|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
70.55.193.247
Why do tube amps have such low slew rates?
Follow Ups:
Slew limit is the maximum rate of change the amplifier (usually the output stage) can produce. At slew limiting, any signal you feed in, is converted to a triangle waveshape who's slopes are at the slew limit.
The issue is that during slew limiting, any information in the input signal is lost other than fundamental frequency of the triangle wave.
Early SS amplifiers before feedback had a very low high frequency corner but they also had a lot of gain .
That gain could be exchanged for bandwidth using negative feed back and so with a sine wave, those amplifiers could go to 20KHz or beyond (although most used a De-coupling choke wound over a resistor to disconnect the load higher up).
Tubes have no bandwidth limit problems, they have bandwidth limiting issues as nearly any audio tube is happy working at least into the AM radio band.
The limiting things are capacitor coupling and especially the transformers which have level and frequency dependent issues and add a unique "spectrum" of harmonics which has the same envelope shape regardless of input freq (kind of weird). In any case, Tube amps rarely use huge amounts of negative feed back and it isn't needed to bump the amps inherent open loop corner response of say 100Hz, out to 40KHz as with SS.
Anyway, Slew limit distortion (was also called TIM once upon a time) slew limit became another hifi sales buzz word where it is assumed more is better.
How much do you need?
Lets pretend one had recordings made with modern electronics, microphones and like all CD's your upper bandwidth ends at 21Khz more or less.
So next take the maximum RMS voltage your amplifier can produce and recall the peak voltage in that sine is that RMS voltage times 1.414.
Then recall the maximum rate of change for a sine wave is 2 *pi times the peak Voltage.
From that info, you can pick a high frequency like 21KHz, figure out it's time period for one cycle, figure out what the rate of change is for that undistorted sine wave at full Voltage swing and then convert the period to Volts per micro-seconds.
The slew-rate of a tube is usually limited by the geometry and dimensions of the grid. The thinner the grid is, the faster the tube. Also, the plate material and plating can have an effect. My understanding is that the grid ultimately limits the HF response in most tubes. If you get a Siemens CCA grey-plate (6922), this has an extended response to probably 50kHz, but it is the exception. It was a limited build binned version of the best of the best.I have compared this to other tubes in the output of a DAC, where there is only capacitive coupling, no transformers. The caps were tin/Teflon.
Edits: 04/25/14
I'm not sure that that 21 kHz sine wave is the true worst case. I would like to see specific assumptions and mathematical calculations before I would believe this. One also needs to handle spurious (non-audio) signals that can appear various ways, including ultrasonic noise on recordings and even RF hash output from DACs.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Hi Tony
This 21KHz thing does not mean the amplifiers response stops there, it only means that it can’t produce a full Voltage rail to rail undistorted sine wave higher than that. I don’t see the need for an audio amplifier to be required to deliver full rated power above 21KHz.
Best,
Tom
Edits: 12/13/13
My requirement is that an amplifier should deliver undistorted output at any volume level of any music signal that is ever played. I am not interested in ratings, which are just advertising BS. My test is that the system must play all my music recordings at 10 dB louder than row 5 concert volume without audible distortion. (This requires going to the far reaches of the room so that my ears don't distort and aren't damaged.) Unless one can get uncompressed dynamics at natural volume levels one will never make reproduced music sound realistic except for a limited set of recordings (e.g. solo harpsichord music).My system will do this for all acoustic music except for 32 foot organ pedal notes. (One could argue that organ music is not really acoustic music unless the wind chest is being hand pumped by a choir boy.) However, my system will not handle artificial sounds, such as amplified rock concerts, not to mention a Space Shuttle launch (140 dB at 10 Hz at the VIP seating area). I suspect you have made systems that can handle these. My late wife loved your "train start" recording. :-)
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Edits: 12/14/13
In many respects, you and I have similar perspectives and understanding of sound quality, whether simply listening, or recording and reproducing sound. But I've noted that you almost always refer to music when discussing sonic performance. This is not a criticism, just an observation.
I am what some people call a sound addict. It doesn't have to be music. I am equally interested in and stimulated by all sounds outside the arena of music, from the closing of a door to the barking of a dog or chirping of the birds, the hum of the refrigerator, and everywhere in between. In this regard, I listen to and hear the sometimes subtle and sometimes obvious differences between a naturally occurring sound and its reproduction. Do you have this same kind of obsession with sound, or, are you mostly interested in musical sounds? Just askin'.
:)
No obsession with sounds for their own sake, or even music. Just for the divine essence of music, when and where it can be found. :-)
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
"One also needs to handle spurious (non-audio) signals that can appear various ways, including ultrasonic noise on recordings and even RF hash output from DACs."
Aye, especially for open systems, and brother is home audio ever one, inputs simply can not allow signals in that succeeding circuitry can't handle. Typically you need to passively filter out RF and the like and have fast enough circuits that the input filters can be gentle and far from the signals. Audio guys don't worry about it and RF guys are apt to use hi-K caps that screw up the sound.
Rick
Hey Tom, a couple thoughts/questions/comments...
Regarding this comment:
"From that info, you can pick a high frequency like 21KHz, figure out it's time period for one cycle, figure out what the rate of change is for that undistorted sine wave at full Voltage swing and then convert the period to Volts per micro-seconds."
Putting that into perspective, I would think that, since upper frequencies typically require less power/current output than lower frequencies, we'd be less concerned about a full power voltage swing to reproduce them. On the other hand, using an amplifier for full-range usage (as opposed to bi- or tri-amping), the higher frequencies "ride on top of" the lower frequencies (when one looks at a complex waveform), and thus are at their mercy with regard to amplifier performance.
I'm no John Curl by any means, but these would seem to be considerations. (Now he's going to come here and bitch-slap me. Which, by the way, I'll learn from! )
Your thoughts?
:)
Hi
Your right, one would never find a full Voltage swing at 20KHz in any music and if one looks at the spectrum of real music, one finds the last upper octave is nearly always well down relative to say 100-400Hz
The full swing at 20KHz, actually is conservative and this doesn’t mean the amplifiers response rolls off there, only that it can’t produce an undistorted sine wave at the full rail to rail Voltage swing above 20KHz (and a signal which would let the magic smoke out of any tweeter with a large amplifier).
Tom
Easy for you to say!
Slew rate is the measure of an amp's capability to change output potential per unit time. Often stated as volts per microsecond.
Tube amps have limited bandwidth (due to output transformers) and high output impedance relative to their solid state brethren. Hence they aren't able to change potential nearly as fast as solid state, driving a comparable speaker impedance.
Insufficient slew rate is mainly an issue with amps of very high power ( > 200 wpc ), where the limit in change of potential per unit time compromises bandwidth at full power. The "power bandwidth" spec is related to output power and slew rate, and is a good spec in regard to real world power capability.
I'm assuming that you mean tube amps with output transformers. In that case, FWIU, it is not the tubes that are limiting the slew rate but rather the output tranny. Basically it is a big challenge to design an output tranny that can handle the deepest bass as well very high frequencies. As such a compromise is reached that keeps the output tranny from going up high enough so that it is not affecting the slew rate. David Berning has a tube amp design that uses a very high carrier frequency on the outputs so that He can use much smaller output trannies and thus not impact the slew rate as much as a traditional output tranny.
Further proof is to look at OTL amps which do not have the bandwith or slew rate issues of a more traditional tube amp using an output tranny.
Thanks. This would mean that the odd transistor amplifiers using output tansformers have a low slew rate also?
A solid state amp that has output trannies could have it's slew rate impacted although it would probably have it impacted less than a comparable tube amp. The reason being, FWIU, is that the output tranny for a solid state amp is easier to design, thus it can have a wider bandwith which equates to a better slew rate.
BTW, it is possible for a poorly designed solid state amp without output trannies to have a terrible slew rate as well.
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: