No, not some discussion of the sonic qualities of different materials 😉
The electrolytic usually used to ensure unity gain at DC, why do we hardly ever see designs with back-to-back electrolytics or bipolar electrolytics in this position? If there is a fault with the amp causing the output to swing to the 'wrong' rail, it would blow up the cap, wouldn't it?
Also, a polarised component can be slightly reverse biased under normal operating conditions, depending on component tolerances causing slight DC offset at the output. Isn't this bad (the reverse bias I mean)?
The electrolytic usually used to ensure unity gain at DC, why do we hardly ever see designs with back-to-back electrolytics or bipolar electrolytics in this position? If there is a fault with the amp causing the output to swing to the 'wrong' rail, it would blow up the cap, wouldn't it?
Also, a polarised component can be slightly reverse biased under normal operating conditions, depending on component tolerances causing slight DC offset at the output. Isn't this bad (the reverse bias I mean)?
"Usually, a diode across the cap solves the 'polarisation' problem."
A diode will cause more problems than it solves.
"Also, a polarised component can be slightly reverse biased under normal operating conditions, depending on component tolerances causing slight DC offset at the output. Isn't this bad (the reverse bias I mean)?"
I have, for years, asked why electrolyic capacitors are being used to couple circuits when there is no d.c. polarizing voltage on either end of the capacitor.
Thanks for bring this up.
A diode will cause more problems than it solves.
"Also, a polarised component can be slightly reverse biased under normal operating conditions, depending on component tolerances causing slight DC offset at the output. Isn't this bad (the reverse bias I mean)?"
I have, for years, asked why electrolyic capacitors are being used to couple circuits when there is no d.c. polarizing voltage on either end of the capacitor.
Thanks for bring this up.
If you are using a bipolar differential the base bias current is usually sufficient in conjunction with the feedback resistor to adequately bias a dc-blocking electrolytic. When using FET differential or dual (complementary) differentials this won't work.
Obviously it is cheaper to use a polarised electrolytic if it is correctly biased. Even better not to use a dc blocker at all if your circuit tolerances are good enough to give an acceptable and acceptably stable dc offset at the output.
Obviously it is cheaper to use a polarised electrolytic if it is correctly biased. Even better not to use a dc blocker at all if your circuit tolerances are good enough to give an acceptable and acceptably stable dc offset at the output.
Frank Berry said:
I played with polarizing caps for a long time, with no results.
You would imagine that they might be better with some
voltage on them (just like some people think of wires) but
they didn't measure different, and subjective commentary
was pretty random.
John Curl has played with this I believe, and maybe he will
offer an insight here.
Leach has used back-to-back electrolytics or bipolars in that application for years. Even better would be back-to-back bipolars, according the the article "The sound of capacitors" in Electronics World.
Say Pooge, I see that you are a patent examiner. Any
insights as to what's going on these days with the
patent office?
insights as to what's going on these days with the
patent office?
We usually use a polarized electrolytic cap as a 'feedback cap' "because we can!". ;-) It is not a perfect solution, BUT it is a cheap solution. Usually, the voltage across the cap is just a few tens of millivolts, SO the cap can handle it. The reason for this is that an ALUMINUM electrolytic cap is actually 2 caps in series: one with the nominal voltage breakdown and nominal capacitance, and the other with 10-100 times more capacitance, but with perhaps only 1.5V breakdown. No matter what the DC offset is, + or - the cap can handle it. Is this good? No, better to servo, or direct couple. Tantalum caps are a separate problem, and we usually avoid them in this situation, today
Caps will tolerate some small amount of overvoltage and reverse voltage with small leakage currents resulting. see the manufacturers' data sheets. That is why the caps seem to survive OK in this position, (until an output blows, invariably shorted).
A DC blocking cap in on the feedback network will in fact explode if one output device is shorted, the typical failure mode. The caps are typically large value, low voltage devices so can't handle much voltage. I have repaired numerous amps with that "collateral damage" after an output device shorted. The caps look like little firecrackers that shred and make a mess.
It's usually not good for the input pair either, so you have to replace them. That's actually more of a pain in the rear, since sometimes they are rare dual transistors and you have to really scrounge for the replacement.
Many high end amps that I have looked at do use a bipolar cap here or diodes across the unipolar one (and typically parallel that with a quality film cap for low impedance at high frequencies). Incidentally, I salvage them from obsolete PC power supplies or motherboards, since the switching regulators on these seem to need oodles of small low ESR 2200 MF/10v caps (just call me cheap).
A double diode in each direction or back to back zeners in parallel with the cap will avoid the problem. I'm concerned that this may have an audible effect as the diodes switch in, but have not really noticed it.
Caps will tolerate some small amount of reverse voltage with small leakage currents resulting. see the manufacturers' data sheets. That is why the caps seem to survive OK in this position, (until an output blows, invariably shorted, then fireworks).
A DC blocking cap in on the feedback network will in fact explode if one output device is shorted, the typical failure mode. The caps are typically large value, low voltage devices so can't handle much voltage. I have repaired numerous amps with that "collateral damage" after an output device shorted. The caps look like little firecrackers that shred and make a mess.
It's usually not good for the input pair either, so you have to replace them. That's actually more of a pain in the rear, since sometimes they are rare dual transistors and you have to really scrounge for the replacement.
Many high end amps that I have looked at do use a bipolar cap here or diodes across the unipolar one (and typically parallel that with a quality film cap for low impedance at high frequencies). Incidentally, I salvage them from obsolete PC power supplies or motherboards, since the switching regulators on these seem to need oodles of small low ESR 2200 MF/10v caps (just call me cheap).
A double diode in each direction or back to back zeners in parallel with the cap will avoid the problem. I'm concerned that this may have an audible effect as the diodes switch in, but have not really noticed it.
Caps will tolerate some small amount of reverse voltage with small leakage currents resulting. see the manufacturers' data sheets. That is why the caps seem to survive OK in this position, (until an output blows, invariably shorted, then fireworks).
Talking about caps in the feedback. Why can´t I find 220uF MKP caps at low voltage say 50 to 100 Volts for the feedback. Mr. Pass what do you use here?
It´s almost like no one produces them but everybody is searching for them.
It´s almost like no one produces them but everybody is searching for them.
Frank Berry wrote:
Frank, what problems are these? This approach appeals, what are the issues?
Cheers,
Hugh
Frank, what problems are these? This approach appeals, what are the issues?
Cheers,
Hugh
promitheus said:
It's economics. Electrolytics are vastly cheaper in that capacitance range. Bipolar electrolytics in that range are used in speaker crossovers -- that is probably the largest market for bipolar caps in that value and voltage range. Film caps are used for smaller values on the tweeter in better crossovers only. Also, the physical size would not be practical, just look at a 1 MF metalized polypropylene cap and scale up.
Maybe these have been custom made for military or aerospace, but I have never seen one.
Promitheus:
The Rubycon SWR is a radial-lead polyester cap available in values of up to 100uF (107) at 35V. It's chunky, but not ridiculously large. The ones that I have here measure 24mm wide x 23mm deep x 37mm tall.
Link to pdf in Japanese:
http://www.rubycon.co.jp/catalog/j_pdfs/film/j_SWR.pdf
Note that the price is considerably more expensive than any comparable electrolytic cap that I am aware of.
hth, jonathan carr
The Rubycon SWR is a radial-lead polyester cap available in values of up to 100uF (107) at 35V. It's chunky, but not ridiculously large. The ones that I have here measure 24mm wide x 23mm deep x 37mm tall.
Link to pdf in Japanese:
http://www.rubycon.co.jp/catalog/j_pdfs/film/j_SWR.pdf
Note that the price is considerably more expensive than any comparable electrolytic cap that I am aware of.
hth, jonathan carr
Nelson Pass said:
I think it's what some cap manufacturer(s) (eg WIMA) have as a name for metallised polypropylene series. So in Europe it's accustomed to call those MKP rather than metallised polypropylene (same as ELKO for ELectrolytic KOndensatoren).
eg: http://www.wima.com/navig/mkp4.htm
promitheus said:
Yep, that's why I put 'manufacturer(s)' in my post. Anyway, in my neck of the woods I have to ask for 'square box caps' to get some of these 🙂
Regarding the feedback cap blowing up, this would be my last concern if the output stage goes to fireworks. And if it blows up (open) it's a really good thing for the diff pair stage.
Any SQ/listening tests with a straight piece of wire in the place of the FB cap? I'd rather have no input cap but unity gain at DC if something goes waaay wrong - a low ESR 220uF-470uF/35V with film bypass in place in the FB loop... (sorry for putting this from the tinkerer point of view...)
Mr.John Curl said that it is better to use servo than feedback caps to get right dc offset. I've built one, with opamp, somehow the sound is less natural (compared without servo). The highs lacks a little, and the bass too. Maybe my servo is not good. Really wants to see what Mr. John Curl uses.
With other design, I have weird problem too. Having put caps in the FB, the DC still high. It comes out that the caps have to be both in the input and FB. What is really wrong so the caps have to be in 2 places, input and FB?
- Status
- Not open for further replies.