How big do the capacitors have to be?

20,000µF

:rolleyes:

Thank you, thank you. I'll be here all week!

The difference is, the power supply with more filtering will have less ripple. The ripple is proportional to current drawn and inversely proportional to filter capacitance. 20,000µF is a very modest amount of filtering for a power amplifier, but you get to choose, since you are designing the thing. You could also use some other method of power supply filtering, using pi filters or some other method. You could even try a regulated supply.
 
At what level does the ripple become acceptable?

That's up to you, of course.

For me if I can sit at my normal listening seat with the gear powered up but no signal playing and I can't hear the hum, that's good enough for me.

Other folks have stated that 1 - 2J per 10W of power is a good rule of thumb -- 0.5*C*V-squared, where V is the voltage across the capacitor and C is the capacitor.

Erik
 
How big etc

MikeW said:
At what level does the ripple become acceptable?

One thing to remember is that the bigger the cap, the lower the ripple amplitude, but the more higher harmonics there will be in the ripple. Power amps normally have power supply rejection ratio that rapidly falls with frequency. You end up with less hum (which should not be audible in the first place, if you do your wiring correctly) but with more higher freq junk, which tends to muddle the sound field.
I would use the 20mF. If you hear hum with a well designed and executed amp, there definitely is something else wrong.
But, indeed, it's your call in the end.

Cheers, Jan Didden
 
MikeW said:
Could somebody explain to me why you need such big caps for the power supplies?

For the same reason you need such big power transformers. Because the AC line that's ultimately feeding them is operating at only 60 Hz (or 50 depending on which country you live in). With a 60 Hz line frequency, and full wave rectification, the caps can only be refreshed once every 1/120th (or 1/100th) of a second.

The longer the time between refresh cycles, the more current can be drawn from the capacitors which translates into a lower voltage across the caps before the next refresh cycle. That's what defines the "ripple voltage."

In switching power supplies, which typically operate in the tens of kilohertz, the time between refresh cycles is much shorter and the caps are drawn down less which translates into lower ripple voltage for a given capcitance and allows you to get by with considerably less capacitance.

se
 
If you a drawing 2 amps from a 40 volt supply and have 20,000 uf. caps or 40,000 uf. caps, is there an audible difference? After reading this forum for the last couple of weeks there seems to be more problems from in-rush current than from ripple. Is it possible they are using to much capacitence?
 
MikeW said:
If you a drawing 2 amps from a 40 volt supply and have 20,000 uf. caps or 40,000 uf. caps, is there an audible difference? After reading this forum for the last couple of weeks there seems to be more problems from in-rush current than from ripple. Is it possible they are using to much capacitence?

Well inrush current isn't a problem as it's transient and only occurs during the initial power up.

However the companion to ripple voltage is ripple current. And as you increase the amount of capacitance, while you reduce ripple voltage, you increase ripple current.

And that can have consequences with regard to raidated magetic fields from the power transformer seeing as the magnitude of the magnetic field around a current carrying conductor (such as the windings in your power transformer) is proportional to the amount of current flowing through it.

So with regard to ripple voltage, its consequences depend on the power supply rejection of the amplifier topology, and with regard to ripple current, its consequences depend on how susceptible your circuit layout is to magnetic field interference.

se
 
Steve Eddy said:
Mmmmm. What have toroids to do with it?

If you have a large toroid, the iron core can hold a magnetic field after the mains power is disconnected. If you next connect the mains power in the opposite phase (over which you have no control, so it's a good chance), there will be a massive rush of current trying to reverse the field in the core. This can often cause intermittent blown fuses at turn-on. I always use a resistor between the mains and the primary to reduce this current inrush.
 
jwb said:
If you have a large toroid, the iron core can hold a magnetic field after the mains power is disconnected. If you next connect the mains power in the opposite phase (over which you have no control, so it's a good chance), there will be a massive rush of current trying to reverse the field in the core. This can often cause intermittent blown fuses at turn-on. I always use a resistor between the mains and the primary to reduce this current inrush.

Ah, ok. Yeah, then the primary appears as nearly a dead short until the domains realign.

Wasn't aware that the cores in toroids had such significant residual magnetization compared to laminated cores. But then I'm getting out of the whole AC power insanity anyway. :)

<center>
<img src="http://www.q-audio.com/images/noac2.jpg">
</center>

se
 
Another issue that effects capacitor requirements is related to the design of the amp. Some amps designs are more susceptical to ripple on the mains than others.

For example, I built a pair of Slone's 250W MOSFET amps running in Class B and got virtually no audible hum with 40mF total capacitance (20 mF per rail voltage). The Zen amp at less than 1/10 the power needs at least as much if not more filter capacitance.

Phil
 
haldor said:
Another issue that effects capacitor requirements is related to the design of the amp. Some amps designs are more susceptical to ripple on the mains than others.

For example, I built a pair of Slone's 250W MOSFET amps running in Class B and got virtually no audible hum with 40mF total capacitance (20 mF per rail voltage). The Zen amp at less than 1/10 the power needs at least as much if not more filter capacitance.

Right. That's owing to the Zen's single-ended topology.

But one advantage of the Zen is that due to its high quiescent current, it makes it more effective for using a choke regulated supply (which requires a certain amount of current be drawn through the choke at all times). The greater the quiescent current, the smaller the choke you can get away with.

se
 
Power supply design information.

I have a pretty good introductory article from an old digital text that I have that addresses the issues of designing a standard linear power supply. It address issues concerning calculations for capacitor values based on desired ripple voltages, etc.

I will try to post it tomorrow in .pdf format or at least provide a link to where you may download it if the file is to big.

Later,
 
If you have a large toroid, the iron core can hold a magnetic field after the mains power is disconnected. If you next connect the mains power in the opposite phase (over which you have no control, so it's a good chance), there will be a massive rush of current trying to reverse the field in the core.


A transformer doesn´t have stored energy in it after you disconnect the ac from it.

When you apply ac to the primary, the toroid has no energy stored in it, so it "reacts" with a great transient till it is charged. A transformer reacts in the opposite way as a capacitor. It kicks back at what you give it.

Usually a 500 VA or more Toroid can drop a 16 A security fuse by itself with no caps. The inrush current is 50 or amps but only for a few msec
 
Other folks have stated that 1 - 2J per 10W of power is a good rule of thumb -- 0.5*C*V-squared, where V is the voltage across the capacitor and C is the capacitor.

When making calculations it is more correct to use Energy as a start point like stated above.
If you see the Aleph amps from Pass the larger wattage amps still have the same amount of capacitance like the small ones around 100mF. BUT the large ones have more energy stored in them due to the larger rail to rail voltage.

It´s energy and power what the speakers use and not current only.
 
promitheus said:



A transformer doesn´t have stored energy in it after you disconnect the ac from it.

When you apply ac to the primary, the toroid has no energy stored in it, so it "reacts" with a great transient till it is charged. A transformer reacts in the opposite way as a capacitor. It kicks back at what you give it.

Usually a 500 VA or more Toroid can drop a 16 A security fuse by itself with no caps. The inrush current is 50 or amps but only for a few msec

I have to agree with Bakmeel. It is not a matter of energy storage, is has to do with the magnetic alinement of the metal particles in the iron at the moment the ac is cut. If you later reconnect in a way that would require a different magnetization direction, there is almost a zero impedance while the magnetic direction is flipped.

Jan Didden
 
How big etc

Steve Eddy said:


Well inrush current isn't a problem as it's transient and only occurs during the initial power up.

However the companion to ripple voltage is ripple current. And as you increase the amount of capacitance, while you reduce ripple voltage, you increase ripple current.

And that can have consequences with regard to raidated magetic fields from the power transformer seeing as the magnitude of the magnetic field around a current carrying conductor (such as the windings in your power transformer) is proportional to the amount of current flowing through it.

So with regard to ripple voltage, its consequences depend on the power supply rejection of the amplifier topology, and with regard to ripple current, its consequences depend on how susceptible your circuit layout is to magnetic field interference.

se

Absolutely. The radiation effect you mentioned is often overlooked, also in the wiring up of the amp. And, as I'm sure you know, not only does the ripple current increase with increased capacitance, the current pulse gets narrower and narrower as well, meaning you rapidly get all kinds of higher harmonics which radiate like hell (or should I say h**l?). Unreasonably high capacitance may increase the shareholders value for cap makers, but that's about the only positive thing I can think of.

Jan Didden;)