Shoog: It is not imbalance as normally intented (i.e. two devices whose specs are not identical) but a linearity issue which is common to most audio power pentodes. You can get ideal PP behavior from pentodes only for a limited range of anode loads. With triodes this doesn't happen as long as it is class A or even class AB if they are linear enough. One reason why I love the 45. You can run it in so many ways without problems. The price to pay is low mu but given its power range it is not an impossible task. One might argue that a triode can be seen like a pentode with internal feeback but in reality it is not same thing.
The good thing I have found about some pentodes (those with medium-to-high gm, say at least 5-6 mA/V for small power devices and about 25-30% more for bigger devices) is that thanks to their sensitivity they can be efficiently used with cathode feedback (with dedicated windings in the output transformer) and this makes them extremely good performers. So good that one might forget about DHT's. EL84, PCL/ECL 82-86, EL34 are some of these.
Ideality is first a consequence of the device characteristics. The fact that most devices were made with efficiency in mind at expense of linearity, which favours PP, doesn't mean a lot to me as linear devices exist. Once you have something like a 211 the argument SE vs PP has very little relevance for practical purposes.
The good thing I have found about some pentodes (those with medium-to-high gm, say at least 5-6 mA/V for small power devices and about 25-30% more for bigger devices) is that thanks to their sensitivity they can be efficiently used with cathode feedback (with dedicated windings in the output transformer) and this makes them extremely good performers. So good that one might forget about DHT's. EL84, PCL/ECL 82-86, EL34 are some of these.
Ideality is first a consequence of the device characteristics. The fact that most devices were made with efficiency in mind at expense of linearity, which favours PP, doesn't mean a lot to me as linear devices exist. Once you have something like a 211 the argument SE vs PP has very little relevance for practical purposes.
Two identical devices fed identical (but one inverted) signals feeding into an ideal perfectly balanced OPT will alway result in cancellation of even-order distortion. This is independent of what the device is, and what the anode load value is. It is simply a result of symmetry. The fact that PP does not always do this shows that the devices are not identical and/or their drive is not perfectly inverted and/or the OPT is not perfectly balanced. It has nothing to do with whether triodes or pentodes are used, or Class A or AB.
Theory is not very useful at this point. I was talking about real dervices and Class A or AB does make a difference! Perfect symmetry never happens with devices where linearity is not great and working in class AB regardless of your efforts in selecting components and tuning the circuit. The Mullard datasheet for the EL84 says this: for the PP class AB UL amplifier, both 20% and 43%, 2nd harmonic is the dominant component up to half Pout at least. In this case 2nd and 3rd hermonics are reported together with the total amount. There is no ambiguity. It might well be the result of a real thing and accurate measurements with tightly matched devices and components. In general, I repeat, the rest of the amplifier is always supposed to be ideal and devices are identical in datasheets but this case shows that theory doesn't apply so well!
If you draw the typical pentode composite curves for class AB you will find out that it is extremely easy to get some imbalance because linearity is really mediocre everywhere even using two identical devices with perfect OT. The less linear the device the more it is true. You might get peferct behaviour only for a very limited range of loads that is not what one needs in practice. So feedback is necessary. As most pentodes make more sense for use with class AB to take advandage of their sensitivity it can happen quite easily. In fact, even in those cases where xover distortion doesn't occur and there is no bump in THD vs Pout characteristic, THD goes up very quickly anyway at low power and then grows very slowly. With a linear device like the 45 (that is in the same power class of the EL84, actually less poweful) it just doesn't happen. THD at 1W is around 0.1% and 3rd harmonic only for Class A PP and it reaches 1% at 6W in the worst case (low frequency around 25-30 Hz) for my real amp.
Last edited:
PP devices are not identical in datasheets. They measured a few real devices in a real circuit and plotted the average. Hence you get some contribution from the driver/PS. They didn't have the computing resources to do anything else. No Spice in those days.
Theory is very useful at this point. It tells you why there is plenty of 2nd in a PP output: either imbalance or from an earlier stage. The amount will of course also depend on the linearity of the devices, as they are the source of the distortion. You can characterise PP even-order distortion as being (drive imbalance+output imbalance)x(device distortion). Perfect balance means no even-order, however poor the device, because zero times anything is still zero.
Theory is very useful at this point. It tells you why there is plenty of 2nd in a PP output: either imbalance or from an earlier stage. The amount will of course also depend on the linearity of the devices, as they are the source of the distortion. You can characterise PP even-order distortion as being (drive imbalance+output imbalance)x(device distortion). Perfect balance means no even-order, however poor the device, because zero times anything is still zero.
I say that theory is not very useful because it doesn't happen most of the time with pentodes in class AB and doens' t need to be satisfied at any rate. If Mullard reported those results I don't think they made it against their interests!
The first important thing for me is the amount of distortion. It has to be as low as possible. Both UL amplifiers, 20% and 43% taps, are better than the pentode even if the pentode had perfect balance because THD is quite lower:
1) The 20% UL has 1% at 8W and stays within 1.2% up to 16W
2) The 43% UL always has less than 0.5% up to 7W amd less than 1% up to the clipping around 12W
At 0.5W the UL amps only have 0.35% and 0.15%, respectively. With the pentode THD at 0.5W is 1% already and 2% at half power!
Generally speaking THD for the UL is almost 3 times lower in the worst case.
If the low enough THD premise is satisfied it will be 99.9% 2nd or 3rd only or a mix of the two. At this point I haven't found any correlation with the sound so there is no reason to get perfect balance at any rate. It just becomes a technical exercise of no use.
Yes, I agree perfect balance is not necessary. This is fortunate, as it is not achievable.
Datasheet performance figures should be treated as being at the better end of the range of what is realistically possible given good design. In some cases they require quiescent dissipation right on the edge of the limits, so a real circuit intended for reliability may not do quite so well.
Datasheet performance figures should be treated as being at the better end of the range of what is realistically possible given good design. In some cases they require quiescent dissipation right on the edge of the limits, so a real circuit intended for reliability may not do quite so well.
I've built maybe a dozen amplifiers over the past five years. Always some one else's design. And from that experience single ended amps sound better. More "alive" with a transparent quality that I love. I've noticed that push pull amps have something like that quality without feedback. And that while feedback decreases distortion and increases linearity it also can damp the life out of the sound. I've looked at Lynn Olson's amplifiers and they seem hideously expensive to build and from his own description nearly impossible to drive.
Therefore, I would suggest that a really good compromise would be to design a push pull amplifier that is low in distortion and linear without feedback. Or at least with the absolute minimum of feedback. Not a flea amp. Not a rock amp. So the choice of tubes would be critical. It seems to me a good starting point would be to use a 6SN7 voltage amplifier in a paraphrase configuration. And drive a pair of EL34's in an ultralinear configuration. MJ praises 6SN7 as a low distortion tube that can be pushed to 400 volts on the plate. And EL34 at about 400 volts has a low distortion spec of 1.3%. So it might be good to direct couple the stages.
I'm wondering if this post is off topic? But it is this thread that has got me to this conclusion, so maybe not.
Therefore, I would suggest that a really good compromise would be to design a push pull amplifier that is low in distortion and linear without feedback. Or at least with the absolute minimum of feedback. Not a flea amp. Not a rock amp. So the choice of tubes would be critical. It seems to me a good starting point would be to use a 6SN7 voltage amplifier in a paraphrase configuration. And drive a pair of EL34's in an ultralinear configuration. MJ praises 6SN7 as a low distortion tube that can be pushed to 400 volts on the plate. And EL34 at about 400 volts has a low distortion spec of 1.3%. So it might be good to direct couple the stages.
I'm wondering if this post is off topic? But it is this thread that has got me to this conclusion, so maybe not.
This depends on whether your approach is science, listening and measurement, or one of religion.
Everyone I have spoken to about SE views it like a religion.
Mine is purely based on what you can hear, what you can measure and how you design.
Specifically, the only reason you would build and SE amp is either (a) to appeal to that religion to make money (b) one of cost to keep it small (c) building a PP amp with feedback is difficult engineering exercise and (d) you are a member of that religion and considering anything else is blasphemy.
Since people build the most ridiculous SE amps out there (which is fun and a curiosity and nothing more), then (a), (c) and (d) are the answer.
Blinded by religion and a lack of science is a common fault these days. But building an SE amp will make you money as a manufacturer because where religion is concerned (in tube concepts) people will not control themselves.
Its the emperors new clothes with a more modern twist.
Build an SE amp if you want to, for fun, for satisfaction, but don't claim in anyway it has better performance than a really good PP amp, Because , simple, measurable and listenable fact is that it does not.
Everyone I have spoken to about SE views it like a religion.
Mine is purely based on what you can hear, what you can measure and how you design.
Specifically, the only reason you would build and SE amp is either (a) to appeal to that religion to make money (b) one of cost to keep it small (c) building a PP amp with feedback is difficult engineering exercise and (d) you are a member of that religion and considering anything else is blasphemy.
Since people build the most ridiculous SE amps out there (which is fun and a curiosity and nothing more), then (a), (c) and (d) are the answer.
Blinded by religion and a lack of science is a common fault these days. But building an SE amp will make you money as a manufacturer because where religion is concerned (in tube concepts) people will not control themselves.
Its the emperors new clothes with a more modern twist.
Build an SE amp if you want to, for fun, for satisfaction, but don't claim in anyway it has better performance than a really good PP amp, Because , simple, measurable and listenable fact is that it does not.
It may of course be that some peoples hearing and perception of sound (or vision ) is different to others so that the one group can only explain the difference by assuming the other is pursuing a pointless exercise and labelling this as religion etc. It's just possible they may have missed the point!
Regards,
gsd
It is the faults and limitations of the engineering components of SE which make it unique. This is demonstrable because when you eliminate or reduce those faults the sound more closely approaches a good PP amp. I am talking about applying judicious feedback to SE to improve damping, using parafeed to eliminate the frequency constraints of a gapped transformer, using a regulated power supply to avoid power supply sag.
I just find it odd that people prefer compromised performance to what can easily be achieved, but if people prefer what that brings to the table then thats great for them. However to claim that the result is superior is difficult to accept.
Shoog
I just find it odd that people prefer compromised performance to what can easily be achieved, but if people prefer what that brings to the table then thats great for them. However to claim that the result is superior is difficult to accept.
Shoog
What you call faults, I call strengths. You don't need feedback with a low plate resistance triode SE amp. And, for me, parafeed is just NOT true SE. You cannot judge SE quality with a parafeed amp! The OPT core needs to be permanently magnetized. Not to mention that big and nasty parafeed capacitor in the signal path...
There are good SE OPTs, with extended frequency response. Yes, they are big and expensive. So what?
SE amps should be paired with very good, transparent, high efficiency speakers. These usually have lightweight cones and don't need excessive damping.
What are you on about?
The only religion I see is the crusade your against SE.
I have not even finished my latest SE, but I know it will sound better then my best PP.
Why because I don't like crossover distortion, odd-order harmonics and the hassle of matching tubes, or mythical class A PP amps?
What earthly benefit would there be for me to need a PP amplifier when I get enough power from SE? Oh ya, bragging rights over noobs who are fresh from the "moor whats is better" SS camp.
What earthly benefit would there be for me to need a PP amplifier when I get enough power from SE? Oh ya, bragging rights over noobs who are fresh from the "moor whats is better" SS camp.
SE has a large cap in the signal path and its called the last filter cap and or the cathode bypass cap. there is nothing intrinsically wrong with the cap used in Parafeed. A normal SE transformer will always have low inductance and as such will always have poor bass and high phase shift. That is why people always say that PP does bass better.
I have exactly the type of speakers you are describing (vintage German alnicos from the era of SE) and very low output impedance finals to drive them - and let me tell you - they still sound better with higher damping.
If you compromise the top end and the bass response of an amp - is it any surprise that what people talk about is the prominent midrange.
The things you like about SE are all about its limitations - thats just the simple reality.
Shoog
Last edited:
As we have discussed at length here - all your complaints against PP are in fact not intrinsic to a PP amplifier - they are totally bogus red herrings. And the reason why you should consider PP is because it implicitly does certain things better.
If you build straw men arguments do not be surprised when you win but prove nothing.
Shoog
Can I suggest you try building the Vacuum State PP1, it ticks all your boxes and can be implemented at modest cost. I think it would make fair comparison to the SE amps you have built.
Shoog
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.