Give this filter a name.......

[Freq Band]

(for a Green Preamp)
I've gutted a Digidesign 442 box, and kept the switching PSU (+15,-15v). The removed old circuit board had two coil/chokes on it where it recieved power. I'm keeping/reusing these, but a friend of mine suggested adding two caps after the coils, both tied to ground...as added filtering.
1) Does this seem applicable?
2) Is there a specific name for this filter?
3) What value and type could these be?
4) Most importantly, how do I place these, POLARITY-wise ?

(. . . dots mean nothing . . . . .)


PSU +15_______Coil_______|__________to preamp plus
.
. . . . . . . . . . . . . . . . . . . . . . Cap
.
PSU ground______________|_____________ground
.
. . . . . . . . . . . . . . . . . . . . . . Cap
.
PSU -15________Coil ______|__________to preamp minus



thx.
=RR=

 

[Viitalahde]

That's simply a low pass filter, removing all the crap switched supplys make.

If you're going to add capacitors, I would go for the polyester/polypropylene route. Have you scoped the rails? Maybe they're pretty clean already, the coils are not the only thing filtering there.

Anyway, adding huge electrolytics won't help, but polarity-wise they would be: On the +15 rail, caps - to ground, + to +15v. On the -15v rail, + to ground, - on the -15v rail. At least on a linear supply.. :roll:

But just stick a, say, 100nF poly cap there, the polarity won't matter.

 

[Freq Band]

Thanks.
I kinda knew that what it was, but searching the net, could not find a single filter diagram employing a "plus, minus, ground" .....configuration.

I am scopeless. :grin:

=RR=

 

[Svart]

it is ripple filtering.

not ALL switched supplies make "crap", just the ones improperly designed, but in this case any PSU with less than adequate filtering will have an element of switching noise from the diodes and potential shifts on the rail output.

you might even add a low value, high wattage R in series after the L and C.

 

[Freq Band]

Hmmm....I've got me two Axon 1uf 250v caps sitting here in front of me...those do the trick??
(earlier suggested were .01uf (100nf))
=FB=

 

[Svart]

probably not.

you want 1000uf per rail, per amp of current being drawn, and at least 1000uf for any PSU less than that.

you also want at least 20% over the rail voltage for your capacitor voltage rating.

so for 15v rails you want at least 25v caps.

so plan on 1000uf/25v being your minimum and you will be fine. the 100nf/.1uf cap is in parallel with the larger electrolytic cap.

I would advise against tantalum caps here.

 

[Viitalahde]

I was thinking more like removing RF crap with 100nF's. I don't really know enough about switched supplys..

 

[NewYorkDave]

Give this filter a name.......

Let's call him "Frank."

 

[Larrchild]

I've gutted a Digidesign 442 box

It already sounds better, even without the power rails!

 

[Freq Band]

[quote author="Svart"]I would advise against tantalum caps here.[/quote]

I was doing some (web)research about this, and read the same recommendation, that tantalums don't like pulse applications, and will fail sooner or later....also that ceramics have tendancies to crack when heat is repeativly cycled through them.....that poly's are best esp the self healing type.
This is not written in gold, it's what I read in one article, regarding low-pass filters after a dc-to-dc converter.

--------------

Off topic...since I'm scopeless, is reading ac ripple from your average DMM, good enough to be usefull?


=FB=

 

[bcarso]

[quote author="Freq Band"]
--------------

Off topic...since I'm scopeless, is reading ac ripple from your average DMM, good enough to be usefull?


=FB=[/quote]

Useful becomes a matter of "useful for what?" For estimating the audible effect of mains frequency power rail ripple, "average" a.c. DVM's can be o.k. for rough measurements. If there is a lot of spiky hash from switching supplies they are less useful.

In AC VOLTS mode "average" DMM's are all over the map. The cheapest simply rectify and lowpass-filter the a.c. component of the voltage. This gives you the average of the resulting waveform. The display is then calibrated to give the right "r.m.s." answer on the assumption that the waveform is a sinusoid*. If the waveform is not, the answer is not accurate.

In addition the frequency response of the rectifier is limited and often unspecified. And as well the highpass filtering that strips off the d.c. component of the applied signal is also often unspecified. But, again, for mains freq work most everyday DMM's are o.k.

Better DMM's perform a squaring operation on the voltage, then lowpass filter that and finally take the square root, to get a d.c. voltage that represents the r.m.s. ("root mean square") value. The squaring circuit has its own bandwidth limitations and also a limit to how well it handles high peak-to-average voltages. But within the limitations you can at least get an accurate value for non-sinusoidal waveforms.

Of course the point of looking at ripple usually is to determine what is going to work or not when the source is used for your circuit. For unregulated supplies you need to know how sensitive the circuit is to the ripple. If regulated, the crucial data is what the lowest-going voltage may do to your regulator's ability to supply a constant output voltage. So for this the combination of the a.c. measurement along with the d.c. average measurement can be helpful. But pretty soon it's clear that the most reassuring approach is the good old oscilloscope, which will reveal at a glance how close the lowest magnitude point on your voltage droop is coming to the regulator dropout threshold.



*The average value of a fullwave rectified sinusoid is 2/Pi (about 0.6366) of the peak voltage, the r.m.s. is (square root 2)/2 (about 0.7071) of the peak. A typical mains a.c. ripple waveform is not a sinusoid, although with multiple filtering sections it comes close. There are probably posted tables of realistic ripple waveforms' r.m.s. and average values in terms of the peak-to-peak values---if not, there should be :razz:

 

[Freq Band]

It's hard....to........reply to....that.....
I'm still ......proscessing..........data
.
.
please wait....
please wait....
pleas............

OK, I'm back.
Bcarso, I believe it's time I enter full geekdom and buy a scope. I know there are other threads here about that....although I see you're from SFV, I'm in Burbank. Perhaps there might be a deal on one at Apex ??

=FB=

 

[bcarso]

Steve Slater is a forum member now---perhaps he will see this and advise.

In any case there are many scopes that come up on evilBay, and some threads about peoples' favorites on this forum.