I know perfectly what he speaks about. The "silence" (we skate on thin ice now) is meant like "silence between tones" or "silence under tones" or after great dynamic peak. No grain and nervosity. This "silence" is an attribute of very low distortion amps, not achievable by moderate products.
PMA said:
In virtually all recordings, there is of course not silence between notes, but reverberation tailing off into silence. This is a low-level signal which carries a great deal of psychoacoustic information; you are quite right that poor-quality electronics (e.g. quantisation or crossover distortion, or instability) tends to mangle this first.
However, when I hit pause on my CD player, what I expect to come out of the speakers is nothing, period. Nothing is a very different sound to sitting in an empty concert hall; not being able to distinguish the two seems like sloppy listening to me...
Cheers
IH
I don't think we are talking here about silence, like silence when you press pause on your CD, but about impression of "silence" when listening to the music. This is still silence, but it should not be taken literally. Once you have a chance to hear it, you will know what the reviewer is talking about.
Pavel REost 17
regirtitation of knowledge from others is why we repeat the same mistakes and lose how big the circumference in our abblities can stretch...my knowledge for what it is has been through practice as the base fundamentals have eroded away through out the years hence most people tip toe in grey areas
DIRT®
ost 17regirtitation of knowledge from others is why we repeat the same mistakes and lose how big the circumference in our abblities can stretch...my knowledge for what it is has been through practice as the base fundamentals have eroded away through out the years hence most people tip toe in grey areas
DIRT®
quote:
"Another boundary-stretching aspect of the Halcro's sound was its silence. Its background was not the outer-space void of a stream of digital zeros but the taut, deep silence of an empty concert hall "
Unh huh..... Even the most florid wine snobs sound more sober minded than this type of audio reviewer. I just wonder if in person the writer looks more like Yoda or Morpheus
"Another boundary-stretching aspect of the Halcro's sound was its silence. Its background was not the outer-space void of a stream of digital zeros but the taut, deep silence of an empty concert hall "
Unh huh..... Even the most florid wine snobs sound more sober minded than this type of audio reviewer. I just wonder if in person the writer looks more like Yoda or Morpheus
Hi Pavel,
I just noticed this thread and wnet to your website to look at your measurements. Those are some very impressing results! Congratulations!
I don't mean to hijack this thread, but I see that you are using Spectralab. I would very much like to know which soundcard and generator you use to get such accurate measurements. Care to share?
I just noticed this thread and wnet to your website to look at your measurements. Those are some very impressing results! Congratulations!
I don't mean to hijack this thread, but I see that you are using Spectralab. I would very much like to know which soundcard and generator you use to get such accurate measurements. Care to share?
Peter Daniel said:Once you have a chance to hear it, you will know what the reviewer is talking about.
I know well the effect he's talking about, thanks. He's just describing it in a particularly misleading way. Anyone who imagines the gaps between notes in a typical recording is "a stream of digital zeroes" is mistaken - just take a quick look with a PC digital audio editor.
The implication of this badly-written passage is that the Halcro amp is somehow producing something other than 0V out when given 0V in - I'm sure Bruce Candy would be aghast at this suggestion.
Cheers
IH
> the whole termination issue - on one hand impedance matching isn't recommended anymore in audio standards, on the other hand it seems to me that a) higher signal current means less influence of picked up RFI and b) reflected RFI in unmatched lines can intermodulate the signal...
So use shielded wire and block that RFI!
For 3-foot interconnects in the average home, RFI should not be much of a problem.
However, at hundreds of MHz, a 3-foot cable may be 50Ω, or Short, or Open, depending on the exact frequency you put in. TV wires have this problem. Matching eliminates this problem.
The same thing can happen at audio, but only on a cable that is many miles or KM long. Even at 20KHz, you need to get up around 10 miles to have cable resonance problems. And the resistance of 10 miles of affordable wire usually damps the resonance.
Back about 1915, we had a phone wire from Boston to Denver USA. No amplifiers in 1915! So it was made with really fat low-loss copper. If the far end was not terminated, it sounded like you were talking in a hard 10-foot room, with 20 milliSecond echoes and comb-filtering throughout the speech range.
But typical home or small-studio lines are FAR shorter and have FAR shorter reflection times. 3 NANOseconds for a typical interconnect. If there is even 1dB of damping on the line, then by 100 nSec the echoes are 100dB down and inaudible. The top of the audio band is MICROseconds: another 1,000:1 lower in time-scale. Unterminated lines of reasonable length cause no audio problem.
In the days when radio networks ran analog lines between cities, they did have to terminate. Lines were often un-shielded and never perfectly balanced, so a nonreactive 150Ω resistor helped damp any RF that got on the cable. It also masked the capacitance (basically by making the driving amp work hard at all frequencies).
However lines thousands of feet long often work fine with low-Z source and hi-Z load.
In fact the current broadcast industry standard for lines (the type that used to be 600Ω matched) is 60Ω source and hi-Z load. It has been found to work very well, even for inter-building links, even when there is a radio transmitter on top of the building. The 60Ω source sucks-up the RF, and the "far end" is not far enough away that it isn't damped.
About the last thing you want to do to audio is terminate. If you have RFI: Use Shielding. Get your source down below 200Ω (and be sure it is 200 all the way up the spectrum). Go Balanced, with proper balancing systems (transformers or multi-amp designs; the 1-opamp "balanced input" is not too great). Move the wire away from the problem. If all of that fails, you may have to really think and experiment, and it may be that Matching is part of the answer. But in 25 years of pro audio, I don't think I have ever been forced to Match.
Before the US telephone long-distance network went all-digital, you would occasionally get a line that "echoed". Mostly this was a sick repeater amp or cross connection along the way, but the time-scale was the length of the trunk wire.
Even today, when we put DSL signals on mile-long wires from home to central office (or concentrator box), the high frequency DSL signal has to calibrate itself to the line's echoes. (Even 56K modems do this a little.)
Baseband Video is very sensitive to reflections, and is always Matched on cables more than a foot or so long.
So use shielded wire and block that RFI!
For 3-foot interconnects in the average home, RFI should not be much of a problem.
However, at hundreds of MHz, a 3-foot cable may be 50Ω, or Short, or Open, depending on the exact frequency you put in. TV wires have this problem. Matching eliminates this problem.
The same thing can happen at audio, but only on a cable that is many miles or KM long. Even at 20KHz, you need to get up around 10 miles to have cable resonance problems. And the resistance of 10 miles of affordable wire usually damps the resonance.
Back about 1915, we had a phone wire from Boston to Denver USA. No amplifiers in 1915! So it was made with really fat low-loss copper. If the far end was not terminated, it sounded like you were talking in a hard 10-foot room, with 20 milliSecond echoes and comb-filtering throughout the speech range.
But typical home or small-studio lines are FAR shorter and have FAR shorter reflection times. 3 NANOseconds for a typical interconnect. If there is even 1dB of damping on the line, then by 100 nSec the echoes are 100dB down and inaudible. The top of the audio band is MICROseconds: another 1,000:1 lower in time-scale. Unterminated lines of reasonable length cause no audio problem.
In the days when radio networks ran analog lines between cities, they did have to terminate. Lines were often un-shielded and never perfectly balanced, so a nonreactive 150Ω resistor helped damp any RF that got on the cable. It also masked the capacitance (basically by making the driving amp work hard at all frequencies).
However lines thousands of feet long often work fine with low-Z source and hi-Z load.
In fact the current broadcast industry standard for lines (the type that used to be 600Ω matched) is 60Ω source and hi-Z load. It has been found to work very well, even for inter-building links, even when there is a radio transmitter on top of the building. The 60Ω source sucks-up the RF, and the "far end" is not far enough away that it isn't damped.
About the last thing you want to do to audio is terminate. If you have RFI: Use Shielding. Get your source down below 200Ω (and be sure it is 200 all the way up the spectrum). Go Balanced, with proper balancing systems (transformers or multi-amp designs; the 1-opamp "balanced input" is not too great). Move the wire away from the problem. If all of that fails, you may have to really think and experiment, and it may be that Matching is part of the answer. But in 25 years of pro audio, I don't think I have ever been forced to Match.
Before the US telephone long-distance network went all-digital, you would occasionally get a line that "echoed". Mostly this was a sick repeater amp or cross connection along the way, but the time-scale was the length of the trunk wire.
Even today, when we put DSL signals on mile-long wires from home to central office (or concentrator box), the high frequency DSL signal has to calibrate itself to the line's echoes. (Even 56K modems do this a little.)
Baseband Video is very sensitive to reflections, and is always Matched on cables more than a foot or so long.
These are well known issues for electrical enginners. The whole think is that termination effect is audible. This is not trivial and I do not find trivial answers. IMHO the case is most probably in properties of materials/dielectrics. I would not have believed it 2 years ago and had been able to bring same evidence as you did.
I also think the issue is not trivial, and much has been written on it.
I use balanced shielded interconnects), about 10 ft long (active speakers), with quality chips for in and output (DRV134 and INA134. I read much about and applied pro audio balanced standard recommendations (as per Rane and Jensen publications). I have L-C line filters (not DIY but commercial units, lower end price I admit).
And I still hear clicks and pops from ... wherever, the line, atmospheric pickup, who knows? ... Sigh.
These days, especially here in Singapore, every other house has a cell phone booster transmitter, and three 1.5 kW A/C units, and zillions of fluorescent lighting with 100 kHz switcher loads. I strongly believe all this has something to do with it and is not trivial to solve.
I use balanced shielded interconnects), about 10 ft long (active speakers), with quality chips for in and output (DRV134 and INA134. I read much about and applied pro audio balanced standard recommendations (as per Rane and Jensen publications). I have L-C line filters (not DIY but commercial units, lower end price I admit).
And I still hear clicks and pops from ... wherever, the line, atmospheric pickup, who knows? ... Sigh.
These days, especially here in Singapore, every other house has a cell phone booster transmitter, and three 1.5 kW A/C units, and zillions of fluorescent lighting with 100 kHz switcher loads. I strongly believe all this has something to do with it and is not trivial to solve.
We often forget that our audio chain works in real world – not only in range of audio frequencies. No cables, transistors, resistors and capacitors which work only in audio range. No ideal filters, shielding or protections for HF and RF frecuriences. Our equipment must interfere with them. We can only try to reduce their influence on the audio stuff. Of course we cannot hear above 20 kHz but our equipment can. This probably one of the main reasons to build simple circuits.
Pavel,
first of all, great site with nice amp designs!
Now, terminated lines. I had a look at your site and saw that you use a 50 ohm input termination resistor. That is, you have added a 3 dB noise figure by doing this. Is it your opinion that the sonic benefits outweigh this increase in system noise?
Secondly, have you compared terminated (do you use a 50 ohm cable BTW?) single-ended transmission to non-terminated differential ("balanced") transmission? I would like to hear your opinion on this too!
Cheers
/Magnus
first of all, great site with nice amp designs!
Now, terminated lines. I had a look at your site and saw that you use a 50 ohm input termination resistor. That is, you have added a 3 dB noise figure by doing this. Is it your opinion that the sonic benefits outweigh this increase in system noise?
Secondly, have you compared terminated (do you use a 50 ohm cable BTW?) single-ended transmission to non-terminated differential ("balanced") transmission? I would like to hear your opinion on this too!
Cheers
/Magnus
Swedish Chef said:
Hello Magnus -
ad1) the answer is yes. This is not only my opinion, but that of quite large audio community here who have the possibility to hear the comparison, even in blind tests. IMHO the increase in system noise is negligigle, especially for high level link signals like CD. In fact - HF noise, probably EMI/RFI generated, is considerably lower (measured).
ad2) again - proved on balanced transmission to (2 buffers with terminating resistors) with same results.
I tend to believe that it is a high power link signal transmission (high signal current into terminating resistor) that makes a difference, compared to standard voltage low power transfer (high Rload, usually 10 kOhm and more).
Regards,
Pavel
Pavel,
Absolutely true. I was thinking preamp -> power amp where it makes a difference. Before an attenuator set to -60 dB it really is negligible...
Have you tried this with both "high bandwidth" devices like the BUF634 and "low bandwidth" devices like a standard opamp with GBW<10 MHz (OPA2132 for example)? I was just thinking that with a very fast buffer the (unterminated) cable reflections could perhaps upset the buffer with a possible recovery time and/or resulted IM distortion that could fall within the audible range. Oh, well just a thought. I have some BUF634 lying around so it would be interesting to play around with them a little.
Regards
/Magnus
Absolutely true. I was thinking preamp -> power amp where it makes a difference. Before an attenuator set to -60 dB it really is negligible...
Have you tried this with both "high bandwidth" devices like the BUF634 and "low bandwidth" devices like a standard opamp with GBW<10 MHz (OPA2132 for example)? I was just thinking that with a very fast buffer the (unterminated) cable reflections could perhaps upset the buffer with a possible recovery time and/or resulted IM distortion that could fall within the audible range. Oh, well just a thought. I have some BUF634 lying around so it would be interesting to play around with them a little.
Regards
/Magnus
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.