Forums

PS Filter C in Series, w Parallel R

Started by Warren December 8, 2011
I am planning to build a small power supply, probably (for 
now) in the range of 18 to 32 VDC after filtering. But I'll be 
using a transformer secondary with no center tap and a bridge 
rectifier. So I was planning to use two filter caps in series, 
grounding the center point.

+)-------+-----+-----> (+)
       + |     |
    C1 -----   R1
       -----   |
         |     |
         +-----+----+
       + |     |    |
       -----   |   Gnd
    C2 -----   R2 
         |     |
-)-------+-----+-----> (-)

This got me to thinking about using parallel resistors to keep 
the voltages across each cap "balanced". In a higher voltage 
PS, I'd also want to avoid exceeding the cap's rated max at 
any point during its operation (like at "turn on"). This last 
aspect is academic for the current project but I'd still like 
to know more about the design process.

How does one normally choose R?

At turn on, you have a high current (low impedance) and low 
voltage for both caps, so initially voltage imbalance 
shouldn't be an issue (assuming both are discharged 
initially).

At full (or nearly full) charge, each C has low/no current 
(high impedance) and high voltages.

Starting at some point in between (midpoint? 60%?) you want 
neither cap not to over charge past it's maximum voltage, 
while the "weaker" cap is catching up.

Given the tolerance for electrolytics is something like +/-
30%,  I would guess you need to look at C1 (electrolytic) as 
being 30% over, and C2 as -30% in value. Then look at the 
voltage curves as they charge to see where the unbalanced risk 
is. Then compute a suitable R around that with some sort of a 
safety margin (10%?).

I can do the above but I wonder if there is a simpler 
procedure to arrive at a suitable value - a method that can be 
used on the back of a envelope over lunch?

Warren.
On Dec 8, 11:05=A0am, Warren <ve3...@gmail.com> wrote:
> I am planning to build a small power supply, probably (for > now) in the range of 18 to 32 VDC after filtering. But I'll be > using a transformer secondary with no center tap and a bridge > rectifier. So I was planning to use two filter caps in series, > grounding the center point. > > +)-------+-----+-----> (+) > =A0 =A0 =A0 =A0+ | =A0 =A0 | > =A0 =A0 C1 ----- =A0 R1 > =A0 =A0 =A0 =A0----- =A0 | > =A0 =A0 =A0 =A0 =A0| =A0 =A0 | > =A0 =A0 =A0 =A0 =A0+-----+----+ > =A0 =A0 =A0 =A0+ | =A0 =A0 | =A0 =A0| > =A0 =A0 =A0 =A0----- =A0 | =A0 Gnd > =A0 =A0 C2 ----- =A0 R2 > =A0 =A0 =A0 =A0 =A0| =A0 =A0 | > -)-------+-----+-----> (-) > > This got me to thinking about using parallel resistors to keep > the voltages across each cap "balanced". In a higher voltage > PS, I'd also want to avoid exceeding the cap's rated max at > any point during its operation (like at "turn on"). This last > aspect is academic for the current project but I'd still like > to know more about the design process. > > How does one normally choose R? > > At turn on, you have a high current (low impedance) and low > voltage for both caps, so initially voltage imbalance > shouldn't be an issue (assuming both are discharged > initially). > > At full (or nearly full) charge, each C has low/no current > (high impedance) and high voltages. > > Starting at some point in between (midpoint? 60%?) you want > neither cap not to over charge past it's maximum voltage, > while the "weaker" cap is catching up. > > Given the tolerance for electrolytics is something like +/- > 30%, =A0I would guess you need to look at C1 (electrolytic) as > being 30% over, and C2 as -30% in value. Then look at the > voltage curves as they charge to see where the unbalanced risk > is. Then compute a suitable R around that with some sort of a > safety margin (10%?). > > I can do the above but I wonder if there is a simpler > procedure to arrive at a suitable value - a method that can be > used on the back of a envelope over lunch? > > Warren.
If you have loads that aren't constant, you won't be able to achieve balance without a lot of power loss or more complexity. Since you're still in the design phase, why not use a voltage doubler topology: ---+---|>|-------+-----+-----> (+) | + | | | C1 ----- R1 | ----- | | | | -----------------+-----+----+ | + | | | | ----- | Gnd | C2 ----- R2 | | | +---|<|-------+-----+-----> (-) You can still use the bridge rectifier though you won't need everything in it. You'll need roughly half the transformer voltage, twice the current, and bigger capacitors, of course. But it's immune to +/- load imbalance.
On Dec 8, 7:05=A0pm, Warren <ve3...@gmail.com> wrote:

> I am planning to build a small power supply, probably (for > now) in the range of 18 to 32 VDC after filtering. But I'll be > using a transformer secondary with no center tap and a bridge > rectifier. So I was planning to use two filter caps in series, > grounding the center point. > > +)-------+-----+-----> (+) > =A0 =A0 =A0 =A0+ | =A0 =A0 | > =A0 =A0 C1 ----- =A0 R1 > =A0 =A0 =A0 =A0----- =A0 | > =A0 =A0 =A0 =A0 =A0| =A0 =A0 | > =A0 =A0 =A0 =A0 =A0+-----+----+ > =A0 =A0 =A0 =A0+ | =A0 =A0 | =A0 =A0| > =A0 =A0 =A0 =A0----- =A0 | =A0 Gnd > =A0 =A0 C2 ----- =A0 R2 > =A0 =A0 =A0 =A0 =A0| =A0 =A0 | > -)-------+-----+-----> (-) > > This got me to thinking about using parallel resistors to keep > the voltages across each cap "balanced". In a higher voltage > PS, I'd also want to avoid exceeding the cap's rated max at > any point during its operation (like at "turn on"). This last > aspect is academic for the current project but I'd still like > to know more about the design process. > > How does one normally choose R? > > At turn on, you have a high current (low impedance) and low > voltage for both caps, so initially voltage imbalance > shouldn't be an issue (assuming both are discharged > initially). > > At full (or nearly full) charge, each C has low/no current > (high impedance) and high voltages. > > Starting at some point in between (midpoint? 60%?) you want > neither cap not to over charge past it's maximum voltage, > while the "weaker" cap is catching up. > > Given the tolerance for electrolytics is something like +/- > 30%, =A0I would guess you need to look at C1 (electrolytic) as > being 30% over, and C2 as -30% in value. Then look at the > voltage curves as they charge to see where the unbalanced risk > is. Then compute a suitable R around that with some sort of a > safety margin (10%?). > > I can do the above but I wonder if there is a simpler > procedure to arrive at a suitable value - a method that can be > used on the back of a envelope over lunch? > > Warren.
I've used this approach before in an odd situation, and it does work, but has its issues. Tolerance is the biggest issue. If you're using +100%-50% caps, the 2 can have a capacitance ratio of 4:1, and voltage balance doesnt even begin to happen. And as caps age, things only get worse. So you really need to use zeners across the caps. That of course puts a pair of zeners across the whole psu, so you check that with max possible line voltage, worst case transformer regulation and no load, you dont end up with both zeners conducting. The next issue is that as the load i changes on one rail, it alters the V in the _other_ capacitor too, since C charging conditions are changed for the other cap. So you get noise/signal fed from one side to the other. Fine in some circuits, not ok in some. Check those points and its fine. NT
cassiope expounded in news:ea92ced9-986c-4951-8534-a376db2ac5d3
@i6g2000vbe.googlegroups.com:

> On Dec 8, 11:05&#2013266080;am, Warren <ve3...@gmail.com> wrote: >> I am planning to build a small power supply, probably (for >> now) in the range of 18 to 32 VDC after filtering. But I'll be >> using a transformer secondary with no center tap and a bridge >> rectifier. So I was planning to use two filter caps in series, >> grounding the center point. >> >> +)-------+-----+-----> (+) >> &#2013266080; &#2013266080; &#2013266080; &#2013266080;+ | &#2013266080; &#2013266080; | >> &#2013266080; &#2013266080; C1 ----- &#2013266080; R1 >> &#2013266080; &#2013266080; &#2013266080; &#2013266080;----- &#2013266080; | >> &#2013266080; &#2013266080; &#2013266080; &#2013266080; &#2013266080;| &#2013266080; &#2013266080; | >> &#2013266080; &#2013266080; &#2013266080; &#2013266080; &#2013266080;+-----+----+ >> &#2013266080; &#2013266080; &#2013266080; &#2013266080;+ | &#2013266080; &#2013266080; | &#2013266080; &#2013266080;| >> &#2013266080; &#2013266080; &#2013266080; &#2013266080;----- &#2013266080; | &#2013266080; Gnd >> &#2013266080; &#2013266080; C2 ----- &#2013266080; R2 >> &#2013266080; &#2013266080; &#2013266080; &#2013266080; &#2013266080;| &#2013266080; &#2013266080; | >> -)-------+-----+-----> (-) >> >> This got me to thinking about using parallel resistors to keep >> the voltages across each cap "balanced". In a higher voltage >> PS, I'd also want to avoid exceeding the cap's rated max at >> any point during its operation (like at "turn on"). This last >> aspect is academic for the current project but I'd still like >> to know more about the design process. >> >> How does one normally choose R?
...
> > If you have loads that aren't constant, you won't be able to achieve > balance without a lot of power loss or more complexity.
They will be balanced mostly, since this is just powering a small 8W rms class-B ss amp. But after reading your post and the next one, I think what I was planning is just a bad idea.
> Since you're > still in the design phase, why not use a voltage doubler topology: > > ---+---|>|-------+-----+-----> (+) > | + | | > | C1 ----- R1 > | ----- | > | | | > -----------------+-----+----+ > | + | | | > | ----- | Gnd > | C2 ----- R2 > | | | > +---|<|-------+-----+-----> (-)
> You can still use the bridge rectifier though you won't need > everything in it. You'll need roughly half the transformer voltage, > twice the current, and bigger capacitors, of course. But it's immune > to +/- load imbalance.
Diodes are no problem, but I think I will spend a little time seeing if I can get a surplus split secondary transformer instead. Some things are probably just not worth messing around with. Warren
NT expounded in news:de743f59-da15-4e37-b9cf-3b4cd72d0bb1
@q9g2000yqe.googlegroups.com:

> On Dec 8, 7:05&#2013266080;pm, Warren <ve3...@gmail.com> wrote: >> I am planning to build a small power supply, probably (for >> now) in the range of 18 to 32 VDC after filtering. But I'll be >> using a transformer secondary with no center tap and a bridge >> rectifier. So I was planning to use two filter caps in series, >> grounding the center point. >> >> +)-------+-----+-----> (+) >> &#2013266080; &#2013266080; &#2013266080; &#2013266080;+ | &#2013266080; &#2013266080; | >> &#2013266080; &#2013266080; C1 ----- &#2013266080; R1 >> &#2013266080; &#2013266080; &#2013266080; &#2013266080;----- &#2013266080; | >> &#2013266080; &#2013266080; &#2013266080; &#2013266080; &#2013266080;| &#2013266080; &#2013266080; | >> &#2013266080; &#2013266080; &#2013266080; &#2013266080; &#2013266080;+-----+----+ >> &#2013266080; &#2013266080; &#2013266080; &#2013266080;+ | &#2013266080; &#2013266080; | &#2013266080; &#2013266080;| >> &#2013266080; &#2013266080; &#2013266080; &#2013266080;----- &#2013266080; | &#2013266080; Gnd >> &#2013266080; &#2013266080; C2 ----- &#2013266080; R2 >> &#2013266080; &#2013266080; &#2013266080; &#2013266080; &#2013266080;| &#2013266080; &#2013266080; | >> -)-------+-----+-----> (-)
..
> I've used this approach before in an odd situation, and it does work, > but has its issues. > > Tolerance is the biggest issue. If you're using +100%-50% caps, the 2 > can have a capacitance ratio of 4:1, and voltage balance doesnt even > begin to happen. And as caps age, things only get worse. So you really > need to use zeners across the caps. That of course puts a pair of > zeners across the whole psu, so you check that with max possible line > voltage, worst case transformer regulation and no load, you dont end > up with both zeners conducting. > > The next issue is that as the load i changes on one rail, it alters > the V in the _other_ capacitor too, since C charging conditions are > changed for the other cap. So you get noise/signal fed from one side > to the other. Fine in some circuits, not ok in some. > > Check those points and its fine. > NT
You've talked me out of it. While it will be just powering a small 8W rms audio amp (class-B), it's not the kind of thing I want to come back to and fix later. I might as well look for a surplus xfrmer with the split secondary and be done with it, once and for all. I hate doing things twice. Warren.
On Fri, 9 Dec 2011 01:53:43 +0000 (UTC), Warren <ve3wwg@gmail.com>
wrote:

>cassiope expounded in news:ea92ced9-986c-4951-8534-a376db2ac5d3 >@i6g2000vbe.googlegroups.com: > >> On Dec 8, 11:05&#2013266080;am, Warren <ve3...@gmail.com> wrote: >>> I am planning to build a small power supply, probably (for >>> now) in the range of 18 to 32 VDC after filtering. But I'll be >>> using a transformer secondary with no center tap and a bridge >>> rectifier. So I was planning to use two filter caps in series, >>> grounding the center point. >>> >>> +)-------+-----+-----> (+) >>> &#2013266080; &#2013266080; &#2013266080; &#2013266080;+ | &#2013266080; &#2013266080; | >>> &#2013266080; &#2013266080; C1 ----- &#2013266080; R1 >>> &#2013266080; &#2013266080; &#2013266080; &#2013266080;----- &#2013266080; | >>> &#2013266080; &#2013266080; &#2013266080; &#2013266080; &#2013266080;| &#2013266080; &#2013266080; | >>> &#2013266080; &#2013266080; &#2013266080; &#2013266080; &#2013266080;+-----+----+ >>> &#2013266080; &#2013266080; &#2013266080; &#2013266080;+ | &#2013266080; &#2013266080; | &#2013266080; &#2013266080;| >>> &#2013266080; &#2013266080; &#2013266080; &#2013266080;----- &#2013266080; | &#2013266080; Gnd >>> &#2013266080; &#2013266080; C2 ----- &#2013266080; R2 >>> &#2013266080; &#2013266080; &#2013266080; &#2013266080; &#2013266080;| &#2013266080; &#2013266080; | >>> -)-------+-----+-----> (-) >>> >>> This got me to thinking about using parallel resistors to keep >>> the voltages across each cap "balanced". In a higher voltage >>> PS, I'd also want to avoid exceeding the cap's rated max at >>> any point during its operation (like at "turn on"). This last >>> aspect is academic for the current project but I'd still like >>> to know more about the design process. >>> >>> How does one normally choose R? >... >> >> If you have loads that aren't constant, you won't be able to achieve >> balance without a lot of power loss or more complexity. > >They will be balanced mostly, since this is just powering a small 8W rms >class-B ss amp. But after reading your post and the next one, I think >what I was planning is just a bad idea. > >> Since you're >> still in the design phase, why not use a voltage doubler topology: >> >> ---+---|>|-------+-----+-----> (+) >> | + | | >> | C1 ----- R1 >> | ----- | >> | | | >> -----------------+-----+----+ >> | + | | | >> | ----- | Gnd >> | C2 ----- R2 >> | | | >> +---|<|-------+-----+-----> (-) > >> You can still use the bridge rectifier though you won't need >> everything in it. You'll need roughly half the transformer voltage, >> twice the current, and bigger capacitors, of course. But it's immune >> to +/- load imbalance. > >Diodes are no problem, but I think I will spend a little time seeing if I >can get a surplus split secondary transformer instead. > >Some things are probably just not worth messing around with. > >Warren
It's the asking of the question that makes the most sense. Another alternative is to make a single supply rail, then capacitively couple the load. RL
On Dec 8, 8:53=A0pm, Warren <ve3...@gmail.com> wrote:

> > On Dec 8, 11:05=A0am, Warren <ve3...@gmail.com> wrote: > >> I am planning to build a small power supply, probably (for > >> now) in the range of 18 to 32 VDC after filtering. But I'll be > >> using a transformer secondary with no center tap and a bridge > > balance without a lot of power loss or more complexity. > > They will be balanced mostly, since this is just powering a small 8W rms > class-B ss amp. But after reading your post and the next one, I think > what I was planning is just a bad idea. >
> > Diodes are no problem, but I think I will spend a little time seeing if I > can get a surplus split secondary transformer instead. > > Some things are probably just not worth messing around with. >
On Dec 8, 8:58=A0pm, Warren <ve3...@gmail.com> wrote:
> > While it will be just powering a small 8W rms audio amp (class-B), it's > not the kind of thing I want to come back to and fix later. I might as > well look for a surplus xfrmer with the split secondary and be done with > it, once and for all. I hate doing things twice. >
You don't need a bipolar power supply if you go with a bridge-tied- load (BTL) amplifier. In your power range, parts like the TDA2050V are cheap, readily available, and have all kinds of built-in protection too.
On Dec 9, 1:58=A0am, Warren <ve3...@gmail.com> wrote:
> NT expounded in news:de743f59-da15-4e37-b9cf-3b4cd72d0bb1 > @q9g2000yqe.googlegroups.com: > > > > > > > On Dec 8, 7:05=A0pm, Warren <ve3...@gmail.com> wrote: > >> I am planning to build a small power supply, probably (for > >> now) in the range of 18 to 32 VDC after filtering. But I'll be > >> using a transformer secondary with no center tap and a bridge > >> rectifier. So I was planning to use two filter caps in series, > >> grounding the center point. > > >> +)-------+-----+-----> (+) > >> =A0 =A0 =A0 =A0+ | =A0 =A0 | > >> =A0 =A0 C1 ----- =A0 R1 > >> =A0 =A0 =A0 =A0----- =A0 | > >> =A0 =A0 =A0 =A0 =A0| =A0 =A0 | > >> =A0 =A0 =A0 =A0 =A0+-----+----+ > >> =A0 =A0 =A0 =A0+ | =A0 =A0 | =A0 =A0| > >> =A0 =A0 =A0 =A0----- =A0 | =A0 Gnd > >> =A0 =A0 C2 ----- =A0 R2 > >> =A0 =A0 =A0 =A0 =A0| =A0 =A0 | > >> -)-------+-----+-----> (-) > .. > > I've used this approach before in an odd situation, and it does work, > > but has its issues. > > > Tolerance is the biggest issue. If you're using +100%-50% caps, the 2 > > can have a capacitance ratio of 4:1, and voltage balance doesnt even > > begin to happen. And as caps age, things only get worse. So you really > > need to use zeners across the caps. That of course puts a pair of > > zeners across the whole psu, so you check that with max possible line > > voltage, worst case transformer regulation and no load, you dont end > > up with both zeners conducting. > > > The next issue is that as the load i changes on one rail, it alters > > the V in the _other_ capacitor too, since C charging conditions are > > changed for the other cap. So you get noise/signal fed from one side > > to the other. Fine in some circuits, not ok in some. > > > Check those points and its fine. > > NT > > You've talked me out of it. > > While it will be just powering a small 8W rms audio amp (class-B), it's > not the kind of thing I want to come back to and fix later. I might as > well look for a surplus xfrmer with the split secondary and be done with > it, once and for all. I hate doing things twice. > > Warren.
Your call. Its not hard to implement though. NT