The start up current of the 625VA transformers is large.
A 625VA running on 230Vac will have a maximum operating draw of ~2.7Aac
But fitting a 2.7A fuse to the primary will not allow reliable start up unless there is a current limiter built in somewhere.
Normally a transformer that is not close rated will use a fuse about 3times higher than the maximum rated current, i.e. a 625VA will use a T8A fuse to allow reliable starting.
That T8A will not blow at 8A.
Even at 16A it may take more than half a minute to rupture.
During that time much damage could be done to whatever is surrounding the transformer.
If one wants to close rate the fuse and use a T3.1A then a soft start circuit will be required. A 20r current limiter will probably allow a 625VA to start on 230Vac with a T3.1A fuse fitted.
Slow charging of smoothing capacitance is a different issue.
A 625VA running on 230Vac will have a maximum operating draw of ~2.7Aac
But fitting a 2.7A fuse to the primary will not allow reliable start up unless there is a current limiter built in somewhere.
Normally a transformer that is not close rated will use a fuse about 3times higher than the maximum rated current, i.e. a 625VA will use a T8A fuse to allow reliable starting.
That T8A will not blow at 8A.
Even at 16A it may take more than half a minute to rupture.
During that time much damage could be done to whatever is surrounding the transformer.
If one wants to close rate the fuse and use a T3.1A then a soft start circuit will be required. A 20r current limiter will probably allow a 625VA to start on 230Vac with a T3.1A fuse fitted.
Slow charging of smoothing capacitance is a different issue.
Read the amplifier power supply articles here:
http://sound.westhost.com/articles.htm
Very respected guy.
http://sound.westhost.com/articles.htm
Very respected guy.
Its class AB.
You can get away with quite a bit more with music than you can applying a test sine wave.
Last edited:
One consideration nobody has mentioned is what kind of music or material you listen to- at what base power level.
I listen to classical music at 1 v pp base level into 8 ohms (about 0.06 W/ch) with peaks about 1% of the time 55 db higher. So my first Solid state class AB amp has 3.3 mf for both channels, sounds totally loud enough to me , and doesn't have any hum. My speakers are 101 db @ 1 W 1 m.
but if you are playing techno or house music in a bar, with almost no volume variation, near the speaker limit of 300 W rms, 10 mf per channel may be a little light. The transformer that will do that weighs >20 lb.
So design for your needs. Amps that run near the upper limit are called "PA" amps, and tend to have more heat sink and fans than "hifi" amps, too. As a lot of lying goes on in power advertising (like those 5 channel TV systems, chock full of **** to me) people buy PA amps based more on brand reputation, than on specifications.
I listen to classical music at 1 v pp base level into 8 ohms (about 0.06 W/ch) with peaks about 1% of the time 55 db higher. So my first Solid state class AB amp has 3.3 mf for both channels, sounds totally loud enough to me , and doesn't have any hum. My speakers are 101 db @ 1 W 1 m.
but if you are playing techno or house music in a bar, with almost no volume variation, near the speaker limit of 300 W rms, 10 mf per channel may be a little light. The transformer that will do that weighs >20 lb.
So design for your needs. Amps that run near the upper limit are called "PA" amps, and tend to have more heat sink and fans than "hifi" amps, too. As a lot of lying goes on in power advertising (like those 5 channel TV systems, chock full of **** to me) people buy PA amps based more on brand reputation, than on specifications.
Wiring and grounding is important, for a lower noise floor. Keeping high ripple currents isolated from sensitive signal grounds with a star on star ground can improve signal to noise ratio at all power levels.
I listen to all kinds of music, but usually I listen to electronic music at low volume. Sometimes I do turn it up when I`m alone at home.
"But what is the power of the amps? "
Doesn't matter.
Power stored is a function of the square of the voltage, so if the voltage of the power supply doubles the energy stored quadruples.
The values for power supply capacitance can be realistically determined by a consideration of the numerical value of W C R where W is the line frequency (377 rad/sec), C is the power supply capacitance, and R is the load resistance. An WCR value of 10 yields about 10 per cent ripple (pk.-pk. ) and a value of 100 has about two percent. Below 10, the power supply will have serious problems and values of about 100 will achieve diminishing performance returns. The minimum value then, for each of the four power supply capacitors should be about 3,000uF and the maximum about 30,00OuF. Capacitances above this value may cause diode bridge failure due to turn-on surges and are not recommended.
(http://www.passdiy.com/pdf/a40.pdf)
3,300µF x2 for one channel at 8Ω with 60hz line (3,900µF for 50hz), double for 4Ω, double again for 2Ω.
Example: Crest CA9 has a tiered supply and 120,000µF total capacitance. This is 15,000µF per channel, per tier, the 10% ripple point.
Remember, stored energy increases with the square of the voltage, so a 3,300µF cap with 100V on it stores 4x the energy that a 3,300µF cap with 50V on it stores.
radian per second = W
f*2Pi=W
377 (60hz)
314(50hz)
Doesn't matter.
Power stored is a function of the square of the voltage, so if the voltage of the power supply doubles the energy stored quadruples.
The values for power supply capacitance can be realistically determined by a consideration of the numerical value of W C R where W is the line frequency (377 rad/sec), C is the power supply capacitance, and R is the load resistance. An WCR value of 10 yields about 10 per cent ripple (pk.-pk. ) and a value of 100 has about two percent. Below 10, the power supply will have serious problems and values of about 100 will achieve diminishing performance returns. The minimum value then, for each of the four power supply capacitors should be about 3,000uF and the maximum about 30,00OuF. Capacitances above this value may cause diode bridge failure due to turn-on surges and are not recommended.
(http://www.passdiy.com/pdf/a40.pdf)
3,300µF x2 for one channel at 8Ω with 60hz line (3,900µF for 50hz), double for 4Ω, double again for 2Ω.
Example: Crest CA9 has a tiered supply and 120,000µF total capacitance. This is 15,000µF per channel, per tier, the 10% ripple point.
Remember, stored energy increases with the square of the voltage, so a 3,300µF cap with 100V on it stores 4x the energy that a 3,300µF cap with 50V on it stores.
radian per second = W
f*2Pi=W
377 (60hz)
314(50hz)
Hi Djk,
I had not seen, or at least not remembered, seeing your WCR rule.
10<WCR<100, seems so simple.
+-20mF on an 8ohms capable amplifier fed with 50Hz Mains results in WCR = 50.
I had not seen, or at least not remembered, seeing your WCR rule.
10<WCR<100, seems so simple.
+-20mF on an 8ohms capable amplifier fed with 50Hz Mains results in WCR = 50.
But does this rule apply for Class A, since PSRR is poor in simple designs atleast and more C along with R or L is required to minimise hum.
I have also seen in PA amps, capacitance being kept conservative, when transformer and output stage are conservative.
I have also seen in PA amps, capacitance being kept conservative, when transformer and output stage are conservative.
Most simple class A amplifiers aren't that high powered. You can put 100kuF on a 300VA trafo and not blow anything with the surge if you use a 25A rectrifier. Don't try this with a 1VA unless you want little pieces of hot rectifier flying around the room.
Many PA amps are capacitor challenged -they put in relatively small ones resulting in a high pole frequency. They essentially can only put out full power down to 40Hz at 8 ohms and below that (ohms or frequency) power capacity drops. The touring models (like DJK's CA9) can put out full power down to about 20Hz at 4 ohms. Bigger capacitors than that aren't needed for typical audio reproduction. For LFE, you may want more. But subwoofer plate amps in particular tend to skimp on power supplies - at just the frequencies where you shouldn't.
Many PA amps are capacitor challenged -they put in relatively small ones resulting in a high pole frequency. They essentially can only put out full power down to 40Hz at 8 ohms and below that (ohms or frequency) power capacity drops. The touring models (like DJK's CA9) can put out full power down to about 20Hz at 4 ohms. Bigger capacitors than that aren't needed for typical audio reproduction. For LFE, you may want more. But subwoofer plate amps in particular tend to skimp on power supplies - at just the frequencies where you shouldn't.
There are PA amps built for newbies that look first at the price, and PA amps built for the experienced that want the volume they paid for, all night. Wattage can be lied about, unless the customer tests for it. I use log resistors to test with a VOM. Brand name even can be counterfeited, but the performance can't. Read the PA thread to see what brands repairmen recommend. My first power amp was bad for losing its rated power to a quarter of rated value in three-four years of >2000 hours a year.
The DJK "rule" was violated by the first solid state amp I owned. It doesn't hum at 1-2 watt average, with 3300 uf for both channels. The trick, I suppose, was regulating the main supply, and using a huge heavy secondary wire transformer, 6.5 amps all day for 120 W music. Dynaco was a transformer builder originally and copper magnet wire was cheap in 1966. It might hum at 120 w average but the heat sinks wouldn't support that; it was a "hifi" amp.
A fast acting switcher supply may allow modern designers to skimp on capacitors on the transistor rail area. 2200 of 200 v capacitors have a lot of energy due to that square law rule. If the music demands a lot of energy while the line wall voltage is low, a switcher supply can deliver it quickly whereas a transformer supply would have to wait for the wall voltage to build up.
Last edited:
Imagine the filter caps getting smaller and smaller until there are only parasitics in the circuit. The supply would get progressively more lumpy, and the amplifier would not work well. Probably wouldn't work at all quite quickly, although most amps have local capacitance which might sustain function. So it matters.
OTOH you can increase the cap size and hum and PSU sag and maybe noise will diminish. To the point where they become imperceptible. After that there's no point in going bigger. Of course sag isn't a problem with class A.
The problem is, just where is that point?
That's without going into the drawbacks.
So you can have bigger and bigger caps until somebody else says stop or other aspects of the equipment become irksome. Some members assemble huge capacitors for a hobby. What the hell, this is a hobby.
Decisions, decisions.
The amount of capacitance you have will allow you to have some more bass.
But the actual sound will be affected by the brand and series of the capacitor. This is true to the point where I'll even use less capacitance for better sound. I haven't had to sacrifice bass for this, but it could happen. This may not be as true with a constant current power supply since they pull on line so hard.
But the actual sound will be affected by the brand and series of the capacitor. This is true to the point where I'll even use less capacitance for better sound. I haven't had to sacrifice bass for this, but it could happen. This may not be as true with a constant current power supply since they pull on line so hard.
I think that the latest versions of my PSU cap-sizing spreadsheet and related documents were posted at:
http://www.diyaudio.com/forums/chip-amps/251159-best-capacitance-gainclone-7.html#post3830213
Cheers,
Tom Gootee
http://www.diyaudio.com/forums/chip-amps/251159-best-capacitance-gainclone-7.html#post3830213
Cheers,
Tom Gootee
Reply to #35 How? Look at a music waveform. Do you see any low freq? (Slow changing signal). No. The composite signal is usually fast and full level (loudness wars). This is what the amp sees. How does it loose cap voltage in just the low freqs? That dosnt make sense.
If there are low frequencies present in the sound, e.g. bass, then yes, it will obviously be seen in the signal, which is the current from the PSU caps, by the way (see link in my sig).
A half cycle (i.e. positive or negative rail's signal component) of a low frequency component of the sound can last longer than the charging period, or even two or more charging periods, of the reservoir and decoupling caps. (Charging period = 1/(2X the mains frequency), in a linear PSU.)
Therefore, with low frequency (bass) signal components, at a high-enough output power level, the caps could literally run out of charge and current, also possibly causing their voltage to fall to the point where clipping could occur.
See the links at the link in Post 36.
Last edited:
It is an interesting experiment to scope the output of an amp and the power rails at the output devices while it is playing music. You see similar wave forms as long as you filter out the DC components...
The PSU current's waveform should look like the amp output's voltage waveform. The PSU's voltage waveform (when caps are not charging) would basically be the integral over time of the output voltage waveform, since i = C(dv/dt) for the PSU and decoupling caps. See the link in my sig, which compares the output voltage waveform to the PSU current waveform, showing that they are virtually identical, in a class AB power amp. I guess your low-pass filter ("...filter out the DC components") was acting as an integrator, as might be expected!
Last edited:
- Status
- Not open for further replies.