Hi Cliff,
I am referring to this section of the PCB, so perhaps it's not the output section. Power supply instead?
In any case, please find attached relevant schematic and circuit diagram.
In terms of lower capacitance, I've already used 3.3uf MKC caps in place of the 4 x 10uf electrolytics surrounding the op-amp to good effect.
View attachment Rotel RC971 Technical Manual.7-8.pdf
Thanks for your help,
Alex
I am referring to this section of the PCB, so perhaps it's not the output section. Power supply instead?
In any case, please find attached relevant schematic and circuit diagram.
In terms of lower capacitance, I've already used 3.3uf MKC caps in place of the 4 x 10uf electrolytics surrounding the op-amp to good effect.
View attachment Rotel RC971 Technical Manual.7-8.pdf
Thanks for your help,
Alex
That is NOT helpful information!
you have the schematic, the components have reference numbers (eg C324), please use that so that we do not have to guess.
Why did you change from 10uF to 3.3uF? The only possible effect is a slight reduction in bass response. Does that sound better to you?
As I suggested, this depends very much on the load circuit. What is it?
you have the schematic, the components have reference numbers (eg C324), please use that so that we do not have to guess.
Why did you change from 10uF to 3.3uF? The only possible effect is a slight reduction in bass response. Does that sound better to you?
As I suggested, this depends very much on the load circuit. What is it?
Hello,
the component numbers are shown in the .png file I attached earlier, but for reference they are C913, C914 and C915 specifically.
The imput impedance of the pre-amp is 24kOhms. Based on this, 3.3uf caps would yield a -3db rolloff at 2Hz, as opposed to 0.66Hz with the 10uf caps. I suppose anywhere inbetween would also be agreeable.
However, my query was more oriented towards the type of capacitor (polypropylene or polycarbonate over electrolytic) as opposed to the values.
Many thanks!
the component numbers are shown in the .png file I attached earlier, but for reference they are C913, C914 and C915 specifically.
The imput impedance of the pre-amp is 24kOhms. Based on this, 3.3uf caps would yield a -3db rolloff at 2Hz, as opposed to 0.66Hz with the 10uf caps. I suppose anywhere inbetween would also be agreeable.
However, my query was more oriented towards the type of capacitor (polypropylene or polycarbonate over electrolytic) as opposed to the values.
Many thanks!
I use a 2uF polypropylene cap (any brand), and call it good. That gives you -3dB at 8 HZ when driving a 10K load Z (the power amp). I don't know of any poweramps that have an input Z lower than 10K, so that should be fine. The trouble with electrolytics is that they need to be biased with DC or they generate distortion, unless they are a non-polarized electrolytic.
For one thing, you seem to be looking at the wrong parts. In the schematic, I'm seeing 10µ/50V at the line-out (C523/524) and 4µ7/25V going to the headphone amp circuit (C601/602).
At the output I wouldn't be going under about 1µ5...2µ2 worth of coupling cap when expecting a ~10k minimum load. I can only guess how high exactly the supply voltages are, but I'd assume the parts should be rated at 16 V for failure-proofing. If film caps like that seem too clunky (not like it would detract many modders), quality low-leakage electrolytics of about 10µ/63V would also be suitable (BCComponents 013 RLC, Elna RLB and similar). Ideally it'd be a bipolar, but I'm not sure whether you can get low-leakage bipolars. If in doubt you can always go for two polars of double capacitance back-to-back.
That headphone amp circuit is a bit kinky. There's a 47k to 100k||47p voltage divider at the input, which seems a tad high in impedance to me, even with the well-behaved 4556A. Maybe that was lab-optimized? I'd rather suggest R601/602 = 4k7, R603/604 = 4k7, C603/604 = 680p. R609/610 could probably be reduced a bit without troubling the chip too much, too - like 47-75 ohms. And whether you really need 20 dB of extra gain after the line-out (minus about 3 dB in the input circuitry, or 6 dB modified) for modern headphones is quite dubious, too - I'd rather suggest R605/606 = 3k3. Total gain stock would have been about 17 + 17 = 34 dB, arguably appropriate for insensitive 600 ohm headphones and 1980s input levels, but usually way too much (and correspondingly noisy) these days.
At the output I wouldn't be going under about 1µ5...2µ2 worth of coupling cap when expecting a ~10k minimum load. I can only guess how high exactly the supply voltages are, but I'd assume the parts should be rated at 16 V for failure-proofing. If film caps like that seem too clunky (not like it would detract many modders), quality low-leakage electrolytics of about 10µ/63V would also be suitable (BCComponents 013 RLC, Elna RLB and similar). Ideally it'd be a bipolar, but I'm not sure whether you can get low-leakage bipolars. If in doubt you can always go for two polars of double capacitance back-to-back.
That headphone amp circuit is a bit kinky. There's a 47k to 100k||47p voltage divider at the input, which seems a tad high in impedance to me, even with the well-behaved 4556A. Maybe that was lab-optimized? I'd rather suggest R601/602 = 4k7, R603/604 = 4k7, C603/604 = 680p. R609/610 could probably be reduced a bit without troubling the chip too much, too - like 47-75 ohms. And whether you really need 20 dB of extra gain after the line-out (minus about 3 dB in the input circuitry, or 6 dB modified) for modern headphones is quite dubious, too - I'd rather suggest R605/606 = 3k3. Total gain stock would have been about 17 + 17 = 34 dB, arguably appropriate for insensitive 600 ohm headphones and 1980s input levels, but usually way too much (and correspondingly noisy) these days.
Thanks for that, much appreciated.
I hadn't considered those at C523/524 seen as it's the tone control section, which I never activate anyway. Does that mean it will be an effective part of the signal path regardless (and therefore open for modding even if tone is not used)?
Cheers
This "bias" recommendation has been superceded by one of our "Capacitor Testers".
The report stated quite clearly (and showed the plots) that the biased electrolytic did not perform as well as the unbiased electrolytic.
Don't forget that with electrolytics it may not be sufficient to consider what the bias is now, but also what it has been over the previous few years. The oxide layer acts something like a combination of a dielectric and a somewhat rough diode, but the thickness/quality of the oxide layer depends on bias history.
As a general rule, a symmetric capacitor with a nonlinear dielectric will generate some 3rd order when unbiased, but more 2nd order when biased*. An asymmetric capacitor can already generate 2nd unbiased, and bias can make this better or worse.
* The reason for this is that a symmetric system can only generate odd order distortion, which will normally be dominated by 3rd and so be proportional to Vsig^3. Add a bias and you can get 2nd which is proportional to Vbias x Vsig^2 - assuming Vbias is greater than Vsig means that this 2nd will be greater than the 3rd.
As a general rule, a symmetric capacitor with a nonlinear dielectric will generate some 3rd order when unbiased, but more 2nd order when biased*. An asymmetric capacitor can already generate 2nd unbiased, and bias can make this better or worse.
* The reason for this is that a symmetric system can only generate odd order distortion, which will normally be dominated by 3rd and so be proportional to Vsig^3. Add a bias and you can get 2nd which is proportional to Vbias x Vsig^2 - assuming Vbias is greater than Vsig means that this 2nd will be greater than the 3rd.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.