When tube manufacturers came up with their datasheets, how exactly did they settle on the values for loads and operating points given in the datasheets?
I've been playing lately with push pull EL84s, and I've noticed that with either fixed or cathode bias, with 300 volts B+, UL with either 43% or 20% taps, essentially any load from 5K to 10K works, with any load from 6K to 8K being "ideal" in a sense that there is no clear winner. Distortion is more or less minimized with the distortion profile of a 6K load with say 43% UL taps not being different from 8K with 20% UL taps, but not nearly different enough to declare a clear winner.
When drawing load lines manually for the EL84 in straight pentode mode, I've noticed the same. It's hard to find an "optimal" load for a given set of conditions as many load lines work and perform essentially the same.
How does one decide on the subjectively "best" load? Randomly? Eeny, meeny, miny, moe?
Moreover, and this may be another question for another subforum, but given the wide variation in impedance with respect to frequency exhibited by many loud-speakers, does it really even matter whether one picks 7K plate to plate vs 8k plate to plate? Especially if you're like me and run ESL-57s where the impedance varies from about 2 ohms all the way up to about 35 ohms within the audible range?
Finally, when I'm working on simulations, I'd like to have a good circuit to simulate the load impedance of the ESL-57. Is there any reason I couldn't substitute the model found on this page for the pure resistance I usually use when working in LTSpice as a substitute for a speaker?
I've been playing lately with push pull EL84s, and I've noticed that with either fixed or cathode bias, with 300 volts B+, UL with either 43% or 20% taps, essentially any load from 5K to 10K works, with any load from 6K to 8K being "ideal" in a sense that there is no clear winner. Distortion is more or less minimized with the distortion profile of a 6K load with say 43% UL taps not being different from 8K with 20% UL taps, but not nearly different enough to declare a clear winner.
When drawing load lines manually for the EL84 in straight pentode mode, I've noticed the same. It's hard to find an "optimal" load for a given set of conditions as many load lines work and perform essentially the same.
How does one decide on the subjectively "best" load? Randomly? Eeny, meeny, miny, moe?
Moreover, and this may be another question for another subforum, but given the wide variation in impedance with respect to frequency exhibited by many loud-speakers, does it really even matter whether one picks 7K plate to plate vs 8k plate to plate? Especially if you're like me and run ESL-57s where the impedance varies from about 2 ohms all the way up to about 35 ohms within the audible range?
Finally, when I'm working on simulations, I'd like to have a good circuit to simulate the load impedance of the ESL-57. Is there any reason I couldn't substitute the model found on this page for the pure resistance I usually use when working in LTSpice as a substitute for a speaker?
Very interesting experiments and findings. I had a similar question about “the ideal EL84 operating condition” wrt a planned amp build using the Baby Huey design. My loudspeakers exibit a near constant impedance of 8 ohms accross the audio band, so I hope that would make better sense to try to find the optimal operation conditions.
I hope experts with more experience will contribute to this thread.
I hope experts with more experience will contribute to this thread.
> how exactly did they settle on the values for loads and operating points given in the datasheets?
#1 rule: do not leave the customer wondering how to use your product! He needs "a number" for his parts. If "it doesn't matter", then pick a number and publish it!
In a range of possible values, pick "attractive numbers".
At any given time, some OT impedances are more common than others. 8K for 6F6 or 6V6 (or EL34), just because it works. 6.6K because that was on the 6L6 sheets. Also 3.8k, only for AB2... but then the 5881's ratings allowed higher Vg2 and pulling 4K in AB1. Bassman 5F6a is new bottles on (at first) old-design OTs.
Electrolytic caps generally run to 450V so any design over 400V is liable to be more expensive, less attractive.
Real high impedance means less bandwidth or more OT winding cost.
Real low impedance means high current to get the power, and this may embarrass your rectifier tube.
In your specific experiments--- the EL84 is a very soft-knee tube and UL connection makes it even more roundy. A wide variety of loadlines fit almost equally as well, if you optimize the B+ as you change loads.
For a different story: 7027 data page 8. This shows a *sharp* "null" at 6.6k. This is mildly bogus: the grid drive should be adjusted for "a low THD" rather than held constant and allowed to clip madly. But the selected conditions were drawn right TO the sharp knee of 7027, and any deviation means re-selecting all the conditions. Raising Vg2 would allow more current/power in low Z; reducing drive would tame that THD rise at high Z more than it would cut power.
#1 rule: do not leave the customer wondering how to use your product! He needs "a number" for his parts. If "it doesn't matter", then pick a number and publish it!
In a range of possible values, pick "attractive numbers".
At any given time, some OT impedances are more common than others. 8K for 6F6 or 6V6 (or EL34), just because it works. 6.6K because that was on the 6L6 sheets. Also 3.8k, only for AB2... but then the 5881's ratings allowed higher Vg2 and pulling 4K in AB1. Bassman 5F6a is new bottles on (at first) old-design OTs.
Electrolytic caps generally run to 450V so any design over 400V is liable to be more expensive, less attractive.
Real high impedance means less bandwidth or more OT winding cost.
Real low impedance means high current to get the power, and this may embarrass your rectifier tube.
In your specific experiments--- the EL84 is a very soft-knee tube and UL connection makes it even more roundy. A wide variety of loadlines fit almost equally as well, if you optimize the B+ as you change loads.
For a different story: 7027 data page 8. This shows a *sharp* "null" at 6.6k. This is mildly bogus: the grid drive should be adjusted for "a low THD" rather than held constant and allowed to clip madly. But the selected conditions were drawn right TO the sharp knee of 7027, and any deviation means re-selecting all the conditions. Raising Vg2 would allow more current/power in low Z; reducing drive would tame that THD rise at high Z more than it would cut power.
Can anyone answer my question regarding the supposed equivalent circuit for the Quad ESL 57s? It's easier to work out some of the feedback arrangements in SPICE and tweak it on the bench with the help of a scope. Getting something that is a good approximation of the ESL in SPICE would be a huge help.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.