I have been searching for definitive data that would explain the crosstalk level between triodes in dual triode tubes. For me, part of this hobby is to understand the physics and not just blindly do what appears acceptable. I have researched books and threads in diyAudio and other websites and could not find concrete information. I would appreciate others to review the test data I have created and comment.
My research suggests that the phono cartridge will dictate the channel separation of -20db at 1kHz as this is the average separation and higher quality cartridges would have -30db at 1kHz separation. Further, it is noted that vintage amplifiers regularly use dual triode tubes (and new designs too); therefore, from the perspective of the designers of these amplifiers, the crosstalk level must be acceptable. This would suggest that a crosstalk between the triodes of at least -30db at 1kHz is acceptable.
I decided to construct a test circuit and measure the crosstalk. The triode amplifier circuits used for all tubes were based on the resistance-coupled amplifier circuits in the RCA Receiving Tube Manual where Ebb=180V and Rp=47kohm. A sine wave generator was used for the input to the driver triode and the test triode was given a 47kohm input impedance. Crosstalk data was recorded for input levels of 50, 100, 500, and 1000 mV rms and at 1, 10 and 20 kilohertz. Separate plate voltage supplies were used for each triode to eliminate crosstalk through the plate power supply and for tests where two separate 6SN7 tubes were used, separate 6.3V DC heater supplies were applied.
6SN7GTB -54dB at 1 kHz, -34dB at 10 kHz, -29 dB at 20 kHz
12AU7A -74dB at 1 kHz, -53dB at 10 kHz, -49 dB at 20 kHz
E88CC -70dB at 1 kHz, -54dB at 10 kHz, -47 dB at 20 kHz
A photo of the test setup and a more detailed test result table is included in the attached files.
It appears that the crosstalk between the triodes within a single tube glass envelop should have minimal affect on the Channel Separation for a phono amplifier.
My research suggests that the phono cartridge will dictate the channel separation of -20db at 1kHz as this is the average separation and higher quality cartridges would have -30db at 1kHz separation. Further, it is noted that vintage amplifiers regularly use dual triode tubes (and new designs too); therefore, from the perspective of the designers of these amplifiers, the crosstalk level must be acceptable. This would suggest that a crosstalk between the triodes of at least -30db at 1kHz is acceptable.
I decided to construct a test circuit and measure the crosstalk. The triode amplifier circuits used for all tubes were based on the resistance-coupled amplifier circuits in the RCA Receiving Tube Manual where Ebb=180V and Rp=47kohm. A sine wave generator was used for the input to the driver triode and the test triode was given a 47kohm input impedance. Crosstalk data was recorded for input levels of 50, 100, 500, and 1000 mV rms and at 1, 10 and 20 kilohertz. Separate plate voltage supplies were used for each triode to eliminate crosstalk through the plate power supply and for tests where two separate 6SN7 tubes were used, separate 6.3V DC heater supplies were applied.
6SN7GTB -54dB at 1 kHz, -34dB at 10 kHz, -29 dB at 20 kHz
12AU7A -74dB at 1 kHz, -53dB at 10 kHz, -49 dB at 20 kHz
E88CC -70dB at 1 kHz, -54dB at 10 kHz, -47 dB at 20 kHz
A photo of the test setup and a more detailed test result table is included in the attached files.
It appears that the crosstalk between the triodes within a single tube glass envelop should have minimal affect on the Channel Separation for a phono amplifier.
Attachments
Interesting contribution! I would also like to see the ECC83/12AX7 added to that list, since it has the highest internal resistance of all the common triodes and might therefore be expected to have the worst crosstalk.
Also, I assume you mean the 50, 100, 500, and 1000 mV test levels were at the input, not the output, right?
Also, I assume you mean the 50, 100, 500, and 1000 mV test levels were at the input, not the output, right?
The fact that it worsens with frequency is the dead giveaway that it is capacitive. If the tests were continued at higher frequencies you should expect the usual 6dB/octave degradation.
could you explain further what you mean. would this mean a smaller Rp? or something else? what physical activity are talking about?
Greetings from FixitLand!
Thanks for this work; very interesting! I would suggest testing the 6CG7 (with center shield), 6FQ7 or 6CG7/6FQ7 (without the shield), 12AT7 (no shield) and 6/12DT8 (virtually same as 12AT7 but has shield). It would be of interest to see how much difference that shield actually makes (for audio use).
Take care,
--
J. E. Knox "The Victor Freak"
Thanks for this work; very interesting! I would suggest testing the 6CG7 (with center shield), 6FQ7 or 6CG7/6FQ7 (without the shield), 12AT7 (no shield) and 6/12DT8 (virtually same as 12AT7 but has shield). It would be of interest to see how much difference that shield actually makes (for audio use).
Take care,
--
J. E. Knox "The Victor Freak"
i could test a whole bunch of different tubes with dual triode construction but i think the results will be roughly the same. note that the E88CC has the internal shield and it improves the crosstalk considerably. i would suggest other tubes with the internal shield will test the same.
the real question i have is what does this mean regarding channel separation. there are amplifiers that offer specifications with absurdly high channel separation. channel separation greater than -30dB at 1kHz appears not required and therefore one should not try to spend a lot of time, money, and effort trying to better -30dB
the real question i have is what does this mean regarding channel separation. there are amplifiers that offer specifications with absurdly high channel separation. channel separation greater than -30dB at 1kHz appears not required and therefore one should not try to spend a lot of time, money, and effort trying to better -30dB
I think that you should also determine the PHASE of the capacitively coupled crosstalk; that is, whether the crosstalk at the second plate
is of the same polarity as the input to the first grid, or is of the opposite polarity.
If it is the same polarity, then the coupling could be from the first plate to the second grid (probably due to wiring/layout).
If it is the opposite polarity, the coupling could be from the first plate to the second plate (probably due to the tube's internal coupling).
Coupling to the cathode is less likely, and it can be bypassed to eliminate that source.
Last edited:
money......the ECC807 with an internal sep screen...now hard to come by, would be my solution. A pity that this tube wasn't continued.
My approach would be to use cheaper tubes, with a series type feedback which nicely lowers the impedances as explained in Morgan Jones 3th ed, page 499 but doesn't appear in my version of the 4th ed. Despite being somewhat more complex, I find this configuration a very useful building block.
richy
My approach would be to use cheaper tubes, with a series type feedback which nicely lowers the impedances as explained in Morgan Jones 3th ed, page 499 but doesn't appear in my version of the 4th ed. Despite being somewhat more complex, I find this configuration a very useful building block.
richy
Did you test it with and without the sheild grounded? I think we'd all like to see that (actually I'm susprised this wasn't your top story...?)
The fact that you got only 75dB separation between triodes in separate bottles suggests to me that most of what you're seeing is capacitance between the external wiring, rather than between the valve elements themselves.
I agree, endless tests of 'basically similar' valves is unecessary.
It is well known that separation of 30dB is more than enough to preserve the stereo image as far as human perception is concerned.
Last edited:
Couldn't disagree more. Proven by AES and other papers that for disconcerning HiFi listening, channel separation should be maintained with rising frequency esp in the critical midrange and above, that is to avoid degrading definition.
The recording industry esp console designs specify not less than -70dB crosstalk at 15KHz between adjacent channels. One can get that with a preamp using a well screened layout philosophy and/or separate tubes.
It seems a pity that most CD players do offer superlative channel separation specs, and to ruin those performance figures by a poorly constructed preamp seems a pity.
A Denon DCD 485 CD player offers 102dB separation at 1KHz.......your 30dB is throwing good performance away. Ignoring the practice of headphone blending where any separation harmonics are deliberately mixed; classicals certainly aren't and want to hear the truth fidelity.
richy
References?
30dB crosstalk is fine for loudspeaker reproduction. Far more mixing of the L+R channels occurs in the room itself. And with headphones, some narrowing of the stereo image is usually considered a good thing, so it does sound so much like it's in the centre of your own head.
Recording equipment has different design goals from home hi-fi. Obviously 30dB would be a tragic figure in a recording situation.
Last edited:
Possibly. However, this thread is about (pre)amplifiers.
BTW, 30dB @ 10KHz was the minimum allowed crosstalk specified in the ancient German DIN 45500 from the 1960's. I would not call this a reference by today's standards.
True, but it shouldn't be achieved by a design flaw.
Last edited:
Pin 9 was grounded on E88CC and 188. didnt test it ungrounded.
I suspect i could have done a lot of work to shield the circuits of the two triodes but does one go to all that trouble when building an actual amplifier? it appears from designs i have reviewed, efforts are taken to keep the circuits separated from each other, but there is no extra care except for the very initial stage, which i would only use a single triode tube anyways.
looking at the results for the 6SN7 a separation of -54dB at 1kHz is very acceptable and the results of -30dB at 10 kHz and higher is not bad. perhaps a little work on further arrangement of the circuit to provide better isolation of the circuits would optimize the channel separation.
Excerpt from article by Rane Corporation (RaneNote 145 by Dennis Bohn) regarding Audio Specifications, "the reason (lower channel separation) usually isnt a problem (at higher frequencies) is that the signal level at high frequencies is also reduced by about the same 6 dB/octave rate". of course i dont Dennis Bohn's credentials, but the article is well written and appears accurate.
I suspect i could have done a lot of work to shield the circuits of the two triodes but does one go to all that trouble when building an actual amplifier? it appears from designs i have reviewed, efforts are taken to keep the circuits separated from each other, but there is no extra care except for the very initial stage, which i would only use a single triode tube anyways.
looking at the results for the 6SN7 a separation of -54dB at 1kHz is very acceptable and the results of -30dB at 10 kHz and higher is not bad. perhaps a little work on further arrangement of the circuit to provide better isolation of the circuits would optimize the channel separation.
Excerpt from article by Rane Corporation (RaneNote 145 by Dennis Bohn) regarding Audio Specifications, "the reason (lower channel separation) usually isnt a problem (at higher frequencies) is that the signal level at high frequencies is also reduced by about the same 6 dB/octave rate". of course i dont Dennis Bohn's credentials, but the article is well written and appears accurate.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.