OT: preferences/toughness: BJT vs FET

[newguy]

Sorry for the off topic post.

I have an automotive product which needs to control many pneumatic solenoids. The solenoids can easily be cycled 500 times/day or more. For that reason, and because of space, I'm going with a solid state drive system. I'm worried about relays burning out in a short period of time too. Most relays are rated for approx. 100,000 cycles. At 500 cycles/day, that's a lifetime of only 200 days.

My first stab at the thing used a PNP power transistor arrangement to energize the 12V solenoids. It worked very well, but the prototype was only in service for a few months before the customer asked me to build a slightly different version. My concern is that the prototype didn't see enough real-world service for me to gauge whether or not the drive system is robust enough for a lifetime of several years.

I'm thinking about changing the drive configuration to a p-channel MOSFET instead of a PNP BJT. Does anyone have any experiences regarding the robustness of FETs over BJTs or vice versa? Of course, all lines will have ESD protection as well as overcurrent protection in the form of resettable fuses.

[Ttelmah]

The key with any such system (relays as well), is 'where does the energy in the coil go, when the drive switches off'?. You talk about ESD, but this is not the same problem. You have a magnetic circuit, containing a lot of energy, and it has to go somewhere. If you don't properly cater for this, parts can be destroyed in nSec. FETs, contain a parasitic diode, as part of their structure, which this energy will flow through. However the rating of these is far below that of the drive ability of the same FET, except in some specific cases designed to handle such drives. You also have to consider what happens to the rails which this energy goes into. On a circuit drawing small amounts of power, if you dump the energy rapidly into the supply, you can easily end up with this going overvoltage...
A combination of a fast recovery trap diode, possibly with the current limited by a resistor, and a snubber network, can be designed to precisely control the rate of voltage change when the drive is released. I have some systems that were built over twenty years ago, and are still running switching in some cases dozens of times an hour. The few that have been retired, have been through obsolescence, or closure of the plants involved, rather than failures in the drives.

Best Wishes

[PCM programmer]

For automotive applications, look at "Smart Switches" instead of
just a transistor or a FET. These switches have many extra features
that are very useful in an automotive environment. Because these
switches are specifically designed to operate "under the hood", I would
assume them to be more reliable than other devices.

To find companies that make these switches, search for the following
string in Google: automotive "high side switch"

Thompson is a well-known maker of these switches. Many other
companies make them as well:

STMicroelectronics Intelligent
http://www.st.com/stonline/products/families/analog_and_mixed_signal/intelligent_power_switch/intelligent_power_switch.htm

[newguy]

TTelmah,

I should have mentioned that I do have transient suppressors to catch the inductive back emf from the solenoids. I will, of course, incorporate the same protection on this next churn of the product.

As an aside, I came across an application note some time ago from one a relay manufacturer. I can't for the life of me find that app note now (I wish I could), but I do remember one thing it recommended. Instead of using a simple diode arrangement across the coil, this app note recommended back-to-back zeners, each with a breakdown voltage = 2x the coil voltage. The app note went into great detail about how this particular arrangement minimized contact "chatter" during switch-off while still effectively killing the inductive transient. The net effect of this arrangement is to extend the life of the relay's electrical contacts. Because of that one app note, I now use a bidirectional zener-based TVS on all relay drives. This works really well.

PCM, thanks - that's exactly the type of thing I was looking for. It helps when you know the proper search terms.

[Ttelmah]

The only problem with the Zener's, is that for fast operation, the recovery times can be uncomfortably slow. However for your application this should not be a problem. I have used some of the 'smart switch' units mentioned by PCM programmer, and if the price is OK, these are great.

Best Wishes

[newguy]

PCM,

Thanks for the tip. However, the prices of the high side switches are too high. I can make one out of discrete components for less $$$. Cost is a big factor since I need to drive 24 solenoids. Not a typo - 24.

Ttelmah,

You're right about speed, but like you said, speed isn't an issue in this case.

[Storic]

I use in most of my application a MOC and a TRIACfor AC switching and a OPTO with a MOSFET for DC. I only use relays because my client has requested it.

In mose cases I get them to have a Relay + base setup that is switched by the above.

(I have not tried switching 500 x per day, the only application I can think of the switches a lot would be aprox 100 per day (on a TRIAC).

ANdrew
_________________
What has been learnt if you make the same mistake?

[Charlie U]

Take a look at the ULN2803 from TI. It has 8 open collector drivers in a package. Check the current capabililty for your application. I have used these in applications that require 1000's of operations per day, controlling solenoids for compressed air, dry nitrogen and vacuum. They have been in operation for several years without failure.

Also check out Allegro micro at the link below:

http://www.allegromicro.com/ic/interface.asp

They have a wide range of peripheral drives including latched and addressable.

[newguy]

[Charlie U]

Pneumatic solenoids that require 2A continuous?!?! Those are some "HEALTHY" solenoids. Dissipating 24 Watts, they should be getting mighty hot.

Still, if you must, you could use some P-Channel FETs on the high side as switches and drive them accordingly.

[newguy]

[Charlie U]

You may have worked this out already, but watch out for the 12A rating. With FETs that usually means that the case is held at 25 degC and the internal junction is at the limit, around 150 deg C or more, which is nearly impossible to achieve in the real world, except in a laboratory. Check the RDSon for the device and the thermal resistance, then work out your heatsink requirements. In most applications, a junction temperature rise of 50 to 75 deg C should be the limit.

Automotive applications are the nastiest, except for possibly space.

[newguy]

[Sergio]

BJT

Though for your aplication I'd use a ULN2003.

Cheers
_________________
Sergio

[newguy]