Star grounding in PSU

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi everybody,

This is my first really serious project and I'm trying to validate what I have done so far.

I'm currently designing a PSU for a tripath based amp (class-T). I need 3 different power supplies. The mosfet supply: -60V, 0V, +60V; the amp logic: 0V, 5V, and finaly the soft start: 0V, 12V.

I would like to have a separate ground for the high power/volatge supply and for the low power/voltage.

After many hours reading different threads on this site and on various external places, I think I have understood something. Here is the solution I came with.

1- 60V, 0V, 60V supply. This one has a separate star ground point. The secondary transformer center taps, the rectifiers and the capacitor bank are all wired directly to it. The capacitor bank would be built using 14AWG solid copper wires. The primary would be connected to the main star ground.

2- 0V, 5V and 0V, 12V supplies. Thoses will be built on pcbs with ground plane. Both will connect to the main star ground. Respective transformers and ground planes would be connected to the main star ground.

3- The separate star gound (-60, 0, +60) would be isolated from the casing. The main one will be connected directly to the case. A connection will be made between the two star grounds using a resistor and a capacitor in parallel. I saw a design that used a pair of diode in parallel wih the resistor and capacitor. Why?

4- The filtered main lines will be connected as follow. The live wire will be connected to the different transformers (throught soft start for the -60,0,+60). The neutral and earth will be connected to the main star ground.

Here is a drawing I did to show this. Note that I plan to have shielded pairs at the output of each power supply. (maybe not for the Soft Start PSU). These shields would be connected to the main star ground. The idea is to have paired wires with opposite current flow so the EMI, RFI would be minimized. Am I right?

All bypass capacitors haven't been included for clarity.

Thanks for looking at it and to share your comments.
 

Attachments

  • ps.jpg
    ps.jpg
    69 KB · Views: 856
I didn't pay much attention when drawing the bridge rectifiers. The idea was more to validate the star grounding configuration. I will make sure they are wired correctly at construction time.

For the main line, would you suggest that the neutral be connected to the primary of the transformer as the live and the earth to the star ground?

thanks
 
Live and neutral connect to the transformer primary, earth goes to the case. Your star point betwen your caps as I mentioned earlier should also have a wire off to the chassis point.

Looking at your diagram again I see now that you have been too adventurous with what you are trying to ground. I don't wish to sound offensive, but I really think it would be wise to do some more reading up before you build something if you got that wrong :)
 
I don't plan to build anything before a couple of weeks at least. I'm looking for information at this point and I just realised 2 days ago that star grounds existed. Yes, I agree, I still have to get more information, that's why I did ask.

Regarding the neutral issue, I think I have been confused by the tripath application note. English not being my mother tongue, I sometime have to read stuff more than once to get everything right. But now this is clear to me.

why should I connect the "caps star ground" to casing if it is already conected to the "earth star ground" through the parallel resistor/capacitor/diodes? Also, what is the purpose of having it in the center of the caps? Wires lenght?
 
10 R between the two grounds is too much.. in addition you have a cap shorting both at high freq, inducing digital noise to your analog part.. an inductor 100uH should be better.. i suggest you used no bars but seperate cabling to the star ground instead (the inductance in series from the bar for the last cap is 10-15 times that of the first).... you should build the psu star on the chassis, connecting there the ground return fom the board. you should plug the mains ground only for safety, although youd be better off without it...
 
ok listen.. the schematic on rod elliots site show a resistor||to a cap connecting the mains ground to the amps star ground... its NOT about connecting an analog to the digital ground.. impedance of a coil rises with frequency, so high freqs flowing through the digital will be cut out from the analog part due to the high impedance node between the analog and the digital ground.. and i am talking about a low dcr rf choke (or similar...)..
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.