guess who bought a power jack inverter

[oztules]

Correct, it is a capacitive drive with the 2k7 in parallel with the capacitor... to hold the voltage high, but high enough in resistance to not let the 60v do too much damage upstream...... so the big cap to turn it on, the resistor to hold it on. There is 20k to ground at the fet gate-source.

The 60v surge is immaterial to me, although the transistors could take it, as they are 2amps and over 60v... but it is only the boards actual integrity i worry about... but yes probably survive. A zenner on the driver side as a shunt will stop it anyway.

There is a 339 further upstream, but it has never failed even when the transistors have toasted and exploded.

Yes worth a try.


.....oztules

[andymack]

Without having really seen the FET board. 

Lower the R gate-drive to 10R 1/8th watt  (sacrificial)  mounted high to make it easier to replace, and use a tranzorb.

[andymack]

if mode of failure is over current;  monitor current by what ever means (hall effect sensor, shunt) and kill the drive at the LM339 output.  (preferably locked requiring reset)

[oztules]

Thanks Andy, the transorb is installed on a test unit... we'll see how much blows up .

Over current will be terribly difficult, as this thing can supposedly pull 60kw surges... and still be fine.
But a false switch or something else it does not like will send it to hell. Usually it is me doing something wrong without thinking. All easy to fix except the board charcoal makes some of them useless really.

So 2 x 16v transorbs are now on the fet board. Just tested, there is no interference at low power.... now we will see if big loads spike enough to cause havoc. I have left the 49r resistors there at this time. Will lift them up later.

In short,........ I like the idea.

The cap resistor thing, did not make much difference to the wave form, but it attenuated it too much to bother trying to go on with it. This is less invasive to the normal operation of the system I think.

..............oztules

[oztules]

Well it seems to work.

I built a fet board up with cheap chinese fets that were no where near spec, and decided we could afford to kill them off........ so i did.

It was impressive with 12 fets blown apart very quickly....... this time there was no sign of stress in the totem pole transistors.... one died quietly, but no cooking or anythinjg else.
I suspect it copped a few millisecs of pain and then died, but did not cause any catastrophic damage....

Rebuilt the board again, and replaced the single transistor on the totem pole.... it was on the side of the fets on both high and low  that blew. The high side drivers did not suffer any problems..

So a quick rebuild, replace all the resistors and fets on those two banks...... and then.... pushed it back into service as a transformer and card set on the bench and nothing else. Ran it for 20 minutes at 2kw... and even with poor fets, the heat sinks did not get above skin temp... so it must be running alright, and the wave form must be fairly decent.

Also found out that 2 liters of water in a 2kw jug heats up 45 degrees in about 4 mins.... must do the math and see if thats close .... or the amp meter I used was out.

The unit was doing 225v@8.5A... so was probably a 2400w element running slightly low voltage. Board is a 230v board, and the regulation is fair, and can be improved upon I think. Unloaded was 228v.. so sagged 3v@2kw..... not too bad I guess.

Tomorrow we over load the system big time to see how the lousy fets hold up under 6-8kw... have no idea what they really are.. the print is IRFB4110... but they test out very much worse than that.... instead of 3.5-4.7mohm... they are 7-15mohm... every one is different.

Hopefully it will blow up, but leave the totems unburnt in any way like todays test..... then I will get some confidence that this may be a way to protect the fibreglass

Yes Tom..... were just animals here....

Oh, and 1 transorb 16v shorted so that other bits did not incinerate.

..............oztules

[rossw]

Also found out that 2 liters of water in a 2kw jug heats up 45 degrees in about 4 mins.... must do the math and see if thats close .... or the amp meter I used was out.

06:30:32-7/05/15| <RossW> !kwheat 2 2
06:30:33-7/05/15| <RossBot> 2.00 Kilowatts input will heat water at 2.00 litres/min by 14.34 deg C

That should have raised the temperature 57 degrees in 4 minutes, or 43 degrees in 3 minutes.
So somethings not QUITE right there (but it's still close enough for government work!)

[off the wall]

I'm currently running on a 24v 10kW solar bank - my own design solar controller into 320 Ah lithium batteries (coming up to 3 years old), 3000W latronics, and just replaced a deceased chinese non transformer inverter with a 24v 8000W powerjack (i'll de-rate to 3000W)

Next year we'll be building a new shed / house on our property and it'll be a 48v 10kW solar, 320Ah Li system , hopefully with 2 x 6kW PJ homebuilt inverters.

You don't need to derate the 8000 to 3000 - it handles 4kW to 4.5 but I have just had significant success adding to my 24V unit a 3rd transformer from an ex 5000LF unit.



For some reason the extra transformer driven by the other inverter produced a lower voltage so I had to add some extra windings to match.

I used the original control board thinking that the temperature control on that would drive the cooling fan, but it's not working so I will take it out and possibly drive the fan in parallel with the fan in the other unit, although the driver transistors for that might not like the extra current.

Anyway, now I don't have to worry about turning the kettle on whilst the washing machine is running.

Best wishes

OTW

[frackers]

Tomorrow we over load the system big time to see how the lousy fets hold up under 6-8kw... have no idea what they really are.. the print is IRFB4110... but they test out very much worse than that.... instead of 3.5-4.7mohm... they are 7-15mohm... every one is different.

I have a couple of hundred fets to sort through to repair a mod-sine inverter that is used for dump load and as a spare so interested in how you measure them.

I was going to use my bench PSU, current limit to a couple of amps and measure source/drain voltage with a suitable volts on the gate to push them hard on. Measure the volts with trusty mv meter and see what occurs!

Got a card left by the postie yesterday to arrange drop off of a 'large parcel' so sounds like my boards have arrived. I'll know when I get home tonight

[oztules]

Frackers.

Current regulated power supply 5A
separate power supply to provide the 12v for the gate

Measure against the fet body for the mv drop across source drain pins.... or the tab and the source right against the body.
R=E/I

If you use 10amps it is instant no calc really.... ie 47mv=4.7mohm.... so .047/10=.0047 ohms

For the voltage Haven't done it at this time.... but the fakes all seem to have the voltage in the range. I believe that for say under 60v, then a 220k resistor in series with the drain, turn fet off ( short gate to source ), and the voltage will be clipped by the mosfet.... use high impedance volt meter.... thats the break down voltage.

For higher voltage units, you should use a breakdown meter..don't know why.


The rds is the slippery figure.
This means also that the capacitance on the gate is different too.
I suspect that there is a problem, with the deposition of the insulator and gate material, and the surface area and more particularly the thickness. The surface area and thickness of the gate will give you the capacitance.... simply measured across the sg pins... assuming the insulator is simple enough for them to do correctly.

It makes sense I suppose that the Rdson will be less, as if we loose some capacitance, then that and the rds will go hand in hand, as the less surface , and  thickeness will limit both the capacitance, and importantly  it's effect on the n channel to conduct.

So the worst of the fakes will have low capacitance, and higher rds on as a rule....  the voltage seems ok though.

 Hope you stuff arrives.

As promised, I blew up the board again this morning... but not as I expected.
The fakes worked particularly well... and remained cool at 2kw, bit warmer at 4kw, and handled the 6kw of highly inductive load of the big mig welder without problem too....

So the whole thing was running far too well to blow up easily.

I put the transformer and board into a box to keep it all together neatly ..... and connected it to the 15a supply to pretest before the battery.... and unbelievably, with no load, it did it's pre test, and promptly blew up..... have no idea how or why..... time for autopsy.

The darndest thing....

...............oztules

[oztules]

06:30:32-7/05/15| <RossW> !kwheat 2 2
06:30:33-7/05/15| <RossBot> 2.00 Kilowatts input will heat water at 2.00 litres/min by 14.34 deg C

That should have raised the temperature 57 degrees in 4 minutes, or 43 degrees in 3 minutes.
So somethings not QUITE right there (but it's still close enough for government work!)
[/quote]

Yes when I did a quickie on it ,I found I had used 456000 joules to do the work on the basis of 1900 watts (225X8.5...not a full 2kw) for 240 seconds
@ 4.18 joules /degree/cc X 2000ccX45c or 376200 joules..... this is a botch up of some 80000 joules could have heated an extra 400 ml or more.... maybe got the timing wrong you think??

Think I'll become an economist... there close enough is amazingly  good  and worthy of a nobel prize....or at least industry leader...

................oztules

[oztules]

OTW,
How do you have that wired up exactly.... are you using two controllers or one.

..........oztules

[andymack]

Good Show

Are you putting just one tranzorb per quadrant?

Is it mostly the case where an upper quadrant and corresponding lower quadrant blows but more damage  (like circuit board tracks) is done to the lower quadrant?

[oztules]

Well autopsy showed nothing. 5 fets decided to go to hell. All the rest fine.
Totems fine
Transorbs fine
Upper driver opto fine
Have not the slightest idea of what or why.... suspicion of a loose joint going to the inductor in series with the main tranny... but more wishing than proof.

So replaced the fets with china wannabees.. of 73mv@10amps... or 7.3mo, and all is well again... now I don't seem to be able to kill it to see the results.... always something.... if only the batteries were connected up this morning instead of just the 10 amp  53v power supply..... may have seen some explosions, and found out how the transorb went again....

Transorbs are only on the lower sides, the upper is isolated and plug ins....never charcoals.


..............oztules

[off the wall]

Interesting news of instructions for explosions above. Very important actually but probably beyond many of us - what are Transorbs?

Yes - I am using one controller simply putting the new transformer in parallel with the two in the other unit. In order to keep down the quiescent current I put a choke in on the return side of going back to the controlling inverter. I simply wired the output of the new transformer through a controller board in case there was an algorithm to switch on the fan automatically on high power draw, but that's not working, nor is the temperature sensor on the transformer.

The second fan is controlled by the second temperature sensor which is a plain thermostat. I took its power supply straight from the AC going into the transformer through a full wave rectifier and 100uf capacitor so that operation of that fan is failsafe whilst there is power available to the transformer.

I ran a current transformer sensor monitoring the output power back to the meter board in the primary inverter and wiring it in parallel with the current transformer in the original unit gets the meter to work accurately monitoring the out put from the three transformers.

Taking the power output directly from the auxillary transformer and not running it back through the primary inverter control card means that the power generation from that transformer is not monitored by the inverter controller and so is extra. This gives me the extra power that is convenient up to 4kW continuous and 5 and 6kW for long enough periods to be useful.

On a 24V system really the current draws on high power are not to be favoured so I wouldn't want to run a bigger inverter capacity on the system. At a draw of 4kW junctions in my busbars and connections get warm so being able to draw higher for periods of time limited by the inverter prevents anything getting hot.

Best wishes

OTW

[frackers]

Got a card left by the postie yesterday to arrange drop off of a 'large parcel' so sounds like my boards have arrived. I'll know when I get home tonight

I was right - big box of stuff!

LF Driver rev1.1 2014.12.08 on one PCB and on the other LF8K Mosboard v2.1 2015.1.4

Still hunting for cores and wire. Might have to see about shipping some ex-Aussie grid ties here on the slow (cheap) boat across the tassie ditch. In the meantime, might put 2-3kw of transformer made up of those 1" thick rings.