At the risk of sounding like Grumpy Cat, I agree, the people who are so strongly opposed to global feedback mostly seem to have a cartoon level understanding of it. Delayed feedback, signals echoing round and round and so on.
Of course I have a vested interest because I survived some similar control engineering classes. 🙂
Excessive growth of hair??
Excessive tolerance of low temperatures??
Whatever it is, it sounds a nasty experience for you.
Last edited:
I presume, therefore, that you are particularly familiar with the following:
Putzeys, Bruno (2011) "The F word, or, why there is no such thing as too much feedback", Linear Audio Volume 1
Hamm, Russell O. "Tubes Versus Transistors-is there an Audible Difference." Journal of the audio engineering society 21.4 (1973): 267-273.
Crowhurst, Norman H. "Some Defects in Amplifier Performance Not Covered by Standard Specifications." Journal of the Audio Engineering Society 5.4 (1957): 195-202.
There's one other on measurement of distortion (in a speaker QA context) which has some interesting things to say about the audibility of various harmonics (60db at 11th if I recall correctly) but I'm away from my notes.
Ps the control world is finally moving away from PID to statistical and model based controllers. (Not every problem is a nail!)
This resonated with me. Just came back from the European Triode Festival, where we had several presentations related to the improvement of tube amplifiers. This is of course nothing new – after all, it is a triode festival – but I noticed a red thread if you will.
Frank Blöhbaum presented some very high power single ended designs with very high linearity (100W @ <0.01% THD from SE). Next Guido Tent from Tentlabs fame presented some intriguing circuit variations on the lowly cathode follower topology, improving it to the point where it was an almost idead output buffer, no distortion and an ohm or less Zout.
Then it was Menno van der Veen’s turn to talk about what he called trans – variations on circuit topologies based on transconductance stages, again coming up with a very simple circuit with very high performance.
What struck me in all three presentations was first, that all the variations resulted in very high linearity, very low distortion circuits that are very uncommon in the tube world. Secondly, all these very high improvements were realized by methodologically identifying weaknesses and mending them with a combination of - drum roll - solid state active devices and judicious amounts of negative feedback.
Now this is interesting I thought. We seem to move on from the usual situation where people design tube circuits with mediocre or even sub-par performance while claiming that they sound very good. It is as if the audio world has decided that there is no reason why tube circuits should not be stellar performers technically, and should sound clean and transparent, letting through the music, all of it, without taking something away or adding something.
And this jump in quality seems to come from engineering types who finally decided to take tubes seriously.
Jan
Yup. Well, I hadn't read the Bruno Putzeys article until now, but the contents were no surprise.
In engineering school you learn to design a circuit to meet a specification. So many watts of output, such and such a bandwidth, so many percent distortion. It is understood that the cheapest solution that meets the spec is the best one.
The problem with hi-fi is not meeting a spec, it is mapping from the user's expectations to a technical spec that will fulfil them. I was going to go off on a rant about this, but Putzeys' F word article actually sums it up better than I could. He mentioned amongst other things, that some people buy overpowered solid-state amplifiers (and some people design them) because they like the sound of crossover distortion.
If you try to build a low-distortion tube amp, you are just trying to beat the transistor people at their own game. Well, I guess the amps you described above probably beat Nelson Pass's solid-state designs on performance, weight and cost. 😉
I haven't read the Putzey article, but that doesn't make any sense.
It is well known that some people like distortion. It's generally considered that even order distortion sounds nice to some people - though when E R Wigan of the BBC research department carefully surveyed people in 1961, as to their perceptions of artificially introduced distortion, he found that people noticed even order as an impairment as much as they noticed odd order.
Cross over distortion that arises in a Class B stage is odd order.
Significantly, it has been found that, for equal percentage distortion, cross-over that occurs at zero crossings sound worse than cross-over that occurs at some offset from zero crossings. So if one obtains an amplifier with a higher output power (but the same THD due to crossover), it will sound better, because the change in transfer slope will occur at higher signal level. Supporting this is a curious fact that most of us that have experience designing solid state public address amps (where the emphasis is NOT extremely low distortion, but one always wants to do a competent design) as a professional have noticed: A bipolar transistor Class B stage has an optimum bias point where the THD measured on instruments is lowest. But that is not how it sounds! By ear, the more bias (& thus the more standing current) the better it sounds.
This (the reduction in percieved cross-over distortion) has been advanced as part of the reason for the rapid increase in power output in solid state amps in the 1960's, when the design engineers had not the experience of later years. Along with simply being able to do it for the fisrt time without a huge cost penalty of course.
True. And it WILL cost more. But there is the engineering challenge. And the fun of building it.
And I can tell you from experience: If I say to guests at my house "Here is my latest solid state design." - regardless of whether they be electronics industry friends or just friends, they typically say something like "Oh, right. Did you see the flodding on the TV news....." But if I say "Here is an old (well known name) tube amp I have just finished a full restoration on", they go something like "Oh! How about that. Can I see in the chassis? Can you put a CD on - I'd like a listen? Gee, nice work on the finish. Nice wiring." And it skews the conversation for a good 15 to 20 minutes sometimes.
I haven't built a tube amp to my own design for quite a few years now, but have no doubt another one would provoke interest much more than a solid state amp.
Perhaps its like vintage cars. If one buys a new car, it barely rates a mention. But if a guest brings a vintage car round, it keeps the interest of other guests for quite a while.
Perhaps it is like painting. If I show a photograph I have taken, people look politely for a second or so. If I show the wife's latest oil painting, a reproduction technically far below any photograph, people show quite a bit of interest. And she gets a measure of pride that I do not.
Last edited:
Here is what Putzeys said:
bruno putzeys said:
I think this needs some expansion. The distortion signature he is describing is that of a classic Blameless type amplifier with feedback that rolls off at 6dB/octave, and an output stage design that favours high power over low crossover distortion. I think it is relevant to the discussion of "Crown Slam" we had some time ago. My old Alexander current feedback amp also sounds like this.
This is the reason I built several tube amps myself. But fun has nothing to do with Blamelessness.
If that is what Putzseys said, he's a dead loss as far as I am concerned. It's the language of unqualified audiophile reviewers. Not that of someone who knows what he's talking about. What does "glassy mid-highs" mean? Whatever you want it to mean I suspect. What has "impossibly controlled bass" got to do with 20 dB/decade rolloff? Answer: Nothing whatsoever.
There's nothing about that quote that has anything to do with cross-over distortion though.
No, that's not right. At the risk of repetition, here's what Doug Self meant by "blameless": An amplifier that contains no sloppy engineering or design errors, and thereby achieves the technically best performance that the particular topology is capable of.
If you have sufficient engineering knowlege, the tools, and the will, you can engineer an error free circuit using whatever devices you like. And I personally find great satisfaction and enjoyment coming up with error free elegant designs in all sorts of things - audio equipment (hobby), large diesel engine and power station control systems (the day job), or whatever. Almost anybody can come up with something that works. But coming up with something that works better than what most others can achieve - that's not so easy, and there's a lot of fun and pride in doing it. Whatever the endeavour.
Last edited:
Objectively of course the high frequency distortion has nothing to do with the bass. But we are in the realm of subjective mappings here. There is an old trick used by rock and metal recording engineers, a credit card is taped to the skin of the kick drum next to the beater, adding a high frequency click to the attack of the drum. Adding this click has no effect on the bass output of the drum, but subjectively it makes it sound harder and tighter.
The kind of distortion described by Putzeys probably has a similar subjective effect on bass notes. Of course the flip side of it is a lot of undesirable high frequency IMD, which probably maps subjectively to the "glassy" and "superglue" impressions.
Here I side with Self (and Putzeys)- a power amplifier is not an effects processor, it should just make the signal bigger and not mess with it otherwise. If we prove that the amplifier adds no audible distortion, then we don't have to worry about subjective mappings, as we know they must be void. Just need to spend a lot of money on the case and buy an ad in the same issue of Stereophile that the review will appear. 🙂
I have some experience in building guitar amps and effects. Here the subjective mappings are the main thing. The objective part of the design process is limited to making sure that the unit will run reliably, though many guitar amp designers don't even seem to bother about that.
Last edited:
Scopeboy,
I think you are being kind to Putzeys. You are reading into his words something that you want it to mean - something that makes sense to you. Putsey's words here are the words of an incompetent magazine reviewer. A technically competent person wouldn't write like that. If indeed he did mean something like what you say, he would have put it like you did - written precisely what he meant.
What you wrote makes sense to me. Both the bit about adding impact to drums, and especially that an amplifier should be a replicator not an effects generator. As to whether the HF distortion in a lag compensated amplifier can improve perceived bass - that is another matter.
I too have dabbled in making guitar amps (40 years ago), so I do understand what you are saying. My experience though, is that for professional musicians, reliability is everything. If you get hired by a hotel, and on the night you can't play, they won't just not pay you for the night, they won't hire you again. And they tell the booking agency. And if you play at weddings, there is just NO excuse. One failure not fixed on the spot immediately and the bride will bad-mouth you to all her friends.
But placing Putzeys with Self? I bet if I emailed Self and drew his attention to this, he'd go ballistic. Not about your technical principles - Self's career was about designing/engineering for the recording studio, he'd know a thing or two as well, but about the audiophile reviewer language. This is going on the quote attributed to Putzeys of course.
On another note, I have gleaned from a number of sources, including the Sun Records Story and the George Martin biog, the following story:-
When Sam Phillip's Sun Studio was just a local bit player, and was lucky to get some then unknowns like Elvis, J L Lewis etc walk in off the street, their studio was just a basic shop with economy radio station equipment (ie not proper professional recording studio equipment) and a couple of cheap radio station microphones. None of them were happy with the sound on playback. Whoever was on drums for some reason put his wallet on the skin of his largest drum during a short break. The others started playing, and the drummer just had to come in too. Philips got it on tape, and on playback they found it had an edge that would likely make a mediocre song recorded with mediocre equipment sell really well. And they were right.
When the Beatles began their recording career, they (and John in particular) were not happy with the sound of the drums on playback. They just couldn't get the tight impact that was on the early Elvis records. John continually blamed EMI techs and EMI equipment, all of which was the very best in England and far better than what Sam Philips had. But none of them knew the wallet-taped-to-the-drum trick. The recording engineer tried putting a mike righ up against the drum so as to overload it, breaking the EMI recording rule book, but it didn't work.
I don't know how true or untrue this is. But the better drum sound on 1950's Elvis & Jerry Lee records is clearly much more exciting than the drum sound on 1960's Beatles records. Maybe not as accurate. But more exciting.
I think you are being kind to Putzeys. You are reading into his words something that you want it to mean - something that makes sense to you. Putsey's words here are the words of an incompetent magazine reviewer. A technically competent person wouldn't write like that. If indeed he did mean something like what you say, he would have put it like you did - written precisely what he meant.
What you wrote makes sense to me. Both the bit about adding impact to drums, and especially that an amplifier should be a replicator not an effects generator. As to whether the HF distortion in a lag compensated amplifier can improve perceived bass - that is another matter.
I too have dabbled in making guitar amps (40 years ago), so I do understand what you are saying. My experience though, is that for professional musicians, reliability is everything. If you get hired by a hotel, and on the night you can't play, they won't just not pay you for the night, they won't hire you again. And they tell the booking agency. And if you play at weddings, there is just NO excuse. One failure not fixed on the spot immediately and the bride will bad-mouth you to all her friends.
But placing Putzeys with Self? I bet if I emailed Self and drew his attention to this, he'd go ballistic. Not about your technical principles - Self's career was about designing/engineering for the recording studio, he'd know a thing or two as well, but about the audiophile reviewer language. This is going on the quote attributed to Putzeys of course.
On another note, I have gleaned from a number of sources, including the Sun Records Story and the George Martin biog, the following story:-
When Sam Phillip's Sun Studio was just a local bit player, and was lucky to get some then unknowns like Elvis, J L Lewis etc walk in off the street, their studio was just a basic shop with economy radio station equipment (ie not proper professional recording studio equipment) and a couple of cheap radio station microphones. None of them were happy with the sound on playback. Whoever was on drums for some reason put his wallet on the skin of his largest drum during a short break. The others started playing, and the drummer just had to come in too. Philips got it on tape, and on playback they found it had an edge that would likely make a mediocre song recorded with mediocre equipment sell really well. And they were right.
When the Beatles began their recording career, they (and John in particular) were not happy with the sound of the drums on playback. They just couldn't get the tight impact that was on the early Elvis records. John continually blamed EMI techs and EMI equipment, all of which was the very best in England and far better than what Sam Philips had. But none of them knew the wallet-taped-to-the-drum trick. The recording engineer tried putting a mike righ up against the drum so as to overload it, breaking the EMI recording rule book, but it didn't work.
I don't know how true or untrue this is. But the better drum sound on 1950's Elvis & Jerry Lee records is clearly much more exciting than the drum sound on 1960's Beatles records. Maybe not as accurate. But more exciting.
Last edited:
Putzeys is the designer of the successful Hypex Class-D amps, so I wouldn't dismiss him out of hand. If you read his F word article, you will see that I wasn't equating him with Douglas Self, though I'm sure he knows as much about Class-D as Self does about Class-AB. I was pointing out that he agrees with Self on the issue of the amplifier as a replicator, though unlike Self he is willing to ponder the subjective implications of objective defects.
If you are emailing Doug Self, ask him about the clipping behaviour of his designs while you are at it.
If you are emailing Doug Self, ask him about the clipping behaviour of his designs while you are at it.
OK, so the lesson is that excellent performance can be had from "tube" circuits, if you're prepared to use semiconductors?
That's a bit like saying vegetarian food can be delicious, but only if you add bacon.
Dunno about that last bit. Sounds more like they finally gave up trying to get good results from tubes alone, and resorted to using semiconductors to straighten out the whole mess.
That is a view I do not share. I believe you CAN get excellent performance from tube circuits, without using semiconductors.
I base that belief on:-
a) experience as a professional electronics engineer, with experience in amplifier design;
b) a knowlege that neither tubes nor transistors are perfect active devices, and knowlege that, within reason, it's not the active device that limits performance, it's the circuit;
c) experience in restoration, to ex-factory standard or better, vintage professional electronics. Professional studio electronics of the highest industry standards reached impeccable performance in the 1950's. The microphones available then had some limitations, and the process of mastering onto disks and stamping out black plastic had very definite limitations.
1950's electronics, especially consumer gear, did not reach the high standard of performance that today's solid state circitry can do. But with the 33&1/3 RPM disks used then, there wasn't any point in doing better. Now with CD's there is, and it can certainly be done.
Not cost effectively though.
Yes that's the obvious question we also posed - why hang on to that tube in the first place.
But the general answer is that design can be fun, and things can be interesting for their own cause, not necessarily because they give the best bang for the buck.
Jan
I would see the analogy as 'adding some bacon gives you a shortcut that otherwise would be very difficult and awkward 😉
I think it is more philosophical. Traditionally in the tube world people have been accepting performance that in the ss world would be considered totally unacceptable. Some things were accepted in tube stuff because there was not much you could do about it, with more or other tubes.
But if you throw away those blinders and embrace all of the options you have, using tubes, bjt's , FETs, what have you, each for the duty it can do best, you can do great things and design circuits that also contain tubes and that perform absolutely impeccable. Review Stuart Yanigers Equal Opportuniy as a good example.
Jan
Looks like Pink Floyd......
When I add Pink Floyd to my SE tube amp, I need to add feedback.....so that the bass from DSOTM can heard in the house across the street!
Note, no GNFB is needed or wanted. Just some local feedback in the output stage, cathode feedback + UL.
I don't know where the "echo" business comes from, especially echoes that you can see, perhaps ringing on square waves from improperly applied feedback.
I have recently (been) retired from a 41 year career as an RF engineer at Motorola. Most of my last 6 or 7 years were spent designing LTE transmitters for the 770 MHz or 2.4 GHz band. These transmitters had a DC to 10 MHz modulation bandwidth and have strict requirements on phase and amplitude linearity. As with a speaker, an antenna on a moving car is not a fixed load in either the amplitude or phase domain. Cartesian feedback, adaptive pre-distortion, feedforward, and other proprietary techniques are used.....so I think I understand feedback as applied to audio amps.
We can debate the feedback thing forever and there will always be strong opinions on both sides. I have been building audio amps for about 50 years, both tube and solid state. My OPINION has always favored just enough feedback to get the output impedance down and reduce the distortion to acceptable levels. I prefer this feedback to be applied to a minimum number of stages, working backward from the output stage. Cranking up the open loop gain and then piling on the GNFB to make the specs look good does affect the dynamics.
Granted we have the HARDWARE needed to analyze an audio amplifier to the Nth degree with impeccable accuracy. Why, then do we still test our amps with steady state signals and dummy loads. Can we hear the dull red glow emitted by a 500 watt 8 ohm resistor being tested? Do we listen to a 1 KHz tone or a 7KHz - 60 Hz pair? I don't, do you?
We need the SOFTWARE to apply real music to the amp, with a real speakers as a load, and analyze the input of the amp VS the output. It is fairly easy today to load up a WAV file into a DAW like SONAR or FLstudio, play this file through the soundcard into the amp while recording the amp's output through the soundcard back into the DAW. Most DAW's have the ability to view both WAV files simultaneously while scrolling forward and backward through time.
Your 0.005% distortion amp will have visible distortion on some musical elements. It's no surprise that drum transients are one of the elements that cause visible edge rounding on some amps. The distortion of the same edge will be different depending on what other musical elements are playing at the same time.
We need the software to make these measurements repeatable and automated. We need to define the "elements" for testing, and specify a speaker or speaker simulation that reveals amp flaws.
Note: DAW = Digital Audio Workstation The "lite" edition provided with many high end soundcards is sufficient for this type of work.
I started recording drums with 16 bit 44.1 KHz hardware and a primitive DAW when my daughter lived at home and played the drums (about 2000). The technology wasn't good enough then, but it is now, and I still have her drum set. I have learned a lot about sound, music, and electronics since then, and I will continue down this path when my new house (with basement studio) is finished, but I lack the software skills needed for this job.
I have convinced myself that a speaker is a much more unpredictable load than most people believe, and I will attempt to document this fact next year. I believe that some speakers when fed with a tube amp having a non zero output impedance can appear to have a near zero or even negative instantaneous impedance when stimulated with a sharp transient that tries to reverse the cone's motion.
If we have several competently designed amps with similar outstanding specs, and similar high levels of feedback, and plenty of available output power reserves, why don't they all sound the same? If we can hear a difference, but can't measure a difference, then we need to improve our methods of measurement. How can we define blameless when there are still things that we hear, but can't (or don't) measure?
Granted we can measure all sorts of steady state differences in tube amps, but many of todays decent solid state amps have specs that should make them near blameless, so they should all sound the same.
I do advocate the use of the right semiconductors in the right places, in the right configuration. No semis in gain producing places, only followers in the signal path, and solid state CCS's where needed.
I would say this is more like putting egg and cheese in my vegie salad, but hold the bacon!
Can I build a tube only amp with solid state like specs, absolutely. It would weigh a ton and cost a bundle, but it is doable....would it sound like a solid state amp? I don't know, maybe it's time to build it.
- Status
- Not open for further replies.