I'm sure it wouldn't. There will be differences in Zout and frequency response to start with, which would make it sound different.
My output transformers are pretty unique(Plitron 1070UC). The manufacturer claims ~1500H primary inductance, so I would expect the lows to have a very different character than an SE amp. Plus, if I were to hook my PP amp to efficient speakers, behavior at very loud volumes will be very different due to the fact that much more power is available from the amp.
The PP amp with the "waterfall" harmonic profile has a sweet sound but I agree, probably not the same sound as an SE amp.
But I actually prefer the "nulled" distortion profile on the PP amp.
Hello All,
Ten Years After was a good old rock and roll band a long time ago. Somebody woke up this thread, it was RIP.
Was this question ever answered? Remember the top tube is not SRPP, the M.I.T. guys of the 1940’s said that the top triode acts as a resistor. Does the half mu front end of the Aikido perform measurably better than a grounded cathode with the same same same tube, same cathode resistor, same supply voltage and a plate resistor equal in resistance to the top triode?
Seems to me that a 5692 is not a good test specimen, remember Morgan Jones did exhaustive tests looking for the valve with the lowest distortion. The 6SN7 easily was the best performer. A 5692 is a dressed up 6SN7. Also seems to me that a pedestrian 12AU7 with a reputation of so so distortion performance is a more likely suspect. More musing, the top and bottom triodes may offer equal and offsetting nonlinearity.
DIYers, this is a challenge, there are more than a few audio analyzers sitting your benches out there.
Any challenge takers?
Any guess at the results?
DT
There was a nice discussion of this circuit in Landee, about 50-60 years ago. The touted advantage was that the output voltage would be relatively immune to changes in heater voltage because of symmetry.
The 5692 was chosen specifically to see how low the distortion could go, and useful for comparing to normal grounded cathode voltage amps (since I had a lot of 5692 data).
Audio Analyzer Tested
Audio Analyzer Tested,
Sy you asked, Years late I have data. Here it is.
On a breadboard I connected up the input stage of an Aikido, you the half mu totem pole thing. This time with a 9AU7 and a U8903B Audio Analyzer.
B+ 258 volts
Rk := 562 ohms
Grid volts := -3.48volts
Output volts := 2.01volts
H2 := -50db H3 := -86db H4 := -110db
H5 and higher are in the weeds.
Using a 20830 ohm resistance in place of the top triode
B+ 258 volts
Rk := 562 ohms
Grid volts := -3.46volts
Output volts := 2.01volts
H2 := -49.5db H3 := -86.5db H4 := -105db H5 := -115db
H6 and higher are in the weeds.
Nice theory: equal and opposite nonlinearity canceling resulting is lower distortion.
These results do not support the theory.
DT
Audio Analyzer Tested,
Sy you asked, Years late I have data. Here it is.
On a breadboard I connected up the input stage of an Aikido, you the half mu totem pole thing. This time with a 9AU7 and a U8903B Audio Analyzer.
B+ 258 volts
Rk := 562 ohms
Grid volts := -3.48volts
Output volts := 2.01volts
H2 := -50db H3 := -86db H4 := -110db
H5 and higher are in the weeds.
Using a 20830 ohm resistance in place of the top triode
B+ 258 volts
Rk := 562 ohms
Grid volts := -3.46volts
Output volts := 2.01volts
H2 := -49.5db H3 := -86.5db H4 := -105db H5 := -115db
H6 and higher are in the weeds.
Nice theory: equal and opposite nonlinearity canceling resulting is lower distortion.
These results do not support the theory.
DT
6DJ8 Audio Analyzer Tested
Audio Analyzer Tested,
On a breadboard I connected up the input stage of an Aikido, you know the half mu totem pole thing. This time with a 6DJ8 and a U8903B Audio Analyzer.
B+ 258 volts
Rk := 562 ohms
Grid volts := -3.48volts
Output volts := 2.01volts
H2 := -60db
H3 and higher are in the weeds. -110db
Using a 20830 ohm resistance in place of the top triode
B+ 258 volts
Rk := 562 ohms
Grid volts := -3.46volts
Output volts := 2.01volts
H2 := -60db
H3 and higher are in the weeds. -110db
Nice theory: equal and opposite nonlinearity canceling resulting is lower distortion.
These results do not support the theory.
Audio Analyzer Tested,
On a breadboard I connected up the input stage of an Aikido, you know the half mu totem pole thing. This time with a 6DJ8 and a U8903B Audio Analyzer.
B+ 258 volts
Rk := 562 ohms
Grid volts := -3.48volts
Output volts := 2.01volts
H2 := -60db
H3 and higher are in the weeds. -110db
Using a 20830 ohm resistance in place of the top triode
B+ 258 volts
Rk := 562 ohms
Grid volts := -3.46volts
Output volts := 2.01volts
H2 := -60db
H3 and higher are in the weeds. -110db
Nice theory: equal and opposite nonlinearity canceling resulting is lower distortion.
These results do not support the theory.
Audio Analyzer Tested
Audio Analyzer Tested,
On a breadboard I connected up the input stage of an Aikido, you know the half mu totem pole thing. This time with a 6CG7, the 9-pin version of the 6SN7, and a U8903B Audio Analyzer.
B+ 258 volts
Rk := 562 ohms
Grid volts := -3.18
Output volts := 2.01volts
H2 := -54db H3-106
H4-116 and higher are in the weeds.
Using a 20830 ohm resistance in place of the top triode
B+ 258 volts
Rk := 562 ohms
Grid volts := -2.9volts
Output volts := 2.01volts
H2 := -54db H3-108db
H4-120db and higher are in the weeds.
Nice theory: equal and opposite nonlinearity canceling resulting is lower distortion.
These results do not support the theory.
A bit interesting, 3, 9-pin tubes; 12AU7, 6DJ8 and 6CG7 tested with no difference between the totem pole version and the grounded cathode version.
The 6DJ8 Tested best!
DT
Audio Analyzer Tested,
On a breadboard I connected up the input stage of an Aikido, you know the half mu totem pole thing. This time with a 6CG7, the 9-pin version of the 6SN7, and a U8903B Audio Analyzer.
B+ 258 volts
Rk := 562 ohms
Grid volts := -3.18
Output volts := 2.01volts
H2 := -54db H3-106
H4-116 and higher are in the weeds.
Using a 20830 ohm resistance in place of the top triode
B+ 258 volts
Rk := 562 ohms
Grid volts := -2.9volts
Output volts := 2.01volts
H2 := -54db H3-108db
H4-120db and higher are in the weeds.
Nice theory: equal and opposite nonlinearity canceling resulting is lower distortion.
These results do not support the theory.
A bit interesting, 3, 9-pin tubes; 12AU7, 6DJ8 and 6CG7 tested with no difference between the totem pole version and the grounded cathode version.
The 6DJ8 Tested best!
DT
Last edited:
How can a 6DJ8, AU7 and 6CG7 top triode all be replaced by the same resistor value? It can only match one of them, or none of them.
Just to be clear, this was the half-mu circuit (i.e. active load, output from lower anode) not an SRPP? High impedance load? Bottom cathode unbypassed?
To get cancellation of even order distortion you would need the anode impedance to vary more strongly with anode voltage than anode current. I suspect the opposite is true for most triodes.
Just to be clear, this was the half-mu circuit (i.e. active load, output from lower anode) not an SRPP? High impedance load? Bottom cathode unbypassed?
To get cancellation of even order distortion you would need the anode impedance to vary more strongly with anode voltage than anode current. I suspect the opposite is true for most triodes.
Having thought about it a little more, I now think that the above statement is true if the output is heavily loaded. In the Aikido it isn't. Further thought suggests that the above statement is not true even under heavy load, as the upper valve is not being driven.DF96 said:
For small load (i.e. high impedance) the condition for distortion cancellation is constant mu i.e. anode impedance depends on anode current alone, not anode voltage.
For the high load case you need the two anodes to combine to give a more linear impedance which is unaffected by anode voltage swings; the load then sees this linearised impedance. This means that they need to balance each other off. This balance will be disturbed by the loading, however.
For the low load case you need the two anodes to present the same impedance, so the lower valve sees the upper valve matching it. The potential divider then has two equal nonlinear impedances so the junction still gets half the applied voltage.
Last edited:
I can also confirm that the first-principle half-mu stage does not enjoy much if any distortion cancellation in practice, as opposed to theory. However, you can encourage distortion cancellation by making the 'upper' load more non-linear, i.e. by bypassing the upper cathode resistor only (of course, it's not really a half-mu stage if you do this!). The degree of cancellation does depend on the external (linear) load too, however. See my book for examples.
Last edited:
"Theory" is getting a bashing here! Whose "theory" is this? Can we correct the theory - there is nothing so practical as a good theory.
To get low distortion from a voltage source with a nonlinear output impedance (e.g. a triode) you need either a high impedance load (so the nonlinearity is swamped) or a matching nonlinearity in the load. The half-mu stage can do this if the triodes are well-behaved and the external load is small. For more external load, you need to make the upper triode more nonlinear by (partially) bypassing the cathode resistor. Alternatively, use the mu follower.
To get low distortion from a voltage source with a nonlinear output impedance (e.g. a triode) you need either a high impedance load (so the nonlinearity is swamped) or a matching nonlinearity in the load. The half-mu stage can do this if the triodes are well-behaved and the external load is small. For more external load, you need to make the upper triode more nonlinear by (partially) bypassing the cathode resistor. Alternatively, use the mu follower.
Small-signal analysis tells us the gain of the unloaded half-mu is (surprise surprise) mu/2. Everything else drops out of the equation. The linearity therefore ought to be as good as the linearity of mu.
Except it ain't!
Was that for the half-mu stage alone, or the Aikido circuit (i.e. cancellation from the cathode follower's contribution)? For the ECC88 half-mu alone, 200V, unbypassed, I got about 0.7% THD+N @ 10Vrms into high-Z loads.
Last edited:
Small-signal analysis only looks at linear terms so it is unsurprising that it gives a linear result. It is a mistake to take that linear result and just look at potential nonlinearities in it, as there may be other terms too. The most that can be said is that nonlinearity in mu will affect the ouput, but exactly how much will depend on other nonlinearities which may add or subtract from the mu terms.Merlinb said:
To put it another way, if you do a calculation assuming that mu is constant then you can't just take the result of the calculation and assume mu is not constant. This is the classic 'engineering' approach to theory: make assumptions, violate them, then complain that 'theory' doesn't work! There may be terms you threw away because they are zero when mu is constant. When making approximations it is very important to keep a careful check on exactly what approximations are made and what effect they have on the final result. One good way to do this is to calculate without approximations and then apply them to the final result; or at least employ fewer approximations if the full theory is intractable.
Anyway, have we established that it "ain't"? Has someone calculated/measured the variation of mu with anode voltage/current and shown that this does not explain the measured nonlinearity?