To many trolls on this site (Bang goes another expensive mosfet)

Status
Not open for further replies.

Thread Starter

fourtytwo

Joined May 2, 2017
32
View attachment 127436 Hi all I am busy designing and building a grid tie inverter, I like challenges!
At the moment I am working on the unfolder or inverse rectifier that connects the unipolar output of the SMPS to the bipolar grid, it is an H bridge that switches at twice the supply frequency and here in the UK with our 240Vrms grid means it deals with about 350Vpeak on both the grid and DC sides. Like all things it works nicely when you are looking at it BUT when you turn your back POP goes another mosfet!
Although the switching speed is low the timing around zero cross of the grid is critical as having the wrong half of the H-bridge conducting tries to short out the grid with only one outcome (smelly electronics) but timing is difficult to ensure given the gate drive has to be (advisable) galvanically isolated from the control logic.
It turned out however the problem was an occasional glitch in one of the gate drives causing a large shoot through by turning on both mosfets on the same side of the bridge simultaneously, the glitch grew and shrank with temperature and was caused by both lower bridge legs sharing the same driver chip.
Really what I am trying to say it is not always the obvious reason that causes a fault :)
I hope everybody knows what an Hbridge is, if not https://en.wikipedia.org/wiki/H_bridge
I enclose the offending glitch, red trace is mosfet gate driveView attachment 127435
 

Thread Starter

fourtytwo

Joined May 2, 2017
32
Are you hoping to pump power back into the grid?
Absolutely :) I have already done so but my MTBF is a bit poor at the moment and my repayment time gets longer everytime I have to replace something! I think I have so many belts and braces around these damm mosfets now I am hopeful (subject to test).
 

crutschow

Joined Mar 14, 2008
23,502
Do you have some deliberate non-overlap in the MOSFET drive signals to insure that the two MOSFETs on the same side of the bridge are never on at the same time?
If not you need to add that in the drive circuits.

Here's a common circuit to do that:
You can vary the capacitors' value to get the non-overlap time you want.
The inverters in parallel are to give a low output impedance to directly drive a MOSFET gate.
If you are only driving a logic gate than you only need one inverter on each output.
upload_2017-5-23_12-44-12.png
 

Thread Starter

fourtytwo

Joined May 2, 2017
32
Thats a very nice circuit and thank you, but in this case the mosfet gates are 350V apart of alternating polarities! The problem I encountered was that although appropriate dead times were/are guaranteed by the control logic this did not necessarily translate through the isolation barrier (opto-couplers) and gate drivers at all times under all conditions.
It's actually quite hard to interlock gate drives in such a circuit, if all else fails (I hope not) I have also played with (in simulation) a self synchronising H-bridge where the entire logic (transistors & diodes) is on the high voltage side and requires no isolation barrier at all HOWEVER I know of some cheap grid tie's that use this and are known to pop mosfets frequently when presented with less than ideal grid conditions. Such a simple function yet so interesting to bullet proof :)
 

crutschow

Joined Mar 14, 2008
23,502
....The problem I encountered was that although appropriate dead times were/are guaranteed by the control logic this did not necessarily translate through the isolation barrier (opto-couplers) and gate drivers at all times under all conditions.
Then it would seem you need to increase the amount of non-overlap to insure there is still some non-overlap under all conditions.
If you show the complete circuit we may be able to offer other suggestions.
 

Thread Starter

fourtytwo

Joined May 2, 2017
32
I'm surprised your electricity provider is ok with having a DIY inverter coupled to the grid. I understand many providers would only sanction an 'approved' commercial unit.
I think perhaps as your contribution seems distinctly negative I would prefer if you did not posts on my threads.
 

Thread Starter

fourtytwo

Joined May 2, 2017
32
Then it would seem you need to increase the amount of non-overlap to insure there is still some non-overlap under all conditions.
If you show the complete circuit we may be able to offer other suggestions.
I am fairly confident I have fixed the problem, the logically defined overlap is 125uS in both directions whereas the maximum logic to mosfet delay is 35uS. As I thought the problem turned out to be sharing a common driver between two mosfets and it being a TC4584B hex schmitt inverter (not really built for the job). There were many reasons for using it including it's threshold voltage but it's now been replaced with a MCP14E3 cross coupled to add a further layer of interlock. This does change the threshold but it turns out there is sufficient slack in the timing to cope.

I am very wary of posting actual schematics as there are many who would copy without understanding and potentially damage themselves or others in the process, instead I prefer to stick to general principles and well known information already published. Also given the other recent posting on this thread it seems not everybody on this site is a genuine enthusiast!
 

dendad

Joined Feb 20, 2016
3,073
It is an interesting project you have there fourtytwo.
BUT....
Alec-t's comment about his surprise on you putting a DIY inverter on the grid is very valid!
I too am of the same opinion, and here in Australia it is definitely illegal to do so.
I think you are treading on thin ice so to speak, and could be heading for trouble if things go wrong.
 

MrAl

Joined Jun 17, 2014
6,610
I have deleted this thread due to excessive trolling!

Hi,

Oh so you finally realized you were a troll and thus tied to get rid of your own thread?
Very commendable :)

Ok seriously you should not take things so strongly like that. Lighten up a little. You'll get more help that way.
Yes some people may not help as much as you would like to see, but others here will.
 

dendad

Joined Feb 20, 2016
3,073
It is disappointing you are reacting this way.
Back feeding to the grid is very dangerous and can kill some innocent person. That is why it is regulated. It is not "Orwellian" but common sense to regulate these sorts of thing to help keep people safe.
This could have been an interesting discussion, and I was going to try to help but that is your decision I suppose.
Please be careful!
 

dendad

Joined Feb 20, 2016
3,073
If you persist in playing with your inverter, I hope you have implemented auto disconnect when the mains fails so you do not feed power into the grid if some poor electrician turn the mains off to work on it, and "knowing" he has removed power, gets killed by your contraption.
If you do not have this working, you are being irresponsible and possibly criminally negligent !
This is the reason you have received these comments, and if you cannot accept them maybe you should stop work on this project, just for the protection of other people.
 
Status
Not open for further replies.
Top