking out of the side of the block!
I did point out in PM both Scott & Bogen managed to make amps, using 7591s (or the 7868) which not only has higher anode & screen volts but doesn't go wrong, and makes a lot more power.
How do they do it?
Look at the LK72.
Low impedance drive.
Scott uses a 6U8 pentode DC coupled into a strong triode... 15k source,
not a 68k, split through a high impedance anode .AX7 thing.
They can safely then let the output valve bias still sit on 200-220k DC source resistance.
Fisher did the low Z thing with their 8417 amp but differently, using instead a dual power pentode strapped as a triode. They don't go wrong and they make good power with low distortion.
In my opinion I cannot understand why Sherwood didn't use a 7247 as driver/phase invertor to drive their output stage.
It's pin for pin plug in.
That would give it the low impedance needed, cure all the ills and make it ultra stable.
I can't even say their power output and distortion is very nice.
As is, originally it's at the bottom end of 'fi.
IMD is very high at higher powers, the amp maxes out at 24Wper ch, & it appears to be limit reliable.
I would suspect the 1v biased 12AX7 input stage is also to blame.
Scott managed easily 35-40W with basically the same line-up, reliably and Scotts output transformers are brilliant, much under-rated kit.
The other silly idea was Sherwood using a series of dropping resistors to feed anode and screen supplies.
OK the quick fix is to remove one and substitute a 30V zener.
I don't reckon fooling with the bias line is neccessary.
In any case with a higher mains voltage, the bias voltage will increase in exact proportion to the line voltage.
Sherwood appears to have done what quite a lot of 60s amps did, use a 12AX7 as an output driver/ phase splitter.
For a low gain valve it works fine (eg. the beautiful Bogen DB110), but 10W was no longer enough in 1963.
They seem to have carried over the circuit from their EL84 which also was a tube killer, chasing high output on the edge of what is sensible.
For a high gain higher performance valve it can rapidly become suicidal.
With a little bit of drift a 150k resistor can become 180k or higher.
I know RCA quote their maximum g1 circuit as 0.3Mohm, but that's at an anode/screen voltage of 300V aa -10V bias.
here we have over 400 and -18>-20 (double), and a repro valve.
so,-
Otherwise, off we go into self bias which rapidly becomes a vicious circle as it drags the entire secondaries down with the extra current.. (I note they derive their bias from an extra secondary winding), but the silicon diode will allow as much current as you like to dump into a dying valve inc a DEAD SHORT direct through the output tranformer primary.
What you would get is the equivalent of running a big end bearing in an engine, max damage and a rod sticking out of the side of the block!
I did point out in PM both Scott & Bogen managed to make amps, using 7591s (or the 7868) which not only has higher anode & screen volts but doesn't go wrong, and makes a lot more power.
How do they do it?
Look at the LK72.
Low impedance drive.
Scott uses a 6U8 pentode DC coupled into a strong triode... 15k source,
not a 68k, split through a high impedance anode .AX7 thing.
They can safely then let the output valve bias still sit on 200-220k DC source resistance.
Fisher did the low Z thing with their 8417 amp but differently, using instead a dual power pentode strapped as a triode. They don't go wrong and they make good power with low distortion.
In my opinion I cannot understand why Sherwood didn't use a 7247 as driver/phase invertor to drive their output stage.
It's pin for pin plug in.
That would give it the low impedance needed, cure all the ills and make it ultra stable.
I can't even say their power output and distortion is very nice.
As is, originally it's at the bottom end of 'fi.
IMD is very high at higher powers, the amp maxes out at 24Wper ch, & it appears to be limit reliable.
I would suspect the 1v biased 12AX7 input stage is also to blame.
Scott managed easily 35-40W with basically the same line-up, reliably and Scotts output transformers are brilliant, much under-rated kit.
The other silly idea was Sherwood using a series of dropping resistors to feed anode and screen supplies.
OK the quick fix is to remove one and substitute a 30V zener.
I don't reckon fooling with the bias line is neccessary.
In any case with a higher mains voltage, the bias voltage will increase in exact proportion to the line voltage.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.