Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Jon Hodgson]

maxdimario wrote on Tue, 18 October 2005 13:35

Quote: And surely you can't consider the tube in isolation from the resistive and capacitive components in the circuit? What is the effect if a reduced plate voltage on the frequency response of the system as a whole? actually I can. I can make an amplifier with a triode, a plate resistor a grid resistor and a dc blocking capacitor, and it will still do it, even though the elements involved do not make the difference. I can make a tube mike soften up by keeping the voltage at 100 V or sound sharper with the B+ at 250 V. both will go way over 30 KHz transformer aside. this is not news...it is 'common' design knowledge for audio electronics.

So what is happening to the transient?

How is its shape (and sound) being affected by the reduced B+?

[Jon Hodgson]

If you fed a low frequency square wave (basically lots of transients) into the amplifier with the two different plate voltages, then what would the output look like?

[maxdimario]

Quote:

On the other hand, how does this translate to sampling? It doesn't. In a constant rate sampling system, transient response is directly connected to frequency response.

my whole point is that oversimplifying the concept of transient response by relating it to 'pure' models has little to do with how the system works in a physical real world.

in reality every kind of ADC DAC has a different transient behaviour being different machines, as I suspect different sample rates and depths do as well, also because of the changes that they imply on the system as a whole.... power supplies...analog stages. switching..the ADC chip's internal architecture..etc.

anyway, there'd be a lot less arguments if people didn't defend digital using pure textbook terms all the time, and actually related their comments to the physical machines that embody those simple theories. (digital converters and routing systems) a little more often.

[maxdimario]

Quote:

If you fed a low frequency square wave (basically lots of transients) into the amplifier with the two different plate voltages, then what would the output look like?

you can see the sharper edge on the square, AND you can hear it speeding up... even though your hearing only goes to 20KHz!

I did not make this up, people who design with tubes for audio have done this forever.

[Jon Hodgson]

maxdimario wrote on Tue, 18 October 2005 13:49

Quote: On the other hand, how does this translate to sampling? It doesn't. In a constant rate sampling system, transient response is directly connected to frequency response. my whole point is that oversimplifying the concept of transient response by relating it to 'pure' models has little to do with how the system works in a physical real world. in reality every kind of ADC DAC has a different transient behaviour being different machines, as I suspect different sample rates and depths do as well, also because of the changes that they imply on the system as a whole.... power supplies...analog stages. switching..the ADC chip's internal architecture..etc. anyway, there'd be a lot less arguments if people didn't defend digital using pure textbook terms all the time, and actually related their comments to the physical machines that embody those simple theories. (digital converters and routing systems) a little more often.

You are quite right that a real system has a multitude of variables.

However in a good design it should be possible to reduce the effect of those variables to the point where any error introduced by them in the converter is negligible.

[Jon Hodgson]

maxdimario wrote on Tue, 18 October 2005 13:53

Quote: If you fed a low frequency square wave (basically lots of transients) into the amplifier with the two different plate voltages, then what would the output look like? you can see the sharper edge on the square, AND you can hear it speeding up... even though your hearing only goes to 20KHz! I did not make this up, people who design with tubes for audio have done this forever.

Ok, well the first thing is that if you can see the sharper edge, then more high frequencies are getting through on the transients, no escape from that one.

As for hearing the difference, that is most likely down to those high frequencies combined with non-linearities in your output stages and transducers generating components in the audible range.

[maxdimario]

Quote:

As for hearing the difference, that is most likely down to those high frequencies combined with non-linearities in your output stages and transducers generating components in the audible range.

absolutely NOT.

I'm afraid I've heard the results too many times to think it's added 'components'.

Whatever it's doing, it's doing BETTER, because the effect is not good sound but REALISM.

You can make a sound more present and in your face by increasing voltage and current (reducing plate resistance).

I was just doing this the other day, and I did look at a 150 KHz square wave on the oscilloscope to see how much it would change..just for curiosity's sake, it does change a little in shape, becoming more squared off (this also depends on the capacitance of the load).

the end result is that the transients are more realistic and 'air-like'.

the voices sound like they are 'in front' drums have a distinct presence.

the fact that the bandwidth is increasing over the hundreds of KHz shows that it's not to do with the 20-20KHz sine response but something to do with sharp impulse response.

...MIND YOU, I didn't say I think that this has to do EXCLUSIVELY with the increased frequency response, I am talking about how the electronics react to transients, and trust me.. every kind of active component reacts differently.


it's easy to get caught up in technical jargon that can make sense out of complex reality.

it's more difficult to deal with reality effectively.. much more difficult and time-consuming..

Anyway... what I said before.

[Jon Hodgson]

maxdimario wrote on Tue, 18 October 2005 16:51

I was just doing this the other day, and I did look at a 150 KHz square wave on the oscilloscope to see how much it would change..just for curiosity's sake, it does change a little in shape, becoming more squared off (this also depends on the capacitance of the load).

150 KHz?? Your fundamental is above audible there.

And the clue appears to be in what you mention about the capacitance of the load, does it not?

So where can I find a technical explanation of this phenomenon? It is happening at the circuit level, so must be measurable, and after 100 years of valve circuitry, I would guess it has been analyzed and explained.

I'm not being fascetious, I am interested in the explanation

[maxdimario]

Quote:

150 KHz?? Your fundamental is above audible there.

I used 150 KHz because it's easy to see any deformation on the 'scope.

I don't know the technical reason why tubes reproduce transients faster and more realistically with higher voltages and currents (up to a point of course).

I imagine it has something to do with the fundamentals of emission etc.  I have not studied it. I am only interested in designing circuits and using them as alternatives to what I can find.. and most people who design with tubes know this as working knowledge.

Transistors react to transients differently too, and it doesn't have to do with freq. response, as most modern audio transistors go past the audio range into the hundreds of K.

frequency response is NOT the only dynamic in determining a system's potential to reproduce transients NOTICEABLY better or worse.

[Jon Hodgson]

Can you have a sharp transient which contains no high frequencies?

Can you have a transsient containing high frequencies which is not sharp?

[Jon Hodgson]

Also, how is the waveform distorted when you reduce the plate voltage?

[zmix]

maxdimario wrote on Mon, 17 October 2005 09:16

One of the tricks you can employ in a tube circuit to make the sound softer and make transients less evident is to lower the plate voltage. the funny thing is that the tube still retains it's bandwidth of over 200 Khz but the transients are slower sounding, imaging becomes a little fuzzy. Sometimes I see people writing about transient response in digital as a function of bandwidth only, which is wrong in any case other than simplistic theory. in the real electronic world, every circuit and component reacts to transients differently...not necessarily related to bandwidth.

When the voices inside your head tell you to make an apple pie and all you have are oranges....

Gee Max, did it ever occur to you that lowering the B+ voltage will reduce the headroom of the circuit, and that this will cause the transients to clip?

What do you mean to say when you state that the "imaging becomes a little fuzzy"? are you refering to the aliasing components (aka imaging) or are you referring two the integrity of the stereo image if listening to two channels? Can you imagine any other factors which come into play here?

[Jon Hodgson]

zmix wrote on Tue, 18 October 2005 18:14

When the voices inside your head tell you to make an apple pie and all you have are oranges....

Now we're talking!!!

For apple pie I definately do advocate the highest sample rate possible  

Can someone please pass the custard?

[C-J]

maxdimario wrote on Tue, 18 October 2005 15:27

Bandwitdh is RELATED to transient reproduction, but it is NOT the only thing that determines it, in a physical, complex, electronic real-world system. it has nothing to do with theoretical limits in digital systems.

Max,
We had a similar discussion on the "Building WC cable" thread, where Dan gave a simple, technical explanation about the relation between transients and bandwidth:

Quote:

Now here is the interesting question: You want to know the maximum slew rate for any signal within a given bandwidth. Say the bandwidth is 20KHz. But you know the slew rate for a 20KHZ sine wave. Clearly, audio is not only a sine wave. Can you tell the maximum slew rate for ANY signal within 20KHz? Will a sudden step yield faster slew rate? The answer is: The maximum slew rate is the slope of the 20KHz full scale signal (it happens when crossing zero volts). There is no other signal within the 20KHz bandwidth with faster slope. One can not have a sudden step when the signal is bandwidth limited. The slope of a band limited square wave is lower then the slope of a sine wave at the highest possible frequency within the bandwidth. It may not be completely intuitive, and it is often not understood, because marketing types like to talk about sudden steps and "high frequency transients", often creating the false impression that there is "fast signal activity" beyond sine waves. It is not so. The fastest signal occurs when the signal is a full scale sine wave at the highest frequency within the bandndwidth. Regards Dan Lavry

C.J.

[maxdimario]

Quote:

Gee Max, did it ever occur to you that lowering the B+ voltage will reduce the headroom of the circuit, and that this will cause the transients to clip?

gee zmix, did you know that tubes have more headroom than transistors, and I have been doing this for 20 years?

maybe you should build 50 or so tube pre's and amplifiers before you comment.

Even at low level signals WELL BELOW clipping this happens.

don't make assumptions 'til you have the experience to back it up.