Pi filters

JaredC79

Banned
2015-09-02 4:32 am
NJ
I have seen a few different ways this has been done. I am using 2 .1 ohm resistors on my power supply. Nelson Pass has done this by putting them in series between the middle of the cap bank per rail serial to the rails polarity - so if it has 10 caps on the positive rail they would be between c5 and c6 on the positive side and running parallel before c6) I have also seen where one resistor was used in the middle of the cap bank on the rails polarity and then the second resistor bridging both the pos and neg after the last cap in the bank. What is the better of the 2 options I need to finish my soft start and trigger system before I can test my amp and a purchased amp module but I figured while I am waiting on parts I may as well build a bunch of power supplies - I can always leave additional traces on them but wondering the differences between the 2 and imagine putting a resistor across the last cap would also limit the capacitance?
 

Mark Johnson

Member
Paid Member
2011-05-27 3:27 pm
Silicon Valley
What does your intuition tell you?

What does circuit simulation with LTSPICE tell you?

What does your budget tell you? (can you afford 6 capacitors and 5 resistors?)

What does your pencil and paper calculation tell you?

Who cares what other people tell you? They might be wrong and you have no way of discerning who is right (if any) and who is wrong.
 

JaredC79

Banned
2015-09-02 4:32 am
NJ
What does your intuition tell you?

What does circuit simulation with LTSPICE tell you?

What does your budget tell you? (can you afford 6 capacitors and 5 resistors?)

What does your pencil and paper calculation tell you?

Who cares what other people tell you? They might be wrong and you have no way of discerning who is right (if any) and who is wrong.

I have not run it in LT spice - i don't know how to use that program yet nor have the time to really play with it at the moment. I do not have all my components or the equivalents loaded up in it yet and it is quite possible that the l will not have to finish everything myself but I would still like to know as much about the topic as possible. This is the last phase of a long product development cycle and the original engineer who was to handle the amps has been flakey bit has just come back into the picture - regardless I would like to hear from others who have tried both before I make my final decision. I should have the rest of my parts today to finish the soft start and I guess there is no harm building a board with extra holes to test both designs on a few different amp modules so I can see if there is any audible differences between designs and then just cut the traces for that one board and run the rest. I was just trying to get as much done as possible in down time between deliveries. I am new to amp design but for some reason am totally addicted to it now and am more fascinated with it than any other aspect of the project but then again it is new to me so maybe that is the reason.

Luckily, I do not have a budget. Basically I am building a total of 19 amps for a home theater. For this PSU I am using a total of 24 Nichicon Golds 2200UF - 12 per rail with full wave rectification from Schottky diodes. All resistors are thin film 1% tolerance. This is the PSU for a 300w 500W and 800W @ 4 ohm hybrid chip/SS amp with laterals and the PSU was inspired by a psu i purchased but I am adding other things to this design to suit my needs as they will be used for powered speakers with active crossing and tri-ampflication all inside the cabs with unique controls. Similar implementation will be used on a rack amp but with different features as well as powered speakers with passive crossing.

I would not know how to do the math on this implementation but my intuition tells me that running a resistor over the last cap on each rail will limit the "bursting" slightly when needed and that is is best to do it as Nelson Pass did running them in parallel between c6 and c7, connecting the resistors serially since the filtering is knocking off the hf and then has additional caps are there to help with bursts to improve dynamics vs a design with resistors crossing the last cap on each rail.

Would the forum agree with this method of thinking?
 
If your amp's a classAB then the load induced noise is probably the major concern. Which means the more capacitance closest to the amp OPS, the better.

I'd ditch the 1ohm resistors and use inductors wound on iron powder toroidal cores. 25mm diameter probably will be sufficient.

For 300W I'll stick my neck out and remark that you're not using enough capacitance overall - from what you write you'll have about 25,000uF per rail.
 
Power supplies have to be designed. Ideally this means doing some calculations based on understanding. Failing that, do simulations. Failing that, do experiments. Failing that, ask someone else what they did (and then carefully apply that to your own different situation - which requires calculations).

Intuition has to be informed by understanding.

There is lots on the web, or in books, on PSU design.
 

JaredC79

Banned
2015-09-02 4:32 am
NJ
The idea of the small value caps and pi filter was from Holton Precision. 12x2200 is 26,400 per rail. His module in 500 and 800 watt 4 ohms in unconditionally stable and 2 ohms stable with as well. He is using all discrete circuitry and 78v rails. I imagined that something similar would push the same wattage as the the driver and input state is being change.