PS-Filter

S

Samuel Groner

Guest
Hi

I want to design a C-L-C (or is L-C-L better?) filter for a 30 V/4 A
linear power supply. What is a good value for L if C is around 1 uF?

Thanks for input
Samuel
 
Samuel Groner wrote:
Hi

I want to design a C-L-C (or is L-C-L better?) filter for a 30 V/4 A
linear power supply. What is a good value for L if C is around 1 uF?

Thanks for input
Samuel
Click to expand...
Something near infinity (assuming you are talking about the rectifier
filter).
If you think of your full load as a resistor, it would have a value of
30/4=7.5 ohms. If your line frequency is 60 Hz, at the fundamental
ripple frequency (120 Hz) of a full wave rectified output, a 1 uf cap
would have an impedance of about 1300 ohms. This is essentially
insignificant as a filter in parallel with 7.5 ohms. So the inductor
is going ot have to do the whole job. I would suggest you start with
capacitors that have a 120 Hz impedance less than 7.5 ohms before you
start to worry about the inductor. I would normally think in terns of
1000 to 5000 uf per amp for a rectifier filter. So a 4700 uf 50 volt
electrolytic would be a ball park start. And the input cap has to
have a ripple current greater than 4 amperes, also, if you use the CLC
form.

What is the design goal of the filter?

--
John Popelish
 
I want to design a C-L-C (or is L-C-L better?) filter for a 30 V/4 A
linear power supply. What is a good value for L if C is around 1 uF?
Click to expand...
Typo - should read "1 mF" or "1000 uF"...

What is the design goal of the filter?
Click to expand...
The filter will sit right after the rectifier and in front of the
regulator (IC & ext. transistors). Goal of the CLC-filter would be
lowest ripple for C=1 mF.

Samuel
 
Samuel Groner wrote:
I want to design a C-L-C (or is L-C-L better?) filter for a 30 V/4 A
linear power supply. What is a good value for L if C is around 1 uF?

Typo - should read "1 mF" or "1000 uF"...

What is the design goal of the filter?

The filter will sit right after the rectifier and in front of the
regulator (IC & ext. transistors). Goal of the CLC-filter would be
lowest ripple for C=1 mF.

Samuel
Click to expand...
I have not seen a linear regulator that likes to be fed with an
inductance, so I think your choices are CLC or LC or C. The first and
last (capacitive input) tend to charge the caps up to the peak voltage
of the rectified AC somewhere approaching 1.4 times the AC voltage of
the secondary, while the LC (inductor input) tends to average the
rectifier output. This averaging effect only takes place if the
inductor current is continuous throughout the rectification cycle. So
for light loads or insufficient inductance, the voltage will rise.
Indictor input filters are good if the transformer voltage is a bit
too high, and they also spread the diode conduction time throughout
the cycle, lowering the transformer copper losses, line harmonics and
diode peak current. The CLC (pi) filter) doesn't put much AC voltage
across the inductor, so can work with smaller inductance. You just
want ot avoid having a resonance at any rectified line harmonics, so
the fundamental resonance should be lower than the second line
harmonic (the fundamental of the rectified line). But in all cases,
more inductance implies lower ripple, so your design goal of minimum
ripple still implies infinite inductance. By the way, what is the
load that you are so worried about rectifier filtration, if you have a
linear regulator? after the filter? The output ripple will be
dependent on both the filter and the regulator, assuming the filter
ripple is not so deep that the regulator has nothing ot discard at the
minimum points on the ripple.

What is the secondary voltage from the transformer you are using to
drive the 30 volt 4 amp linear regulator?

--
John Popelish
 
This site may be useful to you:
http://www.tpub.com/neets/book7/27d.htm

--
John Popelish
 
Thanks for the help, link etc. so far!

The PS is for a multiple headphone amp. I'm not worried about the
performance of this design, but the PS should be a prototyp for a
mic-preamp where it really matters.
Samuel
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top