|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
92.24.58.30
In Reply to: RE: 3B SST... posted by morricab on April 07, 2014 at 08:03:49
Morricab: Triodes versus Bipolar from Humble
You may be interested to learn young man that between Disbeliever and myself there amounts to over 100 years of listening to both valve & transistor amplifiers , we do not essentially disagree with you about sound quality and can concur that many amplifiers that measured well were largely overrated. However you appear to be widening the goalposts, originally stating that triodes had less distortion than bipolar transistors being a blanket statement without any circuit references, so I had to challenge you. You then go on to rubbish any form of feedback as if it were the bane of any design concept in the audio domain.So firstly allow me to clarify a few points, it is true that taken as as an individual device transistors do have more intrinsic distortion than say a high mu triode but the point I have re-iterated is that a solitary transistor is rarely used in isolation as a straight amplifying device in a high quality amplifier, so the point about distortion becomes irrelevant , simply put, a circuit topography is used because the way in which transistors work allows it to be used to cancel out the very distortion that nonlinearities produce, but in such a way that it essentially still remains within the domain of a single device, meaning no phase anomalies an increase in speed thus widening the bandwidth and significant reduction in all distortions, yes it is essentially within the realms of current negative feedback, but the electrical reaction within the topography is without detriment , it,s now a symbiotic unity. The simplest of which is the longtailed differential pair to the more complex Current mirror load and constant current source, the same CCS is often used to feed the next stage the VAS to improve its linearity.
Now it is also true that with this type of topography very high intrinsic gains are achieved and further distortion from the output devices and drivers are also reduced by negative feedback, however many articles have been written about the proper use of feedback and all those other artifacts of SID, TID etc. which includes nested feedback to reduce phase anomalies , but you can not rule out the fact that many very good amplifiers have been designed with both high & low amounts of feedback, and some like Onkyo with no Global feedback which still produce relatively low amounts of distortion because the circuits are an amalgam of topographies used to reduce the additive stage distortion and if I remember correctly the Onkyo produced around 0.02% that Company believing as you do that an amp with limited distortion sounded better but in fact did,nt.
Referring to Geddes and Lee have you considered what happens when when say an audio amplifier of the transistor variety uses differing devices throughout the amplifying stages , for instance a J.Fet front end with it,s slope character followed by a standard B.polar VAS stage then followed by by say Mosfets, here you have three different slopes all adding their own particular distortions , feed all these back into the early stage and you have a disparity of control which results in both odd and even harmonics but more importantly some of the harmonic intervals are not amplitude consistant meaning some harmonics in the higher reaches have a larger than those proceeding, perhaps this is one reason why feedback can have rather unpleasant sounds with some designs , but is not generally known as a casual link,.. but in the end preferences will always be the deciding factor, however for proof of the proverbial pudding just take a well respected modern amplifier of each variety , in the case of the transitor version remove the input filter if it has one , then feed it with a square wave set at the same level as a valve version , then monitor the output into a dummy load , I think you will be very suprised at the result, because all the other factors of the amps will come into play which include the reaction by capacitors in the signal path and the interelectrode capacitance of the triodes CGK+(A+1)CAG, this then becomes a measure of accuracy not necessarily distortion, because it is measuring the filtering effect of both types of amplifier .
I have already conceded that there are many amplifiers of the valve variety out there which sound very sweet, so essentially we are not in any conflict whatever, I simply believe that a properly designed Transistor amplifier will give very satisfactory and pleasing results and can be blameless, I have had many debates with well regarded audio engineers , consultants and other speaker designers. Everyone has a preference for their particular beliefs and experiences , but they would all tell you that sound quality is of paramount importance but more often than not are convinced that their way is the only way forward and so sometimes get caught up in a negative feedback loop. ha ha.
Humble
Edits: 04/09/14 04/09/14 04/09/14 04/09/14Follow Ups:
No need for the condescension. Maybe you guys have learned something during all your years of "experience" but based Disbelievers amps of choice I am inclined not to think so. Experience is not a substitute for talent.
As for the linearity of devices there is no debate, triodes are more linear than bipolar transistors by a long way. So are FETs for that matter...at least running Class A. Adding more non-linear devices in a circuit doesn't make the circuit more linear, it makes it less so. This is only logical and it can also be shown easily if you build any kind of circuit with them and don't wrap it up in negative feedback...fact is that if you go with multiple stages you MUST use some negative feedback and often it won't be stable without it. At least with a FET or SIT you can take one, run it Class A and it will be stable, sound very good and measure ok to a few watts. With a tube amp you need probably at least two stages but the whole thing will run without feedback and still measure ok and sound really good as long as you used good parts.
Making an amp with multiple stages (this is also true for FETS or tubes) only makes it less linear and then NECESSITATING the use of local and/or global feedback just to control the way too high gain and stability issues.
"essentially within the realms of current negative feedback" Heard amps with this topography like Accuphase and well YAWN, they didn't sound very good despite the glowing press...They sound gray and dead.
"believing as you do that an amp with limited distortion sounded better" Their problem is the same as others, the use of the wrong design gives a poor sound not the absolute amount of distortion. The design dictates what the form of that distortion will be and it is that pattern that is important. Find the design parameters that deliver the right pattern and you get good sound...stray into ultra low distortion techniques and you pay in sound quality. It is not as you think that it is because the distortion is low it is because of what was done to the signal to get it low that is the problem.
"Referring to Geddes and Lee have you considered what happens when when say an audio amplifier of the transistor variety uses differing devices throughout the amplifying stages , for instance a J.Fet front end with it,s slope character followed by a standard B.polar VAS stage then followed by by say Mosfets, here you have three different slopes all adding their own particular distortions , feed all these back into the early stage and you have a disparity of control which results in both odd and even harmonics but more importantly some of the harmonic intervals are not amplitude consistant meaning some harmonics in the higher reaches have a larger than those proceeding, perhaps this is one reason why feedback can have rather unpleasant sounds with some designs"
Once you have multiple stages and complex topologies in each stage AND feed all that back to the input I don't think it matters anymore what the elements are. Especially if one of the stages is Class B. Like I said, a tube push/pull amp in Class AB that is not triode and uses negative feedback will suffer many of the same problems because a non-triode tube is also highly non-linear, feedback exacerbates the problem AND it is likely to have transformer saturation issues unless the output iron is very good. I don't cut them any more slack than most transistor designs.
I have found that one thing that works very well is a hybrid design with two stages only, running Class A or very high bias AB (so like Class A 10 or more watts). With a FET output this can be run without any feedback and will sound very good with low absolute and low order distortion primarily. Make this single ended and it will sound even better but the amp will have to be physically huge.
Of course the pure tube variant and reverse hybrids (like KR Audio) can also sound phenomenal.
"I simply believe that a properly designed Transistor amplifier will give very satisfactory and pleasing results and can be blameless,"
And I would simply argue that no such animal exists or has ever existed, nor will exist until a truly linear amplification device is invented. Until that device is invented then I will stay with the psychoacoustically correct path for sound reproduction.
I also don't really care too much what engineers think on these matters...they are usually designing by what they learned is proper circuit design and to price points for the market. The majority of them are not scientists and don't know how to make the proper observations about what their products are doing. I am a scientist, which makes me a professional observer, which means i hear things in their products that they don't or wish not to hear. I may or may not know how to design various cool and tricky circuits but I know how to tell when one is doing the right thing and one is not. THen I bother to find out what was used and how it was implemented and wait to see other similar examples and how they sound. One can build a mental library of what sounds consistently good and what doesn't and start to draw conclusions about various "innovations" in circuits. I have done so and came up with some guidelines for a good sounding amp. I would share them with you but I doubt you would take them seriously. I have posted them before so feel free to search under my name and you might find my ideas.
"in the case of the transitor version remove the input filter if it has one , then feed it with a square wave set at the same level as a valve version , then monitor the output into a dummy load , I think you will be very suprised at the result"
Why on earth would you remove the input filter? That's silly. All systems have finite bandwidth and dynamic range and I think that one of the biggest failures of home audio gear is NOT limiting the input to what the device can swallow. Many audio people think the world stops at 20Kc. and removing input filters just punishes those that know better.
But... I am very much in favor of evaluating systems by using square waves, I'm just not in favor of modifying them first.
Rick
From HUMBLE
Input filters were originally used amongst other things to prevent the main amplifier from suffering SID especially , when the pre-amp was much faster. Bearing in mind that many earlier amplifiers had relatively slow Bi-polar output devices which meant that as the frequencies increased less feedback was available and distortion crept up sometimes significantly, modern out put devices like Sankens and some Toshiba audio devices are extremely fast with an FT up to 40mhz making the amplifier much faster and thereby keeping the distortion lower at much higher frequencies. BY removing the input filter and some amplifiers don,t have them ,enables you to see what the amplifier is capable of without the inhibiting effect of that filter which often comes into effect depending on the designers criterion of frequencies as low as 150kHz, taking the filter out of the Hafler 200 series enables a beautiful square wave to be reproduced above 10kHz, with the filter in the square wave was much more rounded as you would expect.. There is a strong correlation between the speed of transistors and point of roll-off in the treble particularly prevalent with slow output devices a factor that feedback does not totally compensate for, of course Mosfets enable a much faster performance and often noted for a detailed treble especially with supertweeters. When the JL Linsley-Hood 75 watt amplifier was first produced it had 2 t 3 mhz FT O/P devices, later these were substituted with BDY56 10 MHZ devices and immediately the distortion dropped by almost a decade and that amplifier had 78 db of NFB, it also sounded faster and easier on the ear.
Humble
Good Post!
I have no problem with injecting test signals after an input filter, or any place else for that matter to evaluate a system's response. I also believe that using square-waves and a scope are far more likely to get you to the root cause of audible problems than peering in the frequency domain with a spectrum analyzer, especially an FFT one that doesn't go very far out-of-band.
All "properly designed" systems need to make a reasonable attempt to not accept signals that they can't handle. I think the best case is to have enough bandwidth margin to allow a simple RC LPF as the dominant pole and still be high enough to not mange the treble.
Regards, Rick
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: