3-Phase Inverter / Mosfet gate problem

Go To Last Post
34 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hello, i am having serious problem with an inverter im trying make. The driving method is SVM on 23KHz.  

 

The 12V for opto/driver testing is coming from Mastech HY3005F-3 . I cant apply voltage to bridge cause acts like short circuit.

 

Used a programmable power supply to safe apply voltage on bridge and check the current (if has short-circuit) . Graph its ofc without load.

 

Rigol DP832A

 

 

 

I have 3 PWM from  Mega32 that fed into my optocouplers, then to my drivers and mosfets.

 

The problem:  The output from driver is going to Mosfets gate , is realy changed and as you can see with 200Ω resistor i have some "capacity"

problem on gate.

 

Reduce the resistor to 20Ω or remove it fixs a lot my signal but i could have 'ringing' problem.

 

 

 

Circuit schematic (a part of it cause too large to embed it. If need i can attach it)

 

                                    [ Bootstrap capacitors are Tantalum capacitors ]

 

Opto signal:

 

 

 

 

Signal from driver's legs:

 

You can see that signals arent so clear but at least they dont mix ! Also you can see the 0.5uSec DeadTime the drivers is adding to signal.

 

 

 

 

 

Signal that i measure on mosfet gate with 200Ω resistors:

 

Its clear that signal mixes and that gate signal isnt the same as driver's . Seems like capacity problem

 

 

Yellow gate signal without resistor or with 20 Ω

 

Signal is getting realy better compared to blue signal which has 200Ω on gate.But i may have ringing problems

 

 

 

The circuit is on PCB as you see below:

 

 

 

 

 

 

If i give 1KHz PWM signals (via fuction genarator) to optocouplers and i remove Gate resistors :  Mosfets stay cold , current drawn of bridge is 30-40mA. (without LOAD)

If i increase frequency >4Khz with above setup the problems starts again.

 

 

 

 

 

 

 

Distance  may add parasitic industance , specially on those Power-Switching circuits. But even on so low frequency ? 

 

 

 

 

If i missed any serious info about circuit/driving/componets. Just ask

 

Wireless remote car device using hand movements:

https://www.youtube.com/watch?v=uerWyUwLLBo

Last Edited: Wed. Jul 8, 2015 - 06:26 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Why did you put all of those inductors in your circuit? I'm suggesting that your pcb layout, especially aroung the mosfets is not very good. Seek to shorten and fatten up the power around the mosfets.

Note that you need a minimum frequency and pulse width for your highside mosfet driver. You need to charge the boost capacitor. Read the driver specs carefully.
Also note you do not have adequate creepage distance around the optos. Don't run a fill around the optos - you use optos for galvanic isolation but then compromise it with the pcb layout. You want a defined gap of at least 6mm. This is a safety issue.

Last Edited: Wed. Jul 8, 2015 - 10:40 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

 

Kartman wrote:

You want a defined gap of at least 6mm. This is a safety issue.

 

Yes PCB may need some changes. But i had limits during designing . Limits like single face, max 10x15 etc . Created on homemade PCB machine. 

Those kind of circuits need special PCB design. But for my first PCB i think was the best. Next will be better i hope :)

 

Kartman wrote:

Note that you need a minimum frequency and pulse width for your highside mosfet driver. You need to charge the boost capacitor. Read the driver specs carefully.

 

Totally agree, but was solved from first project steps.  

 

 

 

 

I read so much before even try build this project. The gate's signal and its capacity get affect by Rg so much. 

The 200Ω resistors its the problem. Re-reading the datasheet and focus on temperature tests i see that they using 10-30 Ohm resistors.

 

 

I will need new measures. Reduce resistors will fix the signal i guess. I will post results.

 

 

Thanks for reply Kartman.

Wireless remote car device using hand movements:

https://www.youtube.com/watch?v=uerWyUwLLBo

Last Edited: Wed. Jul 8, 2015 - 11:18 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You violated the safety constraint! Hopefully you have other measures in place to protect yourself.
Make sure you wear eye protection. Its not fun getting a bit of epoxy blasted into your eye when something goes pop!
Yes, 200 ohm gate resistors are excessuve. 12 or 22Ohm is more reasonable.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Around the optos is running the ground plane. 

 

GND of opto/driver is connected to PGND right here:

 

Wireless remote car device using hand movements:

https://www.youtube.com/watch?v=uerWyUwLLBo

Last Edited: Wed. Jul 8, 2015 - 11:54 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I think you've missed my point. You don't have sufficient isolation between the low voltage input to the optos and the high voltage side. Pouring a ground plane around the optos is a dangerous thing. Within the constraints of your board dimensions you could have easily created a 6mm gap between the low voltage and high voltage sides.
In this instance, the pcb isn't simply a way to connect and mount your components but rather a vital component in your system to ensure correct operation and safety.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

In this app note http://ww1.microchip.com/downloads/en/AppNotes/00889b.pdf you can see they add protection around the FETs and use a lower resistance to drive the FETs off compared to on.

 

Might be of help.

 

David

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Kartman wrote:

I think you've missed my point.

 

Got your point, safety above all.

 

I had the safety distance in my mind as you see in the board but thought the ground plane could be helpfull

being around my opto to reduce noise.

 

That's all about safety issues. Has nothing to do about my problem. Even that 1.5-2mm is a fine distance for 12-24AC input. 

 

[ if i get the results/measures for low Voltage maybe will need a re-designing of PCB for HV ]

 

DAFlippers wrote:

Might be of help.

 

David

 

Resistors affecting my gate capacity/time switching. The 200Ohm is so high, without resistor i could have ringing problems.

So more measures with 12+ Ohms will clear the things i hope.

 

The driver has special circuit inside and take care about driving the mosfet/igbt. IF you set the components/signals as must be.

 

I saw that article David during searching. I will check again the gate protections. 

 

Thanks for reply David.

 

 

 

If i continue getting wrong measures i will check the option of IGBT, which acts better on those kind of circuits.

Wireless remote car device using hand movements:

https://www.youtube.com/watch?v=uerWyUwLLBo

Last Edited: Thu. Jul 9, 2015 - 08:37 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Reducing the Rg resistors fixing the signal.

With 0 Ohm i have the best signal.As it must be, with the 500nS DeadTime being clear.

With 10 Ohm i have still a good signal but DeadTime seems shorten.Around 250 nS.

So as i increase Rg >0 Ohm , the signal is 'going' to mix. After 30 Ohm about , signal is ready bad.

And the worst.

Powering circuit with 24AC and sending PWM to Phase C mosfets my bridge draws around 30mA current.

If i try send PWM to other Phases A or B , i see my bridge draws 1 A .

Wireless remote car device using hand movements:

https://www.youtube.com/watch?v=uerWyUwLLBo

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

When you say the signal is really bad - how does it look?  The App note I posted turns the FET on via 51 Ohm resistor and of via approx. 17 Ohm sharpening up the off edge.

 

David

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Both rise and fall time increasing. But with 10 Ohm signal is very good. 300nS DeadTime with 10 Ohm resistor should be fine for Mosfet .. PS on App note they use IGBT , mosfet as faster shouldnt have any issues with that DeadTime.

 

On IR2109's datasheet Rg is changing from mosfet to mosfet or IGBT to IGBT, as i see on testing graphs.

 

 

Wireless remote car device using hand movements:

https://www.youtube.com/watch?v=uerWyUwLLBo

Last Edited: Thu. Jul 9, 2015 - 04:48 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Ok.

 

David

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Now that 'seem' signals are OK the strange is why without load i have so much current draw.

Wireless remote car device using hand movements:

https://www.youtube.com/watch?v=uerWyUwLLBo

Last Edited: Thu. Jul 9, 2015 - 05:07 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You have mosfet gate capacitance and increasing your pwm Hz, you will increase the driver current because of the gate capacitance.

One thing is the gate signal problem that can increase the heat in the mosfet and other is when you increase the pwm Hz you will increase the heat too.

Regards,

Bruno Muswieck

Last Edited: Mon. Jul 13, 2015 - 12:31 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I know the problem is capacity.  My main program PWM freq is 23Khz, but also with 4KHz signal from generator i still have the same problem.

 

Yes may freq increase the problem but the aim is to remove this problem.

Wireless remote car device using hand movements:

https://www.youtube.com/watch?v=uerWyUwLLBo

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Measure the current to the mosfets - you'll see if there is any cross conduction on the 'scope.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You mean the current running D-S without load ? Or you mean the gate current ?

Wireless remote car device using hand movements:

https://www.youtube.com/watch?v=uerWyUwLLBo

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The gate current isn't going to tell you much about cross conduction. I'm suggesting measuring the supply to the mosfets. If they cross conduct, you'll see the spike.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I used the programmable supply and he stoped on 0.7V / 1A without LOAD !!

 

Its sure we have some cross conduct like mosfet dont follow the gate signal.I will check it more ( i m on holiday atm :) )

Wireless remote car device using hand movements:

https://www.youtube.com/watch?v=uerWyUwLLBo

Last Edited: Wed. Jul 15, 2015 - 12:24 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Gate signals dont MIX, they have clear 0.4 usec deadtime.

 

Does the boost capacitor (22uF tantalum) makes the trouble and increase gate capacity somehow ?

Wireless remote car device using hand movements:

https://www.youtube.com/watch?v=uerWyUwLLBo

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If you add a 10Ohm between the VDD feed and the top MOSFET then use a scope the voltage at the junction of the resistor and MOSFET should be constantly VDD with no load.  If it is spiking low then you have both MOSFETs on at the same time.

 

David  

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I think i didnt get it David. I cant supply the circuit with VDD, it acts like short circuit !

Wireless remote car device using hand movements:

https://www.youtube.com/watch?v=uerWyUwLLBo

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I mean:

 

So are you saying Q7/Q8 are a dead short?  If so, what happens if you disconnect R7 or R8?

 

David

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

David i test what you said during circuit debugging ( its the right word ? doesnt sound good). Without resistors i had some short circuit, i cant remember in which Phase's mosfets was.

But that might was cause gate was floting without resistors.

 

The most strange is with 10 ohm  gate resistors and supply 24AC.

 

Before apply the 24AC , i disconnect all signal from drivers. 

 

I tried to give pulses on drivers 1 by 1.

 

First i tried Channel C , my programmable power supply shows me that bridge draws 30mA. Not bad is guess, seems PhaseC dont have any short circuit. 

 

After i connect ChannelC and channel A together, bad results. Too much current, the power supply turn it off on 1A.

 

So finally i tested Channel A and B  separate. Seems short circuit on those channels.

 

 

But its so strange. Signals are OK to all gates. Mosfets were working before solder them. (ofc i solder them on professional solder station, on low temperature )

 

And if there's a chance that mosfet are dead , how ? They can draw 8A and the max i let it was 1A.

Also during soldering i had antistatic bracelet and lwo temperature.

 

And as i know there's no some technique to check if is no dead without de-solder it 

 

P.S  internal diode from 6/6 mosfet i measured with DMM seems OK. That's not guarantee they are not burned but ..

Wireless remote car device using hand movements:

https://www.youtube.com/watch?v=uerWyUwLLBo

Last Edited: Wed. Jul 15, 2015 - 03:02 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I continue the post since i still face some problems. I end up that i will need more dead time so i decide to switch from IR2109 to IR21091. The footprint and characteristics are same , the only diff is that IR21091 gives the choice of deadtime selection .Deadtime that i set with resistor between SD-Ground.

 

 

 

My both 3 SD signals gets signal from same opto, so i have them ON or OFF together. With IR2109 , everything works fine.Signals from opto are good,output gate signal are fine. With IR21091 , nothing stays good. Opto signal changes, output signal its a mess  and the worst... SD signal must be 0V or 12V... with IR2109 its stable.. with IR21091 its like waveform with noise...

 

 

Opto signal with IR2109 plug in:

 

Opto signal with IR21091 plug in:

 

 

SD signal with IR21091 ( with IR2109 the signal its stableon 12V or 0V):

 

 

IR2109 output high and low with 500ns delay:

 

 

IR21091 output:

Attachment(s): 

Wireless remote car device using hand movements:

https://www.youtube.com/watch?v=uerWyUwLLBo

Last Edited: Wed. Sep 2, 2015 - 10:23 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

What's happening on your 12V rail?

What do IR say about circuit layout and capacitors?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The 12V rail is stable during tests with IR21091.

 

 

About capacitors, the left ceramic capacitor 100nF at each 12V-GND of driver had no special choice. 

 

The bootstrap its an 10uF ceramic.But for now i think right part of driver is out of discussion since bridge has no V till i found the driving error.Otherwise will keep blow my mosfet. 

 

 

 

Layout its the same with IR2109 , i cant understand why those troubles. And how/why can change my opto signals..

 

Hmmm

 

 

 

 

Wireless remote car device using hand movements:

https://www.youtube.com/watch?v=uerWyUwLLBo

Last Edited: Thu. Sep 3, 2015 - 02:09 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 1

Your rise time is very strange. The selection of your components ( optocoupler and gate driver ) are not matching your requirements ( rise time/ frequency ). optocoupler stage should be eliminated because they are generally slow (except the one used in communications). You should switch towards isolated gate drivers if you want isolation and faster response together. Otherwise this crossing over would create shoot through condition ( short circuit ). Gate driver should be of sufficient approach so that it can provide required current charge and discharge mosfet. You can try using GDA-2A6S1GDS-2A6S1SKHI 61 R gate drivers. The GDS one is perfect for your case as it provides very low propagation delay of 75ns.  Hope this helps for anyone searching over internet with similar query. 

Peter Kibbson

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You do realise the last post in this thread was over two years ago ... ?

 

Or is this just a spam/shill post for those links to taraztechnologies.com ... ?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You probably need to read my reply again. I mentioned that this may help anyone searching over the internet facing similar issue (as we all google for solution these days and this forum is probably to help people). Secondly you can call it spam if i just pasted a link here and didn't provide any technical details, i wrote a very detailed reply after reading all the replies above and proposed an easy solution (please elaborate if my proposed solution was wrong). These gate drivers are from three different brands so i am not spamming/promoting any specific brand. 

Peter Kibbson

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You're right - I missed that. Would've helped to put it right at the start of your reply.

 

We do get a lot of people who don't seem to realise that they are addressing a long-dead question.

 

We do also get quite a lot of "shill" posts.

 

Apologies.

 

EDIT

 

"Shill" posts - see, for example: http://www.avrfreaks.net/comment...

 

And: http://www.avrfreaks.net/comment...

 

Last Edited: Wed. Oct 4, 2017 - 07:46 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Having read Peter's post on a mobile device, I too did think it was spam. Having seen Peter's location which is only visible on a 'proper' browser, then that would've swayed my thinking.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I have opened this thread twice and each time my Avast antivirus goes ape about a url threat (102f.net) presumably from post #28. Anyone else?

 

edit corrected 102f.net

 

Ross McKenzie ValuSoft Melbourne Australia

Last Edited: Wed. Oct 4, 2017 - 12:31 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

No antivirus warnings - I was just going on "observation" - albeit slightly unobservant.

 

blush

 

EDIT

 

I didn't follow the links - just looked at the domain name that Chrome shows

Last Edited: Wed. Oct 4, 2017 - 10:44 AM