[qbe]

Hey what does an integrator do in the Gilmore amp? (Kevin calls his servo an integrator).

If it is trying to cancel out the DC in the output, and if it was an inverter (which I would have thought it was, going in to the inverting input of the OP27 opamp) I could understand. You would get a negative DC signal of the same level as the existing DC signal and that cancels out the existing DC in the output, giving you a resultant zero value.

But, trying to remember my maths from xxx years ago, a DC signal is a constant value. An integral of a constant should be a slope of constant value (ie a rising voltage ramp for a positive DC input signal) and I can't see how that will cancel out the DC.

So, because the circuit works I'm wrong somewhere, please let me know what the integrator does.

 

[jeffreyj]

Quote:

Originally posted by qbe Hey what does an integrator do in the Gilmore amp? (Kevin calls his servo an integrator).

It corrects voltage offsets that are the result of long-term drifts in component values. For example, a mismatch between Vbe's in the input differential pair would be seen as a valid signal as far as the rest of the amp was concerned. Long-term is key, here: an integrator is basically an averaging circuit with a time constant greater than the lowest valid signal frequency of interest. Statistically speaking, the average of the positive parts of the audio passing through the amp should equal the negative excursions; if they don't, then the integrator (or servo) injects a correcting voltage into the inverting input of the amplifier (whether it be discrete or integrated).


Quote:

But, trying to remember my maths from xxx years ago, a DC signal is a constant value...

DC only means the voltage does not alternate polarity, it doesn't mean the voltage remains constant over time.

 

[Chipko]

Quote:

DC only means the voltage does not alternate polarity, it doesn't mean the voltage remains constant over time.

So an AC voltage becomes a DC voltage if you move the ground reference below the lowest point of the original AC voltage?

Better to divide all voltage in to a DC and an AC part and possibly neglect one of them depending on the specific situation.

If I understands Jeffreyj's answer correctly the integrator splits the signal in to a positive and negative part and make sure that the integral (area) of the two are equal over time. Any changes it makes to the reference must be slower than that of the lowest audible frequency or the servo will not be transparent to the listener.

[RANT]All naturally produced soundwaves should have a integral equalling zero since the physical object that generates it strives to return to its natural state, be it a string, a vocal cord or a skin on a drum. Not only does an offset produce DC that's potentially dangerous to a speaker. But since in its simplest mechanical form the speaker is a spring. An offset means the membrane moves less in the direction of the offset for a given amount of energy applied while the opposite is also true. This also applies to the eardrum but here the offset is a difference in airpressure between the outside and inside of the head.

Since most recordings are done with several microphones this whole thing fails at the mixerboard already since the signals from the sources are added electronically not taking in to the account the effect of superpoisitioning of frequencies.[/RANT]

 

[jeffreyj]

Quote:

Originally posted by Chipko So an AC voltage becomes a DC voltage if you move the ground reference below the lowest point of the original AC voltage?

Correct. If the voltage troughs reach zero then it is called pulsating DC; if they don't go below the average value then it is called ripple. These are somewhat arbitrary distinctions...


Quote:

Better to divide all voltage in to a DC and an AC part and possibly neglect one of them depending on the specific situation.

You have to apply common sense, rather than rules, as to when a signal should be treated as DC or AC. The small amount of ripple that makes it past a regulated DC supply does not render the output of the supply AC, right? Yet neither does that mean it is totally ignored. Same with the voltage fluctuations of a battery when subjected to differing loads; one doesn't normally refer to it as pulsating DC, even if in the strictest sense it is.


Quote:

If I understands Jeffreyj's answer correctly the integrator splits the signal in to a positive and negative part and make sure that the integral (area) of the two are equal over time. Any changes it makes to the reference must be slower than that of the lowest audible frequency or the servo will not be transparent to the listener.

Correct, again. This is only a slightly more mathematical way of putting it (and very similar to my original answer).

Quote:

[RANT]... Since most recordings are done with several microphones this whole thing fails at the mixerboard already since the signals from the sources are added electronically not taking in to the account the effect of superpoisitioning of frequencies.[/RANT]

Not quite. If the individual signals integrate to zero over time t then the sum of those integrals will also be 0 over time t. Only for instaneous points will the integral be non-zero, which is as it should be.

 

[aeberbach]

jeffreyj wouldn't it make sense to say that every signal has both AC and DC components? A DC offset and an AC frequency and amplitude. In practise either component can come close to zero but rarely is.

 

[jeffreyj]

Quote:

Originally posted by aeberbach jeffreyj wouldn't it make sense to say that every signal has both AC and DC components? A DC offset and an AC frequency and amplitude. In practise either component can come close to zero but rarely is.

Err... This is exactly why I said some common sense needs to be applied here. In the strictest sense, yes, all signals have a DC offset - no matter how miniscule it may be. The DC component might be from asymmetrical reproduction of a signal (even harmonics) or leakage through a coupling capacitor. Whatever the source, it will be there, but if it is 0.001% of the AC level, who cares?