Jiiim,
I agree that a phase shifted sinewave looks the same as a time delayed sinewave, on the screen of a scope for instance. But there is no time delay involved. If you consider the general case of a square wave or impulse this model breaks down.
I make this point because all too often real understanding is lost by incorrectly extrapolating a specific model. This is what led to ALW's question about whether a fixed time delay across all frequencies makes any difference. There is no fixed time delay in Naim's filter so the question has a false premise.
BAM
I agree that a phase shifted sinewave looks the same as a time delayed sinewave, on the screen of a scope for instance. But there is no time delay involved. If you consider the general case of a square wave or impulse this model breaks down.
I make this point because all too often real understanding is lost by incorrectly extrapolating a specific model. This is what led to ALW's question about whether a fixed time delay across all frequencies makes any difference. There is no fixed time delay in Naim's filter so the question has a false premise.
BAM
>>>But there is no time delay involved. If you consider the general case of a square wave or impulse this model breaks down.
Could you explain what you are thinking here?
ALL analog filters have a time delay associated with them.
(or read that as phase delay if you like, as the time delay only results in a delay of a fraction of a wavelength in a real world filter). There are probably some filters where the time (or phase if you like) delay is zero at a particular frequency.
Group delay is simply a method of describing how this time (or phase if you like) delay changes with frequency.
As previously discussed, the group delay is the pertinent factor in this discussion. Overall time delay is pretty irrelevant.
Jim.
Could you explain what you are thinking here?
ALL analog filters have a time delay associated with them.
(or read that as phase delay if you like, as the time delay only results in a delay of a fraction of a wavelength in a real world filter). There are probably some filters where the time (or phase if you like) delay is zero at a particular frequency.
Group delay is simply a method of describing how this time (or phase if you like) delay changes with frequency.
As previously discussed, the group delay is the pertinent factor in this discussion. Overall time delay is pretty irrelevant.
Jim.
This may be deteriorating into a semantic debate. Provided designers use maths or a simulator to work out the behaviour then all will be ok. 
My argument is simply that a time delay (aka propagation delay) is a specific function that is not the cause of sinusoidal phase shift in an analogue filter. Consider these examples:
a) If you feed an impulse into the filter the output will respond immediately. There will be no delay in the output response.
b) Suppose you have a perfect low pass filter of knee frequency F. You inject a sinewave of frequency f and at the output you observe a lower amplitude sinewave with a 45 deg phase lag. You deduce that there is a time delay. Then you add a perfect high pass filter of frequency F to the output and then observe the output of the new filter. You observe the original input waveform is restored. Do you then deduce that the HPF has a negative time delay? Has time been reversed?
c) The Laplace transform for a filter is quite different from the Laplace transform for a time delay.
I accept that it is often handy to describe a phase shift as a time delay at a specific frequency.
My argument is simply that a time delay (aka propagation delay) is a specific function that is not the cause of sinusoidal phase shift in an analogue filter. Consider these examples:
a) If you feed an impulse into the filter the output will respond immediately. There will be no delay in the output response.
b) Suppose you have a perfect low pass filter of knee frequency F. You inject a sinewave of frequency f and at the output you observe a lower amplitude sinewave with a 45 deg phase lag. You deduce that there is a time delay. Then you add a perfect high pass filter of frequency F to the output and then observe the output of the new filter. You observe the original input waveform is restored. Do you then deduce that the HPF has a negative time delay? Has time been reversed?
c) The Laplace transform for a filter is quite different from the Laplace transform for a time delay.
I accept that it is often handy to describe a phase shift as a time delay at a specific frequency.
Hi Bam,
Yes, I'm of the feeling that we are arguing semantics more than anything else.
Here's my response to your points:
a) so there is a frequency (or range thereof) where the delay is zero.
b) Hmmm. got me on that one for the mo. Need time to think about it. However, a constant sine wave is a confusing way to analyse this behaviour as it is repetitve - ie the final output could be n2PI out of phase. A single cycle of a sine wave would be a better way to investigate.
c) I dont think laplace transforms are a good way to analyse the phase behaviour of a filter, as they throw away the phase information. Or is it the fourier transform that does that? it's been so long since I've had to think about transforms, i'm a bit rusty.
Jim.
Yes, I'm of the feeling that we are arguing semantics more than anything else.
Here's my response to your points:
a) so there is a frequency (or range thereof) where the delay is zero.
b) Hmmm. got me on that one for the mo. Need time to think about it. However, a constant sine wave is a confusing way to analyse this behaviour as it is repetitve - ie the final output could be n2PI out of phase. A single cycle of a sine wave would be a better way to investigate.
c) I dont think laplace transforms are a good way to analyse the phase behaviour of a filter, as they throw away the phase information. Or is it the fourier transform that does that? it's been so long since I've had to think about transforms, i'm a bit rusty.
Jim.
Hmmmm - dragging this up from the dead. I am looking for info on what the NA729 TA boards do differently from the normal Naim input buffers (I've a NAC 62 on the way, my first Naim, that I intend to mod a bit). My previous experience of lower end hi-fi is to remove buffering/filtering where ever possible - I assume Naim do things a little better from the outset, but don't understand this time align thing.
Can anyone recommend any reading or posts that describe it in laymans terms? The previous 4 pages of this thread was a little too detailled for me! FWIW - IMO you shouldn't delay anything in a signal. And if you do it is critical to apply the same across the entire frequency range. But that's basics, no?
Thanks, Richard
Can anyone recommend any reading or posts that describe it in laymans terms? The previous 4 pages of this thread was a little too detailled for me! FWIW - IMO you shouldn't delay anything in a signal. And if you do it is critical to apply the same across the entire frequency range. But that's basics, no?
Thanks, Richard
Maybe me confusing buffer with filter then I've seen reference to bypssing the TA boards in some threads too? What do these boards actually do?
I used to be a technician who moved to networking many, many moons ago for the money. I played with audio output stages of CD players and preferred the sound with a few discrete components replacing the op-amps and filtering.
Richard
I've seen reference to bypssing the TA boards in some threads too? What do these boards actually do?I used to be a technician who moved to networking many, many moons ago for the money. I played with audio output stages of CD players and preferred the sound with a few discrete components replacing the op-amps and filtering.
Richard
I have read this again with interest after looking at naim circuits.
For stability, dominant pole first of all, my limited knowledge, I note a few things.
Having looked at the circuits of 100s of amplifiers, I possess and have on the web looked at lots,. the stability of 3 stage amplifiers, varies from the downright simple to the very very complex, with little inbetween.
Loooknig at some japanese amplifiers, for example, a rather large rotel, I can't recall the name, the very biggest one, 360watts, its just about similar to all rotel circuits.
In the beginning, rotel, based on the 820 types, simply were not mirror push pull inputs, whavever you would call them.
they are a single diff pair input.
later one, michi style, 1060, R1062 and all amps were mirrored, ie 2 diff pairs, a balanced? input stage, but with a single ended, RCA input connection.
reason I am briging up the big one is the stability is far more complex.
http://www.audiogon.com/cgi-bin/cls.pl?ampstran&1255897847&/Rotel-RB-1090-380-wpc-powerhou
There are caps all over, ihnput colloector to collector, ie across diff pair, base-base, just about everywhere in this amp there are stability caps.
many 70s jap amps have quite a few, eg, either a network across the input pair collector or base.
As regarding caps around the VAS, afaik, there are around 3 ways to do it.
Active feedback, std. miller compensation from VAS collector -base.
In mirrored amplifiers, I havne't seen this used much, more often is a capacitor to ground from collector, lag compensation?
which is usually in conjunction with a step network across the input collector load resistor, RC to power rails.
And finally, there is the rare, I hardly ever see, of VAS output to input diff. pari negative, linsley hood was a lover of this, I have only seen on about 2 amplifers his and another.
Lastly, Randy slone talks about pole splitting in his book. ie 2 miller caps.
90% of amps I see have miller comp, ie active.
most duals, mirrors don't
and a very few have the linsley hood method.
I hope this brings some fruitful discussion.....
For stability, dominant pole first of all, my limited knowledge, I note a few things.
Having looked at the circuits of 100s of amplifiers, I possess and have on the web looked at lots,. the stability of 3 stage amplifiers, varies from the downright simple to the very very complex, with little inbetween.
Loooknig at some japanese amplifiers, for example, a rather large rotel, I can't recall the name, the very biggest one, 360watts, its just about similar to all rotel circuits.
In the beginning, rotel, based on the 820 types, simply were not mirror push pull inputs, whavever you would call them.
they are a single diff pair input.
later one, michi style, 1060, R1062 and all amps were mirrored, ie 2 diff pairs, a balanced? input stage, but with a single ended, RCA input connection.
reason I am briging up the big one is the stability is far more complex.
http://www.audiogon.com/cgi-bin/cls.pl?ampstran&1255897847&/Rotel-RB-1090-380-wpc-powerhou
There are caps all over, ihnput colloector to collector, ie across diff pair, base-base, just about everywhere in this amp there are stability caps.
many 70s jap amps have quite a few, eg, either a network across the input pair collector or base.
As regarding caps around the VAS, afaik, there are around 3 ways to do it.
Active feedback, std. miller compensation from VAS collector -base.
In mirrored amplifiers, I havne't seen this used much, more often is a capacitor to ground from collector, lag compensation?
which is usually in conjunction with a step network across the input collector load resistor, RC to power rails.
And finally, there is the rare, I hardly ever see, of VAS output to input diff. pari negative, linsley hood was a lover of this, I have only seen on about 2 amplifers his and another.
Lastly, Randy slone talks about pole splitting in his book. ie 2 miller caps.
90% of amps I see have miller comp, ie active.
most duals, mirrors don't
and a very few have the linsley hood method.
I hope this brings some fruitful discussion.....
Naim do not release schematics - period. The only one available is that generic schematic for the NAP250, which has been around for many years and is posted all over the internet and often seen in guides for NAP140 clone kits, for which the supply voltages and some resistor values are incorrect. All early NAP models share the same basic electrical design though, with only small changes to resistors setting bias currents according to the various power supply rail voltages of each model.
The NAP200 clone PCB has the parts values and types marked on the overlay for the best guide. Some kit information here may help:
Black Box ?Clone Naim NAP200 Amplifier kit Diy Power amp kit 75W+75W naim nap200 | eBay
Sorry, but you won't find a simple answer or complete schematic unless you reverse-engineer it for yourself from the available information. Reading the more recent discussion in the big NAP140 thread may also help with determining the specific semis used, as these are all important to the sound quality, which, if you are interested in the product, will be essential to its performance, just as the special PCB is.
The NAP200 clone PCB has the parts values and types marked on the overlay for the best guide. Some kit information here may help:
Black Box ?Clone Naim NAP200 Amplifier kit Diy Power amp kit 75W+75W naim nap200 | eBay
Sorry, but you won't find a simple answer or complete schematic unless you reverse-engineer it for yourself from the available information. Reading the more recent discussion in the big NAP140 thread may also help with determining the specific semis used, as these are all important to the sound quality, which, if you are interested in the product, will be essential to its performance, just as the special PCB is.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.