4kW SMPS RFI problem

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Pierre said:
Thanks, I will try.
So you haven't found any problems with that configuration, even with very low output current? Any ripple in the output rails?


No, I did not.
Actually there are only two problems with my design.
1. Lots of switching noise going out of the case through holes, mains cable and input audio signal terminals, what I plan to solve by better case design (holes not longer than 2 cm), mains off-the-shelf input filter (EPCOS) and input audio signal transformers (probably JENSEN).
2. Lack of any power supply protection against overload. Switching mosfets have blown several times yet. I plan to mount small current transformer at the primary winding which (through rectifier, filter and voltage divider) will drive the input of an auxilitary op-amp which will drive the SHUTDOWN pin of the controller.
 
I can't read your files but if you don't have any Y1-caps between primary gnd and secondary gnd I suggest you mount a 2,2nF Y1. That will reduce common mode EMI a lot.

If you don't have a mains input with a mains earth connector you can go all the way up to 4,7nF but if you allready have Y-caps between mains and earth you must not exceed a total capacitance of 4,7nF.
 
Mine don't have any filtering as well, but it is only a prototype, I will go with that after it works well. Is your switching noise perturbating the audio? (is it audible)

About protection, I added a small current sense transformer (from Coilcraft), that is then rectified and filtered, it attacks the SD pin of the SG3525 but I had to disconnect that because it triggered erratically, I have to insert an opamp or comparator before as you say.

I will keep you updated on my findings in that area.
 
Pierre said:
Mine don't have any filtering as well, but it is only a prototype, I will go with that after it works well. Is your switching noise perturbating the audio? (is it audible)

About protection, I added a small current sense transformer (from Coilcraft), that is then rectified and filtered, it attacks the SD pin of the SG3525 but I had to disconnect that because it triggered erratically, I have to insert an opamp or comparator before as you say.

I will keep you updated on my findings in that area.

The amplifier itself, is not picking up interference from the supply. But the whole device is emitting lots of EMI and it interfers with other devices close to it and on the same mains phase and ground, and input terminals.

Originally I also did not plan to put an OP-AMP into the protection circuitry, but as soon as I saw that trigger voltage range (0,6-1V) for SHUTDOWN pin, I decided so.
 
e96mlo said:
I can't read your files but if you don't have any Y1-caps between primary gnd and secondary gnd I suggest you mount a 2,2nF Y1. That will reduce common mode EMI a lot.

If you don't have a mains input with a mains earth connector you can go all the way up to 4,7nF but if you allready have Y-caps between mains and earth you must not exceed a total capacitance of 4,7nF.

first, please explain what are "Y1" caps. I have directly connected primary and secondary ground. Should I break this connection and put that 2,2nF cap between?

Currently I only have two 1uF/400V capacitors at mains AC rectifier (one at the input, one at output). But soon I am going to mount Epcos 10A off-the-shelf mains filter (B84111-A-B110), so I hope this will help reducing RFI and EMI.
 
Disney_SK said:
I have directly connected primary and secondary ground.

Does this mean that you have no galvanic insulation between your primary and secondary side on an off-line (230/115VAC) supply?

Disney_SK said:
Should I break this connection and put that 2,2nF cap between?

If the answer is yes on my question above, then YES - BREAK that connection. Otherwise you could end up killing someone...
 
e96mlo said:


Does this mean that you have no galvanic insulation between your primary and secondary side on an off-line (230/115VAC) supply?



If the answer is yes on my question above, then YES - BREAK that connection. Otherwise you could end up killing someone...
What are you talking about? standard practice in in plenty of smps units is to have prim and secondary grounds directly connected. As a matter of fact separating them can cause extra safety risks as then transformer and everything else has to be desinged double-insulated.

Maybe you mean neutral by ground as you are talking about Y-caps? as between neutral-ground would be a right place for a Y-caps

disneyland, check this for cap explanations:

http://www.justradios.com/safetytips.html
There your can also find pic of typical EMI-filter with Y and X caps and CM-choke.
 
e96mlo said:


Does this mean that you have no galvanic insulation between your primary and secondary side on an off-line (230/115VAC) supply?

NO. I have.
There is pretty well isolated transformer, and the case is firmly connected to mains ground terminal. On the same terminal, a connection to transformer's secondary "zero-point" is made.
Tripath recommends this (along with other anti-EMI tweaks)

If the answer is yes on my question above, then YES - BREAK that connection. Otherwise you could end up killing someone...

As far as I know, I could ONLY kill someone if whole metal case would NOT be connected to mains ground terminal. Primary-to-secondary ground connections have nothing to do with safety.
Please correct me if I am wrong.


mzzj said:


http://www.justradios.com/safetytips.html
There your can also find pic of typical EMI-filter with Y and X caps and CM-choke.

I took a short look at your link and also on the label on this epcos filter, and it's clear to me.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.