rotel 820 stan curtis

[lt cdr data]

I read somewhere that curtis explained how he tweaked teh bass end of this amplifier in the power supply.

Anyone any idea how precisely?

a specific time constant of the caps and the load? ( however, that load will vary depending on what speakers are used)

or a combination of resonance between the transformer and smoothing caps?

thanks!!

[AndrewT]

Hi,
the relationship I use (but not universally accepted by other builders) is:-
Set input filter time constants to about, Low pass 0.5uS to 1.5uS
High pass 80mS to 100mS for sub-bass or 20mS for universal amp that does NOT pass bass signals.
Set the NFB time constant at least half an octave lower i.e>140mS
Set the PSU time constant at least half an octave lower than the NFB. i.e.>200mS.
This last recommendation may be something similar to what S.Curtis was referring to.

My interpretation of those RC time constants leads to +-20mF for the PSU in an 8ohm single channel amp or +-40mF for 4ohm.
That has led to considerable disagreement between builders on what affects bass performance.

Read other's and decide/experiment for your self.

[lt cdr data]

thanks andrew, where also would RC power rail decoupling come in that scheme?

best wishes

[janneman]

Originally posted by AndrewT
Hi,
the relationship I use (but not universally accepted by other builders) is:-
Set input filter time constants to about, Low pass 0.5uS to 1.5uS
High pass 80mS to 100mS for sub-bass or 20mS for universal amp that does NOT pass bass signals.
Set the NFB time constant at least half an octave lower i.e>140mS
Set the PSU time constant at least half an octave lower than the NFB. i.e.>200mS.
This last recommendation may be something similar to what S.Curtis was referring to.

My interpretation of those RC time constants leads to +-20mF for the PSU in an 8ohm single channel amp or +-40mF for 4ohm.
That has led to considerable disagreement between builders on what affects bass performance.

Read other's and decide/experiment for your self.

Andrew,

This is quite interesting. I know Stan Curtis did 'something' but was not aware it involved tweaking time constant ratios. Is there any other stuff available on the net where this is addressed that you know of?

Jan Didden

[AndrewT]

Originally posted by lt cdr data
where also would RC power rail decoupling come in that scheme?

What I do and it may not be correct, is find the current fed from a particular decoupling and calculate the equivalent R to match the current with the supply voltage. Then check that the RC time constant is below the frequency I am interested in. I suspect it should be more complicated than that but it appears to work or at least identify if the cap has significant ripple and then investigate it's purpose.

[AndrewT]

Originally posted by janneman
I know Stan Curtis did 'something' but was not aware it involved tweaking time constant ratios. Is there any other stuff available on the net where this is addressed that you know of?

Hi Janneman,
I have seen a few papers that use the RC ratios suggestion. The main purpose is to guarantee avoidance of motorboating that used to be more common in valve circuits, but I can't see why the same logic should not be applied to SS circuitry.

Once one accepts that the staggered ratios are important, it follows that the HP turnover is directly related to the capacitance fitted into the smoothing of the PSU. This of course only applies to the simple unregulated PSU.

There are many that laugh at my recommendation of +-20mF for an 8ohm amplifier and point to the thousands of amps that work "perfectly" well with 25% to 50% of this level of smoothing. Few seem willing to even experiment.

[janneman]

No, your numbers make perfect sense. I once read a story by Stan Curtis that he was able to "tune" an amp so that it didn't measure particularly well on the testbench, but boy it sounded good!

He didn;
't give details but I read between the lines that it had to do with the interaction of real loudspeakers with 'time constants' in the amp.

Jan Didden

[lt cdr data]

well, I recall the 'tuning' of the rotel was to do with bass guitar, so if you look at the fundamentals and harmonics of that instrument and work out the frequencies.

also, say the rhythm of music, say 120bpm, you may be able to coincide the discharge current and values of smoothing caps

problem is like I mentioned is that it will vary obviously depending on the speaker used.

its a particular interest of mine how so things are 'tuned'/voiced, I am sure its very simple, basic equations, but there are really no references to it.

there may be a way to do it with possible compression things like attack and release times, and also phase manipulation, as that changes bass character, too, but I havne'e mangaged to fit any of the pieces of the puzzle together yet properly

I don't know if you could use cap values to resonate at a certain bass frequency with the secondary inductance of the mains tx ( likely to be small, but that's isolated by the bridge rectifer??)

[AndrewT]

Hi,
try to find Dr Cherry's article on tuning the high pass filter in the NFB loop.
It produces a hump (Q>0.7?) in the low end response just at the turn over frequency.
This frequency could easily be chosen to coincide with audio signals and thus become very audible.
Much the same as tuning a speaker to have a bass hump to mimic bass notes and what they are doing is exaggerating the second harmonic to trick the ear/brain into hearing the bass.

[lt cdr data]

I once heard julian vereker, of naim, say that he increased the time constants in the power supply to put a hump in the bass.

I am sure there is no 'magic' to it, just a posh way of saying he increased cap values. I think they try to camaflauge it in posh words to make them look cleverer than they really are, and there is a simple explanation for it.