Project Journals > Oztules
testing the egs002 inverter board
oztules:
I have been far too darn busy for the last few weeks... and the salt water RO machine has taken me longer to do the electronics and computer programming for.. it is always too long between playing with the programming, and I forget all the syntax, and spend too much time wondering why it wont work as I wanted.. or even compile etc... wish I was better at this stuff.
Now have a arduino nano with 30 wires coming away from it... then wondering where they all go .....etc. etc.
I will get to it...if you don't get there first, then you can tell me....
Where does the time go....
.................oztules
oztules:
Finally got some time to play (* albeit small).
Wound 25 turns onto a 3kw aerosharp torroid to get a 48v transformer to drive a bit harder... looks like this:
Then a 2100w stove element , and we get this:
Voltage was this
Current this:
test unit looks like this:
Ok, things we found out.
The unit ran quite cool considering 40 amps over only 2 fets per bridge for 10 mins. :)
The voltage was controllable, very smooth. :)
The current control worked.... but there was a problem. :o When I used a welding transformer ( EI type) the current feedback seemed to work fine, but the torroid blew the thing to bits when I then turned the pot to make the current sense more acute and make the current regulation start....instant fusion reaction actually, and fire bombed the fets.... the smoke was well and truly released. :(
So it looks like another system of current shut off is necessary for big torroids... thats the only reason I can see for the blow up..... other ideas welcome.
It seems that the shut down current control must change the frequency or some other rapid wave change problem that the transformer reacts badly too... don't know, but will try to find a reason.... more fets must die in the name of progress I think..... :-[
Torroids are very good, but take no prisoners... they are electrically stiff.
Apart from that, the little test was impressive to say the least. A 3 second soft start was good, the current regulation was good, the voltage rock steady throughout, mostly happy... but would be ecstatic if the current control had not have blown 8 more fets to hell.... with more to come I can bet.
Anyone else been playing with these things yet?
............oztules
RFburns:
Me think's its pin nine set to turning on the bridge resulting in FET smoke (or maybe for a torroid the dead time needs to be extended?). I have all the bits but havn't done anything yet (to hot to go out to the shed on days off) From the data sheet
"Pin IFB measures the output load current for overcurrent protection detection. As the current sampling and feedback circuit shown in figure 8.1a, pin IFB’s reference peak voltage is 0.5V and overcurrent detection time is 600mS. If current is higher than inverter’s lad current by some reason, EG8010 will set the electrical level of SPWMOUT1 to SPWMOUT4 at “0” or “1” and shut down all
power MOSFET to decrease the voltage to zero depending on pin (9)PWMTYP’s setting. This function mainly protects power MOSFET and the load. Sixteen seconds after overcurrent protection
activates, EG8010 will turn on power MOSFET for 100ms to re-determine load current. If overcurrent issue still exists, EG8010 will repeat the process above every sixteen seconds. If EG8010
runs regularly for more than one minute, it will zero the counter of overcurrent. However, if EG8010 does not function regularly after five 16-second cycle, it will complete turn off the output of SPWM unit. It needs a hard reset to start again. If in some scenarios the starting current is relatively high and it takes longer timeto start, which overcurrent protection is not suitable, Pin IFB can beconnected to the ground."
9 PWMTYP I
PWM type select
“0”: positive polarity PWM type?MOSFET on when SPWMOUT is high
“1” positive polarity PWM type?MOSFET on when SPWMOUT is low
Best configuring pin according to driver device and referring to the typical
application schematic below, otherwise will result in both sides of MOS
tubes conducting at the same time.
and dead time could be an issue with a stiff inductor
Pin DT1, DT0 controls the dead time. Dead time control is one of the important characteristics
of power MOSFET. Lack of enough dead time will result in the damage of MOSFET due to
conduction. If the dead time is too long, it will lead to distortion of waveform and overheating of
MOSFET. Figure 8.5a is EG8010’s four dead time control settings.“00”= 300nS. “01”= 500nS.
“10”= 1uS. “11”= 1.5uS.
Just my 2c worth. Stu
oztules:
Thanks Stu.
I have pondered those bits of the data sheet for some time.
I will use the big EI tranny again, to check if this is really the problem.
It looks like it is unipolar operation on my boards.... which is another problem with the EG002 boards for me.
This means the high switch on one side is working in high frequency, and one side is 50 hz...... I want to switch the high side on both bridges at 50hz, and this will make the driver much simpler.
As it is, the 50hz high side using the current pump is sloppy, and bleeds off too quickly. It cant maintain current for long enough unless you use a bigger cap on the charge pump..
Add this to the current regulation with torroids, and I think it will be better to just buy the chips for a few dollars and make my own board.
So I want 50hz high side switching, a small circuit that will unambiguously shut the thing off, or maybe a nano arduino to monitor, and make decisions on time and action.
I want isolated high side power supply..... so really there is nothing on the commercial offering I need to use ... apart from the crystal and chip..... don't want their drivers either.
I will persist with the tests on these boards, but feel i will probably move on at some stage and go it alone.
Hope the weather cools so you can do some tests of your own.
..............oztules
oztules:
The big EI tranny works fine........ with the current control ..... although only barely surviving in fact.... lots of noise on current shut off... and the scope gave the reasons why.
So it appears that the IFB pin on the board goes to not just the ifb pin on the chip... which would shut it down properly, but also to two differential amplifiers. These intercept the signal from the CT, and then process it. One signal goes to the inhibit of the driver chips, and one to the inhibit of the main chip.
What this really means, is there is a short period where three things are competing for the right to shut the thing down on an over current event.... this leads to a train of small pulses from one of them, that causes the disasters with the torroid.... which one....... I don't care! ... the system has to go.....
I don't know how others have circumvented this behavior, but I decided to use the temp input as the over load input instead, and get rid of the other offending signals completely... ie effectively short the ifb signal pin out, and instead, use the CT output and send it to the temp input.
The good thing about the temp input, is the moment it goes above 4.3v, it shuts down instantly... no train pulses to be seen.. so it causes no problems with either the torroid or the big EI transformer...... the other beauty of it, is that it is programmed to restart with a 3 second soft start...every time, not like the unit does form low voltage etc etc.... this suits the torroid fine..... and it also means you dont make changes to the commercial board, they stay interchangeable and complete.
I am also using this input for the on/off, as it never fails to soft start.
Cooling will be done by my system anyway, so no loss of usability, and you can still use this input as the temperature shut down as well if you so choose to have a doomsday heat shut off.
Currently designing a from scratch board, with isolated driver voltages for the high side, and so that also means pwm control of the DC side.... safer from spikes in the battery line, and also the egs8010 chip is only a few dollars anyway.... so the complete boards will be very cheap, with opto drivers for high and low.. this means a blow up will only take out the optos and fets.... never the board..... I like that sort of thing... especially for testing.
These things are unipolar in the type i bought, but the 50hz is on both a high side switch, and a low side switch.... not what I expected.
The high frequency high side switch gets warmer than the 50hz high side switch,( as expected ), but for this little 3kw unit, will be fine.. little heat even at 2-2.5kw, and thats with lousy fet drivers driving two fets per leg running 40-50 amps or so The commercial driver chips are not up to the 50hz, as they sag badly with the pump system on the high side drivers. They seem to be only able to handle 2 fets of the 4110 type, as the capacitance is quite high on the gates for hex fets.... not sure I'm game to use 3 or 6 fets per leg on these.
The other thing is that measuring newer PJ boards, they are bipolar outputs, so the opto's are driving the high side at 20khz just fine... I'm sure on the original one I played with, it was unipolar, and the high side was only 50 hz... maybe I was wrong.....
( unipolar runs 20khz for the spwm modulated legs and fundamental frequency for the other two ( 50 hz square waves ), the bipolar uses spwm modulation on all four )
...............oztules
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version