MOSFET wont turn off

Thread Starter

theblackpanther

Joined Feb 8, 2008
8
Hi

Attached are schematics of a bidirectional AC voltage MOSFET based regulator.

dg1.jpg:

This is the power side. I've attached IRF840 datasheet for your immediate reference. This arrangement might work without the diodes. The diodes are present in the design for a future development when the MOSFETS will be replaced by IGBTs. These diodes will counter the internal parasitic diode effect of the IGBTs. But for now its only MOSFETs.

dg2.jpg:

This is the gate drive circuit. 4N35 isolates power from MCU. The TC4431 is a MOSFET driver from Microchip. I've attached that datasheet also. In that, Pins 6 and 7 are said to be sink and source outputs respectively. I've used the source output. Correct me if I'm wrong.
The MCU is progrmmed to give switching pulse for delay angle based control.

Experimental observation:

For testing purposes, I did not give more than 60Vrms as input.

The sequence of events:

1. With correct switching pattern across gate source(I checked these on a scope), both MOSFETS refuse to turn off. The Input voltage appears unmodified across the resistor.

2. Next I left the gates floating. Still the same result.

3. Suspecting that the gate is precharged, I shorted gate source (with the driver unit disconnected) with a wire. The MOSFET turns off instantly. But the moment I remove the wire, the device turns on gradually, within 2 seconds.

4. Previously I assumed that TC4431 has a discharge path in the off-state. Suspecting this assumption, I attached a 10k across gate-source, in addition to the driver. Now the devices wont turn on at all, irrespective of gate source voltage.

5. I replaced the MOSFETS. Same results.

Querries:

1. What is wrong?
2. Am I missing out anything from the 4431 datasheet? Am I using the chip correctly?
3. This is my first high-side switching. Should I add any other device for the special purpose of high-side switching? (TC4431 is a high side driver)

Thank You.
 

Attachments

Ron H

Joined Apr 14, 2005
7,063
You need to connect pins 6 and 7 on the 4431 to the gate. Otherwise, you can turn the MOSFET on, but you have no active turn-off. I'm not sure why the 10k from gate to source didn't work (albeit slowly).
 

Thread Starter

theblackpanther

Joined Feb 8, 2008
8
Thank You.

jpanhalt, I quite didn't get you. The source of the mosfet is already connected to the driver (4431) IC's ground. Regarding your second suggestion, the 10k I added in step 4 is a turn off resistor, is it not? Correct me if i'm wrong.

Thanks Ron H for that detail. But does anyone have an idea why the 10k didn't work?
 

Ron H

Joined Apr 14, 2005
7,063
Thank You.

jpanhalt, I quite didn't get you. The source of the mosfet is already connected to the driver (4431) IC's ground. Regarding your second suggestion, the 10k I added in step 4 is a turn off resistor, is it not? Correct me if i'm wrong.

Thanks Ron H for that detail. But does anyone have an idea why the 10k didn't work?
I would guess that, for some reason, there is leakage from the pullup pin, when it is supposed to be off. With 10k, it only takes a few hundred microamps (which seems excessive) to turn on the FET. Try a 1k resistor, but don't use it in your final version. Hard, fast pulldown is much preferred.
 

Thread Starter

theblackpanther

Joined Feb 8, 2008
8
Just to clarify, I'll state the 10k problem once more. Without the 10k, the Mosfet doesnt turn off. With the 10k across gate source, it is always off, even when there is a gate voltage.

I'll try the 1k,though I dont understand the concept you explained. :D
 

Thread Starter

theblackpanther

Joined Feb 8, 2008
8
Thank You John. I'll take a look at the app notes.

Reg your questions, yes, there are separate power supplies. On the whole there are 3 transformers on board. One connected to a rectifer and 7805 to power the MCU, and the other two connected to 2 other rectifiers and 7812s. All the three are completely independent. I use two driver ics.

And I just connect the driver out to the gate and driver ground to the source of mosfet.

dg.jpg attached contains whatever I told above.

Thanks again.
 

Attachments

Thread Starter

theblackpanther

Joined Feb 8, 2008
8
Its Working!

Thanks a lot people for all your help. I shorted pins 6 & 7 of the 4431. And I also found out why the 10k didn't work the last time. Its because the pulses were reversed. That is, the wrong devices were triggered at each half cycle, when they were reverse biased. i swapped the gates, and they're working fine now.

Thank you again to all the people who took their time in viewing this post.
 
Top