|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
98.144.114.174
We know it's inefficient - of course. But does this inefficiency come with a SONIC penalty?
This is what I mean: The inefficiency of high bias requires *more output devices* for equivalent output power. There's really little doubt that increasing the number of output devices has some sonic penalty; there's a reason why both tube & transistor amps with a single pair of devices per channel are known for sonic purity. (Sure, extremely close matching of devices may reduce the penalty to zero or near it, but, at the least, that comes at great expense.)
Am I all wet? I just spilled my scotch all over myself, so, yes, I am, but am I intellectually as well as literally?
Follow Ups:
There is no sonic disadvantage assuming you are comparing the Class A amp to a similarly configured Class AB amp ... same power transistors, for example. However that is rarely the case, or put another way you typically can choose from an array of options with very dis-similar topologies. So it's Apples and Oranges. But limiting it to two varieties of Apple, so to speak, no.
nt
nt
I'm not sure your premise is true. It's been a very long time, but AFAIK class A doesn't require more devices, it requires a great deal more heat dissipation. Keep the heat controlled, and you can use the same number of devices.
Your limiting factor is still the rail to rail voltage and the current capability of the devices. That doesn't change going class A, I seem to recall.
However, practical limitations on the effectiveness of heat sinks may be a limiting factor. This is one main reason the early Krell's had built in fans. :)
Best,
Erik
I have noted that the class A push-pull transistor amps I'm familiar with have more output devices than what I'd consider the very best non-class-A push-pull transistor amps.
Put a better way, I'm not familiar with a Class A push-pull transistor amp with a single pair of transistors per channel that puts out 60-80W/ch, while there are several AB examples (LFD, Einstein, even a mid-fi Denon that Tony of Tonian Labs used to rave about). In fact there are probably all kinds, and perhaps many sound not so good.
I'm completely willing to accept that this is little more than a coincidence. Indeed, I pretty much assumed my premise, which is apparent in retrospect. I just figured there was reason related to Class A operation that Pass XA amps, for example, use so many parallel devices.
Of course, I realize there are theoretical reasons to *prefer* more output devices and that there are many variables here.
I'm glad you came along and pointed this out. :) Thanks.
You may want to check out First Watt designs, I'd be surprised if you did not find a Class A amp there with a single pair of output devices.
Some of Pass' FW designs have just 1 output transistor. :)
All of the FW amps are severely power constrained though.
Best,
Erik
Been discussed here already. I've owned almost all the FW amps. All are excellent. None produce 60-80W into eight ohms, which is the minimum I can get away with now.
Nor do I claim a FW would have enough power. You just mentioned you had not seen a class A amp with a single pair of transistors. :)
Best,
Erik
I qualified it with the power rating, but no matter.
Very sorry. I'm afraid I'm going to be very foggy headed for a while, so I beg your patience.
Best,
Erik
I don't think there is any obvious negative correlation between the number of output devices and sonic purity. There are a lot of performance gains from paralleling output devices, e.g. lower output impedance and higher damping factor without using more negative feedback, better performance and stability into lower impedance loads, more power, possibly even lower distortion if it results in operating the output devices in a more linear region. Component tolerances are critical even if you have a single pair of output transistors per channel. I don't see why the tolerances need to be tighter if you're using 16 output devices per stereo amp vs. 4.
Well, the point is that when you have one push-pull pair per channel no matching whatsoever is necessary. Ok - that's not quite true - you still have the pair itself.
But when you have that last bit of air flowing through the trumpet, decaying into silence, is it better to have the signal going through 10 pairs, or one?
(And yes then we can go one step further and assert that single-ended (with one device) should be even better.)
I'm just musing here. I am not a designer. My technical knowledge is limited. But I know, for example, that in the world of DHSETs, paralleling tubes is widely thought to carry a detriment, and I agree. And the same goes for paralleled pairs in push-pull valve amps as well. Then again, because tubes drift, the problem would seem to be much greater there.
I was just after some opinions, so thank you.
Well, you want to match the two channels as well, so there is a minimum of 4 output transistors that have to be matched. I've read that transistor tolerances are very tight these days due to modern manufacturing methods, so maybe there isn't much to be gained from hand matching anymore. Also, if you're using paralleled BJTs in the typical arrangement, the emitter resistors inherently balance the output current across the devices, so if the devices aren't perfectly matched their variations kind of average out. So I would think that the more output pairs you connect in parallel, the less critical it is to match. One other thing I forgot to mention was noise. The self-noise from each transistor is uncorrelated with the self-noise from other transistors, so when you parallel them up their noise adds incoherently while the signal adds coherently. This improves SNR. In the best case where the output noise is dominated by the self-noise of the output transistors, you could net up to 3 dB SNR improvement by doubling the number of output devices.
My real-world experience with minimal output stages mostly comes from owning a Simaudio i-5 LE, which has a pair of BJTs per channel and is rated for 75W into 8 ohms and 120W into 4 ohms. I actually owned two of them. I ended up regretting selling the first and bought a second one on Agon, which I still have but am not currently using. I also own a Simaudio W-5 LE and Titan. At one point, I connected the W-5 LE to the preamp outputs of the i-5 LE so I could make a direct comparison of the amplifier sections. These share a similar amplifier topology and use the same devices, but the W-5 has 4 pairs of BJTs per channel vs. one and much greater power supply capacity. The little i-5 has slightly warmer and sweeter midrange tone, and seems a little faster and more rhythmically engaging, but the W-5 has greater dynamics, slightly greater transparency, and deeper tighter bass, though it's a bit leaner through the upper bass & lower midrange. I could not say the i-5 is sonically purer.
I've also heard Naim amps a number of times including a couple of auditions with my speakers. Naim uses a single pair of output transistors in their designs, but they are not among the most sonically pure solid state amplifiers I've heard. Naim amps are a bit grainy to my ears. However, they are rhythmically engaging and good at timing, so that is something they share with my Simaudio.
Better thru a fully balanced amp like spectral .....
People make class A for it's sonic benefits. If paralleling output devices is sonic suicide then all you have is good room heater.
Or, maybe there is a lot of doubt that increasing the number of output devices has some sonic penalty?
13DoW.
.
we would see class AB preamps.
Even switching amp advocate Bruno Putzey's Makua preamp is class A. :)
That's because very wide range highly linear low power devices are available. It's when you get into higher power devices that linearity is more challanged.
Paul-
I see nor hear any disadvantage(s) of Class A power amps.
Recently I swapped out a Denon POA-2200 amp (class AB, 200wpc) for a Bedini 25. (class A, 25wpc). Neither are a high end product even by the most liberal definition, but I much preferred the Bedini. Yes it's a fraction of the rated output power but with a high gain preamp and reasonable efficient speakers 25 watts is plenty. I don't see the downside at least in this case.
Bedini not high-end? I know that amp has a VERY good rep.
A factor may be that it actually puts out much more than 25W, in class AB. It seems many Class A amps work that way. For example, the "30 watt" Pass XA30.5 actually clips at something like 110W into 8 ohms.
Honestly asking. I wouldn't think all of them do and I don't know if the Bedini 25/25 does.
The Bedini was high end back in the day ....
Almost all transition especially as they have to drive lower impedances. At best you can only be reasonably sure they are class A up to their rated 8 ohm power. And I believe many are really just rich class AB, say up to 25 watts since you can claim it's very rare that such an amp goes class B. About the only push/pull amp that really was the old Levinson ML2 amp rated at 25 watts at 8 ohms and stayed class A even at 2 ohms(100 watts).
I always assumed true class A amps remained in that mode regardless of the output demanded.
A traditional AB amp operates in class A only within limited output, up to 5 watts for example, then shifts to class B when higher demands are made. Some amps, such as the Parasound JC-1, include a switch to select the point for transition. From memory, normal bias remains in class A up to about 10 watts while the high bias selection retains class A to 25 watts. But those amps are rated at 400 wpc into 8 ohms so that is still significant operation in class B.
Hopefully someone with appropriate knowledge of electronics will answer your question.
"The piano ain't got no wrong notes." Thelonious Monk
I believe there will always be some AB headroom in a *push-pull* class A amp though I'm not certain what determines it. (Single-ended, of course, is class A by definition.)
SO pure class A means the BIAS voltage is 100% ALL THE TIME.
So if it is a say 50 watt amp, the bias is over 50 watts all the time. A 200 watt class A amp would be a 200watts plus on 100% of the time.
A LOT of heat to get rid of!!!
In fact a class A amp is using the most power AT IDLE. And uses less when it is making music.
With a partial class A, the constant voltage is actually about whatever the A rating is, If the A rating is a moderate percent of the total AB rating. the heat sinks will be sized to fit that A idle wattage use. More so than the actual full AB rating of the amp.
On the other hand... ALL preamps ae usually 100% class A. and plenty of them do not run very warm..
Not quite , they are 25% efficient and correct they can only be full biased into class-a at 8 ohm, well for SS amplfiers.As to multiple outputs , this is a necessity when dealing with SOA , Beta droop , et al. If only doing for 8 ohm load and low power, one could get away with one or 2 pr , Nelson does with his F series of amplfiers. 3Pr is my min for good sonics and SOA , but the amount of necessary class-a is usually determined by speaker load and sensitivity , this is critical , as our ear/ brain calculus will pick up the transition if too soon or too late.
Never heard a small Amp sound better than a big Amp , all things being equal , same for speakers ...
Class A toobs are different , one reason due to the output Transformers ...
Edits: 10/11/16
Class A apparently needs more robust power supply (at least per output) than A/B; this adds substantially to cost.Class A will probably need to be powered-on longer before reaching optimal sound quality than A/B or (of course) Class D. My Pass Labs X150.5 need 30+ minutes to nearly reach its best sound -- and its Class A only to 10 watts. (Some might say leave the Pass 'On' 24/7 but, No, because it idles at 200 watts which is just too much.)
I love the music of Dmitri Shostakovich ...
Edits: 10/11/16
Class A is very snooty. It looks down on Class B, and Class D becomes resigned to life in the gutter.
Edits: 10/11/16
More electricity consumption means higher electric bills.
It is also less green as a result thereby creating a negative for the environment. But since I don't drive and don't have kids (which is a huge negative impact on the environment as we don't need more people sucking the valuable air and pooping everywhere) I am comfortable with the elevated electricity usage. Heck, having kids is far worse than driving a Hummer.
Did you get that from your parents , i guess your pop would have preferred a Hummer :)
...low power output.
If you like switching distortion, class A doesn't have it so it could be a disadvantage :-)
mt
Yes, we all know that one already.
Whole construction of amp has to take into account the constant heat. Chassis, board construction, sockets, etc. In general, a good Class "A" should cost more than a comparable Class AB. And may have more reliability issues.
Ventilation in the home setup is also more important.
8^)
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: