Grounding a power supply and filtering.

[si2030]

Change of Tac..

Hi MrAl others,

I thought rather than muddle through with the IR2183 chip I simply ordered a different chip that works the same... in this case NCP5181. I have attached the datasheet.

Initially all I want to do is drive a half bridge with a high frequency transformer connected. First with 40 volts and get a transformed voltage of ~4 volts - then with mains voltage via a variac. Thats been the plan however I struck a snag in that I was unable to make the IR2183 and I still dont know..

I have decided to build up a board with all the basic components on it for testing the above.

Here it is:

View attachment 62974

I have included the design and would like to know if this is correct before building it.... any obvious mistakes?

From what I have read so far etc it seems sound.

Simon

 

[MrAl]

Hi again,


Hey you made it back! I thought maybe you disappeared or something.

So you are using another chip, a less confusing one at that which is always nice

The connections look right, but the way you confirm this is to run the simulation and check each drive signal. You should see that one transistor turns off and then after some small delay the other transistor turns on, and then they switch roles. You should do this yourself to get a feel for it really, because that is what you should do with the real circuit too once you get that far.

 

[si2030]

Hi MrAl

I have already found a model for the NCP5181 and run the sim but I will post it tomorrow afternoon for you to see.

 

[si2030]

Hi MrAl and others,

OK I have adjusted the spice simulation to closely resemble the circuit I created a couple of days ago.

Here is a picture of the spice sim plus the traces.

View attachment 63019

I note that once the output had settled down it ranged from ~41V to ~47V. The thing is the LT1083 can only handle an input to output differential of 30V maximum. The fact that the rectified low voltage is 41-47V should not in theory matter as its tied to the output voltage by a matter of a few volts via the error amp.

Even if I move the voltage adjustment suddenly it should still not matter as long as I set the upper limit of the linear regulator to be under 30volts at maximum..

My question here... does it really matter if the prereg without any error amp input - just running flat out is at 41-47Volts? That the error amp will force it never to be in this region anyway.. but instead it will only reach a maximum of ~2-3 volts above the maximum output voltage of the LT1083 set in this case to ~29volts... Am I correct in this assumption?

Simon

 

[MrAl]

Hello again,

Actually you must have the output go high enough to supply the voltage to the linear regulator that will be required to meet the maximum output spec of the entire power supply. So if your power supply has to put out 20 volts, then your switcher has to be able to put out (roughly) 23 volts, but if you prefer you can have the switcher go all the way up to 30 volts. This has to work at low input line too, so that might be 15 percent less of the usual nominal line input voltage, so that means you always need the capability of being able to get more output from the switcher than you really need normally.

The catch of course is that at high line you have a lot of extra voltage that you have to deal with. The whole ball game then becomes that of control. The circuit has to be able to control the output no matter what. This means of course a slow start circuit is required (which you already know) and the time constant has to be long enough to allow the rest of the circuit to catch up. This also means some sort of secondary control might also be a good idea, to catch the output if it starts to go dangerously high...like above 25 volts or something. This could come in the form of another error amplifier or some other trick to force a faster response to cut back the pulse width quickly.

I've seen higher power converters take out very expensive computer system power supplies because they didnt have this kind of safety feature, although it does really depend on the design whether or not it needs it or not. One way to tell is to actually try it in simulation by increasing the input line voltage quickly or removing the load quickly and see what the switcher does. If the switcher EVER forces more than 30v across the linear regulator then there's going to be a problem. You'll have to correct that by design sooner or later.