Like I wrote earlier, my experience has been that it's never a bad idea to reduce distortion, even if there is a bottleneck (usually the speakers) in the signal chain.
Of course, not all ways of reducing distortion are equal. My experience has been - very strongly - that increasing open loop linearity is the way to go for best possible results.
Distortion on distortion sounds worst. Even if you have 5% THD added in your speakers, it's worthwhile to decrease your amp's THD below 0.1%, as long as you do it right.
My main thing is headphones, and there you really can hear the difference very easily. Very unforgiving.
I find a big pentode anode will do it as long as the current is high. I sometimes have a current source to help. This also means I can do what I call East to West Ultra-Linear. I tested this idea with a certain pentode when I have many makes and examples of worn out ones. As long as the current isn't near the maximum the results are very reliable.
A source follower is almost a resistor. A transistor away from Early effect also. MJE340/350 are about 6 MHz. That is not too bad. A MOSFET vastly faster. On that basis the MOSFET wins although in terms of doing nothing a transistor current amp is very linear. Above 300 VDC MOSFET's seem the only choice.
Class-A2 and a power opamp driver
A source follower is almost a resistor. A transistor away from Early effect also. MJE340/350 are about 6 MHz. That is not too bad. A MOSFET vastly faster. On that basis the MOSFET wins although in terms of doing nothing a transistor current amp is very linear. Above 300 VDC MOSFET's seem the only choice.
Class-A2 and a power opamp driver
If you take a picture with a camera that has dirt on the lense, and you want to see the picture you took as clearly as possible; is it then smart to develop the picture sloppily and put it under a glass that is also dirty?
Or is it smart to develop it anyway with best possible quality, and then use only the best glass to frame it?
I repeat my argument; any problems or compromises or bottlenecks in the signal chain, from start to your ears, are not a reason to make further compromises, if best results are the goal. They are a reason to improve other parts of the chain further.
Keep the discussion on technical terms.
Well, for example, I make fully balanced amps with no feedback, and distortion is extremely low. I go to great lenghts to avoid ever using gNFB or any kind of feedback between gain stages.
Subjective evaluations are of course subjective, and I read them as such, but I would strongly repeat my experience that it is not at all meaningless HOW you get low THD.
I would categorically say that the best designs should start with as inherently linear devices as possible (that is the sole reason I build with tubes). Great curves are the foundation of good results; then all you have to do is let the tubes do what they do best. I.e. make sure they have enough current, fast enough, and flat loadlines.
What are your design philosophies?
It is not at all difficult to make a tube amp that doesn't compress at all.
Look at say 4P1L curves, and tell me if you'd find it difficult to get perfectly linear gain out of it?
There is absolutely no need for tubes to be distortion effect generators. They can be used very linearly, if one wants to.
For best results; drive output tubes' grids with source followers. Problem solved.
Also, for best results don't put the OT in series with the output tubes and B+.
This is a DIY board. Looking at a number of commercial schematics, there's a lot of stuff discussed on this board that simply is not found in production equipment.
It doesn't matter what people sell and buy. It matters what you design and build.
Most stuff sold as high end have rather poor PSUs in my opinion, as far as I've seen (and heard).
I can't know that, and I have to assume the marketing hype is at least semi based in reality.
In many domains, (and this is my experience with other "resonant" phenomenae) there is a very big difference between what is says in the blurb - ie. what it "says on the tin", and what it actually delivers*.
Ie. Surely by now the Chinese should be capable of making a simple power supply?
I know this general rule* is the case for any so called hi-performance road car, if you ever try to drive it on a race circuit.
It's just a totally frustrating waste of time, and will be left standing by something which is purpose designed for that task.
I know this general rule, because I design entirely unrelated things which rely on a very good working knowledge of temperature, acoustics, and power output.
They invariably work better than anyone else's competing design, so I have to assume this is normal.
Often people who work on those kind of things,often have a better practical knowledge "in the critical window", than the mathematicians & theorists who don't have to work on those things for a living.
I recently had to listen to, and sort out the room acoustics with some "top of the range" Tannoy speakers with dual concentrics and Alnico magnets.
Ahum. You know, the things those London studios have etc and said were so good.
"Underwhelming" was actually me saying something polite, but I'm not really allowed to say much more because I was in touch with the factory, and it would be 🤐 to say what I really think. 😱
I made my own calibration test system.
It's called the "pig test".
It was odd to reveal a 10hz difference in resonance between 2 ostensibly identical speakers costing 40,000USD (It's just an example!)
Suffice to say,- you can get a good idea as to the potential of a speaker if you look up the frequency it registers at double the theoretic 8 ohm value it's supposed to have (say 15). Imagine then what your amplifier is going to have to cope with.
I had to do all that last week, because I have 2 systems to sort out in the next weeks, and I don't want to have to mess with them again in the foreseeable future.
Last edited:
My introduction to SE amps was I was given one by a friend to try. I was very pleasently surprised. It seemed a zero distortion amplifier. Even where the weaknesses could be found they were minor. One was not sounding like a steam locomotive. I think that can be put to one side. Mostly it did if truth be known. The Hitachi type could make the sound of the whistle as sounding in the station roof. Mostly the amps were very alike.
By accident Mrs Kron thought I was someone who owed her husband $20 000. After two hours of me not giving a dam she gave in and introduced me to her husband. It turns out I do know the person as a close friend and the confussion was over that. Ricardo sadly not with us now said this . " Surely transistors are better " ? To which I said " They are ". Ricardo " That's what I thought ". He then said about how he had made a new 300 B type with dull emmision. " I couldn't give them away ". It was that type I had borrowed. I compared my own amp which uses just EL 34 to the other 300 B design I borrowed. With some courage I said to my friend " I remember your 300 B amp sounding better ". He agreed. It was with the " couldn't give it away " 300B. Ricardo said it was stupid on the one hand to love valves and on the other hand never accept new ones.
Unlike most people what valves do which seems to excite people makes me dislike them. I don't like the dull glow or the smell. It's only the sound that is interesting. My speakers I just built would suit valve very well except for the bass end. Without a plan that way I suspect I will go for that option.
By accident Mrs Kron thought I was someone who owed her husband $20 000. After two hours of me not giving a dam she gave in and introduced me to her husband. It turns out I do know the person as a close friend and the confussion was over that. Ricardo sadly not with us now said this . " Surely transistors are better " ? To which I said " They are ". Ricardo " That's what I thought ". He then said about how he had made a new 300 B type with dull emmision. " I couldn't give them away ". It was that type I had borrowed. I compared my own amp which uses just EL 34 to the other 300 B design I borrowed. With some courage I said to my friend " I remember your 300 B amp sounding better ". He agreed. It was with the " couldn't give it away " 300B. Ricardo said it was stupid on the one hand to love valves and on the other hand never accept new ones.
Unlike most people what valves do which seems to excite people makes me dislike them. I don't like the dull glow or the smell. It's only the sound that is interesting. My speakers I just built would suit valve very well except for the bass end. Without a plan that way I suspect I will go for that option.
Member
Joined 2009
Paid Member
I did change the compensation. The Pioneer does not use Cdom, but I added some when I increased the feedback factor. It's all documented in the SS forum under 'TGM6'.
I wouldn't worry about increasing Cdom to ensure stability. Yes, I've found that some amplifiers are sensitive the value of Cdom (e.g. my TGM1) and there is a range of values that are 'just right' for the perceived sonics whilst also being stable. I don't think the average HT amplifier is worth worrying over to that extent.
Somebody commented on feedback - I've found that my best SS amplifier (it's good enough that I shall never design another SS amp) is one which uses copious amounts of feedback. But I also find it very enjoyable listening to a no-feedback triode SET amplifier.
Last edited:
I will drink to that. I think the Hypex completes the picture. I will try the Hitachi that way. Even the tail can be a resistor. - 92 db or - 127 dB best possible CCS. The problem is the CCS degrades to -92 dB at HF. Split the resistor and add a cap is almost better. People forget that the LTP hardy moves. A resistor running almost constant voltage is also at almost constant current.
Just looked. Very nice project. The distortion spectrum is excellent and no current mirror degeneration. The BCV 61 can be used as two in reverse to make a Wilson Quad if wanting cheap and special. At 2 V there should be no sitting too low . Even at 0.6V you should be OK ( single VAS no Re ).
mullered:
Could you directly adress my questions, specifically:
Could you share some more concrete ideas and things you've done regarding actual circuit design? What kinds of things have you built?
Regarding you claims on tube linearity:
Look at say 4P1L curves, and tell me if you'd find it difficult to get perfectly linear gain out of it?
Could you directly adress my questions, specifically:
Could you share some more concrete ideas and things you've done regarding actual circuit design? What kinds of things have you built?
Regarding you claims on tube linearity:
Look at say 4P1L curves, and tell me if you'd find it difficult to get perfectly linear gain out of it?
I'm going to apply what you wrote to a tube LTP here.
Why do you use a CCS? I use a CCS in my balanced circuits to keep the balance. If you bypass it with a cap, it doesn't do that.
The LTP won't move much as mostly it pumps current into the VAS . Thus it has very little effect on the measurements. Hum or whatever might become of prime importance. You might run that tail on 5 V . It possibly wont be much worse. That aside I am surprised someone didn't say the current mirror is pointless on the LTP to a balanced VAS. As the two times 2SD756 forms two identical voltage clamps. It won't really do much to the balance as that is already about 2% typical. All it would do is make more current available. This would as I said be to able to say it slews better. The Wilson Quad is simple and has almost perfect balance. I did use a CCS to the tail. Measuring as best I could it was the same as the 47 K !
I showed the Hitachi as it is a near perfect design well above any human ear to say it isn't. Add MOSFET's if you feel the load is horrible. Up to six devices the VAS will cope ( reset 12K 6n8 using spectrum analyser if so at 50 kHz ). This is because the 1 nF gate load is bootstraped to the source so it is not hard to drive even at 20 kHz ( distortion still at 0.003% 80 watts 20 kHz, These are Hitachi graphs ). The gate drain is lower and the real load . This amplifier ticks any box. Bob Stewart's copying losses. Very low distortion even at 100 kHz. Easy bias setting ( ears , 100 mA should be fine , 20 mA is already good , not like bipolar mouse trap switch on ). Not unlike valve sound. Not obviously a high feedback design.
If anyone knows. I was starting to sketch a 75 us passive RIAA where the Denon DL 110's 160R + 380 uH forms the 75 uS. This makes the 3180/318 is possible in one op amp with ideal overload margin. Anyone done this ? The output needs to be about 250 mV from 1.6 mV in or 44 dB .
I showed the Hitachi as it is a near perfect design well above any human ear to say it isn't. Add MOSFET's if you feel the load is horrible. Up to six devices the VAS will cope ( reset 12K 6n8 using spectrum analyser if so at 50 kHz ). This is because the 1 nF gate load is bootstraped to the source so it is not hard to drive even at 20 kHz ( distortion still at 0.003% 80 watts 20 kHz, These are Hitachi graphs ). The gate drain is lower and the real load . This amplifier ticks any box. Bob Stewart's copying losses. Very low distortion even at 100 kHz. Easy bias setting ( ears , 100 mA should be fine , 20 mA is already good , not like bipolar mouse trap switch on ). Not unlike valve sound. Not obviously a high feedback design.
If anyone knows. I was starting to sketch a 75 us passive RIAA where the Denon DL 110's 160R + 380 uH forms the 75 uS. This makes the 3180/318 is possible in one op amp with ideal overload margin. Anyone done this ? The output needs to be about 250 mV from 1.6 mV in or 44 dB .
If you checked, you would find I've been stuck in the entrails of *Mr Bogen's old amplifiers recently, which I got dirt cheap.
This came out of neccessity, as nobody could provide me with a monitor system any good in the shops.
It comes after a 30+year absence from having anything to do with this stuff.
(My old s-s amplifier & Matrix H quad system dates from 1977).
This improves them* out of all recognition, but is based on doing some stuff with valves that weren't imagined to be used like that, and the OPT & valves are brilliant.
Again, hi voltage, hi gain valves,- low impedances.
Years ago we could get Parmeko, Woden, Partridge etc stuff for pence from junk shops.Now it's no longer the case.
The other thing I'm working on is SRPP bridges and some odd stuff with multiple seperated screen windings, so it's all about Russian submin valves low voltages and toroids today. You can do all this thanks to the internet today, which allows you to locate and order components from all over the world.
The only thing I'm going to have to pay serious money for is from Oppo.
Does that answer your question?
Thanks.
Is there a stereo compatible recording available?..
I might understand the sentiment of the other audio guys. When I studied at SAE (and about that time) they were introducing it, and a few couldnt quite see the point, me included. But anyway.. interesting recordings are something that I am looking for.
Sun.
When you work somewhere close to the concert management of important concert venues, those things turn up, and nobody notices a DAT machine dropped somewhere into the output of the sono bank.
Of course 15yrs ago, there weren't that many reliable DTD PC multi channel audio systems at a decent price working on NT4 (I had to use 2 sound cards!), so yet again my sound engineer (from Euronews) said WTF is this all about?!!
One surround recording was U Reinemann doing Winterreise live in an excellent concert hall.
He's dead now, so it's the only one ever done.
I have both versions, and again no compression at all.
You can swop between the DTS version, the Stereo version and the 24bit uncompressed -A version on a DVD-A for Cambridge-A/OPPO player.
When you go back to stereo after listening for 15-20 mins in surround, it sounds as flat and featureless as a pancake.
To me compressed stereo is a travesty of the original, but it's what you are all listening to.
The Brahms requiem surround version,- again with the public LIVE, and a choir made entirely of young solists.
Again, the surround version is simply night and day compared with stereo.
(N.W. said it was the closest thing he had heard to being there...)
Maybe the most interesting LIVE were Natacha Atlas and Asian Dub Foundation
We did a lot of "fusion" stuff just then.
The Gladiators was just one of a number of surround recordings.
I'm still waiting for the re-edited/reformatted version to come back as I write.
Alan Stivell live was the BIG highlight in one concert, as we had Beta-num (HD) and I supplied the stereo sound. The video mix guy was the best in the region, (no editing!) and still works for the European parliament/Arte/TF1 in the local capital.
Lots of this stuff ended up under lock and key a bit like Radio France.
A great pity.
I can't find anything in those "hi end" mega-buck systems that can reproduce them, so I had to make it myself, hence why I end up here a few times.
For me it simply HAS to sound pretty close to the original.
That's the acid test.
That rules out SET stuff because they don't come even close to the amps of current required at the business end.
If a reggae band has no bass when played back on a sound system (with compression), then it's some wimpy, tame, hyped up, useless gutless version of the original.
Some speakers I heard, said to be the best in the world.
Phase perfect, omnidirectional but NO BASS.
I don't want that kind of stuff.
Last edited:
At last someone shows an ideal SE Viridian. EF184 looks almost like a ECC81 if simple triode strapped the usual way with 100R g2 to anode. If it's pentode curve is exploited the Alex Kitic RH34 amp can be enhanced with lower distortion and more gain, EL34 being the better valve to my ears. The high Rp of ECC81 can be even higher so allowing better V to I conversion. If the RH design is tested for function the V to I function is truely how it works. If an ECC82 is used the V to I is transformed to standard working as Rp is too low. The sad truth is there are no free lunches this way and the ECC81 only gives marginal advantages. To be honest it's a bit hyped. However when a pentode or MOSFET is used it becomes a true story. The MOSFET has an advantage or two. One is what it gives today it will give forever. EF184 if I am right never was an audio valve? Some TV driving function. By pure chance workable ( 7199 also was an enhanced TV type I think ? ). Here is a variation below with thanks to it's author. Note the beauty of the biasing and very low distortion. At 1 watt perhaps ? My own amp although in detail different is similar. I got to it by a very long route. Wish I had seen this earlier. I don't use the so called plate to plate feedback. One can not feed to the plate as feedback, it is just the Rp allows it to pass without shunting too much. Remember the original question was dynamics of amplifiers. The two I have shown should be better examples. Note how one has massive feedback and this by usual ways of thinking has none. They will sound like two peas in a pod if the output transformer is not pushed too hard at LF. The Hitachi will reduce distortion to absurdity yet do no harm. That is very rare. The fact that no human can detect an overal level of 3 % THD at lets say 90 dB SPL if Hiraga in type seems an uncomfortable truth many will not allow to be said. With the Hitachi I will accept one can have ones cake and eat it. I too would like the low distortion as it seems the right thing to do. When people say valve amps are for fools. Sorry the fool is the one who said it. Even the science says so. That isn't hi fi it is people who study how we hear. Hi fi guru's as an analogy can very easilly see infra red and ultra violet.
- Status
- Not open for further replies.