Guru
Joined: 02/02/2017
Location: AustraliaPosts: 1432
| Posted: 04:28pm 21 Oct 2017 |
Part 15: When 47uH is not the same as 47uH
I have had some good times and bad times with my home brew, most poorly finished inverter. The main power board is one of these
3 x Hy4008 210 Amp mosfets per leg of the H bridge.
I enjoyed approximately 2 1/2 weeks life from the first unit.
I bought another one and it went for less than 24 hrs.
The bag of spare HY4008’s arrived shortly afterwards so I replaced all, checked gate diode and resistors, etc, replaced the EGS002 since it had blown both the IR2110 and the EG8010 and gave it another go.
It lasted not quite 2 days.
Something is wrong.
I modify the EGS002 so that there is no current limit control, the only way it switches on is via the 5V supply coming up, causing the 3 sec slow start. The only way it goes off is when the 5V comes down. No cause for concern.
I use an Aerosharp 1500VA inductor core, with 13 turns of 3 x 6mm cable.
This gives me 47uH and a very low ESR.
The board feeds a 37V primary on a 3kVA Areosharp toroid. There is a 47uH choke in series. The secondary has a 5uF 450V cap, then via a 16A CB to a dual GPO.
After rebuilding the better looking of the two boards, again needing a new EGS002, I put the DSO on the primary winding, NOT the output of the inverter board. This means the two probe points are the two primary winding terminals, one of which is where the choke connects.
I was wondering what is causing the failure.
The boards last for days then fail. This means it’s not a simple design fault.
The problem is more to do with operating limits of components exceeded.
Yellow is low side gate drive, used for sync.
Blue is primary winding voltage.
First we see the Aerosharp choke at idle and under approx 400W load.
See the overshoot and undershoot on each transition from on-off and off-on.
idle
load
Next is the E65/32/27 ferrite core with 1.5mm gap, also 47uH.
There is no over/undershoot at all.
idle
load
It does not seem much but it is clear to me there is a difference when I connect one or the other 47uH choke.
Last night I saw the inverter deal with a large load, bordering on short circuit and survive when the defective blower fan of the wood heater was switched in. I doubt it would have made it through that event without blowing up.
I used the Areosharp choke on the bench when bringing up another but different inverter board, this time using 5 x HY1908 (90 Amp) per leg. Usual test conditions (EGS002 with soft start, no current control, 27V DC, 1500VA toriod, 140W load)
It started OK with no load. I switched in the 140W test load a few times and one leg at least has blown. I have not damaged any boards on the bench before.
It’s so easy to damage things by exceeding something. I think it’s not about current limits being exceeded, I think it’s a excessive voltage thing.
The E core ferrite choke is appearing to me to be the magic sauce in the home brew inverter. They reduce the speed of the fast voltage changes seen by the 4 legs of the H bridge. These voltages correspond to Vds or the voltage between drain and source. Admittedly most modern mosfets have construction such that they can widthstand voltage larger than Vds (called “avalanche”) but I think there are limits to how much energy can be absorbed before failure. Fast voltage changes also contribute to cross conduction thanks to dv/dt inducted turn-on. I wonder..
In summary, I think I found the cause of the 3 failures of the inverter boards.
Iron cored chokes. Use Ferrite cores due to their capacity to reduce high frequency energy passing through. Even if the iron core choke inductance seems enough, they pass too much high frequency energy. This characteristic is not shown with the inductor testing we do. I attempted to show how inductance changes with frequency testing at 1kHz, 10kHz, 100kHz and 1MHz but my skills were not up to the task. I saw small changes in inductance but not enough at 1MHz to explain the iron core's difference. MAybe I need to go to 10MHz..dunno.
I may have spoken too soon and this board no.4 will fail tonight. (Murphy’s Law)
I’m OK with that, it will force me to look harder, think better and keep learning.
wronger than a phone book full of wrong phone numbers