If I were going to prove this to myself one way or the other I would build a nice little triode strapped SPUD E55L push pull amplifier with an input phase splitting transformer and a cheap toroidal transformer for OT - employing garter bias to take care of DC balance.
I would then use exactly the same components to build a SE E55L spud with input transformer and a CCS load in parafeed.
I would have a fairly definitive answer as to the intrinsic difference between PP and SE avoiding all the usual differences - and I would not have had to spend a whole packet of money on building two dissimilar amps.
Shoog
I would then use exactly the same components to build a SE E55L spud with input transformer and a CCS load in parafeed.
I would have a fairly definitive answer as to the intrinsic difference between PP and SE avoiding all the usual differences - and I would not have had to spend a whole packet of money on building two dissimilar amps.
Shoog
Well, they are different topologies, so an equal comparison is difficult. I'm not sure how using some cheap components proves anything however. Your PP stage is a fairly classic arrangement, but you're not compensating for AC balance, only DC balance.
Using a CCS loaded stage with capacitive coupling to the OPT is not a classic arrangement and quite different than the PP arrangement, so that doesn't pan out. Finally, using a triode strapped pentode isn't ideal either, so I can't agree with your proposed test as being very useful.
A more "equal" comparison of SE to PP would be using proper OPTs for each topology and a real set of triodes, like the 45. Use the published datasheet for load and biasing for each output stage, then you have a chance of comparing apples and oranges.
Regards, KM
Using a CCS loaded stage with capacitive coupling to the OPT is not a classic arrangement and quite different than the PP arrangement, so that doesn't pan out. Finally, using a triode strapped pentode isn't ideal either, so I can't agree with your proposed test as being very useful.
A more "equal" comparison of SE to PP would be using proper OPTs for each topology and a real set of triodes, like the 45. Use the published datasheet for load and biasing for each output stage, then you have a chance of comparing apples and oranges.
Regards, KM
Lundhall has a US distributor, K&K Audio. I bought from them at numerous occasion. Great, knowlegable guy. You won't have to pay import tax.
Lundhall are superb transformers, way better than Hammond. SB
Lundhall are superb transformers, way better than Hammond. SB
I wouldn't call E55L a cheap component, check Ebay. Maybe try the 12HL7 for only a few $.
E55L is a frame grid tube that has a Mu curve that would put pretty much any triode, including 300B, to shame. See below, page 9. Ruler strait. It might be lacking some of the usual "distortion", I mean coloration for SE. Depends on loading.
http://frank.pocnet.net/sheets/009/e/E55L.pdf
E55L is a frame grid tube that has a Mu curve that would put pretty much any triode, including 300B, to shame. See below, page 9. Ruler strait. It might be lacking some of the usual "distortion", I mean coloration for SE. Depends on loading.
http://frank.pocnet.net/sheets/009/e/E55L.pdf
If you consider a frequency response of between 10hz and 40khz as an inadequate test of what each can do then thats up to yourself. I consider it a fair test. What using the same transformer without DC bias allows is for the transformer to be taken out of the equation as the dominant sound element since its behaving in the same magnetic domain in both. Parallel two E55L trioded for the SE for the comparison to get us into about the right impedance. The transformer in this situation is far more comparable than comparing a gapped SE to an ungapped PP tranformer. As I said this is a test of the intrinsic differences of the two topologies - not a test of different transformers effects on the end result. I suspect that the reason why many would refuse this particular challenge is because like like the sonic impact of the SE transformers limitations - but that not the issue here.
If the cheap transformer is the weak element then it hobbles both equally.
What i am talking about here is testing the same valve in both configurations - which is almost never done by those who advocate SE as been superior to PP. How many advocates out there have actually built a PP 45 amp for a fair comparison ?? Not more than a fraction I would say.
I am making a suggestion here which can allow anyone to build two roughly equal amps - compare them and ultimately make a choice which they like best and build it out without spending on two distinct builds. That certainly is worth a try.
I have not done this exact comparison myself, but I have come close. I built a nice little 3W E55L parafeed amp and I can report it sounds remarkably similar to my three PP amps, built with very similar components. I haven't found the magic in SE.
The question for me is if I went to the trouble of building a 45 PP (zero GNF) amp and compared it to a parafeed 45 PSE (zero GNF) amp would I hear a significant difference. I personally don't think I would. I specify parafeed because of the not insignificant impact of the magnetic domain.
I would ask people with actual real experience of such a direct comparison to prove me wrong.
Shoog
If the cheap transformer is the weak element then it hobbles both equally.
What i am talking about here is testing the same valve in both configurations - which is almost never done by those who advocate SE as been superior to PP. How many advocates out there have actually built a PP 45 amp for a fair comparison ?? Not more than a fraction I would say.
I am making a suggestion here which can allow anyone to build two roughly equal amps - compare them and ultimately make a choice which they like best and build it out without spending on two distinct builds. That certainly is worth a try.
I have not done this exact comparison myself, but I have come close. I built a nice little 3W E55L parafeed amp and I can report it sounds remarkably similar to my three PP amps, built with very similar components. I haven't found the magic in SE.
The question for me is if I went to the trouble of building a 45 PP (zero GNF) amp and compared it to a parafeed 45 PSE (zero GNF) amp would I hear a significant difference. I personally don't think I would. I specify parafeed because of the not insignificant impact of the magnetic domain.
I would ask people with actual real experience of such a direct comparison to prove me wrong.
Shoog
Surprisingly at the moment I am only listening to one valve amp, and its my little E55L SE. Its nice and powerful enough for my bedroom system where it is never asked to deliver more than a watt of power. Its not because its my best or the nicest sounding amp (that prize rests with a complex PP EL86 amp) but simply because its the most efficient amp for its application.
My main system is now running a Push Pull Le Monstre Class A transistor amp - very nice.
Shoog
My main system is now running a Push Pull Le Monstre Class A transistor amp - very nice.
Shoog
Yup +1 choke input power supply in a tube amp sounds very good. The rest is simple and fun.
The problem I have with this line of reasoning is two fold
a) that I've yet to see any experimental data supporting it.
b) that it doesn't usefully model the distortion products produced.
If the Rp-is-feedback model was valid we'd see pentode feedback based amplifiers with distortion spectra that'd look like a NOS 45 - (and for a fraction of the price. )
But we don't - in fact the opposite happens - you get less 2nd and more 7th and 9th (etc) as you add feedback.
Clearly not. That's called "plate resistance". Every active device can (and is) modelled as a Thévenin equivalent at a particular operating point. Which is a rather more useful model of how a device behaves.
There is only so much gain in a tube, no reason for the pentode to do any better when brought down to the same closed loop gain as an equivalent triode.
The 3/2 power law at grid 1 is actually quite distortive. The triode plate has a counter 3/2 law non-linear feedback that is unusually effective at cleaning up the grid's peculiar mess. Using simple --linear-- resistive external feedback around a pentode does not work nearly so effectively. A finite amount of properly configured non-linear feedback can clean up the problem, where as an infinite amount of linear feedback is required to do the same.
Rp does not necessarily have anything to do with feedback, one could just put a resistor across the output. But for the triode it does. The internal feedback, affecting cathode current, determines the Rp effect.
The 3/2 power law at grid 1 is actually quite distortive. The triode plate has a counter 3/2 law non-linear feedback that is unusually effective at cleaning up the grid's peculiar mess. Using simple --linear-- resistive external feedback around a pentode does not work nearly so effectively. A finite amount of properly configured non-linear feedback can clean up the problem, where as an infinite amount of linear feedback is required to do the same.
Rp does not necessarily have anything to do with feedback, one could just put a resistor across the output. But for the triode it does. The internal feedback, affecting cathode current, determines the Rp effect.
Last edited:
Another issue when speaking of NFdbk is the simplified notion that higher harmonics universally sound bad. This actually is NOT true.
Lets take an ideal Mosfet with perfect square law (2.00 power) behavior, so it only generates 2nd harmonic distortion in SE mode. Now lets straighten out its transfer curve ever so slightly to a 1.99 power law. This is actually more linear than 2.00, but so close that it will sound indistinguishable from the 2.00 power law. But its FFT spectrum will now have endless small high harmonics. So what happened....
The new FFT sequence will have alternating + and - terms that essentially cancel sonically, but are needed mathematically to model the slightly straightened curve. So as a result we have these endless arguments in audio based on erroneous simplifications. One can not give a math course every time we go off on stupid tangents. The real criteria for smooth sound is probably more like the sharpness of the curvature radius of a waveform. Jagged stuff sounds bad, not higher harmonics if curvature balanced. Most audio FFTs unfortunately only give the absolute magnitude of the harmonics, so you can not tell whether it is sounding bad or not. A further curvature calculation App. is really needed.
So statements about pentodes having higher harmonics are not real informative. That just says they don't conform to some simple integer power law. But generally, linear external feedback around pentodes won't perform as well as clever internal non-linear triode feedback.
Lets take an ideal Mosfet with perfect square law (2.00 power) behavior, so it only generates 2nd harmonic distortion in SE mode. Now lets straighten out its transfer curve ever so slightly to a 1.99 power law. This is actually more linear than 2.00, but so close that it will sound indistinguishable from the 2.00 power law. But its FFT spectrum will now have endless small high harmonics. So what happened....
The new FFT sequence will have alternating + and - terms that essentially cancel sonically, but are needed mathematically to model the slightly straightened curve. So as a result we have these endless arguments in audio based on erroneous simplifications. One can not give a math course every time we go off on stupid tangents. The real criteria for smooth sound is probably more like the sharpness of the curvature radius of a waveform. Jagged stuff sounds bad, not higher harmonics if curvature balanced. Most audio FFTs unfortunately only give the absolute magnitude of the harmonics, so you can not tell whether it is sounding bad or not. A further curvature calculation App. is really needed.
So statements about pentodes having higher harmonics are not real informative. That just says they don't conform to some simple integer power law. But generally, linear external feedback around pentodes won't perform as well as clever internal non-linear triode feedback.
Last edited:
Re cheap components: Shoog noted a cheap output transformer. As for the E55L tube... commodity pricing on ebay is not a sound source for value (yea, bad pun). I've seen single plate 2A3 tubes fetch over $1100 on ebay... that's obscene, just because someone paid the price, it doesn't declare it's true value.
Back to the comparison, I still don't see it as a valid comparison. As you're specifying a CCS for the SE amp, you have two active devices in the output stage, so it's not truly single-ended. You also have an extra coupling capacitor in the circuit, so that changes the comparison as well.
My point is (Shoog) is recommending parts designed for a PP output circuit and testing as such, then making a very different topology and stating the same parts are perfectly suited for the task and declaring it okay.
I would agree that having a wide bandwidth is a plus for either topology however.... I noted that in my earlier post, albeit a cheap OPT was chosen with no mention of specifications. A cheap OPT generally results in poor performance.
I'm not advocating one versus the other, nor am I declaring either as having or lacking magic. The simple fact is that you can design, engineer and build either for very good performance. I've done dozens of designs for both over the past 4+ decades and have gotten excellent performance from both topologies. We're just in disagreement about your proposed comparison method.
Regards, KM
Back to the comparison, I still don't see it as a valid comparison. As you're specifying a CCS for the SE amp, you have two active devices in the output stage, so it's not truly single-ended. You also have an extra coupling capacitor in the circuit, so that changes the comparison as well.
My point is (Shoog) is recommending parts designed for a PP output circuit and testing as such, then making a very different topology and stating the same parts are perfectly suited for the task and declaring it okay.
I would agree that having a wide bandwidth is a plus for either topology however.... I noted that in my earlier post, albeit a cheap OPT was chosen with no mention of specifications. A cheap OPT generally results in poor performance.
I'm not advocating one versus the other, nor am I declaring either as having or lacking magic. The simple fact is that you can design, engineer and build either for very good performance. I've done dozens of designs for both over the past 4+ decades and have gotten excellent performance from both topologies. We're just in disagreement about your proposed comparison method.
Regards, KM
I am going to have to disagree again. The CCS is transparent and even helps to improve the performance of the triode rather than degrade it. It seems that its this improvement which people are agianst. The cap in parafeed is similarly a red herring since there are plenty of caps directly in the signal path of a SE amp (think of the cathode bypass cap and the final cap in the power supply), in my proposal the final power supply cap is effectively out of the picture due to the CCS which will mean less cap sound overall.
My proposal takes the essential sound of the amplifying component and highlights it by reducing the influence of the various other components. I think its a very fair way of deciding whether the sound of a given triode in SE or PP is different.
Other than that we will have to agree to differ.
Shoog
My proposal takes the essential sound of the amplifying component and highlights it by reducing the influence of the various other components. I think its a very fair way of deciding whether the sound of a given triode in SE or PP is different.
Other than that we will have to agree to differ.
Shoog
The CCS is an active component (or assembly of components), so it can't be transparent as it directly influences the operation of the tube it sources. The PP version is operating the tubes sans the CCS, hence they're operating as two triodes in an SE-like mode (from idle to increase or decrease current flow) with a common OPT. Bottom line, the operating parameters are different between your proposed method which invalidates the comparison.
The caps (bypass and power supply) are also present in the PP topology you described, but you're not attempting to mask their effect as you claim to be in the SE topology, hence it's not an equal and/or fair comparison of PP vs SE.
As our view and opinion differs, I will agree that we'll simply have to agree to disagree.
Regards, KM
The caps (bypass and power supply) are also present in the PP topology you described, but you're not attempting to mask their effect as you claim to be in the SE topology, hence it's not an equal and/or fair comparison of PP vs SE.
As our view and opinion differs, I will agree that we'll simply have to agree to disagree.
Regards, KM
Lets just say that I disagree with your analysis here - the CCS is a high impedance in parallel with a low impedance (the OT) and so any effect of the CCS is swamped by the dominant behaviour of the OT.
Active components are used because they are transparent in comparison to the imperfect components such as triodes and Output transformers. They allow the imperfections of the parts we desire to listen to to shine through.
Shoog
Active components are used because they are transparent in comparison to the imperfect components such as triodes and Output transformers. They allow the imperfections of the parts we desire to listen to to shine through.
Shoog
I tri-amp and would not use any other amp than a SET for the midrange horns. I think it really matches well with the human voice. I cross from 500hz to 3500hz, then go to a OTL from 3500hz on up. Right now I am driving my woofs with a Carver TFM45.
I find the thread title puzzling, since now that I think about it, all my SE circuits have been cheaper and easier to assemble (less parts) than my balanced amps.
I'm fully aware of what a CCS is and what they do, I designed several back in the 70's and 80's using both discrete parts and ICs and employed them in multiple designs. I do find them quite useful in a LTP phase inverter and in certain power supply regulators.
However, I still disagree with your use for the comparison as it is an active device and will have an effect regardless of what you claim. In short, you're not comparing similar (output tube) operating parameters in PP vs SE, that's the flaw in my view.
Still, you (and all others on the forum) are free to do as you please. We don't have to agree on anything actually, but we're mostly here for the same reason... sharing a common interest. Enjoy.
Regards, KM
However, I still disagree with your use for the comparison as it is an active device and will have an effect regardless of what you claim. In short, you're not comparing similar (output tube) operating parameters in PP vs SE, that's the flaw in my view.
Still, you (and all others on the forum) are free to do as you please. We don't have to agree on anything actually, but we're mostly here for the same reason... sharing a common interest. Enjoy.
Regards, KM
I don't think the OPT behavior is different between SE and PP. Smoking-Amp put it very nicely. The concept of SE never run through zero flux crossing is very misleading to say the least. He can explain a lot better. But in my interpretation, EACH magnetic dipole has a hysteresis to flip regardless of the net total polarity of the complete core. That is regardless of the core has zero flux, or if it has DC flux. There is always a hysteresis to flip a dipole.
Well, the hysteresis may be a mute point between SE and P-P, and it never was a big issue anyway. Probably just got picked up mistakenly for marketing.
The SE OT does however have an effect where the permeability is varying over a range for - to + signal, which would have some 2nd harmonic effect. While the P-P OT has variation of permeability centered around the zero crossing, which would have some 3rd harmonic effect.
A well designed amplifier (low Zout) would again not suffer from these artifacts.
The SE OT does however have an effect where the permeability is varying over a range for - to + signal, which would have some 2nd harmonic effect. While the P-P OT has variation of permeability centered around the zero crossing, which would have some 3rd harmonic effect.
A well designed amplifier (low Zout) would again not suffer from these artifacts.
If there is any advantage of having some net DC current, it's easy!!! Get a good OPT that can take some current, then DC imbalance the PP output pair and you get some 2nd harmonics!!! Believe me, I thought of this already!!!! I am willing to pay more on the OPT as this is the single most important piece in the amp.
Do you know what is the output impedance of EL34 or KT88 in either triode or UL configuration. I just posted the question of choosing between mu-follower or CCS differential power tubes in my thread. If the output impedance of EL34 and KT88 is low enough in triode or UL, with the step down OPT, it might not be that important to go mu-follower to drive the OPT.
Thanks
Last edited:
- Status
- Not open for further replies.