Improving linear P/S efficiency

[AnDe]

Hi guys!
Stumbled upon your forum while searchng for something on the net - looks like a great place! Laughed my head off with a thread on resistance-reactance-conductance - etc. Anyway, I'd be very grateful for some help.

This is the problem:

In a classic linear power supply with standard rectifiers, the angle during which the diodes conduct, is quite small. This creates at least three problems:
(1) high peak currents v.s. moderate average values;
(2) the transformer is actually idling between current pulses and is (in my opinion) under-utilized;
(3) poor power factor.

I am referring to a typical off-the-mains setup @50/60 cps, involving a transformer, bridge rectifier, reservoir caps and a load.

Is there a way to increase the angle during which current is passed to the caps, possibly leading to a higher efficiency than the plain-vanilla setup?

The primary objective of this exercise would be to make better use of the transformer, possibly achieving a higher power to weight ratio, without reverting to a high-frequency SMPS. PWM-driven rectifiers probably? The typical power output would be say 2kW average, 5kW peak.

Can someone point me in the right direction?

p.s. hope I hit the right forum. If not, feel free to move this thread where appropriate.

 

[mneary]

inductive input. Probably a few Henries.

 

[crutschow]

To amplify on mneary's comments you could add a choke inductor between the rectifier output and the filter capacitor. They make special "swinging chokes" for this which have a higher inductance at a low current since you need more inductance at low currents to keep the inductor current from dropping to zero between the current pulses out of the rectifier.

Of course since the rectifiers are now conducting to through more of the AC waveform, the DC output will now be close to the average of the AC voltage, not it's peak. You thus would need a higher transformer output voltage for a given required DC voltage.

 

[AnDe]

Thanks! That would definitely work, however, there are some more considerations:

1. The load is not constant, varying from just a few watts, right up to the specified values;
2. The voltage output at full load has to remain within 10% of idle, the less sag, the better. Overvoltage would kill the electronics.
3. Minimal weight/volume added.

 

[Hero999]

Add active power factor correction.

The basic idea is to use a boost converter to increase the voltage to above the peak voltage of the input waveform. The difference between a normal boost converter and an active power correction circuit is the latter does its best to ensure the input current is directly proportional to the voltage. The simpler active power factor correction units just use a low pass filter, with a cut off frequency much lower than the mains frequency, in the feedback loop which gives reasonable results. The more complected active power factor correction units use an analogue multiplier to ensure that I = V/R which gives a near-perfect power factor.

What's the output voltage from the transformer?

If the peak voltage is below 60V then you could probably use a standard switching regulator IC to make a simple active power factor correction circuit.

 

[AnDe]

The output of the transformer is 2x55Vac + 2x88Vac, fed to two rectifiers, to produce a tiered dual supply of +/- 77Vdc and +/- 123Vdc. This feeds a class H amp, driving as low as 2 ohms in some cases.