roger-k said:
Overkill? Your kidding right? This is DIY audio man! We aren't happy until the power amp turn-on surge pulls down our sector of the power grid
.Seriously, I too am interested in the answer to this question. Something like "X x Class A watts = N uF" as a rule of thumb would come in handy.
Bill
Hi
You ask for a rule of thumb and not a calculation for the size of electrolytics in power supplies.
I do normally try to calculate my needs but this is my rule of thumb;
A class A amplifier of 100 watt continuous output would need 100,000uf to give a ripple of 0.6 volt on the amplifier power rails. You can also reduce ripple in the power supply by using inductors in the power supply.
When I listen to a typical 50 watt class A amplifier I normally find that I can not distinguish any change in sound if the power supply capacitance rises above 50,000uf. Also I can detect only a little change in sound for the same amplifier if the capacitance drops to 25,000uf. Others may have more sensative hearing and may need more capacitance.
You did not mention the transformer, but I think that the same pure class A 100watt amplifier would draw 500 watt all the time and would need a 1,500 watt transformer.
I recommend you read the article by Nelson Pass on Pass DIY entitled "Power supplies"
Don
You ask for a rule of thumb and not a calculation for the size of electrolytics in power supplies.
I do normally try to calculate my needs but this is my rule of thumb;
A class A amplifier of 100 watt continuous output would need 100,000uf to give a ripple of 0.6 volt on the amplifier power rails. You can also reduce ripple in the power supply by using inductors in the power supply.
When I listen to a typical 50 watt class A amplifier I normally find that I can not distinguish any change in sound if the power supply capacitance rises above 50,000uf. Also I can detect only a little change in sound for the same amplifier if the capacitance drops to 25,000uf. Others may have more sensative hearing and may need more capacitance.
You did not mention the transformer, but I think that the same pure class A 100watt amplifier would draw 500 watt all the time and would need a 1,500 watt transformer.
I recommend you read the article by Nelson Pass on Pass DIY entitled "Power supplies"
Don
Well, he could also go all the way to voltage regulation. It's just not very practical with all this additional heat - especially for ClassA. That and the additional parts that require another pcb usually scare people away and let them revert to adding caps instead - it's much easier. Oh yeah, there are also people that decline voltage regulation per se.
Have fun, Hannes
http://www.tnt-audio.com/clinica/ssps3_e.html
There is a list of tables to towards the bottom of this page with some suggested minimums...
On page 2 of that document it offers another solution
"There is another approach. It is accepted that one needs 1-2 joules of energy per every 10W of output power. For a 50W/8 Ohms amplifier, we need 10-20 joules of energy to be stored. We can use a formula, 1/2CVsquared (where C equals capacitance and V equals voltage, the voltage being squared) to calculate that 15,000uF, fed by say 33V (worst case, full load on) allows for 8.16 joules per capacitor, or 16.3 joules per channel - enough to fry quite a few unsuspecting speakers in the low end class, and even some in the midrange class. If 33V is our worst case, we can assume 36-38V supplies with the load off (say 37V), which means that just before a transient, we will in fact have some 20.5 joules stored in each channel's capacitors. "
There is a list of tables to towards the bottom of this page with some suggested minimums...
On page 2 of that document it offers another solution
"There is another approach. It is accepted that one needs 1-2 joules of energy per every 10W of output power. For a 50W/8 Ohms amplifier, we need 10-20 joules of energy to be stored. We can use a formula, 1/2CVsquared (where C equals capacitance and V equals voltage, the voltage being squared) to calculate that 15,000uF, fed by say 33V (worst case, full load on) allows for 8.16 joules per capacitor, or 16.3 joules per channel - enough to fry quite a few unsuspecting speakers in the low end class, and even some in the midrange class. If 33V is our worst case, we can assume 36-38V supplies with the load off (say 37V), which means that just before a transient, we will in fact have some 20.5 joules stored in each channel's capacitors. "
Hi
Whay don't you use capacitance multiplier PSU?
I do sometimes use a capacitance multiplier. Typically that would be in pre amps where I need a higher voltage than my transformers could provide and in the front ends of power amps where I need a higher voltage than in the output stages. They are very effective.
I have tried/used a voltage multiplier for the output stage of a power amp. JLH was a believer in this concept.
However I have found that a capacitance multiplier in the output stage of the power amp is not as convenient or as effective as simply using capacitors. ( or capacitors plus inductors if you need less ripple ). For high power output amps the arrangement becomes more complicated than simply using capacitors and does not work as effectively in clearing up detail on complex music passages. But for lower power amps like the JLH it works OK.
Just my observations.
Don
Whay don't you use capacitance multiplier PSU?
I do sometimes use a capacitance multiplier. Typically that would be in pre amps where I need a higher voltage than my transformers could provide and in the front ends of power amps where I need a higher voltage than in the output stages. They are very effective.
I have tried/used a voltage multiplier for the output stage of a power amp. JLH was a believer in this concept.
However I have found that a capacitance multiplier in the output stage of the power amp is not as convenient or as effective as simply using capacitors. ( or capacitors plus inductors if you need less ripple ). For high power output amps the arrangement becomes more complicated than simply using capacitors and does not work as effectively in clearing up detail on complex music passages. But for lower power amps like the JLH it works OK.
Just my observations.
Don
The LT1083 voltage regulator needs a differential voltage of 1.5V* and I can't see that a cap multiplier would need less as the transistor needs to stay in the saturation region.
Multiplied with some amperes you definitely need a heatsink. Mind you, I didn't say it's not manageable - in fact I use voltage regulation myself - it's just not as convenient as adding caps.
If I miss something, please tell me.
Have fun, Hannes
*+ additional margin for ripple, transformer regulation....
roger-k said:
The idea that there is some arbitrary level of capacitance that somehow qualifies as too much is incomplete. There is more than one way to look at the capacitors in a power supply. Most people think either in terms of ripple reduction or energy (properly measured in Joules, as per Nordic's post above--not in simple quantities of uF) storage. For some reason, this literally does seem to be an either/or proposition for most people; they don't seem to be able to keep both concepts going simultaneously.
Just to mess with your minds, let me toss in the idea of the capacitor bank as a low pass filter...for the signal coming back up the rails.
If you follow the current in an amplifier all the way around the circuit, you will see that the current goes straight through the power supply capacitors. To the extent that the capacitors form a dynamically changing low pass filter modulated by the reflections of the signal in the rails, there are some frequencies that will not pass...or do so only in greatly attenuated form. What happens then is that you begin modulating the rail voltage itself; frequencies above the filter frequency pass through the caps.
So what's the dynamic part mean in this view? Well...
Pretend that you're standing at an imaginary point on the rail, right on top of the power supply caps. Now look "downstream" at the circuit. When the circuit is at idle, you can't tell the difference between the output devices and a resistor that happens to have some nominal value that draws that same amount of current. As a concrete example, let's use the Hafler DH-200, which happens to be the first amplifier that comes to mind that I can remember both the idle current and the rail voltage. In the case of the DH-200, the rails are +60V and the stock bias was .25A (assuming that I haven't slipped a mental cog). Now, if you're standing atop the power supply cap, you're not going to be able to tell the difference between the output MOSFETs and a (60V/.25A=) 240 Ohm resistor. But all that changes once music begins to play. The current draw suddenly begins to change. Accordingly, the apparent resistance changes (okay, okay, say impedance if it makes you happier). And as the apparent resistance changes, the filter frequency begins to change. Higher. Lower. All over the place. You can no longer assume that you know what the low pass frequency is.
So what do you do? You make the filter frequency much, much lower than it would be if you were analyzing things at idle. That means using more capacitance.
No, looking at both rails simultaneously doesn't help. Musical waveforms aren't symmetrical like test waveforms. Again, look at the Hafler. It's class AB. At some point, one bank of MOSFETs (either N-ch or P-ch) are going to shut off completely. They no longer contribute to the output, which at that point is coming entirely from the other bank of MOSFETs. Which only draws from its rail and completes its circuit through ground, not through the other rail. (I will neatly sidestep the temptation to point out that this is a really good reason to have a low impedance connection to ground...) As a result, the entire (assumed to be low frequency) signal is drawing from one bank of MOSFETs--which by definition are now "low resistance"--which raises the low frequency rolloff considerably.
Use lots and lots of capacitance.
Grey
some usual criterias
for circuits are:
Main filter capacitors:
Regulated supply (before the regulator):
2000 uF / Ampere
Unregulated supply:
5000 uF / Ampere
In the case of True Class A we will know the current and this average current drawn will be same all the time, hopefully
As GRollins has explained, thing get a lot more complicated when Class AB.
Not only will the current drawn be assymmetrical, but the level will be different at different power output levels & different frequencies of output will demand different spikes/peaks.
But we can use the worst case = max current drawn at max output.
Estimate how high is the PEAK CURRENT needed.
Then use at least 5000 uF per Ampere drawn.
Or why not 10000 uF per Ampere.
As GRollins also tells us, there is nothing like too Much capacitance.
Except that it will cost more .... and more.
---------------------
The transformer size and inner impedance of transformer is also one issue.
Bigger is better!
Another thing is:
There is a difference in Electrolytic Capacitor Quality.
Two 10000 uF / 50 VDC caps can be very different in electrical quality.
Even if the capacitance value and voltage rating is the same.
So, a very high quality 4700 uF can often do as good job as one lower quality 10000 uF.
Data specification methods & testing can be very different for different manufacturers.
There are some wellknown Electrolytic Capacitors makers
that always have very high quality.
Lineup
for circuits are:
Main filter capacitors:
Regulated supply (before the regulator):
2000 uF / Ampere
Unregulated supply:
5000 uF / Ampere
In the case of True Class A we will know the current and this average current drawn will be same all the time, hopefully
As GRollins has explained, thing get a lot more complicated when Class AB.
Not only will the current drawn be assymmetrical, but the level will be different at different power output levels & different frequencies of output will demand different spikes/peaks.
But we can use the worst case = max current drawn at max output.
Estimate how high is the PEAK CURRENT needed.
Then use at least 5000 uF per Ampere drawn.
Or why not 10000 uF per Ampere.
As GRollins also tells us, there is nothing like too Much capacitance.
Except that it will cost more .... and more.
---------------------
The transformer size and inner impedance of transformer is also one issue.
Bigger is better!
Another thing is:
There is a difference in Electrolytic Capacitor Quality.
Two 10000 uF / 50 VDC caps can be very different in electrical quality.
Even if the capacitance value and voltage rating is the same.
So, a very high quality 4700 uF can often do as good job as one lower quality 10000 uF.
Data specification methods & testing can be very different for different manufacturers.
There are some wellknown Electrolytic Capacitors makers
that always have very high quality.
In the long run ,
buying High Quality CAPS, to a somewhat higher price,
may turn out to be the most economical thing we can do
Lineup
I have tried to use 4x250.000uF pr 150watt mono amp, and it didnt do it any good
After that I went fore best quality and I used just a single pair of 2x15.000 Jensen 4-pole pr 50watt mono, and it works perfectly, VERY smooth and powerfull
Next amp will be a single pair of 2x47.000 pr 150watt mono, and it will probably be Rifa PEH as Mundorf M-Lytics are too expencive and probably no better
After that I went fore best quality and I used just a single pair of 2x15.000 Jensen 4-pole pr 50watt mono, and it works perfectly, VERY smooth and powerfull
Next amp will be a single pair of 2x47.000 pr 150watt mono, and it will probably be Rifa PEH as Mundorf M-Lytics are too expencive and probably no better
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.