No, I don't see your point.
200ms delay is a soft start for the transformer, to allow use of a close rated fuse.
It hardly makes any difference to the rate of charge of the main smoothing bank. I can hear the transformer hum/growl with the load of the capacitors AFTER the 1/5th second delay. In my experience ±45mF for a 4ohm speaker load (or two 8ohm speaker loads) does not need any slow charge circuit.
200ms delay is a soft start for the transformer, to allow use of a close rated fuse.
It hardly makes any difference to the rate of charge of the main smoothing bank. I can hear the transformer hum/growl with the load of the capacitors AFTER the 1/5th second delay. In my experience ±45mF for a 4ohm speaker load (or two 8ohm speaker loads) does not need any slow charge circuit.
Last edited:
I recommend a high pass filter set to 2Hz and use ±20mF per 8ohms channel for good deep bass. This becomes ±40mF for each 4ohms channel.[/QUOTE said:
I never went as far as that.
More evidence that adequate smoothing capacitance is required for good deep bass performance.
oh my!! Ciss - Input Capacitance 9000 pF .. anyway i got two numbers of IRFP250.
thanks Hugh
Dacz
Andrew,
I have found the same as Ranchu. I found that using large filter caps cause problems for rectifiers and transformers. In the power supplies on the FetZilla I suggested two caps for each rails - and two 0.15R resistors between the two parallel caps. One cap smoothes the 100Hz output ripple directly and the next cap accommodates the signal earth current for the output stage. The charge pulses in the first cap are highest at startup but also high when the amp is running because the caps are 'topped up' only when their voltage dropped momentarily under the rectifier pulse.
So consider when this high current from the rectifier pulse meets a high signal current in the filter cap. They mix - intermodulation. It would be better to reduce the rectifier pulses - minimise the size of the cap - and quarantine the speaker/output stage current - use another filter cap. We achieve this with TWO caps per rail, separating them with two 2W 0.15R power resistors between the caps neg to neg and pos to pos.
It works well and you can use a 4700uF for the rectifier and 2200uF for the output stage. There are some advantages this way; you can use smaller caps, you can hugely reduve the artefacts of the ripple, and you can use smaller, faster rectifiers.
Huge cap banks on an amps are expensive, require very stout, slow bridges, and cause interference between charge pulses and output stage earth currents.
HD
I have found the same as Ranchu. I found that using large filter caps cause problems for rectifiers and transformers. In the power supplies on the FetZilla I suggested two caps for each rails - and two 0.15R resistors between the two parallel caps. One cap smoothes the 100Hz output ripple directly and the next cap accommodates the signal earth current for the output stage. The charge pulses in the first cap are highest at startup but also high when the amp is running because the caps are 'topped up' only when their voltage dropped momentarily under the rectifier pulse.
So consider when this high current from the rectifier pulse meets a high signal current in the filter cap. They mix - intermodulation. It would be better to reduce the rectifier pulses - minimise the size of the cap - and quarantine the speaker/output stage current - use another filter cap. We achieve this with TWO caps per rail, separating them with two 2W 0.15R power resistors between the caps neg to neg and pos to pos.
It works well and you can use a 4700uF for the rectifier and 2200uF for the output stage. There are some advantages this way; you can use smaller caps, you can hugely reduve the artefacts of the ripple, and you can use smaller, faster rectifiers.
Huge cap banks on an amps are expensive, require very stout, slow bridges, and cause interference between charge pulses and output stage earth currents.
HD
Last edited:
I don't think the Fetzilla makes a good example for your explanation. As you know from my previous posts when I was building the Fetzilla with your integrated PSU, it turned out to be the noisiest amplifier I had built. I have since had one more noisy amplifier. It too does not use an input LTP.
I don't accept your explanation for adopting low levels of smoothing capacitance.
I certainly do accept that attenuating mains charging pulses from the audio circuits should be a benefit. I have posted a few times that using slightly higher resistance cabling between the transformer through the stages of the PSU to the on board supply rail decoupling introduces rC attenuation. This added attenuation is free and is better than using big cross-sectional areas of cabling.
A typical power amplifier + PSU has 4 cascaded stages of RC attenuation.
Transformer through rectfier to smoothing RC > second smoothing RC > electrolytic RC > final HF decoupling RC.
No problem Andrew.
Remember that if you do not listen to others experiences they will respond likewise. I tried!!
HD
Remember that if you do not listen to others experiences they will respond likewise. I tried!!
HD
I am reading and effectively listening.
It's just that my experiences give different results.
eg
It is different from the opinion of Ranchu, where he said:
It's just that my experiences give different results.
eg
I am reporting my experience.
It is different from the opinion of Ranchu, where he said:
and then followed up with:
Last edited:
Now I'm not one to talk as "other pressing matters" have meant my soldering iron remains stone cold and my p-spice is now years out of date.
But I'm really surprised that no-one has had a decent go at choke input power supplies for high power solid state amplifiers. Particularly those circuits with enough bias current to avoid needing a specific swinging choke design.
Last edited:
Choke input PSU are used in Solid State Low voltage high Current demand amplifiers.
But they are very expensive and as a result are quite rare.
I think I have seen a couple of proposals in this Forum.
But they are very expensive and as a result are quite rare.
I think I have seen a couple of proposals in this Forum.
The rectifier feeds through an inductor, then across a capacitor. It is an LC PSU.
But the voltage output of this LC varies from ~1.5Vac, when current is very low, down to 0.9Vac, when current is in the range that the LC works. i.e. you need a transformer voltage that is ~160% compared to the normal transformer PSU.
One must avoid that rising output voltage by ensuring that adequate current is drawn out of the LC at all times during any signal condition.
But the voltage output of this LC varies from ~1.5Vac, when current is very low, down to 0.9Vac, when current is in the range that the LC works. i.e. you need a transformer voltage that is ~160% compared to the normal transformer PSU.
One must avoid that rising output voltage by ensuring that adequate current is drawn out of the LC at all times during any signal condition.
I`m surprised that no one uses PSU with artificial or virtual ground point , where main transformers secondary coil usually don`t have center tap .
Isn't a typical choke loaded supply CLC rather than LC. In fact, one can build a reasonable amount of R into the choke to drop the DC voltage, when there is such a need. In this case you place the choke in the unrectified Secondary AC lines.
Last edited:
One may also prevent destruction of Output stages which are undersized for a given load, by limiting the capacitor bank and current rating of the transformer. Have seen this technique in PA class AB amps. I have decided to use a 30-0-30, 300VA toroid to power my standard quasi-complimentary MOSFET amp, driving 4 ohm speakers by limiting the capacitor bank.
In my experience, this is a compromise whichever way you look at it, but a way of escape too.
In my experience, this is a compromise whichever way you look at it, but a way of escape too.
Last edited: