Capacitors : the embarasting choice... :xeye:

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
As some may have seen, I'm currently improving my bridged LM4780 amplifier.

I'm mainly working on the PSU, as I noticed quite fast that it made the biggest differences.

The fact is I currently only have 5*1500mfd per rail, so 7500mfd all in all.

Panasonic FC capacitors are only available in small values, and they do cost a lot when you use many of them. :hot:

I have read some things about PSU design over the Internet, and came with this : you need ~ 10000mfd per rail for 10A of current at the terminals of the PSU. So, the LM4780 can pass (2*) 10A max. I have two of them so say I need 20000mfd at least, maybe twice as I'm running bridged.

Little problem : 10000mfd caps can cost from 6€ to 30€ !

I have found some interesting products :

- 10000mfd 50V with 5A ripple @ 100hz priced 8€ (panasonic TSUP)
- 10000mfd 40V with 7A _________________ 6€ (epcos B41303)
- 15000mfd 40V ____8.4A _______________ 8.5€ (epcos B41303)

Looking at this, I find that the best choice would be 4*10000mfd.

It allows 14A ripple, and enough juice for high current without hurting my bank account too much.

But when looking at the 40V rating it may be frightening. Some say it isn't wise to use 40V caps on 35V rails ( no load, my rails are 34.9 - 34.9). Caps allow surges up to 1.15*the nominal rating.

Lifetime is 2000h given @ 85°C but 100'000h @ 40°C, so lifetime may not be an issue in either cases.

To complicate things more, last versions of Carlosfm PSU use 4700mfd caps before the big ones.
Is it possible to remove those without affecting the snubber values ? Or should I use 4700mfd + 15000mfd on each rail to have the 20000mfd ?

Last thing : should I save the Panasonic FC for other chip amps, or keep them in the PSU anyway ? (considering their price)

What would you do if you where me ?

Sorry for this noobie stuff, but I don't have deep pockets :D (I'm student lol)
 
youyoung21147 said:
To complicate things more, last versions of Carlosfm PSU use 4700mfd caps before the big ones.
Is it possible to remove those without affecting the snubber values ? Or should I use 4700mfd + 15000mfd on each rail to have the 20000mfd ?

That 4,700uf cap before the series 1R resistors can be smaller, like 2,200uF or even your 1,500uF caps.
After the series resistors, I recommend around 20,000uF minimum.
This is per rail, single PSU for a stereo amp.
 
BTW thanks tweeker for the link. I where at school when writing the last post and had to do it when nobody was watching over me :D

I've thought about the PSU and I can see the role of the 1R resistor : it reduces the ripple caused by the capacitors calling current and that may damage them.

As the caps I plan to use have more ripple capacity ( 14 ~ 16A) than my transformer allows (6.5A), I think that this resistor can be suppressed, or is it another reason why it is here ?

But something I don't really understand is the role of the 4700mfd cap. Is it here to filter high harmonics caused by the rectifiers ?

Concerning the voltage rating of the caps, I found nothing about it in the link to ESP. Is it safe to use the 40V caps ? I remesured my rails today, they are at +-34.5V when amp is idle. I will also use 100V transil diodes from V+ to V- in the final PSU.
 
All of the caps perform filtering / energy storage. Exceeding ripple current rating just isnt an issue with even 1 pair of 10,000uF caps in a music amplifier, industrial uses might need a closer look.

Dont confuse ripple current with output current, ripple is the AC waveform imposed on your DC by the incomplete smoothing. The elaborate PSU schemes you see here are about sound quality.
 
OK I'll follow your advice :cool:

As you said, it's not for the few € involved, but I rather wonder about the resistor because in my mind resistance always means losses and lower damping factor.

As you'll probably answer me, I'll give it a try, with and without the res/caps.

Could you explain me the purpose of the 4700mfd cap before the resistors ? I'm reading the 50+ pages of the high-cap PSU thread, and am only at the 38th :(

Also, I'll give a try to the 100nf before the rectifiers, it won't harm at all anyway I think.

Just good for ordering new components :D
 
youyoung21147 said:
Could you explain me the purpose of the 4700mfd cap before the resistors ? I'm reading the 50+ pages of the high-cap PSU thread, and am only at the 38th :(

This is a CRC PSU, you'll have lower ripple if you use that cap and resistor.
Also, the peak at power-on is lower and less stress on the diodes, as the first cap is smaller and the other big ones are charged a little slower because of the small resistance.
No worries about losses, the resistor is small and the high capacitance after it will take care of the transients.
 
OK, I see.

So the main aim of a CRC PSU is to protect components from current surges, overload, etc...

But if my trafo is small (300VA 25-0-25V, 6A ), my rectifier bridge is double ( 8 MSR1560 diodes, 15A 600V, ultra soft recovery) and my caps can handle 17A of ripple current, is there really a problem with current ripple ?

I assume the dynamic response will be a bit improved without the resistor won't it ?
 
OK I see the problem with the price.

My PSU is in a separated metal housing, but it may perturbate the PSU itself. Does a choke radiates like a trafo, through the center ?

Also, the use of ferrite beneath the coil increases the inductance very much, like in a trafo. But ferrites tend to saturate above a given current.

Is there a formula for calculating the max intensity the ferrrite can handle ? I didn't find any on the net :(
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.