speaker dc offset question

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Here's the way my mind is seeing the output of that circuit, a rare insight into my diabolical thought processes which always seek the simplest model! :D

An externally hosted image should be here but it was not working when we last tested it.


So, yes, I like 4,700uF on the power supply. The trick is to get the current source approximated...:cool:

cf.

An externally hosted image should be here but it was not working when we last tested it.
 
Last edited:
The PSU choke might look like a current source at audio frequencies, but for subsonics such as the music envelope its true nature will emerge: an inductance in series with a resistor, fed from a voltage source.
You're quite right. Seems it crosses over to current source at around the resonance @ 23 Hz, if I've done the LC calculation right for 10mH/4700uF. Below that it's voltage source. Here's the model I've been using:

An externally hosted image should be here but it was not working when we last tested it.


Not quite the 2X 10000uF/10mh Pi network of the original, but I can hang with it, and allow for other stuff. FWIW you do need the first capacitor before the coil here, because the 10mH coil is below the critical inductance (27mH here) you need with a bridge rectifier on the choke-input filter.

An externally hosted image should be here but it was not working when we last tested it.


I've forgotten why I'm doing this...Oh, yes, we're examining the optimum value for the RC constant of the last power supply capacitor. Seems this amp distorts bass a bit when you turn up the wick. :D
 
I think what that is telling you is:
For the components shown the critical frequency is 23Hz.
The amplifier will misbehave due to supply rail voltage modulation when significant current passes that has frequencies <23Hz.

I would conclude that the input filter should be adjusted to prevent/attenuate these low frequencies from entering the amplifier.

Are we back to the "rule" I stated earlier?
Set the input filters to determine the frequency response of the amplifier.
 
We're not disagreeing on the input filter at all. I was seeing horrors at 23 Hz too. It's the power supply got me scratching my head. :D

The bass issue is actually occurring with his more sophisticated solid-state power supply according to hameay:

100743d1200626821-class-mosfet-amplifier-ps-cap_mult_2.gif


FWIW, the original amplifier board has a 470uF smoothing capacitor too. For all we know, the distortion is just MOSFET non-linearity too.
 
Quite right. I guess we got interested in what could be done with a few simple practical changes. But it's always going to distort at high volume. :)

So we've got a much smaller capacitor on the input, so far. Based on the desired time constant, around 30Hz, I suppose.

I've just fallen in that the power supply is shared between two channels on a design that is probably very sensitive to crosstalk from the shared power supply.

An externally hosted image should be here but it was not working when we last tested it.


I'd quite fancy separate power supplies to each channel. 2X4700uF and 10 mH. :cool:
 
Amplifier distortion increases as one allows the output voltage to approach the supply rails. Negative FeedBack (NFB) will make a very good job of hiding that increase in distortion until NFB can no longer correct the gross error. A single ended open loop BJT shows the massive increase in 2nd harmonic that an NFB tries to hide.

If one let's the supply rail voltage modulate then the onset of that distortion when Vsupply>>Vout is transgressed will happen earlier. The worse the modulation the earlier the distortion happens.

Sort the supply rails and the distortion at LF will go back to normal.

We are still ruled by post21.

The RC time constants (or RLC where inductors are used) must be staggered in the correct proportions to allow the amplifier to perform with the intended frequency band.
 
Last edited:
I'd suggest that for an input filter set to F-3dB=30Hz, that the output filter F-3dB be 15Hz and the PSU filter be <10Hz.

If one wants a lower frequency in the passband then the other filter must be altered to suit.

If one want can accept a higher frequency for the LF roll off, then the other filters can be altered (upwards in frequency) without crippling the amplifier performance.
 
My amp has certainly had its time in the spot light. When I developed it I would never have thought it would be so unbelievably popular. I have had requests for a ZCA MKII. I have some thoughts so keep the faith.

But this thread is about a PS for a 5W amp. And at 5W its possibly getting well into 10% distortion. Just recently I built a 2W 6EM7 integrated amp (Paris) that sounded so good a friend traded his new $3.5K 300BSE amp (which I also built) for it. So let's now dwell on watts.

I have only just been alluded to this thread but will pick-up on a few more recent posting. I have read quite a few of the later posting but seeing I'm "Johny Come Lately" I'll cast my "gut feel"comments on a few latter ones.

I like chokes. All my tube amps have chokes and tube rec. where I can. A SS map module seller and developer friend of mine from Aspen amps agrees a choked PS is a good PS. I think chokes kill XTalk. So you may still get very good results (big sound stage) from a single PS if it is choked.

NFB. I'm just OK with local NFB BUT global NFB is the work of Satan (who was not an audiophile for those who didn't know). I am very well aware of all the benefits of GNFB but the derogation to overall sound is just not worth it.

We have one FET here and a simple PS which is at the heart of a fab. sounding low wattage amp. Don't expect any more from it. Like (now) hundreds of others, enjoy it for what it is. Pure and simple.

I would love to see some images of builds. Please?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.