I am thinking primarily in terms of SET but I suspect the issues involved are kind of universal. Here is my thinking; please critique.
Load impedance on the output tube affects power output and distortion. Starting at some "optimum" load if one reduces the load impedance the maximum power delivery goes up but so does distortion. The question is simple but I suspect the answer is less so. Let us say that 5K is the normal load which gives good all around performance. If you were to reduce the load to say 4K but add local feedback such as plate to cathode could you get a best of both worlds situation with higher max output power but similar or even improved distortion due to the FB? Of course I am assuming sufficient gain.
On the one hand the plate curves are not changed (I don't think) but the local FB reduces effective driving impedance and reduces distortion in a brute force way.
So is this a reasonable way to make use of lower than optimal load trannies? Dumb idea?
Load impedance on the output tube affects power output and distortion. Starting at some "optimum" load if one reduces the load impedance the maximum power delivery goes up but so does distortion. The question is simple but I suspect the answer is less so. Let us say that 5K is the normal load which gives good all around performance. If you were to reduce the load to say 4K but add local feedback such as plate to cathode could you get a best of both worlds situation with higher max output power but similar or even improved distortion due to the FB? Of course I am assuming sufficient gain.
On the one hand the plate curves are not changed (I don't think) but the local FB reduces effective driving impedance and reduces distortion in a brute force way.
So is this a reasonable way to make use of lower than optimal load trannies? Dumb idea?
I believe the local cathode feedback reduces available power because the max available Vak is reduced.
Lowering distortion with fb always exacts some cost, all other things remaining the same.
Jan
Lowering distortion with fb always exacts some cost, all other things remaining the same.
Jan
I should have been a bit more clear. I am thinking in terms of feedback to the driver such as output plate to driver cathode.
OTL does it all the time to extremes (usually exceeding some ratings). More drive voltage (or current) needed. The outputs and B+ have to be able to support the extra current and Watts for the lower Z load plus the N Fdbk will be knocking the tube around some to get distortion to behave. The distortion spectrum will likely become less benign, even though overall being held within bounds. (not just 2nd H )
At least for some tubes. The GE 6L6GC datasheet does NOT support this. It seems to be optimized for max power output already. Normally going to a higher Zload improves power output capability (for a max'd out tube), but screen current distortion goes up if the plate curves have some curvature. So the optimum is found at a lower Z than max power. I guess if you are starting out from a max'd out power condition, then a lower Zload would not help.
At least for some tubes. The GE 6L6GC datasheet does NOT support this. It seems to be optimized for max power output already. Normally going to a higher Zload improves power output capability (for a max'd out tube), but screen current distortion goes up if the plate curves have some curvature. So the optimum is found at a lower Z than max power. I guess if you are starting out from a max'd out power condition, then a lower Zload would not help.
Last edited:
The form of the original question leaves too many unknowns. Is the B+ going to be lowered along with the Zload to stay within tube ratings, for example.
Just lowering Zpri without lowering B+ would increase power out, but would also increase tube dissipation a little more than proportionally.
The simple answer is that tube efficiency improves with higher B+ and an OT Zpri optimized to use that voltage. Distortion may be optimum at a different Zpri however. N Fdbk can fix that if needed, but at some penalty in harmonic distribution.
Just lowering Zpri without lowering B+ would increase power out, but would also increase tube dissipation a little more than proportionally.
The simple answer is that tube efficiency improves with higher B+ and an OT Zpri optimized to use that voltage. Distortion may be optimum at a different Zpri however. N Fdbk can fix that if needed, but at some penalty in harmonic distribution.
I am talking triode. So if I am following correctly the tube ratings also come into play. Being class A (SET) the maximum dissipation is at idle and assuming that any primary DC resistance difference is accounted for in the bias the lower dynamic impedance should not be a factor.
A good point about the change in HD spectrum. The sense I am getting is "it depends". My WAG is that a drastic difference in load would put you at a great disadvantage WRT the AC load line but that perhaps a small difference could be workable. Will have to take a close look at load lines but maybe for something like a 6EM7 you might be able to go to 4K-4.5K but not 2.5K.
A good point about the change in HD spectrum. The sense I am getting is "it depends". My WAG is that a drastic difference in load would put you at a great disadvantage WRT the AC load line but that perhaps a small difference could be workable. Will have to take a close look at load lines but maybe for something like a 6EM7 you might be able to go to 4K-4.5K but not 2.5K.
The optimum load given in datasheets is where you get max power and lowest distortions ... your statements are very debatable .
This is kind of what I did with my Corona amp. I'm using high impedance transmitting triodes in A2 with a 5k transformer and some local feedback. Normally you would use a 10k+ output transformer for these tubes and Zout would still be high.
The tubes do fine with the load (assuming I do my homework and get a good operating point) and I get super-low distortion with higher output power.
The tubes do fine with the load (assuming I do my homework and get a good operating point) and I get super-low distortion with higher output power.
There are 4 or 5 "optimums". Max power, max gain, min THD, min 2nd HD, min 3rd HD....
The differences may not be large. Which is "more optimum": 5 Watts at 5% or 3 Watts at 3%? Is 5% 2nd HD even offensive? On solo flute or dense orchestral?
Consider that on average, amps of this size run more like 1/10th Watt 95% of the time.
Build and enjoy.
Yes that was kind of my thinking and thus the quotes. Trying to get a feel for the amount of flexibility we have. It may be fairly broad especially if we are not equipped to make measurements that might disappoint us. 😀
Lots of things and tradeoffs to consider . . .
Distortion, max power @ x% distortion (Harmonic and Intermodulation), damping factor, and frequency response, etc.
6EM7 sounds good, my friend has a stereo amp with those, I was pleasently surprised, I liked the sound.
No negative feedback. On the bench, it tested OK, but not something that Marketing would publish.
6EM7 u is low, but the plate impedance is low too. Which is a good reason that negative feedback will probably not improve the bench measurements by much, and possibly will not make the sound much better either.
If you are concerned by plate load impedance, then remember to check your loudspeakers.
Most Loudspeakers impedance versus frequency varies over a very large range.
Most of them do not act anything at all like a proper Load Resistor.
If the DCR of the primary and secondary of the output transformer are very low, you can get a moderate damping factor with the 6EM7 750 Ohm plate impedance and a 4k primary.
4000/750 = 5.3 damping factor. Reasonable output transformer DCRs and the results might be a damping factor of 4.0, without negative feedback.
The Low impedance of a typical 8 Ohm rated speaker is often the same as the DCR, as checked with a DMM or Ohmmeter, perhaps 6 Ohms, or even 4 Ohms.
Some examples:
Low impedance below woofer resonance, high impedance at woofer resonance, and low impedance at about 200 or 400 Hz (closed box).
Low impedance way below Port tuning, then a rise, then low impedance at port tuning, then a rise, and low impedance at about 200 or 400 Hz (ported box).
If you have an 8 Ohm rated loudspeaker, you can try it on the 8 Ohm tap, and then try it on the 4 Ohm tap (if available).
I would probably try an output transformer that has a primary impedance between 3k and 4k.
Your B+ voltage and plate current will have to be within the 10W plate dissipation of the 6EM7.
Really low plate current will increase the plate impedance far above the 750 Ohm rating.
So I might tend to try a 3K or 3.5k primary, and use higher current and lower B+ voltage.
10 Watt plate dissipation is the same as a type 45 triode, but the 45 has over 2 times the plate impedance of the 6EM7, so the 6EM7 transformer can have 3 or 3.5k, instead of the 45 that drives 4800 Ohm primaries.
Occasionally, you can get a combination of amplifier and loudspeaker that sound bad.
But more often, they will sound good, as long as you do not play the system too loud.
You probably will not fill a 600 square foot room with a single ended 6EM7 and 85dB sensitivity loudspeakers, no matter what the loudspeaker impedance is.
As always, just my opinions.
"All Generalizations Have Exceptions"
If you build the 6EM7 amplifier, be sure to report the results back to us.
Happy building and Happy listening.
Distortion, max power @ x% distortion (Harmonic and Intermodulation), damping factor, and frequency response, etc.
6EM7 sounds good, my friend has a stereo amp with those, I was pleasently surprised, I liked the sound.
No negative feedback. On the bench, it tested OK, but not something that Marketing would publish.
6EM7 u is low, but the plate impedance is low too. Which is a good reason that negative feedback will probably not improve the bench measurements by much, and possibly will not make the sound much better either.
If you are concerned by plate load impedance, then remember to check your loudspeakers.
Most Loudspeakers impedance versus frequency varies over a very large range.
Most of them do not act anything at all like a proper Load Resistor.
If the DCR of the primary and secondary of the output transformer are very low, you can get a moderate damping factor with the 6EM7 750 Ohm plate impedance and a 4k primary.
4000/750 = 5.3 damping factor. Reasonable output transformer DCRs and the results might be a damping factor of 4.0, without negative feedback.
The Low impedance of a typical 8 Ohm rated speaker is often the same as the DCR, as checked with a DMM or Ohmmeter, perhaps 6 Ohms, or even 4 Ohms.
Some examples:
Low impedance below woofer resonance, high impedance at woofer resonance, and low impedance at about 200 or 400 Hz (closed box).
Low impedance way below Port tuning, then a rise, then low impedance at port tuning, then a rise, and low impedance at about 200 or 400 Hz (ported box).
If you have an 8 Ohm rated loudspeaker, you can try it on the 8 Ohm tap, and then try it on the 4 Ohm tap (if available).
I would probably try an output transformer that has a primary impedance between 3k and 4k.
Your B+ voltage and plate current will have to be within the 10W plate dissipation of the 6EM7.
Really low plate current will increase the plate impedance far above the 750 Ohm rating.
So I might tend to try a 3K or 3.5k primary, and use higher current and lower B+ voltage.
10 Watt plate dissipation is the same as a type 45 triode, but the 45 has over 2 times the plate impedance of the 6EM7, so the 6EM7 transformer can have 3 or 3.5k, instead of the 45 that drives 4800 Ohm primaries.
Occasionally, you can get a combination of amplifier and loudspeaker that sound bad.
But more often, they will sound good, as long as you do not play the system too loud.
You probably will not fill a 600 square foot room with a single ended 6EM7 and 85dB sensitivity loudspeakers, no matter what the loudspeaker impedance is.
As always, just my opinions.
"All Generalizations Have Exceptions"
If you build the 6EM7 amplifier, be sure to report the results back to us.
Happy building and Happy listening.
Last edited: