I plan to use a DIY-SMPS with an inductive switched voltage balancer to get rid of rail pumping effects. Is the UcD so insensitive to frequency beating on the voltage lines that I dont even need to keep the SMPS and the voltage balancer synchronized? (Would save much trouble to let smps and balancer have their own oscillators with whatever frequency they like and thus avoid needing to pass gate drive signals over the isolation barrier)
Indeed there should be no reason to synchronise the voltage balancer. Quite apart from the fact that UcD amplifiers are self-oscillating and hence variable frequency, so you couldn't synchronise anything to it even if you wanted.
Simply grab a scope and look at the disturbances caused by the amplifier itself. If your voltage balance circuit isn't noisier than that you're safe.
Simply grab a scope and look at the disturbances caused by the amplifier itself. If your voltage balance circuit isn't noisier than that you're safe.
Apart from stability issues - how much of the supply rail noise will be visible on the UcD output?
I have a 1 uH/2200uF (also serves as tank for the output switches) filter on each rail to prevent the UcD channels from conducting high frequency disturbance between each other and keep HF current ripple local to each channel, but nothing more. Added to this will be the LC-filter of the SMPS, values still do be decided upon.
My greatest worry is that some harmonic of the SMPS+balancer+UcD switching will come through the LC filter of the UcD and end up in the audible range (the switches will conduct all noise during their on-time)
I have a 1 uH/2200uF (also serves as tank for the output switches) filter on each rail to prevent the UcD channels from conducting high frequency disturbance between each other and keep HF current ripple local to each channel, but nothing more. Added to this will be the LC-filter of the SMPS, values still do be decided upon.
My greatest worry is that some harmonic of the SMPS+balancer+UcD switching will come through the LC filter of the UcD and end up in the audible range (the switches will conduct all noise during their on-time)
The same filter that takes care of preventing any HF noise from the UcD power stage getting out into the power rails is equally effective at stopping noise from the power supply getting into the amplifier. The error control system (the bit that's actually called UcD) takes care of remaining noise.
Again, keep in mind that multiple UcD modules on one supply don't bother eachother, so as long as your circuit doesn't cause more noise than a UcD amplifier module, you're safe.
Again, keep in mind that multiple UcD modules on one supply don't bother eachother, so as long as your circuit doesn't cause more noise than a UcD amplifier module, you're safe.
ghemink said:
I got this in another thread,
http://www.diyaudio.com/forums/attachment.php?s=&postid=864196&stamp=1141781588
I am a bit surprised about why people really cares so much about bus supply pumping. IMHO it is not a problem in practice.
I have been testing a lot of class-D amplifiers at quite high power levels (up to 1300W in bridge-mode) with reasonable capacitance in the (linear) PSU (2x20000uF or so) and NEVER had any problems with bus pumping, even with large power at low frequency (_BIG_ PA subwoofers)
¿¿¿ ???
With a SMPS you won't have problems either if it is regulated and is properly designed, I bet you definitely don't need the switched voltage balancer. Our SPS80 PSU (that is being produced and will become available in a month or so
performs very well in this aspect too and you won't see the rails increasing. It has 6600uF per rail output capacitance.
Perhaps Ghemink can also tell us his experience, as he did some deep tests with a SMPS on audio amplifiers, and the PSU he used (K6 SMPS) had very low output capacitance (2x470uF per rail if I recall ok).
Any comments?
I have been testing a lot of class-D amplifiers at quite high power levels (up to 1300W in bridge-mode) with reasonable capacitance in the (linear) PSU (2x20000uF or so) and NEVER had any problems with bus pumping, even with large power at low frequency (_BIG_ PA subwoofers)
¿¿¿ ???
With a SMPS you won't have problems either if it is regulated and is properly designed, I bet you definitely don't need the switched voltage balancer. Our SPS80 PSU (that is being produced and will become available in a month or so
performs very well in this aspect too and you won't see the rails increasing. It has 6600uF per rail output capacitance.Perhaps Ghemink can also tell us his experience, as he did some deep tests with a SMPS on audio amplifiers, and the PSU he used (K6 SMPS) had very low output capacitance (2x470uF per rail if I recall ok).
Any comments?
I find in practice that having 2x4700uF power supply capacitance is sufficient to support 10Hz without adverse effects. Most people round here have much more than that in their supplies!
Where it makes sense to have some countermeasure is when using an SMPS that has all storage capacitance on the primary side. Many of these supplies don't even like having anything more than a few 100uF on the secondary side. In this case, pumping is a real issue.
Where it makes sense to have some countermeasure is when using an SMPS that has all storage capacitance on the primary side. Many of these supplies don't even like having anything more than a few 100uF on the secondary side. In this case, pumping is a real issue.
I fully agree with Bruno (as almost ever)
You can see lots of SMPS's out there where output capacitance is ridiculous and the designers feel justified by just adding feedback (or conversely, they add only a few uF at the output because they are not able to make the feedback loop stable with more capacitance).
You can see lots of SMPS's out there where output capacitance is ridiculous and the designers feel justified by just adding feedback (or conversely, they add only a few uF at the output because they are not able to make the feedback loop stable with more capacitance).
ssanmor said:I am a bit surprised about why people really cares so much about bus supply pumping. IMHO it is not a problem in practice.
I have been testing a lot of class-D amplifiers at quite high power levels (up to 1300W in bridge-mode) with reasonable capacitance in the (linear) PSU (2x20000uF or so) and NEVER had any problems with bus pumping, even with large power at low frequency (_BIG_ PA subwoofers)
¿¿¿ ???
With a SMPS you won't have problems either if it is regulated and is properly designed, I bet you definitely don't need the switched voltage balancer. Our SPS80 PSU (that is being produced and will become available in a month or so
Perhaps Ghemink can also tell us his experience, as he did some deep tests with a SMPS on audio amplifiers, and the PSU he used (K6 SMPS) had very low output capacitance (2x470uF per rail if I recall ok).
Any comments?
Yes, a comment from my side. I actually never used the K6 SMPS supply with that 2x470uF, I modified the feedback loop and added more caps. Besides that, I also used the amps for the woofers in a kind of bridged way. Meaning, the left and right woofer channel sharing the same supply but I'm running both amps out of phase, so in most of the cases (at low frequencies), each amp will pump in the opposite direction. In the my final setup, I plan to use 2 amps for each channel to drive the two woofers per channel. Either in bridge or out of phase to avoid any risk of pumping. I know, I maybe overly cautious, anyway it won't hurt and hypex won't complain as they could sell me more modules that way
In your case with 6600uF per rail there should be no problem. Can't wait to check out your SMPS.
Best regards
Gertjan
Bruno Putzeys said:
Hi Bruno,
Seems like going balanced output would be a better choice than doing something like a load balancer?
The pumping issue seems to be a mess for a off line switcher since it will go out of regulation unless you can suck current out of the UCD supplies.
I'd think if sticking to single ended, that generating two loosly regulated outputs (perhaps clark converters for PFC) and then a sync buck regulator for each rail would be ideal. That's lots of stuff though. Seems like a balanced/bridged output would be a better cleaner solution.
At any rate, I seems you need some way to store pumping energy and PFC might be a way.
I look forward to seeing SMPS implementation that beats big iron though.
Maybe you should just make a UCD that runs directly off the line! ;>)
Mike
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.