Zener clamp against power supply pumping
Hi all,
I have general class d question, but since I own a Zappulse (actually 2) and I cannot find a recent, more dedicated thread, I am posting it here.
De question is: why is it not possible to simply use a zener clamp on the PSU rails to protect the PSU (possibly unregulated smps) and the amp against the effects of power supply pumping?
TIA, JZ
Hi all,
I have general class d question, but since I own a Zappulse (actually 2) and I cannot find a recent, more dedicated thread, I am posting it here.
De question is: why is it not possible to simply use a zener clamp on the PSU rails to protect the PSU (possibly unregulated smps) and the amp against the effects of power supply pumping?
TIA, JZ
Re: Zener clamp against power supply pumping
You would need a zener that can consume a lot of power. You could make something with a power transistor that starts conducting from rail to GND when the + and - rails are getting out of balance (one rail higher than the other). Actually have been thinking about that for my SMPS but I think I will just go for using dummy resistors (to balance + and - rail current) and using lots of caps at the output of the SMPS so that power supply pumping is no issue.
Best regards
Gertjan
Joep Zonnebloem said:
You would need a zener that can consume a lot of power. You could make something with a power transistor that starts conducting from rail to GND when the + and - rails are getting out of balance (one rail higher than the other). Actually have been thinking about that for my SMPS but I think I will just go for using dummy resistors (to balance + and - rail current) and using lots of caps at the output of the SMPS so that power supply pumping is no issue.
Best regards
Gertjan
Nice layout indeed. Very curious if its finished by now.Zvon said:
One little thing, I would prefer the caps vertically.
My Zapcase (in progress) is not so compact.
Re: Re: Zener clamp against power supply pumping
Hi Gertjan,
I once saw a datasheet of a zener that could handle up to someting like 200W with stand-off voltage of 80V; unfortunately I cannot find the link anymore. In your case, with the UCD causing such an inbalance on the rail, your current drain with the power resistors seems to be a nice and simple solution (I just read something from your latest thread on the K6, I had missed that one...). I don't know what the ZAPs are doing to that respect; I just took my setup apart, I'll measure it soon but I have confidence that this is exemplary for de UCD
.
Anyway, though it looks a bit silly to have linear voltage regulation on an SMPS, I would like to try something like you suggested. But all I can think of just now is standard linear regulators, with power transistors an a zener, on both rails, for a fixed maximum voltage (with a considerable marging on top of the nominal voltage per rail).
How do you compare both rails to drive the transistors? With a comparator opamp, or is there some standard controlling IC? What should be the best way to do this?
Regards, JZ
ghemink said:
Hi Gertjan,
I once saw a datasheet of a zener that could handle up to someting like 200W with stand-off voltage of 80V; unfortunately I cannot find the link anymore. In your case, with the UCD causing such an inbalance on the rail, your current drain with the power resistors seems to be a nice and simple solution (I just read something from your latest thread on the K6, I had missed that one...). I don't know what the ZAPs are doing to that respect; I just took my setup apart, I'll measure it soon but I have confidence that this is exemplary for de UCD
. Anyway, though it looks a bit silly to have linear voltage regulation on an SMPS, I would like to try something like you suggested. But all I can think of just now is standard linear regulators, with power transistors an a zener, on both rails, for a fixed maximum voltage (with a considerable marging on top of the nominal voltage per rail).
How do you compare both rails to drive the transistors? With a comparator opamp, or is there some standard controlling IC? What should be the best way to do this?
Regards, JZ
Re: Re: Re: Zener clamp against power supply pumping
I was thinking of using two bipolar darlington power transistors (an NPN for the + rail and a PNP for the - rail biased in class B with their emittors connected to GND via a resistor. Then a voltage divider between + and - rail generates a voltage that should be equal to GND, if it is higher than GND because the + rail voltage is higher than the - rail, then the NPN would draw more current than the PNP and would thus pull down the + rail. To achieve this, the output of that divider has to be used to bias the base of those transistors and the collectors go to the + and - rails respectively. I have never tried it but it should work (I think).
Best regards
Gertjan
Joep Zonnebloem said:
Hi Gertjan,
I once saw a datasheet of a zener that could handle up to someting like 200W with stand-off voltage of 80V; unfortunately I cannot find the link anymore. In your case, with the UCD causing such an inbalance on the rail, your current drain with the power resistors seems to be a nice and simple solution (I just read something from your latest thread on the K6, I had missed that one...). I don't know what the ZAPs are doing to that respect; I just took my setup apart, I'll measure it soon but I have confidence that this is exemplary for de UCD
Anyway, though it looks a bit silly to have linear voltage regulation on an SMPS, I would like to try something like you suggested. But all I can think of just now is standard linear regulators, with power transistors an a zener, on both rails, for a fixed maximum voltage (with a considerable marging on top of the nominal voltage per rail).
How do you compare both rails to drive the transistors? With a comparator opamp, or is there some standard controlling IC? What should be the best way to do this?
Regards, JZ
I was thinking of using two bipolar darlington power transistors (an NPN for the + rail and a PNP for the - rail biased in class B with their emittors connected to GND via a resistor. Then a voltage divider between + and - rail generates a voltage that should be equal to GND, if it is higher than GND because the + rail voltage is higher than the - rail, then the NPN would draw more current than the PNP and would thus pull down the + rail. To achieve this, the output of that divider has to be used to bias the base of those transistors and the collectors go to the + and - rails respectively. I have never tried it but it should work (I think).
Best regards
Gertjan
Hi Gertjan,
I like your solution to the pumping problem, but in may be simpler to employ a TVS or TVSS. Can get em bi-directional (rail to rail) or Uni-directional for rail to ground, good for ~1000W, one should do the trick.
I do believe pumping is not a great concern though, in a normal PSU.
In case anyone wants to search for info on them, TVS=transient voltage suppressor, and TVSS= +surge, same thing.
Regards,
Chris
I like your solution to the pumping problem, but in may be simpler to employ a TVS or TVSS. Can get em bi-directional (rail to rail) or Uni-directional for rail to ground, good for ~1000W, one should do the trick.
I do believe pumping is not a great concern though, in a normal PSU.
In case anyone wants to search for info on them, TVS=transient voltage suppressor, and TVSS= +surge, same thing.
Regards,
Chris
classd4sure said:
Pumping is not much of a problem with a conventional supply but it can be with a regulated SMPS as I noticed. In case of UcD, and likely ZAP as well if you don't use a seperate supply for the gate driver, one of the rails draws more current than the other. This difference of something in the order of 50mA per module is enough to cause an imbalance in the + and - rail of my regulated SMPS. Now I use dummy resistors to draw some extra current from the + rail to balance the + and - rails. This works perfect and is by far the cheapest solution, although I have to admit, not so elegant. I also don't really like a solution with transistors or zeners or any other clipping devices as it introduces non-linear behavior (especially those hard clipping solutions). I may end up to modify the SMPSs (I have two of them) to each supply one rail so that each real is completely seperate regulated. This is the most elegant (and expensive) solution as it should always work independent of any current differences on + and - rail, of course lots of caps at the output are still needed to prevent pumping due to LF signals. In the latter case (each rail an SMPS) I will feed the left and right amps from the same supply but all amps will have their own decoupling caps in the form of parallel connected RCR networks forming virtual 4-pole caps to isolate the power supply of all amps from eachother. Most likely I will go for that solution.
Best regards
Gertjan
That's alot of complexity for a very simple problem isn't it?
Those are all killer ideas however.. I guess I fail to see how the most expensive solution could possibly qualify as being the most elegant?
Certainly if a resistor or two does the trick... that's alot more elegant?
I think I'd opt for a TVS in series with a limiting resistor myself, bit more elegant than two smps! Same as you have now but it would only conduct during overvoltage and so increase efficiency.
Since you already have that extra smps though, why not use it huh? Nothing wrong there at all!
Gertjan, cool projects!
Regards,
Chris
PS: It's good to have options
Those are all killer ideas however.. I guess I fail to see how the most expensive solution could possibly qualify as being the most elegant?
Certainly if a resistor or two does the trick... that's alot more elegant?
I think I'd opt for a TVS in series with a limiting resistor myself, bit more elegant than two smps! Same as you have now but it would only conduct during overvoltage and so increase efficiency.
Since you already have that extra smps though, why not use it huh? Nothing wrong there at all!
Gertjan, cool projects!
Regards,
Chris
PS: It's good to have options
classd4sure said:That's alot of complexity for a very simple problem isn't it?
Those are all killer ideas however.. I guess I fail to see how the most expensive solution could possibly qualify as being the most elegant?
Certainly if a resistor or two does the trick... that's alot more elegant?
I think I'd opt for a TVS in series with a limiting resistor myself, bit more elegant than two smps! Same as you have now but it would only conduct during overvoltage and so increase efficiency.
Since you already have that extra smps though, why not use it huh? Nothing wrong there at all!
Gertjan, cool projects!
Regards,
Chris
PS: It's good to have options
The original plan was to use an SMPS for each channel. So I have two options, each channel their own SMPS or each rail their own SMPS.
I probably would not advise to use an SMPS for UcDs as it needs a lot of optimization to get it working correctly. However, I have them and I'm going to use them as they are performing very well after all the tweaking. More tweaking planned but first I need cases to built everything in. I like the fact that they are regulated so that I have no problems with weak mains lines and mains variation of +- 10% or so.
Best regards
Gertjan
Just try to imagine that a transformer transforms in all possible directions: Primary to any secondary, any secondary to primary and any secondary to the other secondaries.
If synchronous rectification is used then the voltage/winding ratios are enforced during the switching period if the synchronous rectifiers are made as two-way switches.
Regards
Charles
If synchronous rectification is used then the voltage/winding ratios are enforced during the switching period if the synchronous rectifiers are made as two-way switches.
Regards
Charles
http://www.lcaudio.com/index.php?page=1
[July 21st 2005] L C Audio announces a new Ultra Low Distortion Class D amplifier, the ZAPpulse 700XE! [CLICK] to read more!
But the link seems dead?
yet..
[July 21st 2005] L C Audio announces a new Ultra Low Distortion Class D amplifier, the ZAPpulse 700XE! [CLICK] to read more!
But the link seems dead?
yet..
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.