max668 problems

Hi,

Anyone know anything about this part. I have a design with 2 of these. The first is a standard 5V boost, 0.3A. That seems to work fine. The second is 10W flyback giving +/-70V on the output. This uses a transformer. I'm blowing up max668s like they are going out of fashion :-(. One board now has damaged tracks so I've moved on to board 2. On this board I made sure the 70V psu was working before fitting the CPU etc. It was working fine. I loaded the supply and took measurements. But first time I tried the supply after fitting the CPU it has blown up.

The odd thing about the way it fails is that the feedback pin takes too much current. The current "blown" part takes about 2mA into the feedback pin. I've taken out the external mosfet and put a 10K to the feedback node to a bench supply. The max668 output is running at 80% as expected. When I increase the bench supply the feedback pin voltage increases until it hits about 1.25V (Vref) then the output drops to 0%. All as expected, except that I'm pushing ~2mA in and the FB resistors are 18K and 1M and the feedback pin should only need

20nA.

It is the high current that causes the problems. In effect the feedback fails, the transformer is driven flat out and the mosfet blows up.

I don't understand what is causing this failure and it is odd that the max668 works, except for the feedback pin current. The only thing different from when I had it working is the addition of the CPU. The on/off pin is now driven hard from a CPU port rather than a 100k pull down and 10K link to 5V. Could a port blow a max668?

TIA

Malcolm

--
Malcolm

 Malcolm Reeves BSc CEng MIEE MIRSE, Full Circuit Ltd, Chippenham, UK
 Click to see the full signature
Reply to
Malcolm Reeves
Loading thread data ...

Is the CPU "chattering" the ON/OFF pin?

Have you tried 100K between CPU and ON/OFF pin?

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

What compensation capacitor are you using on the FB pin?. If this is not right, the circuit could oscillate beautifully...

Best Wishes

Reply to
Roger Hamlett

The CPU shouldn't be "chattering" the ON/OFF as I'm running with an emulator and switching it by hand at the moment. Plus, the shutdown pin is also sync so shouldn't mind chattering.

To answer roger

Currently 1nF+18k is the feedback but it was working before I added the CPU.

I've added a 33k in series when I replaced the blown part. My theory is that perhaps the GND at the MAX668 is lifting due to high currents and that means CPU 0V is below MAX668 0V so shutdown is taken negative and that blows the chip.

I've just replaced the blown part and check for shorts/continuity with a DVM so make sure it is all soldered right. I've left the mosfet out and use a bench supply via a 10K to feedback node to check the part is working before any big currents fly about. As expected enable it and output is 80% odd pulse train. Raise FB to Vref and it is a 0% pulse train. BUT, FB pin is taking 1-2mA and I haven't run this part in anger yet. No mosfet, no large currents. I work on a static mat area and the DVM I used to check shorts/continuity measures 0.4V, 0.6uA on another meter so why is this part broken?

Very odd.

--
Malcolm

 Malcolm Reeves BSc CEng MIEE MIRSE, Full Circuit Ltd, Chippenham, UK
 Click to see the full signature
Reply to
Malcolm Reeves

Its easy to check the compensation: get a unit running, and use a FET to apply a step load (on and off) and look at the output voltage. If its somewhere between a critically-damped 2nd-order and a first-order response, the compensation is fine.

if you think your layout might be pulling FB below ground and breaking it (a distinct possibility), slap a schottky directly across the chip.

Cheers Terry

Reply to
Terry Given

I don't see how. The layout is good and it is a 4 layer board with planes.

I have it working again. I built it very careful measuring with a DVM as I went. I especially check FB leakage before any voltage was applied. On this unit I did find 1k5 leakage on the CS pin. I checked a new part and that was ok so either this part had it before I soldered it in or soldering it damaged it. On the CS pin the leakage does not affect the performance as impedances are low. They are tiny parts and so tricky to solder but it is an odd failure mode and the are loads of other parts I have soldered without any problem. I even tried putting the iron on a new part in the same way to see if I could reproduce the failure, but there was no effect.

Very odd.

--
Malcolm

 Malcolm Reeves BSc CEng MIEE MIRSE, Full Circuit Ltd, Chippenham, UK
 Click to see the full signature
Reply to
Malcolm Reeves

you may have fried it with static or heat then.

Cheers Terry

Reply to
Terry Given

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.