Piezo injection and high side FET drivers

[twelveeyedfish]

Morning all,

There seems to be some interest in automotive stuff on here so I thought I would present my problem.

I have been working on a design for a trivial piezo electric diesel injector driver. Basing this on a stock application note produced by Fairchild here:
https://www.fairchildsemi.com/applications/automotive/injection/

The top picture depicts the precise circuit in use. The boost is provided by a linear technology device at about 150V and seems to work well. The issue I am having however is that the HS FET driver (driver1 in the diagram) keeps breaking!

I am using the FAN7081MX device (at Farnell part no. 2290268) and have been through literally tens of these devices.

On my test circuit I have limted current through the FET chain using a 10W 8R2 resistor (ie after D1) and all the appropriate pins on the HS FET driver (ie not the power pins). Just tiny values to improve survival chances. Before I uprated my FETs somewhat I also had the issue that the FETs in the main chain (the boost and the purge FETs) would also blow. I have never had any issue with the injector select FETs going pop. I also have a couple of tiny caps across the FETs to remove the spikes at switching time. They work well.

Sometimes (and often) the system works. Well. However I am failing to place my finger on why these HS FET devices keep breaking. If anyone has a little more experience with these and may have a few pointers I would be very appreciative. If I need to specify more components please say although nothing is underrated! It may be useful to know that the series inductor L1 is currently 150uH however.

Best of thanks

 

[Fish4Fun]

If I had to guess, I would suspect that the switching frequency is too low. Assuming an 8 cylinder 2-stroke diesel @ 3600 RPM any given injector would fire @:

3600 RPM / 60 Seconds = 60 times per second, or 60hz.
1/60 =  16.6mS

In the world of high-side FET drivers this is an eternity. If the engine is a more typical 4-stroke diesel operating @ 600 RPM, the time between On pulses is 200mS. High Side Drivers typically use a charge pump to develop a floating potential of Vcc + ~12V. This potential is refreshed during each on/off cycle, but must be enough to fully charge the capacitors and the capacitors must be large enough to sustain the voltage for the entire on cycle AND overcome any leakage in the capacitors. If the charge is NOT maintained in the capacitors then your FETs will not switch properly and will overheat rapidly. I would suggest doing away with the high side drivers and replacing them with a simple boost buck regulator. You are likely working with a nominal 12V system, creating a 24V rail should be fairly trivial. From there simply use an opto-coupler to drive the FETs.

Fish

 

[twelveeyedfish]

Hello,

Thanks for the comment - yes I agree with what you say. However for a piezo injector the injection cycle takes the form of two miniscule injections which are about 200 uS long. The injection does not last for the whole cycle.

That would manifest itself as a non working driver though I am specifically seeing failed devices, not ones that are simply not operating properly. I always welcome a second sanity check though so thanks very much.

Incidentally, piezo type injectors operate at around 120-200 volts and the high side is literally floating.

 

[Fish4Fun]

Ok, I was unfamiliar with the high voltage of piezo injectors. Assuming Diagram1 (semi-resonant) this brings up the value of L1. I assume this is being used as a boost-buck converter ramping your voltage up to 120V-200V using driver1 & driver3? Or are you using the resonant configuration in diagram3?

In either case which high side driver chips are failing? Are you actually using drivers for the low side of the injectors? Again, I think you might be better off simply designing a boost-buck Power Supply and keep any high side driver voltage ~12V above the rail you use for powering the injectors. A couple low leakage capacitors and a 15V zener should keep the "driver rail" ~12V higher than the injector Vcc. Then driver chips become a moot point and you can focus on finding switches that can handle the 5kHz switching into the piezo load.

My personal experience with high side driver ICs is that they are quirky and difficult to get to work as advertised. I am certainly no expert, but unless it is a really high volume production run it is easier to just create a Vcc + 12V rail.

Fish

 

[SA Electronics]

How is your project?
The link to the circuit does not work anymore. Please send it to me.

 

[GPG]

The link to the circuit does not work anymore.

Nor for me