Well that's what I assumed.
The OP linked caps rated at 1000 volts, so I was a little curious why, that's all.
jeff
That was a safe bet. It is solid state. I don't need 1000 volts... its all they sell.
For higher DC voltage rated capacitors the thickness of the dielectric is greater, hence there is also a consequential greater surface area required for a specific capacitance, hence more dielectric material in contact with the conducting surfaces. My concern is that higher DC voltage rated capacitors can exhibit more dielectric effects as could result in positive sonic attributes to mitigate some undesirable quality, perhaps edginess or aggression.
For higher DC voltage rated capacitors the thickness of the dielectric is greater, hence there is also a consequential greater surface area required for a specific capacitance, hence more dielectric material in contact with the conducting surfaces. My concern is that higher DC voltage rated capacitors can exhibit more dielectric effects as could result in positive sonic attributes to mitigate some undesirable quality, perhaps edginess or aggression.
Or, more likely, is just being used in a badly designed circuit. A good design is usually quite forgiving of minor component variations. Major ones, as you pointed out, are a different story.
If we talk about Baxandall's superb RIAA design, where the lower time constants are in the feedback loop, and the 75us one is passive on the output, his neatly calculated values are 68k, 6.8k, and 750 ohms. The capacitor values are 47n and 100n. In other words absolutely standard values.
If those are bang on, the response (from a spice sym) is +/-0.01dB.
Of course practical components have a tolerance. Setting the resistors to be +/-1% (ie E96 values) and the capacitors by +/-2%, the result is a maximum deviation at 500Hz of +/-0.15dB at 150Hz, and at 10kHz somewhat more at +/-0.3dB.
Incidentally, you can only trim values if you have a method of measuring them accurately - and I don't mean a cheapo DVM - I mean a high grade audio component bridge. Which most people do not have (I do, I actually have two, and I write the value to at least 3 places of decimals, or more, on the capacitors).
If those are bang on, the response (from a spice sym) is +/-0.01dB.
Of course practical components have a tolerance. Setting the resistors to be +/-1% (ie E96 values) and the capacitors by +/-2%, the result is a maximum deviation at 500Hz of +/-0.15dB at 150Hz, and at 10kHz somewhat more at +/-0.3dB.
Incidentally, you can only trim values if you have a method of measuring them accurately - and I don't mean a cheapo DVM - I mean a high grade audio component bridge. Which most people do not have (I do, I actually have two, and I write the value to at least 3 places of decimals, or more, on the capacitors).
Sure. Getting the 1% types I can also check the accuracy of my capacitance meter as well. 100nF appears a good value to begin designing around. I could check out the COG/NPO types as well (although these are 5%)
https://www.digikey.ca/en/products/...imIlKH43dRJIcgBWZnFJEH9wpSIRPSR0VnQkWyI+bGigA
The plan is to implement some low noise regulators on the supplies in lieu of some zener shunt regulators being used at the moment. The following devices from aliexpress appear suitable, though appear to require a change in a surface mount resistor to obtain a higher set voltage other than 5 volts. Doesn't appear much of a problem.
https://www.aliexpress.com/item/100...der_detail.order_detail_item.4.1245f19cUTnOIV
https://www.aliexpress.com/item/100...der_detail.order_detail_item.4.1245f19cUTnOIV
While 1% (or 0.1% for the resistors) is nice and obtainable, I find that for good spatial resolution and imaging, it really pays to match the resistors and capacitors as closely as possible. That said, not all RIAA components have the same influence - I found it quite interesting to play with the values in simulation and understand the sensitivity of the response for each variation. IMHO, for tonality and midrange, it is the bigger cap, and for imaging, the smaller cap.....
Please do not consider ceramic caps in the RIAA, I tried and the sound was horrible..... Even though measurements and distortion values would suggest otherwise, it seems there are things going on at -40dB...-70dB that we can hear well.....
I suggest that you read the late great Cyril Bateman's Capacitor Sound articles, in which he measures capacitor distortion .
https://linearaudio.net/cyril-batemans-capacitor-sound-articles
Of particular interest to the above post:
"With no bias this tiny
COG 10 nF 50 volt
multilayer capacitor
measured just 0.00006%.
Second harmonic was
-128.5 dB, the other
levels remained as shown."
Perhaps your choice was for some reason an outlier. Or perhaps you used X7R, which indeed produce horrendous distortion.
https://linearaudio.net/cyril-batemans-capacitor-sound-articles
Of particular interest to the above post:
"With no bias this tiny
COG 10 nF 50 volt
multilayer capacitor
measured just 0.00006%.
Second harmonic was
-128.5 dB, the other
levels remained as shown."
Perhaps your choice was for some reason an outlier. Or perhaps you used X7R, which indeed produce horrendous distortion.
I have already tried a 0.1uF cap of the non-COG variety in the 50/500 slot and it didn't sound terrible. I don't trust it though... hence the reason for questioning the use of the COG types of ceramics and the cost savings from teflon. I don't mind paying the price for the teflon's if that is one of the only high cost items in the network. I don't currently require 4 units since the 2122Hz pole uses <1000pF or thereabouts.
Using devices that require creating poles in the middle of the audio band seem most problematic and can't be mitigated out with long time constants to minimize their effects. They need to react at specific frequencies. IMO there is still a disparity between sonics and measurements hence I don't really trust anything (not even myself... thats why I am into therapy).
Using devices that require creating poles in the middle of the audio band seem most problematic and can't be mitigated out with long time constants to minimize their effects. They need to react at specific frequencies. IMO there is still a disparity between sonics and measurements hence I don't really trust anything (not even myself... thats why I am into therapy).
Last edited: