I bet you must be an .... Engineer!!!!
And what does that mean?:
"An engineer can do for a dollar what any fool can do for two"
Back on topic, I think that the OP would be fine with a SE 6L6GC amp if he/she can live with around 6-8W output, so long as there is some negative feedback applied. If more power is needed, either a SE KT88 or parallel 6L6s would be needed.
As far as 6L6 vs EL34, I think it's a toss up. The difference in sound is going to come down more to implementation than the individual tube. You can blow a bunch of money on a 300B amp if you want, and it may sound better, or it may not. I personally love the sound of a good 6L6 amp.
Again, just watch the parallel 6L6s. Grid stoppers, screen grid stoppers, and even a parasitic trap on the plate aren't a bad idea, especially with feedback involved. 6L6s have been used in many successful HF transmitters, and I get the distinct feeling that this power amp wasn't intended as a gateway into amateur radio.
All that said, I'm partial to the 6L6 family of tubes and love the sound of them.
As far as 6L6 vs EL34, I think it's a toss up. The difference in sound is going to come down more to implementation than the individual tube. You can blow a bunch of money on a 300B amp if you want, and it may sound better, or it may not. I personally love the sound of a good 6L6 amp.
Again, just watch the parallel 6L6s. Grid stoppers, screen grid stoppers, and even a parasitic trap on the plate aren't a bad idea, especially with feedback involved. 6L6s have been used in many successful HF transmitters, and I get the distinct feeling that this power amp wasn't intended as a gateway into amateur radio.
All that said, I'm partial to the 6L6 family of tubes and love the sound of them.
Lots of good advice so far, but if I could add a general note: for low distortion in both the output valves and the output transformer you want the loading on the valves to be as light as possible. (This is also true of modern transistor amplifiers). This also optimizes loudspeaker damping, so for more performance in these areas, lighter loading.
Because engineering is just balancing competing goals, lighter loading has costs to pay: power output will be comparatively lower, thermal efficiency lower, output transformers will need larger cores to work at the larger idle currents (in a single-ended amplifier context) for the same output capabilities, driver stages will need to be more robust to pump current into Miller capacitances, etc.
Folks are often quite happy with amplifiers (especially if they've just built one and it didn't oscillate or hum terribly or melt - been there) that stray a factor of "several" away from optimum. But this impedance ratio, source to load, is the beginning of an amplifier design. A ratio of one to two is a very poor choice IMO, although it would be expected in a commercial design. One to ten would begin to cause serious problems in the design of the output transformer. Start with an expected load, and work out from there.
All good fortune,
Chris
Because engineering is just balancing competing goals, lighter loading has costs to pay: power output will be comparatively lower, thermal efficiency lower, output transformers will need larger cores to work at the larger idle currents (in a single-ended amplifier context) for the same output capabilities, driver stages will need to be more robust to pump current into Miller capacitances, etc.
Folks are often quite happy with amplifiers (especially if they've just built one and it didn't oscillate or hum terribly or melt - been there) that stray a factor of "several" away from optimum. But this impedance ratio, source to load, is the beginning of an amplifier design. A ratio of one to two is a very poor choice IMO, although it would be expected in a commercial design. One to ten would begin to cause serious problems in the design of the output transformer. Start with an expected load, and work out from there.
All good fortune,
Chris
Agreed. If I could have obtained good iron for a decent price, I would have used an 8k primary transformer for my PP 1625 amplifier. Unfortunately, short of ordering custom winds from Electra-Print, there isn't much good 8-10k iron in suitable wattages if you dont want to pay a fortune for it.
Check out One Electron's offerings... they've got some nice stuff.
Check out One Electron's offerings... they've got some nice stuff.
Chris Hornbeck,
Just a few comments on your post # 50:
A. Matching loudspeaker impedances by using an output transformer: (There Is hope, it is not as bad as it seems; by Calculation the bumblebee can Not fly, But . . .)
Many loudspeakers have impedance ratings of 8, 6, or 4 Ohms. Many of those same loudspeakers have actual impedance that is up to 16, 20, or 40 Ohms at the one bass impedance resonance (closed box) or two bass impedance resonances (ported box). That may be the high impedance extreme of the loudspeaker. (**)
Many of those same loudspeakers have impedances at some frequencies that are nearly as low as their DCR. With bass resonance at 50Hz, the impedance from DC up to 25Hz may about equal to their DCR. The DCR may be about as low as 2/3 or even only 1/2 of the manufacturer’s rated impedance (that could be 2, 3, and 4 Ohms for the 4, 6, and 8 Ohm rated loudspeakers). Sometimes, the loudspeaker impedance decreases at frequencies above the bass resonance(s), but below the crossover frequency impedance peak. This might occur at perhaps 300Hz, and be as low as the DCR (i.e. 4 Ohms for that ‘8’ Ohm loudspeaker). This frequency range is more important than that same 4 Ohms occurring below 25Hz.
Many of those same loudspeakers may have impedances as high as 15, 20, or 30 Ohms at the crossover frequency. That also may be the extreme high impedance of the loudspeaker (**). A good 2-way speaker with a 6dB/octave crossover might have this impedance resonance peak at 1.5kHz. This is near to the range of the ear’s most sensitive frequencies. This frequency range may be the most important to pay attention to how it interacts with the amplifier.
1. Tubes: Now . . . we are going to try and get a reasonable damping factor, low distortion, etc. So we will consider the output tube’s plate impedance, and the range of the loudspeaker impedances (which might range from 4, to 40, to 4, to 20, to 8 Ohms from 20 to 20,000 Hz). The important loudspeaker impedances to take into account are the lowest impedances and highest impedances for the range of frequencies you want the amplifier to cover. You can see one of the reasons that some go to the trouble of creating Bi Amp and Tri Amp systems.
2. Solid State: As to using an output transformer on a real Hi Fi Transistor amp, the only times I have ever seen them . . . they were on MacIntosh Transistor amps. This year, I was surprised to see Output transformers on a current production MacIntosh Transistor amp, at the same Hi Fi shop where I saw the old design MacIntosh Transistor amp 50 years ago.
There are some transistor amplifiers that change their + and - DC voltages, according to the 4 or 8 Ohm setting switch. With the switch set to 4 Ohm, the voltage reduced to about 70% of the 8 Ohm DC voltage switch setting. That is because for the same power into 4 Ohms, you need 1.414 times the current, but only 0.707 times the voltage. That uses an engineering technique of some Lab DC power supplies, which change the tap on the power transformer secondary, according to the DC output setting (keeping the power dissipation of the regulator transistor relatively constant).
Output impedance of an emitter follower: The first order calculation of output impedance is 26 / mA emitter current. So, a standing current of 52mA = 1/2 Ohm output impedance. 52mA barely tickles an 8 Ohm loudspeaker (so little class A power, we go into AB almost immediately).
Let us use a PNP and NPN in a vertical series connection totem pole emitter follower output: With 52 mA in the output stage, the two 1/2 Ohm emitter impedances are in parallel, the output impedance is 1/4 Ohm (0.25 Ohm). For a ‘4’ Ohm loudspeaker, 4 Ohm / .025 Ohm = a damping factor of 16, without (and before) any global negative feedback. Applying negative feedback makes the output impedance go even lower, & the damping factor to increase. With + and - 40V supplies, and 52mA, we have 4.16 Watts quiescent power in the output transistors, a class AB amp, almost a class B amp, which is why we need to have global feedback. With 52mA peak current into an 8 Ohm load, we only get 22mW of class A power.
B. A single ended output transformer that has a higher impedance needs less current to get the same output power as the lower impedance transformer.
P = Isquared x R
P = 50 mA x 5k = 12.5 W Call the ‘Amp x turns’ 100% 0.050A x 1 turn = 0.050 Amp Turn
P = 35.355339 mA x 10k = 12.5 W 10k requires Root(2) more turns than 5k Root(2) x 35.355339 mA = 1.414213562 x 35.355339 mA = 0.050 Amp Turn
The ‘Amp x turns’ of a 5k transformer at 50mA is the same as the ‘Amp x turns’ of a 10k transformer at 35.3mA. This means the single ended core of both of these transformers is exactly the same for the same 12.5 Watt SE transformer. And both of these transformers require an Air Gap.
A push pull transformer neither needs nor has the Air Gap, and the balanced idle currents of push and pull keeps the core from saturating at the quiescent currents. Those two factors means we can use a smaller core, just as you said. But if there is less idle current than in the single ended amp, it must be remembered that the push pull amp is only class AB or class B, not class A like the single ended amp (the push pull is only class A for the first region of its power range).
C. Miller capacitance is negligible in a pentode output stage (including a single ended pentode with Schade feedback). The Schade feedback resistor does require the driver stage to put out more current, but the miller capacitance of the pentode does not. We can neglect the miller capacitance of a 6L6 or EL34 operating as a pentode. But I usually use those tubes in triode wired mode, and then we do need to take miller capacitance into account. I am now investigating Ultra Linear mode, where miller capacitance is less than triode operation, but more than pentode operation.
Just a few comments on your post # 50:
A. Matching loudspeaker impedances by using an output transformer: (There Is hope, it is not as bad as it seems; by Calculation the bumblebee can Not fly, But . . .)
Many loudspeakers have impedance ratings of 8, 6, or 4 Ohms. Many of those same loudspeakers have actual impedance that is up to 16, 20, or 40 Ohms at the one bass impedance resonance (closed box) or two bass impedance resonances (ported box). That may be the high impedance extreme of the loudspeaker. (**)
Many of those same loudspeakers have impedances at some frequencies that are nearly as low as their DCR. With bass resonance at 50Hz, the impedance from DC up to 25Hz may about equal to their DCR. The DCR may be about as low as 2/3 or even only 1/2 of the manufacturer’s rated impedance (that could be 2, 3, and 4 Ohms for the 4, 6, and 8 Ohm rated loudspeakers). Sometimes, the loudspeaker impedance decreases at frequencies above the bass resonance(s), but below the crossover frequency impedance peak. This might occur at perhaps 300Hz, and be as low as the DCR (i.e. 4 Ohms for that ‘8’ Ohm loudspeaker). This frequency range is more important than that same 4 Ohms occurring below 25Hz.
Many of those same loudspeakers may have impedances as high as 15, 20, or 30 Ohms at the crossover frequency. That also may be the extreme high impedance of the loudspeaker (**). A good 2-way speaker with a 6dB/octave crossover might have this impedance resonance peak at 1.5kHz. This is near to the range of the ear’s most sensitive frequencies. This frequency range may be the most important to pay attention to how it interacts with the amplifier.
1. Tubes: Now . . . we are going to try and get a reasonable damping factor, low distortion, etc. So we will consider the output tube’s plate impedance, and the range of the loudspeaker impedances (which might range from 4, to 40, to 4, to 20, to 8 Ohms from 20 to 20,000 Hz). The important loudspeaker impedances to take into account are the lowest impedances and highest impedances for the range of frequencies you want the amplifier to cover. You can see one of the reasons that some go to the trouble of creating Bi Amp and Tri Amp systems.
2. Solid State: As to using an output transformer on a real Hi Fi Transistor amp, the only times I have ever seen them . . . they were on MacIntosh Transistor amps. This year, I was surprised to see Output transformers on a current production MacIntosh Transistor amp, at the same Hi Fi shop where I saw the old design MacIntosh Transistor amp 50 years ago.
There are some transistor amplifiers that change their + and - DC voltages, according to the 4 or 8 Ohm setting switch. With the switch set to 4 Ohm, the voltage reduced to about 70% of the 8 Ohm DC voltage switch setting. That is because for the same power into 4 Ohms, you need 1.414 times the current, but only 0.707 times the voltage. That uses an engineering technique of some Lab DC power supplies, which change the tap on the power transformer secondary, according to the DC output setting (keeping the power dissipation of the regulator transistor relatively constant).
Output impedance of an emitter follower: The first order calculation of output impedance is 26 / mA emitter current. So, a standing current of 52mA = 1/2 Ohm output impedance. 52mA barely tickles an 8 Ohm loudspeaker (so little class A power, we go into AB almost immediately).
Let us use a PNP and NPN in a vertical series connection totem pole emitter follower output: With 52 mA in the output stage, the two 1/2 Ohm emitter impedances are in parallel, the output impedance is 1/4 Ohm (0.25 Ohm). For a ‘4’ Ohm loudspeaker, 4 Ohm / .025 Ohm = a damping factor of 16, without (and before) any global negative feedback. Applying negative feedback makes the output impedance go even lower, & the damping factor to increase. With + and - 40V supplies, and 52mA, we have 4.16 Watts quiescent power in the output transistors, a class AB amp, almost a class B amp, which is why we need to have global feedback. With 52mA peak current into an 8 Ohm load, we only get 22mW of class A power.
B. A single ended output transformer that has a higher impedance needs less current to get the same output power as the lower impedance transformer.
P = Isquared x R
P = 50 mA x 5k = 12.5 W Call the ‘Amp x turns’ 100% 0.050A x 1 turn = 0.050 Amp Turn
P = 35.355339 mA x 10k = 12.5 W 10k requires Root(2) more turns than 5k Root(2) x 35.355339 mA = 1.414213562 x 35.355339 mA = 0.050 Amp Turn
The ‘Amp x turns’ of a 5k transformer at 50mA is the same as the ‘Amp x turns’ of a 10k transformer at 35.3mA. This means the single ended core of both of these transformers is exactly the same for the same 12.5 Watt SE transformer. And both of these transformers require an Air Gap.
A push pull transformer neither needs nor has the Air Gap, and the balanced idle currents of push and pull keeps the core from saturating at the quiescent currents. Those two factors means we can use a smaller core, just as you said. But if there is less idle current than in the single ended amp, it must be remembered that the push pull amp is only class AB or class B, not class A like the single ended amp (the push pull is only class A for the first region of its power range).
C. Miller capacitance is negligible in a pentode output stage (including a single ended pentode with Schade feedback). The Schade feedback resistor does require the driver stage to put out more current, but the miller capacitance of the pentode does not. We can neglect the miller capacitance of a 6L6 or EL34 operating as a pentode. But I usually use those tubes in triode wired mode, and then we do need to take miller capacitance into account. I am now investigating Ultra Linear mode, where miller capacitance is less than triode operation, but more than pentode operation.
Last edited:
Thanks for your comments. Isn't it unfortunate that current DIY designs so commonly follow design goals of the 1950's, optimizing for an extra dB or two of output at clipping, rather than best performance within their working range? Some output transformers from the past designed for use with type 7591 valves have nominal primary impedances of 7500 Ohms or so. For example Heathkit A-100 or Eico ST-70. Maybe not the ultimate, DTN Williamson standard, but good stuff if found.
And I can strongly concur about One Electron's transformers. Excellent stuff, especially their single-ended types, and at what seems to me very fair prices.
All good fortune,
Chris
Thanks for your comments, heavily excerpted above. It's important to remember that real loudspeakers aren't 8 Ohm Dale non-inductive load resistors, and worse yet, aren't even all the same (or even close!), and I glossed over it in my earlier comments.
Maybe I shouldn't have included the aside about solid-state amplifiers. Just a different can of worms. McIntosh is a force unto themselves, beholden to none. Nuf said.
Thanks again, and all good fortune,
Chris
santitrucco,
Just a few generalizations:
Using the same single ended output transformer for each configuration, the same plate voltage, the same cathode current, and no negative feedback.
EL34 Pentode mode:
Highest power out
Highest 2nd Harmonic distortion
Lowest Damping Factor
Poorest Frequency response with most loudspeakers
EL34 Ultra Linear mode:
Lower power out
Better 2nd Harmonic distortion
Better Damping Factor
Better Frequency response with most loudspeakers
EL34 Triode Wired mode:
Lowest power out
Best 2nd Harmonic distortion
Best Damping Factor
Best Frequency response with most loudspeakers
Negative feedback reduces the amplifier gain (how much signal is needed at the input), and may require another gain stage. Negative feedback reduces the Total Harmonic Distortion
Although Negative feedback reduces 2nd and 3rd Harmonic distortion, it Increases the levels of the Higher order Harmonics (4, 5, 6, 7 . . .).
Negative feedback increases the Damping Factor
Negative feedback improves the frequency response with most loudspeakers
and . . .
Negative feedback can be hard to adjust for best performance, and makes it difficult to make the amplifier stable, especially with loudspeaker loads (instead of resistor loads).
But restricting the EL34 Pentode, Ultra Linear, and Triode Wired Modes to use the same output transformer, same plate voltage, same cathode current, and use or not use negative feedback might not get the optimum performance for each Mode.
Just a few generalizations:
Using the same single ended output transformer for each configuration, the same plate voltage, the same cathode current, and no negative feedback.
EL34 Pentode mode:
Highest power out
Highest 2nd Harmonic distortion
Lowest Damping Factor
Poorest Frequency response with most loudspeakers
EL34 Ultra Linear mode:
Lower power out
Better 2nd Harmonic distortion
Better Damping Factor
Better Frequency response with most loudspeakers
EL34 Triode Wired mode:
Lowest power out
Best 2nd Harmonic distortion
Best Damping Factor
Best Frequency response with most loudspeakers
Negative feedback reduces the amplifier gain (how much signal is needed at the input), and may require another gain stage. Negative feedback reduces the Total Harmonic Distortion
Although Negative feedback reduces 2nd and 3rd Harmonic distortion, it Increases the levels of the Higher order Harmonics (4, 5, 6, 7 . . .).
Negative feedback increases the Damping Factor
Negative feedback improves the frequency response with most loudspeakers
and . . .
Negative feedback can be hard to adjust for best performance, and makes it difficult to make the amplifier stable, especially with loudspeaker loads (instead of resistor loads).
But restricting the EL34 Pentode, Ultra Linear, and Triode Wired Modes to use the same output transformer, same plate voltage, same cathode current, and use or not use negative feedback might not get the optimum performance for each Mode.
Last edited:
I haven't found any increase in higher order harmonics upon applying fb. I used 16 dB. Maybe that happens in amplifiers that are not properly engineered. All harmonics went down and showing up again only at the onset of clipping respect to zero fb configuration. Little to no advantage in distortion figures if going to triode mode with nearly halved power. At 1W I could only see little 2nd harmonic and a tiny peak of 3rd just popping out of the noise floor.
Having put a jumper that allowed me to switch from pentode to UL or triode I decided that pentode with fb was the one I liked more. After bench and listening tests I mean.
Actually the triode configuration needs the feedback anyway IMHO because EL34 is not very linear as other pentodes in triode mode. Triode mode without fb is good only for PP. Better if Class A PP. I tested it with original/like new Philips from 1960's, JJ EL34 and current production Gold Lion. Similar results although the original Philips have slightly more performance.
One 12AX7 shared between the two channels was enough to make a power amp with 1.4 V r.m.s for full output. A 12AX7 as driver is not really happy when in triode mode. To get best performance need to use a more "complex" driver.
P.S. If one wants to use cheaper output transformers and is worried about stability with feedback I suggest taking the feedback signal directly from the plate to the cathode of the driver. This will leave the transformer out of the feedback loop. Is not exactly the same but works very well too.
Having put a jumper that allowed me to switch from pentode to UL or triode I decided that pentode with fb was the one I liked more. After bench and listening tests I mean.
Actually the triode configuration needs the feedback anyway IMHO because EL34 is not very linear as other pentodes in triode mode. Triode mode without fb is good only for PP. Better if Class A PP. I tested it with original/like new Philips from 1960's, JJ EL34 and current production Gold Lion. Similar results although the original Philips have slightly more performance.
One 12AX7 shared between the two channels was enough to make a power amp with 1.4 V r.m.s for full output. A 12AX7 as driver is not really happy when in triode mode. To get best performance need to use a more "complex" driver.
P.S. If one wants to use cheaper output transformers and is worried about stability with feedback I suggest taking the feedback signal directly from the plate to the cathode of the driver. This will leave the transformer out of the feedback loop. Is not exactly the same but works very well too.
The values depend on the actual circuit you use and the amount of fbk you want to use.
Last summer I used it in a 6V6 SE amp. This was an amp built by someone else that did not work out well. It was UL which is already uncommon for 6V6 SE and the assembly was no good enough. It was noisy and unstable. So I re-soldered most of it, improved the PSU and applied the fbk leaving the OPT out of the loop. The input stage/driver was another pentode - a 6SJ7 - using a simple un-bypassed 1.5K cathode resistor if I remember well. So I just connected something like 100-120K resistor to the 6V6 plate put some 10 uF cap in series and connected the other end of the cap to the 6SJ7 cathode. I didn't measure anything because I didn't (or better he didn't ) have the tools on his bench. Just trial and test until it worked (sounded) satisfactorily.
Last summer I used it in a 6V6 SE amp. This was an amp built by someone else that did not work out well. It was UL which is already uncommon for 6V6 SE and the assembly was no good enough. It was noisy and unstable. So I re-soldered most of it, improved the PSU and applied the fbk leaving the OPT out of the loop. The input stage/driver was another pentode - a 6SJ7 - using a simple un-bypassed 1.5K cathode resistor if I remember well. So I just connected something like 100-120K resistor to the 6V6 plate put some 10 uF cap in series and connected the other end of the cap to the 6SJ7 cathode. I didn't measure anything because I didn't (or better he didn't ) have the tools on his bench. Just trial and test until it worked (sounded) satisfactorily.
45,
Good idea on the output stage plate to driver cathode for the negative feedback.
I have seen that on schematics. I guess I need to get busy and try that on one of my test amplifiers.
I did use Schade feedback as another method to avoid having the OPT in the loop.
I do not prefer to use this method, but I know others have real good results with Schade FB.
Good idea on the output stage plate to driver cathode for the negative feedback.
I have seen that on schematics. I guess I need to get busy and try that on one of my test amplifiers.
I did use Schade feedback as another method to avoid having the OPT in the loop.
I do not prefer to use this method, but I know others have real good results with Schade FB.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.