Member
Joined 2009
Paid Member
DF96,
Vr = Vmax / f RC
f = rectifier output frequency
With an increasing capacitance the diode must provide the charge in less time.
I = C * V / T (more correctly I = C * dV / dT)
I agree with your remarks, you have made correct statements.
OK, it should have been average load current. T1 = power supply (rectifier input) frequency. Rectifier output frequency has relevance only for the power supply output, where the ripple voltage depends on the input voltage magnitude and frequency, the load resistance and the filter capacitance:
Vr = Vmax / f RC
f = rectifier output frequency
With an increasing capacitance the diode must provide the charge in less time.
I = C * V / T (more correctly I = C * dV / dT)
I agree with your remarks, you have made correct statements.
Last edited:
No, T1 = power supply (rectifier input) period. Period is the inverse of frequency. This is the danger with quoting formulas, with incorrect supporting information: people might believe you and simply plug in some numbers. If they know enough to spot the error, they know enough not to need the formula.
Approximation. Vr is ripple peak-to-peak. Vmax should be Vdc i.e. average DC voltage. The formula assumes that the charging time is a very small fraction of 1/f, which might not be true if there is much resistance present.
Seconded!
Not only is the charge not rectangular, it's also limited by transformer Rdc, Ls and saturation. You can't just put in larger caps and plug the data into the formula, PERIOD. The larger they are, the less this approximation holds, in fact, as you approach saturation, things get very complicated.
Even for a state far below saturation, Lstray, Rdc and the capacitor form a resonant circuit, which, given 'ideal' diodes would produce a sort of truncated half-sine current waveform, since we are essentially talking energy transfer here, one should look at the area below it. It is possible to approximate this with constant current (square waveform) using corrective factors but there are a lot of assumptions that would need to be made.In the old days, this resulted in large transformers and relatively small capacitors (although at the time the capacitor/transformer cost figure was far higher than it is today, that should also be taken into account). Toroidal transformers have low Ls, and because there is much less copper, low Rdc - hence charge current is less limited on that account, but but because of that you run into problems with saturation. In either case, the formula only holds up well for a low PSU power load in W and high transformer VA. It should also be noted that even with an 'infinite' transformer, the problems hten shift to the capacitors and their parasitics, WRT charge current. In real capacitors there is a maximum ripple current rating, which should be adhered to.
Finally, diodes are not ideal. Assuming you can get higher currents out of the transformer and into the capacitor without unwanted effects, shortening the charging pulse while at the same time increasing it's height (necessary to maintain the same transferred charge), increases problems with diode reverse recovery. While the diode is going through that, you are discharging the cap, at very short conduction times this period can become significant. What happens with various parasitics during switch on and off times, which is often a source of 'unexpected' problems, is a story that gets considered only after all the basics I outlined above have been solved, and it's not a simple one either.
IMHO you can't treat them as just caps. Their basic role is to emulate a voltage source together with the rest of the PSU. Since it's not only audio current that passes through, you can't reduce them to being in series or parallel as it's normally defined - you probably could do that if they were perfect and infinite capacitance, which is not going to happen in real life
. That would be the basic conclusion. The secondary conclusion is tat the only cap literally in series with the load is the case of a capacitor coupled output.Then in a "normal" amplifier where the load and PSU caps are tied together in the GND point the psu cap's are current vise in series with the signal, but voltage vise they keep the spread of the rail and serves to reduce the ripple..
I know it more complex that this, but is that how you can look at it from an ideal simplified view point...??
I know it more complex that this, but is that how you can look at it from an ideal simplified view point...??
Hello, a little late to this thread, but I built a Bobozen (Zen v5 with Jfet inputs) with four laptop power supplies and no power caps. It sounds quite nice. About as simple a CLass A amp you can get. No caps anywhere. Gave it to a friend with Levinson equipment and he really liked it a lot. Posted it a few months back...
http://www.diyaudio.com/forums/pass-labs/86012-papa-i-want-have-zen-v5-29.html#post2379427
http://www.diyaudio.com/forums/pass-labs/86012-papa-i-want-have-zen-v5-29.html#post2379427
Last edited:
Think of it as the smoothing caps absorb the ripple current (low AC impedance) and thus keep voltage variations (voltage ripple) low.
The ripple currents can be
a.) charging currents when the charging is active and immediately after charging abruptly stops and
b.) the amplifier current demand. Those half wave audio signals that pass along alternate rails to the speaker and
c.) interference signals that develop noise on the speaker output. These extra speaker currents must come from the decoupling and smoothing capacitors, even though the source of interference is from upstream in the amplifier.
When the charging is off and the plug is pulled, case a.) drops to zero.
Last edited:
Just because you can't see them does not mean they are not there. Have you checked what is inside the laptop PSUs?
We have a saying: "What the eye does not see, the heart does not grieve over." I guess if you don't like caps, then it is best to hide them away in the PSU. We will gloss over the fact that a typical laptop SMPS will be much more complicated and possibly less audio-friendly than a few decent caps.
Ok, ok...
I just like the fact that they were $8 a piece, small, able to deliver constant required current (and yes, less than it's intended use powering a laptop) and extremely quiet.
Also able to produce deep bass when asked, just like a conventional power supply with a ton of capacitance.
Isn't this exactly what the OP asked?
I just like the fact that they were $8 a piece, small, able to deliver constant required current (and yes, less than it's intended use powering a laptop) and extremely quiet.
Also able to produce deep bass when asked, just like a conventional power supply with a ton of capacitance.
Isn't this exactly what the OP asked?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.