DIYHiFi.org

Hi,

I was working on PSU for LM3886 based amplifier and which can also be used with similar Main Audio amplifiers.

after extensive google search I came across many many articles, schematics, documents from various DIY authors and application notes, some talked about differences in 60Hz and 50Hz Mains, transformer capacitance, and inductance all things taken into consideration while designing PSU for Audio Amps.

I used CarlosFM PSU schematic, and National Application Note 1849
and here is the outcome with some alteration and combination.

I need help to make PSU with excellent noise reduction.
Please suggest how can I make this PSU better, using appropriate values, types.

Will it make a difference :

1 . using 2 x 12000uf instead of 10,000 x 2 and 4700uf.

2. Use Metal Film Resistor for Bleeder (R1,R2,R3.R4) and Snubber R5 and R6.

3. Use of Ceramic Capacitor after bridge rectifier (before 12,000uf Capacitor)

4. what would be appropriate value for C8 ( 0.01uf, or 0.33uf )

I don't understand why the two "bleeder" resistors per rail.
You just need one - the first one.
The value of R5 and R6 on the snubber (470R) is way too high. Should be lower than 1R.
You need some series resistance for more effective filtering.
C1 and C2 should have a lower salue (say, up to 4,700uF) or else you need a small resistor value right after the rectifier diodes. A value between 0.22R and 0.47R will do.
Between the two big caps you can insert another series resistor, up to 1R.

_________________
Carlos Filipe

"Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius - and a lot of courage - to move in the opposite direction." Albert Einstein

yes, RC filtering is better at removing source hum and other mains borne interference than simply increasing the smoothing capacitance.
RCRC is better again.

Consider adding a small inductor instead of the small series resistor.
LCRC or RCLC
50turns of 0.6mm diam enameled copper wound on a 7mm bobbin (pencil) will do wonders for attenuating RF.

_________________
regards
Andrew T.

This PS worked VERY well for me.

Regards,
Dan

Did you run a power amp with only 4.7 ~ 100uF capacitance on the PSU? Nothing added elsewhere?

_________________
Carlos Filipe

"Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius - and a lot of courage - to move in the opposite direction." Albert Einstein

I don't understand why the two "bleeder" resistors per rail.
You just need one - the first one.

Reading many docs online, I read bleeder should be close to capacitor, in my first design I had one Bleeder then later added one more. will use only one in circuit. (thanks)

The value of R5 and R6 on the snubber (470R) is way too high. Should be lower than 1R.
You need some series resistance for more effective filtering.

This calculated according to : Calculating Optimum Snubbers - Hagerman Technology

C1 and C2 should have a lower salue (say, up to 4,700uF) or else you need a small resistor value right after the rectifier diodes. A value between 0.22R and 0.47R will do.
Between the two big caps you can insert another series resistor, up to 1R.

Since PCB is already in production cannot change this, will incorporate resistor or Inductor in next version.

Values I choose is 4700uf /63V and 15000uf /63V, not yet determined the Snubber values.
(I am not that good at Maths )

The idea is to make PSU which would be suitable for most Power Amps
I know each Power Amp will have its own characteristics, and requirements but I am trying make PSU for very Budget limited Audio Projects.

How about adding LC filer (EMI/RF) on mains side (Primary of transformer) will that make big difference ?

Mains => EMI Filter => Transformer

Thanks
suds

Idea is to use Metal film resistor but max 2Watt is only available here, so using 2Watt Bleeder at each capacitor, making it 4 Watt total.

I feel Metal Film / Metal oxide resistor are better that carbon film, Please correct me if its not true..

( Sorry - missed this out in earlier reply)

Suds

_________________
Carlos Filipe

"Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius - and a lot of courage - to move in the opposite direction." Albert Einstein

Suds, if that´s what you want, you need to double that resistor value. Two 4.7k 2W resistors in parallel makes one 2.35k 4W.

Thanks
since both values were same, ignored on calculations !

But what about snubber calculations, I am confused..
as per Hagerman document, snubber values depend on mains frequency and transformer inductance.. how can I choose optimal values ?

I used Jim's paper as a guide and tested every transformer I had, I came up with the values of 330ohms and .022uf, they worked well for everyone of them. So now I just use those values for everything. I use a carbon comp for the resistor and a polypropylene for the cap. Put them in series and put that network across each secondary. It can make a big difference.

My low voltage PS of choice these days is to use pi filter (cap, choke, cap) after the bridge, with a small first cap. If the load is highly variable I put a bleeder across the output so I'm putting several watts through the supply no matter what.

This behaves in between a cap input filter and a choke input filter. By careful choice of values you can get low current spikes (MUCH lower than a bridge to high value cap), very good transient recovery time and low ripple. Its bigger, more expensive, but usually way better sonically.

The big drawback is there is no simple formula for calculating the values. I usually do several iterations in a simulator to get things to work out well. The actual builds have matched the simulations quite well.

John S.

But most suggest to use Metal film resistors of 1% or 0.1% , has any one checked the differences ? does it make major difference ?

It really doesn't matter, the carbon comps have a slightly lower inductance so will work better at very high frequencies, but most of the affect of the circuit is at frequencies less than 1MHz so there is not much of a difference. The accuracy of the resistor is completely unimportant, you are not trying to build a precisely tuned circuit but add very broad band damping. The resistor itself is what does the job, the cap is just there to prevent a large DC current from flowing through the resistor and frying it.

John S.

John,

Not to be too dense here (heh), but does the snubber go before or after the bridge or rectifier diodes? I'm guessing after, but assumptions...

Snubber goes after the bridge rectifier..

I have come across very few circuits where snubber is placed before Electrolytic filter capacitors. and in most circuits snubber is places near wire pad of load. so I suppose snubber should be near to load..

-suds