Hi all,
I built a little single supply 5 Watt amp (8 ohm) with the intention of boosting the power rails and using BTL to boost the power to ~40Watts. However the 5 Watts is plenty loud enough for where it's going to be used so I'm going to leave it. Presently I'm using Rubycon 63PK3300MEFC18X40 63V 3.3mF output capacitors and the results sound good with thunderous bass.
According to the internet, output caps should cause distortion but the number of data points is limited. Ie from Douglas Self's site a 6.8mf causes lots of distortion when driving at 40 Watts but a 100mF is OK. Does anybody have any info on what the happens when driving at low power? Does having a 63V capacitor with a 24V power supply improve/make worse the distortion levels?
Thanks...
I built a little single supply 5 Watt amp (8 ohm) with the intention of boosting the power rails and using BTL to boost the power to ~40Watts. However the 5 Watts is plenty loud enough for where it's going to be used so I'm going to leave it. Presently I'm using Rubycon 63PK3300MEFC18X40 63V 3.3mF output capacitors and the results sound good with thunderous bass.
According to the internet, output caps should cause distortion but the number of data points is limited. Ie from Douglas Self's site a 6.8mf causes lots of distortion when driving at 40 Watts but a 100mF is OK. Does anybody have any info on what the happens when driving at low power? Does having a 63V capacitor with a 24V power supply improve/make worse the distortion levels?
Thanks...
Merlin Blencow's testing in his book Designing High-Fidelity Tube Preamps, shows that higher voltage rating capacitors introduce less distortion. 3300 uF for 5 W into 8 Ohm is more than ample, and it's pole is at 6 Hz. You should have the same value cap. in the power supply.
Total malarky. Single supply amps with an output cap of proper size induce distortion in the 3rd decimal place. 6.8 uf is too small for 8 ohm speakers.
The output cap can be included in the feedback loop. See Apex AX6 . https://www.diyaudio.com/forums/solid-state/236256-retro-amp-50w-single-supply-42.html
Thanks for the replies. Fortunately the sample pages of "Designing High-Fidelity Tube Preamps" give the plot showing reduced distortion with increased capacitor max voltage, so overrated caps for such a low power amp is the way to go. The Apex amp looks interesting, forgive me I haven't had chance to look through all 45 pages of discussion on it. I have come across AC coupled feedback before, after looking at the Apex I'm tempted to give it a try.
Do you have a link with measurements of this? Douglas Self's site shows a plot of 0.01% distortion for 6.8mf at 10 Hz, and no visible distortion for 100mf in one of his books. 2 measurements of unknown caps of course doesn't give enough info the expected distortion from another cap in a different situation.
Some people think "6.8mf" is "six point eight millifarads" which equals 6.8E-3 farads. Other people think "6.8mf" is "six point eight microfarads" which equals 6.8E-6 farads.
Here's what I think.
I think what I want, what makes me super comfortable, is a low frequency rollover corner frequency of 1.0 Hertz (6.283 radians/sec) when the output load is 8 ohms. In that case:
(8 ohms) x (Capacitance_in_Farads) x 6.283 = 1
A few key presses reveal: Capacitance_in_Farads = 1.99E-2 farads. That's 19.9 millifarads or 19,900 microfarads.
I'm sure you'll want to calculate the impedance of a 19.9 millifarad capacitor at 20 Hertz. I think you'll discover that Z=0.4 ohms at 20 Hertz, for a 19,900 microfarad capacitor. Like other capacitors, its impedance falls as frequency rises, so the worst case maximum impedance occurs at the worst case minimum frequency: 20 Hertz.
Here's what I think.
I think what I want, what makes me super comfortable, is a low frequency rollover corner frequency of 1.0 Hertz (6.283 radians/sec) when the output load is 8 ohms. In that case:
(8 ohms) x (Capacitance_in_Farads) x 6.283 = 1
A few key presses reveal: Capacitance_in_Farads = 1.99E-2 farads. That's 19.9 millifarads or 19,900 microfarads.
I'm sure you'll want to calculate the impedance of a 19.9 millifarad capacitor at 20 Hertz. I think you'll discover that Z=0.4 ohms at 20 Hertz, for a 19,900 microfarad capacitor. Like other capacitors, its impedance falls as frequency rises, so the worst case maximum impedance occurs at the worst case minimum frequency: 20 Hertz.
Me thinks: What if speaker impedance isn't purely rezistive (XCcout and XLspeaker tend to "cancel" each other)? Does anybody remembers the KEF C75 crossover?
I suppose one could always (eventually) build something that does not have AC (sound) coupling capacitors in the direct signal path; that should include the feedback loop(s).
You'll have to forgive the old timers who assume mF means micro-Farad. There was a time when 1000uF (1mF) was about as big as capacitors came. But this is 2024 and 1 Farad capacitors are commonly used by automotive subwoofer amps. If you tell SPICE a cap is 1mF, it assumes you mean 1000uF. After all, milli- Henrys, milli-Watts and milli-Ohms are not unusual values, even milli-Hertz.
When I was a kid, 1000uF output capacitors were very common. After all, that's 20Hz for an 8 Ohm speaker and nothing but special purpose sub woofers actually handle 20Hz. Cheap little 5W amps often used 470uF output capacitors. Most of us do not listen to a $10K sound system. So, your 3.3mF (3300uF) output cap is just fine. It will protect your inexpensive but nevertheless, precious, 10W speakers from burning up when a DC failure happens.
When I was a kid, 1000uF output capacitors were very common. After all, that's 20Hz for an 8 Ohm speaker and nothing but special purpose sub woofers actually handle 20Hz. Cheap little 5W amps often used 470uF output capacitors. Most of us do not listen to a $10K sound system. So, your 3.3mF (3300uF) output cap is just fine. It will protect your inexpensive but nevertheless, precious, 10W speakers from burning up when a DC failure happens.
One possible caveat with too much capacitance is that DC leakage could be high. Check the datasheets.
You may find that electrolytic caps prefer to run close to their rated voltage, or else they could drive excessive DC through the speaker, which kind-of defeats their purpose. And in this application, the rated voltage needs to withstand twice the idle voltage. Maybe 16V rated caps running at 12V with occasional peaks up to 20V is OK, I don't know.
You may find that electrolytic caps prefer to run close to their rated voltage, or else they could drive excessive DC through the speaker, which kind-of defeats their purpose. And in this application, the rated voltage needs to withstand twice the idle voltage. Maybe 16V rated caps running at 12V with occasional peaks up to 20V is OK, I don't know.
The only reason that capacitors are run near their rated voltage is money. "Failure rate is proportional to the cube of the ratio of the applied to rated voltage.", ie at the rated voltage (or above), failure is guaranteed. When I was working, our MTBF calculations flagged any capacitor running anywhere near the rated voltage. The most common failure of electrolytic caps is a low voltage decoupling cap in the negative feedback that is potentially exposed to the rail voltage. The fact that the normal DC on this cap is negligible is irrelevant.
Failure is guaranteed at every voltage, even 0V, its a question of lifetime... Using better temperature rated caps and/or running them cooler greatly extends lifetime, as does keeping ripple current down, as well as conservatively rating voltage - its quite possible to become limited by shelf life with careful choices, even running at full rated voltage, but often those choices include forced air cooling, often precluded in audio equipment.
In his power amplifier book, Douglas Self pats himself on the back quite vigorously for having the insight to add a quartet of cheap 1N4148 diodes, that prevent this very failure.
But doesn't that kind-of defeat the purpose of modulating the cut-off frequency depending on the chosen polarity?
I think I read that you are interested in good sound. The usual distortion measurement values, or dielectric strength and capacitance, have no indication of how distorted the capacitor sounds, for example. The only thing to do is to buy many different ones and listen to them. They sound as different as beers and wines taste. I would start from the minimum value for dielectric strength or capacitance;-)