I agree with you.
Though, i want to point out that a cap is a cap and it has not several ways of working.
My drawing shows the example of the single supply amp.
If you "supply" the output cap externally at Vcc/2, it will behave exactely as the PSU one. DC will pass again. The signal current flows through it the same way it flows through the psu cap.
In the real life single supply amp the output cap is not "auxiliary supplied" though it has a reserve of Vcc/2 across it, which is not too bad.
This to say there ise only one way for a signal to flow through a cap, psu or output one.
Apart from the low freq roll off, the output cap should not affect the sound more than the supply ones.
Though, i want to point out that a cap is a cap and it has not several ways of working.
My drawing shows the example of the single supply amp.
If you "supply" the output cap externally at Vcc/2, it will behave exactely as the PSU one. DC will pass again. The signal current flows through it the same way it flows through the psu cap.
In the real life single supply amp the output cap is not "auxiliary supplied" though it has a reserve of Vcc/2 across it, which is not too bad.
This to say there ise only one way for a signal to flow through a cap, psu or output one.
Apart from the low freq roll off, the output cap should not affect the sound more than the supply ones.
Attachments
Last edited:
Bobo,
Why don't you build what you have drawn and do the measurements and tests
For a theoretical cap I agree, real ones have properties other than ideal though, you have only to work with HF circuits and pulse and HF power circuitry to run up against this. For the purposes of this though your statement is essentially correct.
Your single supply amp is in green, it's PSU in black.
I don't understand what you are showing here... the extra PSU in green connected in parallel across the output cap.
A single rail amp already has the output cap at Vcc/2 under quiescent conditions. You mention that later
"DC will pass again"... sorry, I don't understand you... DC will flow where ?
Why don't you build what you have drawn and do the measurements and tests
For a theoretical cap I agree, real ones have properties other than ideal though, you have only to work with HF circuits and pulse and HF power circuitry to run up against this. For the purposes of this though your statement is essentially correct.
Your single supply amp is in green, it's PSU in black.
I don't understand what you are showing here... the extra PSU in green connected in parallel across the output cap.
A single rail amp already has the output cap at Vcc/2 under quiescent conditions. You mention that later
"DC will pass again"... sorry, I don't understand you... DC will flow where ?
This Vcc/2 is not maintained all the signal cycle long.
During the negative half wave, it is disconnected from the supply.
It is left free wheel.
I see nothing wrong to auxiliary supply it. (or to use a battery instead of it)
In my diagram, one can see that the output cap, when supplied, processes the signal current flow the same way the psu cap does.
Into the loop: psu cap, output transistor, supplied output cap, load, one unique current flows. So, the two devices, psu and supplied output cap being identical have to behave the same way with the signal current flow.
The usual output cap is not kept filled, that is the only difference with the psu one.
Unfortunately, there is not anymore scope in my house.
I mean the amp passes dc.
Last edited:
Bobo,
I'm sorry I don't follow what you are getting at.
Split supply,
PSU caps are not influencing the output signal... that's proved by the 'scope shots, they are outside the sphere of influence of the amp and it's feedback network, or in the case of my example of a buffer, outside the influence of a simple follower. The ripple current in the cap is a mixture of signal, ripple, mains fluctuations etc etc.
Single rail AC coupled amp,
The output is totally dependant on the output cap as all the output current flows through it. This assumes the cap is not in the main feedback loop. You can incorporate an output cap within global feedback, but that adds extra complications as you then need both a DC and AC feedback loop so as to maintain the DC conditions within the amp.
As in the split supply amp the ripple current is a mixture of all the above mentioned.
I think we have to call it a day at that
I'm sorry I don't follow what you are getting at.
Split supply,
PSU caps are not influencing the output signal... that's proved by the 'scope shots, they are outside the sphere of influence of the amp and it's feedback network, or in the case of my example of a buffer, outside the influence of a simple follower. The ripple current in the cap is a mixture of signal, ripple, mains fluctuations etc etc.
Single rail AC coupled amp,
The output is totally dependant on the output cap as all the output current flows through it. This assumes the cap is not in the main feedback loop. You can incorporate an output cap within global feedback, but that adds extra complications as you then need both a DC and AC feedback loop so as to maintain the DC conditions within the amp.
As in the split supply amp the ripple current is a mixture of all the above mentioned.
I think we have to call it a day at that
Hi Gordy,
I don't quite see that you ever found out what sort of cap would be best in the output but if you're looking for bipolar at a value large enough to admit some bass frequencies, a min. of 2,200 uF was usual for respectable amps. These days, some would insist on double or as much as you could afford. I wouldn't speculate on that.
It's probably obvious that you won't find readily available bipolars at such high values. The quickest and cheapest minimum would be a back-to-back pair of inexpensive 4,700 uF rated at about 35 V or so. That should accommodate most medium power balanced rail amps for experiments unless you really want high SPL too. If you want to get the most from caps in this role, I understand from this and other other discussions that high ripple rated is best.... Is it?.. Why?..anyone....?
I hope you have some fun trying this.
Last edited:
Ian, I've a little source follower project on-going so have decided to wait until that comes together. As it will have a DC offset I am planning to use polarised 2200 uF caps. First just one, then two in parallel to see if there is a measurable difference in distortion and then to see if I can perceive a subjective difference. I am not sure which brand to buy.
WRT ripple current I assume that the higher the rating the more temperature stable the cap will be when passing proportionally lower current, hence more consistent in performance. Just an assumption, not a fact.
WRT ripple current I assume that the higher the rating the more temperature stable the cap will be when passing proportionally lower current, hence more consistent in performance. Just an assumption, not a fact.
series capacitor
Ian Finch, the most famous single supply amp with output caps establishing the switch point for the two quasi-comp output transistors was the dynakit ST120. That came with a 3300 uf 60 WVDC cap on each channel, which was rate a wimpy 60 watts. It was an enormous cap, probably low ESR. Mine never failed, but I replaced them on general principals that old electrolytics leak or explode after 25 years. Modern 3300 uf 80 V caps have an ESR of about .012 and cost about $3. For $18, you can have a 10000 uf 80v non polar electrolytic cap suitable for connecting to a center tap transformer amp with the speaker return connected to the center tap. Film caps would have less interesting chemical events crossing over zero volts, but the most economic enormous ones are about 50 uf at 250 VAC at $11 each. At 3300 uf, my dynakit amps enormous sub sonic thuds caused by my walking across the floor away from the turntable, so I consider 3300 uf to be entirely adequate for a 60 w per channel amp into 8 ohm speakers. Am ordering some non-polar caps to put in series with my CS800S output, for another AB test. I might cobble up a 555 circuit to show a square wave response on my scope, since so many people on this board post that as a amp test.
Ian Finch, the most famous single supply amp with output caps establishing the switch point for the two quasi-comp output transistors was the dynakit ST120. That came with a 3300 uf 60 WVDC cap on each channel, which was rate a wimpy 60 watts. It was an enormous cap, probably low ESR. Mine never failed, but I replaced them on general principals that old electrolytics leak or explode after 25 years. Modern 3300 uf 80 V caps have an ESR of about .012 and cost about $3. For $18, you can have a 10000 uf 80v non polar electrolytic cap suitable for connecting to a center tap transformer amp with the speaker return connected to the center tap. Film caps would have less interesting chemical events crossing over zero volts, but the most economic enormous ones are about 50 uf at 250 VAC at $11 each. At 3300 uf, my dynakit amps enormous sub sonic thuds caused by my walking across the floor away from the turntable, so I consider 3300 uf to be entirely adequate for a 60 w per channel amp into 8 ohm speakers. Am ordering some non-polar caps to put in series with my CS800S output, for another AB test. I might cobble up a 555 circuit to show a square wave response on my scope, since so many people on this board post that as a amp test.
Last edited:
Hi,Indianajo
I recall drooling over Dynakit (and Heath) amps when quite young. They always appealed but I never heard one until about 10 years ago. Someone had done much as you had in replacing caps but the output cap was nothing special. Still, it certainly wasn't "bass shy".
I'm surprised that amongst all your equipment you don't have an old function generator or even the scope's calibrator to simplify things.
Anyways, it is very interesting to hear your thoughts on suitable cap types. I had thought about mains motor start and phase correction caps before but the sheer bulk and often high cost is not encouraging. I admit, I haven't come across values like 10,000 uF 80 V bipolar in popular local distributor's cats. The usual bipolar reference is to cheap loudspeaker crossover caps which only run to 1000 uF.
Low ESR, you think? Well that's reasonable. I guess your'e considering HF performance and that should also show up in power bandwidth testing. I'm planning on making up better dummy loads soon and might get a chance to look at it unless Gordy checks it out when he's comparing cap values.
Thanks for taking on a bit of groundwork. We could all benefit. Still wondering about the real merits of all those "wonder" caps and brands that go for absurdium prices.
I recall drooling over Dynakit (and Heath) amps when quite young. They always appealed but I never heard one until about 10 years ago. Someone had done much as you had in replacing caps but the output cap was nothing special. Still, it certainly wasn't "bass shy".
I'm surprised that amongst all your equipment you don't have an old function generator or even the scope's calibrator to simplify things.
Anyways, it is very interesting to hear your thoughts on suitable cap types. I had thought about mains motor start and phase correction caps before but the sheer bulk and often high cost is not encouraging. I admit, I haven't come across values like 10,000 uF 80 V bipolar in popular local distributor's cats. The usual bipolar reference is to cheap loudspeaker crossover caps which only run to 1000 uF.
Low ESR, you think? Well that's reasonable. I guess your'e considering HF performance and that should also show up in power bandwidth testing. I'm planning on making up better dummy loads soon and might get a chance to look at it unless Gordy checks it out when he's comparing cap values.
Thanks for taking on a bit of groundwork. We could all benefit. Still wondering about the real merits of all those "wonder" caps and brands that go for absurdium prices.
non-polar caps
There will be a slight stocking delay to the output capacitor experiment on a split supply amp. 100v rated 3300 uf capacitors from Cornell Dublier, the SPLX series, are in fact poorly marked and incorrectly advertised 100VDC rated units. They do have a higher surge rating, I'm sure they would protest, but that rating does not include minus voltage. I realize I could take 4 of them and make one 3300 uf 100 VAC capacitor, but I only bought six, so an ear test of a stereo amp is not possible.
My B&K scope does not have a calibrator except I use frequencies of the Hammond organ to calibrate the time bse of the scope. So square wave tests will have to wait until I have a working square wave oscillator circuit.
Thank you for your patience. Considering the speed of chemical reactions, I'm not as confident of good results of an output capacitor on a split supply amp as seems to be the case on a single supply amp like the ST120.
There will be a slight stocking delay to the output capacitor experiment on a split supply amp. 100v rated 3300 uf capacitors from Cornell Dublier, the SPLX series, are in fact poorly marked and incorrectly advertised 100VDC rated units. They do have a higher surge rating, I'm sure they would protest, but that rating does not include minus voltage. I realize I could take 4 of them and make one 3300 uf 100 VAC capacitor, but I only bought six, so an ear test of a stereo amp is not possible.
My B&K scope does not have a calibrator except I use frequencies of the Hammond organ to calibrate the time bse of the scope. So square wave tests will have to wait until I have a working square wave oscillator circuit.
Thank you for your patience. Considering the speed of chemical reactions, I'm not as confident of good results of an output capacitor on a split supply amp as seems to be the case on a single supply amp like the ST120.
split supply output capacitor
Okay, I've done the experiment of running a split supply amp with an output capacitor in series with the speaker. The amp is a Peavey CS800S with +-79 V rails. The speakers are Peavey SP2 piezo horn+15" woofer packages. The capacitors were 2008 manufacture pairs of 3300 uf 100v Cornell Dublier electrolytics back to back on one side (SPLX series) on one side making 3300 uf bipolar total, and a 10000 uf 200v Mallory electrolytic dated 1978 backed up to a 9500 uf 50v 1983 manufacture electrolytic on the other side. Neither old electrolytic leaked more than 1 mv/3 sec after being charged up with the DVM ohms scacle to 1.8v. (Took forever, the capacitance is huge). All wiring was 12 ga SO cord (multistrand). On ZZ Top Afterburner with it's intense bass drum on "Woke up with ****", no audible difference. On Peter Nero Warm & Wonderful Album, "When I Fall in Love" and "Secret Love" cuts, the top octave of the piano is a bit fuzzier with the capacitors. I suppose I could bypass the electrolytics with a 5 uf 50v ceramic cap, but that is all for today. I'm going to assume that Mr. Peavey has done his homework on protecting these speakers with his triac clamp, current sensing transformers, heat sink thermistor, and microprocessor, and leave the amp direct coupled to the speakers.
Okay, I've done the experiment of running a split supply amp with an output capacitor in series with the speaker. The amp is a Peavey CS800S with +-79 V rails. The speakers are Peavey SP2 piezo horn+15" woofer packages. The capacitors were 2008 manufacture pairs of 3300 uf 100v Cornell Dublier electrolytics back to back on one side (SPLX series) on one side making 3300 uf bipolar total, and a 10000 uf 200v Mallory electrolytic dated 1978 backed up to a 9500 uf 50v 1983 manufacture electrolytic on the other side. Neither old electrolytic leaked more than 1 mv/3 sec after being charged up with the DVM ohms scacle to 1.8v. (Took forever, the capacitance is huge). All wiring was 12 ga SO cord (multistrand). On ZZ Top Afterburner with it's intense bass drum on "Woke up with ****", no audible difference. On Peter Nero Warm & Wonderful Album, "When I Fall in Love" and "Secret Love" cuts, the top octave of the piano is a bit fuzzier with the capacitors. I suppose I could bypass the electrolytics with a 5 uf 50v ceramic cap, but that is all for today. I'm going to assume that Mr. Peavey has done his homework on protecting these speakers with his triac clamp, current sensing transformers, heat sink thermistor, and microprocessor, and leave the amp direct coupled to the speakers.
ceramic caps
Well I hate to dispute the terse Mr. Curl. He was right about split supply output caps being undesirable (unless you have a cheapo split supply amp that might fry your speakers. The sound wasn't THAT bad and would be unnoticable on anything that didn't have mostly high frequency energy- like solo grand piano high octave) I just recapped my Hammond H182 organ and replaced all the 10 uf electrolytics on the preamps, input to transistor DC blockers, and hum filters to ground, with 10 uf 50v Aerovox ceramic caps. +80-20% +50 temp coeff. I love the sound. I also have replaced the 5uf tantalum input DC blocker cap on my Dynakit ST120 with a pair of 5 uf 50 v +-20% ceramic Aerovox cap also cheap temp coeff. I also love the sound of that amp after that, all new electrolytic caps, and the DJoffe mod. I suppose on output DC blocker for split supply power amp, the low inductance of the ceramic cap causes a risk of oscillation???? I'm sure the +80 -20% tolerance nor the temp coefficient are not the objection in parallel with a ~5000 uf +80-20% electrolytic on one side and a 3300 uf +80-20 electrolytic on the other side, which is not very precise.
Well I hate to dispute the terse Mr. Curl. He was right about split supply output caps being undesirable (unless you have a cheapo split supply amp that might fry your speakers. The sound wasn't THAT bad and would be unnoticable on anything that didn't have mostly high frequency energy- like solo grand piano high octave) I just recapped my Hammond H182 organ and replaced all the 10 uf electrolytics on the preamps, input to transistor DC blockers, and hum filters to ground, with 10 uf 50v Aerovox ceramic caps. +80-20% +50 temp coeff. I love the sound. I also have replaced the 5uf tantalum input DC blocker cap on my Dynakit ST120 with a pair of 5 uf 50 v +-20% ceramic Aerovox cap also cheap temp coeff. I also love the sound of that amp after that, all new electrolytic caps, and the DJoffe mod. I suppose on output DC blocker for split supply power amp, the low inductance of the ceramic cap causes a risk of oscillation???? I'm sure the +80 -20% tolerance nor the temp coefficient are not the objection in parallel with a ~5000 uf +80-20% electrolytic on one side and a 3300 uf +80-20 electrolytic on the other side, which is not very precise.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.