Hi All,
I am interested to hear about any user experiences or advice regarding output transformer impedance and B+ combinations for KT88 singled ended ultralinear operation.
My current amp is running EL34's with a 5k transformer right on the limit, with 412V across the tube and 60mA bias. I am very happy with the performance.
For the next amp I was contemplating running KT88's with a 5k transformer, about 400V across the tube and 100mA. However, looking at the KT88 datasheet it seems this tube might be most comfortable with 2.5 to 3k and 250V-300V across the tube.
I'm intending to run local cathode feedback on the output stage.
Interested to hear what you're all running.
Also can someone please explain the significance of the dashed lines on the curves below? I take it 425V is the highest voltage I can run across the tube and ther other one is presumably showing the grid 2 dissipation limit under some conditions I don't fully understand.
Many thanks,
Greg
I am interested to hear about any user experiences or advice regarding output transformer impedance and B+ combinations for KT88 singled ended ultralinear operation.
My current amp is running EL34's with a 5k transformer right on the limit, with 412V across the tube and 60mA bias. I am very happy with the performance.
For the next amp I was contemplating running KT88's with a 5k transformer, about 400V across the tube and 100mA. However, looking at the KT88 datasheet it seems this tube might be most comfortable with 2.5 to 3k and 250V-300V across the tube.
I'm intending to run local cathode feedback on the output stage.
Interested to hear what you're all running.
Also can someone please explain the significance of the dashed lines on the curves below? I take it 425V is the highest voltage I can run across the tube and ther other one is presumably showing the grid 2 dissipation limit under some conditions I don't fully understand.
Many thanks,
Greg
I attached 2 snapshots see what you can make of them. The screen voltage can be equal but not more than anode voltage , provided the total screen dissipation (under DC and AC condition) is not exceeded, so if Vg2 is 425V the screen current is about 11mA. Reduce cathode current will reduce screen/anode current proportionally (when Vg2 is same as Va) so they are both within dissipation limit.
I try to sim UL mode SE, OT impedance can be as low as 1.6K but the gain drops, increase distortion by 0.5%, power out reduced by 2 (6-4) Watts, but damping factor improved.
I try to sim UL mode SE, OT impedance can be as low as 1.6K but the gain drops, increase distortion by 0.5%, power out reduced by 2 (6-4) Watts, but damping factor improved.
Attachments
Last edited:
Thanks Koonw, that makes sense.
I wish I could use LT spice that well! Any chance you can share the ASC files here?
There is no problem using a 5k opt, you just get a little less power and lower distortion as compared to a 3.5k unit. Make sure you’re not running it too hot and you’re fine. I run mine at 400v on a 5k opt and love the sound in both triode and UL.
Member
Joined 2009
Paid Member
When I was researching SE KT8k designs I invariably found that many well regarded units were using 5k:8r
Sure they're attached here. Remove ext .txt and open plot in correspinding simulation.
Thanks, that's identical to what I was thinking of using so good to hear.
Excellent news, thanks!
That's 40W or 114% over the tube's rating for a KT88. I've run KT88SE at 80ma with 410V across the tube into a 5K load and it sounded great.
Thanks Steph,
The absolute maximum dissipation is 42W but I see the genalex datasheet does list a design maximum of 35. In that case I'll run it at 33-34 or so. Still tossing up the exact voltage and OPT configuration, it's tempting to try going for more output power this time and run some feedback....just to see what it's like. Hmmm....
You could also use KT120s I the same amp if you are worried about power dissipation. I often use the 120s in a amp that had 88s in it. Just make sure your power tra can handle a little extra heater current.
Well if you want to get picky, it's 41.2W... I was just pointing out it will likely red plate at what you intended to run it at.
I reckon I can get by with an 88 but that 120 sure is a beast of a tube!
Yep, I agree. As I said I missed the recommmended design maximum of 35W on the genalex datasheet - the JJ one I had been using doesn't mention it as pe below...
Actually....looking at it again are you sure it can't do 42W? It even provides grid resistor values for use over 35W....I think?
Different manufacturers have (slightly) different absolute ratings. There are also three different ratings confusing many. The last common practise beofre the demise of mass produced tubes was to use the absolute maximum ratings which is approx 10% above the design center rating. If you want longevity then you might want to stay close to the design center rating.
https://www.tubecad.com/december2000/page17.html
https://www.tubecad.com/december2000/page17.html
Last edited:
Take an amplifier that can output 10 Watts.
Modify that amplifier to put out +1 dB more power. That is 12.6 Watts.
Modify that amplifier to put out +2 dB more power. That is 15.8 Watts.
What is your Perceived increase in sound intensity?
Squeezing just a little bit more power at the expense of reliability and tube life is your choice.
A very old rule-of-thumb for purchasing a more powerful amplifier was that it should have at least double the power (+3 dB).
(10 Watts increased to 20 watts).
Parallel Single Ended, anybody?
"Paralleling Tubes Effects" the cover article of Glass Audio, Volume 12, Number 5, 2000 (the very last issue of Glass Audio;
After that Glass Audio magazine was combined with two other publications by Ed Dell).
Parallel tubes done properly, gives 2X the power, with the same distortion, damping factor, and the same sound character as the 1X amplifier.
Modify that amplifier to put out +1 dB more power. That is 12.6 Watts.
Modify that amplifier to put out +2 dB more power. That is 15.8 Watts.
What is your Perceived increase in sound intensity?
Squeezing just a little bit more power at the expense of reliability and tube life is your choice.
A very old rule-of-thumb for purchasing a more powerful amplifier was that it should have at least double the power (+3 dB).
(10 Watts increased to 20 watts).
Parallel Single Ended, anybody?
"Paralleling Tubes Effects" the cover article of Glass Audio, Volume 12, Number 5, 2000 (the very last issue of Glass Audio;
After that Glass Audio magazine was combined with two other publications by Ed Dell).
Parallel tubes done properly, gives 2X the power, with the same distortion, damping factor, and the same sound character as the 1X amplifier.
If you have a multi tap output transformer (e.g. the ISO (Tango) FC-20S) then you can implement cathode feedback with global negative feedback. Extensive modelling on this with the 7868 (EL506) shows pentode mode with CFB and GNFB is preferable over UL & CFB or UL & GNFB. YMMV.
Experienced listeners can just (barely) perceive a 2dB difference in volume. Output transformers can loose a substantial amount, it is a rare OPT that has less than 0.5dB loss, the more "affordable" have more like 1dB+ loss.
EL506,
Thanks for reminding us of output transformer insertion loss.
I have written about it several times, in multiple threads of Tubes / Valves.
Your research of the 7868 / EL506 seems to indicate that tube is not very well suited for Ultra Linear operation.
There are other tubes that are not well suited for UL, or that at least need special treatment, like reducing the quiescent DC voltage from the UL tap (which then with downward signal swing, makes the screen voltage so low that the plate current starts to not increase as much as it should).
Thanks for reminding us of output transformer insertion loss.
I have written about it several times, in multiple threads of Tubes / Valves.
Your research of the 7868 / EL506 seems to indicate that tube is not very well suited for Ultra Linear operation.
There are other tubes that are not well suited for UL, or that at least need special treatment, like reducing the quiescent DC voltage from the UL tap (which then with downward signal swing, makes the screen voltage so low that the plate current starts to not increase as much as it should).
The EL506 was the last iteration of the 7591 /7868 / 6GM5 and was designed for 40% (50%) PP UL. Using CFB, GNFB and a (cap bypassed) 22V zener to drop screen 20V below plate changes a lot when running in SE. In UL and CFB about 5.5 ~ 6 Watt is possible where with pentode - CFB - GNFB it is 6.5 ~ 7 Watts at slighlty (!) higher distortion levels. (about 4dB GNFB) Dennis Grimwood (https://www.oestex.com/tubes/ul.html) considers it one of the better tubes for UL. At same (lower) output levels it is much the same with the CFB-GNFB having the edge in the 1mW - 10mW region. (about 40dB down).
The difference is about 0.5% distortion in UL-CFB vs about 0.8% in CFB - GNFB mode. But in UL-CFB drive gets right to grid current whereas in CFB-GNFB there is still some extra capacity to drive further (without increase in output, only distortion goes up) before grid current happens. Due to the grid so close to the cathode the 7591 and derivatives do not take kindly to grid current, i.e. red plating is not uncommon, actually thermal runaway is almost garanteed.
The difference is about 0.5% distortion in UL-CFB vs about 0.8% in CFB - GNFB mode. But in UL-CFB drive gets right to grid current whereas in CFB-GNFB there is still some extra capacity to drive further (without increase in output, only distortion goes up) before grid current happens. Due to the grid so close to the cathode the 7591 and derivatives do not take kindly to grid current, i.e. red plating is not uncommon, actually thermal runaway is almost garanteed.
Last edited: