Guru
Joined: 09/08/2007
Location: AustraliaPosts: 4406
| Posted: 06:14pm 15 Sep 2017 |
I have just received a couple of EGS002 boards from China.
Have been feeling a bit left out of it all, and cannot let you guys have all the fun...
Fired up one EGS002 only just now, still have a great deal to learn about these, but a few initial observations about current limit generally.
The Ifb (pin1) appears to do two things, it disables the two IR2110 drivers, which will be a fairly fast and violent on/off action.
It also connects to the EG8010 chip via R27 which is not fitted to either of my boards.
There is also no value for R29 on the schematic, and I notice from many pictures I have found of the EGS002 board on the internet, R27 always seems to be missing from all the boards.
Shorting out R27 has the effect of allowing the EGS002 board to soft start once the Ifb input has been removed. Ifb then has a fast off (via shutting down the IR2110 drivers) and a soft slow recovery similar to a normal power up. That has to be good.
I cannot imagine why this soft recovery has been deliberately disconnected by the manufacturer, but it appears that it has now been a long standing production change.
Anyhow, I think linking R27 and driving Ifb from an SCR triggered by a current transformer as previously suggested here on the Forum, may be the best way to go.
The Tfb input worries me a bit, as it may not offer a fast enough shutdown. Using just Tfb does not shut down the IR2110 drivers which I think is a necessity.
Over temperature shutdown need not be fast.
Shutting off the IR2110 drivers will be very fast indeed but I don't trust the EG8010 board for fast turn off, but the soft start it provides is certainly very nice.
We need to think about the rate of current rise through the mosfets under fault conditions. If both upper and lower mosfets turn on simultaneously, its obviously all over in a giant flash and a bang.
Insufficient dead time is unlikely to be the problem there.
We may also get spurious simultaneous turn on from noise, especially if the EG8010 does not get very clean stable power, or there are serious grounding issues in the whole system.
That all needs to be thought through very carefully.
The other issue is high fault current from excessive loading on the output.
If we crowbar the output, the current rise through the mosfets will not be instantaneous, but will be checked by any series inductance.
That is the value of fitting a large serious non saturating choke immediately after the mosfets. From the inductance of that choke, and the applied dc voltage, we can come up with a figure of amps rise per microsecond, under worst case dead shorted output.
Assuming a nominal supply of 48v and a nominal 48uH choke, the maximum rate of fault current rise can not exceed 1 amp per microsecond. And may well be slower.
So current limit needs not be super fast, say a few microseconds or tens of microseconds. We certainly don't need to stop the fun in nanoseconds.
Even the simplest over current shut down HARDWARE should be well up to that, assuming there is a suitable non saturating series choke fitted !
But software, depending how it has been written may fall sadly short. An over temperature shutdown has no natural urgency about it, it may only be tested once each program loop. In fact it could be worse, it may need to be tested several times before it decides its a real over temperature event.
Looking at that last picture in the previous post, Tfb appears to need about half a mains cycle before it shuts down, maybe 10mS ? That is worryingly slow.
Cheers, Tony.