I've done a lot of modifying of gear with the intention of attenuating the amount of high frequency, non-audio signals, noise, moving around inside the enclosure. High frequency signals are notorious for being able to couple to just about anything, one way or the other, and I have always heard benefit by reducing this element. PSRR always falls off in circuits with increasing frequency, so you're certainly doing the component a favour by reducing that type of noise.
Typically, for me, better snubbing of high frequency glitches reduces the "dirtiness" of the sound, improves the sweetness and subjective extension of the treble, enhances apparent dynamics.
Frank
Last edited:
This is consistant with a zillion people's comments over decades on film capacitors' being used for pwr supply bypass. Similar affect occures with a wide band, low Zo regulated pwr supply. -RNM
I thought you said the heights of sidebands on IMD test had varied. So the intermodulation behaviour had altered; adding the 4uF capacitor changed the supply's impedance characteristics, as RNM stated, sufficient for audible effects ...
Frank
Yes. Upper and lower sidebands became of slightly different heights that indicated phase intermodulation. I could not understand why, but found a working hypothesis that it is because of semiconductor properties of oxide film dielectric that play with inductance and capacitance modulating reactance with current and voltage.
It was a tube amp. Primary impedance of output transformers was 10K. Impedance of 4 microfarad capacitor on 20 kHz is about 2 Ohms. 20 Ohms on 2 KHz, 200 Ohms on 200 Hz, and so on. It shunted 4 caps in series, 2,000 microfarad each. 500 microfarads total.
If to remember about experiments with spark gaps, ten microsecond delay is perceived as audible shift of sound source image. I can't think about another explanation.
Last edited:
You have 4 caps in series, order of 20-30nH ESL each, so probably over 100nH total. The paralleled cap will be of the order of a couple of nH, definitely changes impedance characteristics there.
Plus, screw terminals on the electro's? These are as dodgy as hell, my first decent, bought amp used these and the quality of the connections here caused all sorts of audible problems. I ended up ripping them out and replacing with leaded caps throughout.
Frank
Plus, screw terminals on the electro's? These are as dodgy as hell, my first decent, bought amp used these and the quality of the connections here caused all sorts of audible problems. I ended up ripping them out and replacing with leaded caps throughout.
Frank
Yes, but you should not add up each curve because they are not separate lumps of power but the same lump of power decaying with time. It would be like taking temperature measurements in my room every minute and then adding them up to 'prove' that it is hot enough in here to melt steel! I'm sure you can do better than that.RNMarsh said:
Those graphs tell us very little without more information about exactly what was done. Lets assume someone applied a step change in voltage. The exponential decay of the bipolar suggests some series resistance. The slight ringing on the other one suggests some inductance. No great surprises? Stop Press: I see you have just confirmed this in a later post. Why do people always talk about 'time smearing' when they see a first order rolloff in the time domain? if you put a perfect resistor in series with a perfect capacitor you would get graphs like that.RNMarsh said:
Indeed, some people here seem to miss your point.
Your point is valid, of course. Possibly it's difficult for some people to break through their long held thinking patterns.
Being an audiophile and retired electronics technician, I found out that published measurements of audio gear, as long as they are not grossly off, are absolutely meaningless to the sound quality as perceived by me.
I'd love to hear from you if you had come up with measurement that are more correlated to the realism of reproduced music than frequency response, THD and IMD.
There seems to be very little effort here to separate solid engineering practice from extraordinary claims. I find discussions like Rod Elliot's far more objective. He even removes pages as he learns better. I can't claim it was me, but after I posted somewhere that he had the distortion characteristics of microphone capsules in constant charge vs constant voltage mode reversed he fixed it.
Capacitor Characteristics
I still have not seen an explanation of the claim of a device that distorts a complex waveform and not a sine wave of the same peak amplitude. "Sounds like" and measurable are different.
Capacitor Characteristics
I still have not seen an explanation of the claim of a device that distorts a complex waveform and not a sine wave of the same peak amplitude. "Sounds like" and measurable are different.
Last edited:
I have just had a second look at the plots shown by RNMarsh in post 24421. It is probably an 8th-order filter (my rough estimate) and not even maximally-flat (although nearly). If you read a book on filter synthesis you will find that a high order filter can be made by cascading appropriate lower order filters. To get near maximally flat some of these will have highish Q and some lowish Q. The highish Q ones will ring. This ringing may be be heard as ringing or smearing, even though there might be little or no actual peak in the overall response because of the flattening effect from the lowish Q ones (which won't ring).
I didn't spot that this was a high order filter at first. The poster did not draw attention to this. Apart from anti-aliasing etc., how many people include high order filters in their audio systems? An ordinary first order rolloff won't do this. Exactly what point was being made?
I didn't spot that this was a high order filter at first. The poster did not draw attention to this. Apart from anti-aliasing etc., how many people include high order filters in their audio systems? An ordinary first order rolloff won't do this. Exactly what point was being made?
I would love to see someone recreate the signal chain for some of the 1959-1962 recordings of, say, the Chicago Symphony and then subject them to the new metric just to prove how "bad" all those RCA Reiner recordings are supposed to sound.
All the above is fine and valid, only, it doesn't address what RNM wrote here repeatedly:
(Post #3091122)
(Post #3091476)
(Post #3091519)
Also, the question is, was it passive, or active filter. Active components are non-linear by definition.
http://downloads.hindawi.com/journals/apec/1988/070678.pdf
Yes, talk about music and good recordings, after all Hi - Fi starts there.
A couple points -- it isnt an 8 pole... much less. --- the perception of ANY filter with a cutoff better than 6dB/octave will exhibit a longer decay time near the cutoff (amoungst other artifacts). The total energy is another aspect of what is heard. That affect on hearing is what many people describe and not ringing which is a different description.
If memory serves me -- ringing would show up as a ripple in the decay time data.
Thx,
Richard
It was not about filtering of 100 Hz pulsations by people. It was about dynamic resistance of power supply to power 2x100W class AB amplifier that had to show high-end properties of sound reproduction. Class AB means current consumption highly modulated by envelope that causes dynamic distortions.
Last edited:
- Status
- Not open for further replies.