A new project: a parallel 4x 6V6 with triode, pentode, and UL switch. Plus, adjustable global negative feedback from 0 to 9db. back to the cathode of driver tube.
Each 6v6 will be individually cathode biased and have its own coupling cap from the driver tube.
OPT will be a 1.2K. Edcor, Sowter, Lundhall or equivalent.
Power supply... no idea. Just a low noise one.
I am trying to figure out driver tube. Pentode, Triode? has to have enough gain for 9db of feedback plus decent input sensitivity and be able to drive a 100K load and the input capacitance of 4 6V6's. Thoughts are a EL84 in pentode mode. 4-8K resistive load. 150v screen, 300v V+, 150v plate, 37mA plate current.
Thoughts? Or should I go with a triode driver?
probably 6W triode, 16W UL, 20W Pentode .... maybe. lol.
All of this is still in very early design stages. But I am pretty fixed on using the parallel 6V6s.
Thank you sooo much!
Each 6v6 will be individually cathode biased and have its own coupling cap from the driver tube.
OPT will be a 1.2K. Edcor, Sowter, Lundhall or equivalent.
Power supply... no idea. Just a low noise one.
I am trying to figure out driver tube. Pentode, Triode? has to have enough gain for 9db of feedback plus decent input sensitivity and be able to drive a 100K load and the input capacitance of 4 6V6's. Thoughts are a EL84 in pentode mode. 4-8K resistive load. 150v screen, 300v V+, 150v plate, 37mA plate current.
Thoughts? Or should I go with a triode driver?
probably 6W triode, 16W UL, 20W Pentode .... maybe. lol.
All of this is still in very early design stages. But I am pretty fixed on using the parallel 6V6s.
Thank you sooo much!
Is that 4 per channel? Assuming, yes. Mono or stereo?,.... with 1 PT?
Last edited:
Do we really need that powerful driver for 4x6V6s?
Cin = Cgk + Cgp*(A+1) = 9 + 0.7*(9.8+1) = 16.56p
4*Cin = 66.24p
Impedance at 20Khz: ~120K. In parallel with the grid leak (470k/4, not sure whether we need to divide by 4).
Something like an ECC88 should be enough, right? Even if we take into account stray capacitances.
Cin = Cgk + Cgp*(A+1) = 9 + 0.7*(9.8+1) = 16.56p
4*Cin = 66.24p
Impedance at 20Khz: ~120K. In parallel with the grid leak (470k/4, not sure whether we need to divide by 4).
Something like an ECC88 should be enough, right? Even if we take into account stray capacitances.
4 6V6 per channel. probably 2 monoblocs. 6V6 is an option. It seems like I can better gain etc. out of the EL84. I am open to less powerful drivers. It would need to fit the requirements listed. Preferably an input tube where I can switch out different brands and new vs nos. etc. thats why i was looking at the EL84 etc.
Is that Cin calculation for pentode connection only? Or Triode?
Is that Cin calculation for pentode connection only? Or Triode?
I believe for triode. Pentode capacitance should be lower.
About the driver, you could use an ECL86 or 84, those have a high gain triode and a power pentode. Use the triode for gain, and the pentode triode-strapped as a cathode follower.
I have a 6V6 driving a 6V6 with plate to plate feedback. Sounds really good to me. Got the design from here:
https://hifihaven.org/index.php?threads/sakuma-schade-6v6-amp.8961/page-2#post-200479
https://hifihaven.org/index.php?threads/sakuma-schade-6v6-amp.8961/page-2#post-200479
Paralleled sections of a 12ax7 would be easy and have all the gain headroom you'd need for adding GNFB. The two sections together would drive 100K no sweat.
Keep your total idle current in mind in your planning. Your OPT will need to carry 150mA or more to have a good centered operating point on the load line. If you have to squeeze the idle current you'll be closer to the cutoff point and can't push the input signal as high for total power you might want.
Last edited:
or paralleled 6SL7 sections. 25K Rl. 1.6v bias. 3.8mA . gain is 77. This option seems to be more likeable. Gain of 77 should be plenty even with 9db of feedback.
How much feedback is recommended for a SEP output stage?
How much feedback is recommended for a SEP output stage?
I use DHT drivers and so I need an output tube with a gain of around 18. I use EL12 or EL12n. You could use 7591 or PSE EL84s. What you then get is what I call an "Inverted DHT Amp" where the DHT is the driver tube not the output. You get all that special DHT clarity. I use a 26 but you could use a 10Y or 2P29L/4P1L in triode. True triodes do sound better. I can use any tube with a gain or around 9, like 27, 37 or 6V6/6G6G in triode. Of those I'd recommend a 27 or 37 as being the closest to a 26. Personally I use 26 and rotate with 37, I like both.
With PSE 6V6 you can use a driver tube with a gain of 14 like 76, 56 or 6P5. Very nice tubes and identical except for the base and heaters. Or you can just use a 6SN7 or 6J5. All these are better than 9 pin tubes in my listening experience. I haven't used 9 pin tubes for a long time with rare exceptions like the EL84.
I never use feedback with SE amps. I'm a bit of a purist. Hence the lower mu drivers.
With PSE 6V6 you can use a driver tube with a gain of 14 like 76, 56 or 6P5. Very nice tubes and identical except for the base and heaters. Or you can just use a 6SN7 or 6J5. All these are better than 9 pin tubes in my listening experience. I haven't used 9 pin tubes for a long time with rare exceptions like the EL84.
I never use feedback with SE amps. I'm a bit of a purist. Hence the lower mu drivers.
AudioGeek,
Great idea!
Just a friendly reminder to all.
Keep your amplifier, speakers, and ears from harm:
Never switch from one mode to another mode, when the amplifier is powered (Pentode, Ultra Linear, Triode).
Transients can blow out amplifiers, tweeters, and ears.
g1 to g2 capacitance < 9pF. 4 x <9pF = < 36pF
Ultra Linear mode might, or might not, have more gain from g1 to the screen g2; versus Triode mode gain from g1 to g2.
Gain affects Miller Effect Capacitance.
And the UL tap % determines the plate to screen gain. 75% or 80% UL tap may have more Miller Effect capacitance than Triode mode.
I once used 80% UL tap for single ended.
Plate to Plate feedback often affects the distortion of the driver stage.
Triode mode has less gain and lower plate impedance, so . . .
In triode mode, with the correct ratio of primary impedance to 4 parallel plates impedance, the need for negative feedback is questionable.
With the total amplifier gain lower, not needing negative feedback is a plus, because with 9 dB of negative feedback, the gain of 2 stages will probably not be adequate.
Negative feedback needed?
Triode mode - probably Not
Ultra Linear mode - Borderline
Pentode mode - Certainly
I have done all those modes on a single chassis, with the rest of the circuit the same, only the output mode was different.
Just my opinions
Have Fun!
Great idea!
Just a friendly reminder to all.
Keep your amplifier, speakers, and ears from harm:
Never switch from one mode to another mode, when the amplifier is powered (Pentode, Ultra Linear, Triode).
Transients can blow out amplifiers, tweeters, and ears.
g1 to g2 capacitance < 9pF. 4 x <9pF = < 36pF
Ultra Linear mode might, or might not, have more gain from g1 to the screen g2; versus Triode mode gain from g1 to g2.
Gain affects Miller Effect Capacitance.
And the UL tap % determines the plate to screen gain. 75% or 80% UL tap may have more Miller Effect capacitance than Triode mode.
I once used 80% UL tap for single ended.
Plate to Plate feedback often affects the distortion of the driver stage.
Triode mode has less gain and lower plate impedance, so . . .
In triode mode, with the correct ratio of primary impedance to 4 parallel plates impedance, the need for negative feedback is questionable.
With the total amplifier gain lower, not needing negative feedback is a plus, because with 9 dB of negative feedback, the gain of 2 stages will probably not be adequate.
Negative feedback needed?
Triode mode - probably Not
Ultra Linear mode - Borderline
Pentode mode - Certainly
I have done all those modes on a single chassis, with the rest of the circuit the same, only the output mode was different.
Just my opinions
Have Fun!
I do want to be able to use the amp as a PSE Pentode also... so hence the option of some GNFB. I think i should be fine with a paralleled 6SL7 as the input tube. I agree no feedback needed in triode, but for sure in pentode.
Should I design the amp for the correct gain for medium feedback and UL? 1V input for full output. Then the assumption is that triode / pentode mode will either get no feedback or full feedback. i.e. triode mode will have very high gain overall and pentode mode very low gain... due to amount of gnfb on each.
I will not be using plate to plate feedback.
Should I design the amp for the correct gain for medium feedback and UL? 1V input for full output. Then the assumption is that triode / pentode mode will either get no feedback or full feedback. i.e. triode mode will have very high gain overall and pentode mode very low gain... due to amount of gnfb on each.
I will not be using plate to plate feedback.
nevermind its not gain of 70 its at the most 50 if its a resistive load. But still I think the 6SL7 would work fine. with 9db of feedback should reduce the gain of the input stage to 17 which is plenty to drive the 6V6 in pentode mode. Could even run the 6SL7 with no cathode cap and still have plenty of gain for pentode and 9db. Reduces it to a gain of 12.
Thoughts?
Thoughts?
Just thinking out loud, I like to calculate for a single ouput tube; after that to convert for 2, 3, 4 tubes in parallel . . .
Round numbers off the top of the head calculations:
Output Stage Gain:
First convert output transformer number for a single tube:
1.2k 4-parallel = 4.8k. 5k for ease of calculation.
6V6 transconductance 4mA/V Pentode mode
Gain = 4mA x 5k = 20, pentode mode gain is 20
Plate impedance, rp, 50k Ohms
Damping factor = 5k/50k = 0.1, far less than Unity
Negative feedback needed
6V6 u = 10 Triode mode
6V6 plate impedance, rp, 2k Ohms
Gain = u x (5k/(2k + 5k). 10 x (5k/7k) = 7 Triode mode gain is 7
Damping Factor = 5k/2k = 2.5
Ultra Linear mode's gain and damping factor sits in the middle of Pentode and Triode modes.
Where in the middle? Depends on the UL tap percentage (%)
Pentode = 0%
Triode = 100%
Ultra Linear = 20% to 80% (as I have seen in the literature).
Have fun!
Round numbers off the top of the head calculations:
Output Stage Gain:
First convert output transformer number for a single tube:
1.2k 4-parallel = 4.8k. 5k for ease of calculation.
6V6 transconductance 4mA/V Pentode mode
Gain = 4mA x 5k = 20, pentode mode gain is 20
Plate impedance, rp, 50k Ohms
Damping factor = 5k/50k = 0.1, far less than Unity
Negative feedback needed
6V6 u = 10 Triode mode
6V6 plate impedance, rp, 2k Ohms
Gain = u x (5k/(2k + 5k). 10 x (5k/7k) = 7 Triode mode gain is 7
Damping Factor = 5k/2k = 2.5
Ultra Linear mode's gain and damping factor sits in the middle of Pentode and Triode modes.
Where in the middle? Depends on the UL tap percentage (%)
Pentode = 0%
Triode = 100%
Ultra Linear = 20% to 80% (as I have seen in the literature).
Have fun!
9 dB. OK
Please look at that in voltage gain or voltage loss terms:
20 x Log (2.82) = + 9.00dB
20 x Log (1/2.82) = - 9.00dB
The difference in gains . . . with, and without, 9 dB negative feedback is: 2.82
Thanks for reading.
Tell me if I am wrong, please
Please look at that in voltage gain or voltage loss terms:
20 x Log (2.82) = + 9.00dB
20 x Log (1/2.82) = - 9.00dB
The difference in gains . . . with, and without, 9 dB negative feedback is: 2.82
Thanks for reading.
Tell me if I am wrong, please
yes thats what I get 2.82. So if we have a 6SL7 with a gain of 50 with 9 db of feedback from output then the equivalent gain of it would be 50/2.82 or about 17. With an input of 1V and a gain of 17 on input we can drive the pentode to full output.
Is my math correct?
How much feedback does a SEP need in reality?
Is my math correct?
How much feedback does a SEP need in reality?
I do not have a number of how much negative feedback an SEP needs.
Want a number? Try one dB amount of NFB.
Then do 2 things:
1. Measure the results of Distortion (H and IM), and Damping Factor, on a test bench, with a proper load resistor.
2. Connect the loudspeaker you are going to use, and listen.
Bass flabby? Distorted sounding mids and highs? Bad frequency response of the amplifier/loudspeaker combination? Then increase the amount of negative feedback.
No life in the sound (dead sounding)? then reduce the amount of negative feedback.
Measure again; listen again.
Iteration is the key!
Before I ever learned the longhand method to take the square root of a number, I used Iteration.
Then along came the slide rule, and later the HP 11 (thank God! I have more important problems to solve)
HP 11
JStewart9, are you listening, is this microphone on (are you reading)? I Love my HP 11 (and I am a Tektronix man; but I love some HP T & M gear too!).
Too many variables is why there is no magic dB number of negative feedback.
Want a number? Try one dB amount of NFB.
Then do 2 things:
1. Measure the results of Distortion (H and IM), and Damping Factor, on a test bench, with a proper load resistor.
2. Connect the loudspeaker you are going to use, and listen.
Bass flabby? Distorted sounding mids and highs? Bad frequency response of the amplifier/loudspeaker combination? Then increase the amount of negative feedback.
No life in the sound (dead sounding)? then reduce the amount of negative feedback.
Measure again; listen again.
Iteration is the key!
Before I ever learned the longhand method to take the square root of a number, I used Iteration.
Then along came the slide rule, and later the HP 11 (thank God! I have more important problems to solve)
HP 11
JStewart9, are you listening, is this microphone on (are you reading)? I Love my HP 11 (and I am a Tektronix man; but I love some HP T & M gear too!).
Too many variables is why there is no magic dB number of negative feedback.
~6db NFB will get you Damping Factor about the same as SET.
And does that without too many problems regarding instability.
Much depends on the quality of the OPT.👍