next topic »

[Reply #15 on:]

You don't need to buy 3000 mosfets to get a good price unless you choose a type that the distributor doesn't stock; you can buy stock items in small quantities, usually individually.

If you use the SN754410 to drive 4 solenoids at 0.5A each concurrently, you will need to do something about heat dissipation. which will be 2W typical, 2.8W if you are unlucky. The device will fry unless you remove some of the heat through the ground tabs.

[Reply #16 on:]

Would this work?

http://search.digikey.com/us/en/products/BS170_D75Z/BS170_D75ZCT-ND/1587840

Unfortunately, it's not logic level (Rds(on) is only quoted @ Vgs=10v).

OK, the OP has a better solution so this is purely academic.  At the risk of hijacking his thread, I would still like to know:  Does that spec matter?  Threshold voltage is 0.8-2.1-3 (min, typ, max).  So it looks to me like it would work at TTL levels.  Rds is variable throughout this range, but driving the gate at 5v means (at 0.5A) it would have an Rds of about 1.75 ohms, vs. 1.2 ohms at 10v gate, 0.2A load.  Still inconsequential driving a 30 ohm solenoid, isn't it?

I'm asking, not professing, BTW.  Transistors are certainly more complicated than they seem on the surface, and I have a lot to learn.  (I'm sure I'm not the only one.)

[Reply #17 on:]

At the risk of hijacking his thread, I would still like to know:  Does that spec matter?  Threshold voltage is 0.8-2.1-3 (min, typ, max).  So it looks to me like it would work at TTL levels.  Rds is variable throughout this range, but driving the gate at 5v means (at 0.5A) it would have an Rds of about 1.75 ohms, vs. 1.2 ohms at 10v gate, 0.2A load.  Still inconsequential driving a 30 ohm solenoid, isn't it?

Fig. 1 of the datasheet at http://www.fairchildsemi.com/ds/BS/BS170.pdf suggests that with 5v gate drive, it would drop around 1v at 0.5A. But that's typical behaviour, not worst case, and leaves no margin for the gate drive being a little less than 5V.

I didf eventually find a suitable mosfet in an overgrown TO92 package, see http://uk.farnell.com/toshiba/2sk2989-tpe6-f/mosfet-n-ch-5a-50v-to92/dp/1714337, but it's expensive compared to SMD devices.

[Reply #18 on:]

Hmm..  OK, Figure 1 (On-Region Characteristics) seems to be saying that with 5v on the gate, anything over 1v through drain and source would exceed 0.5A without any external current limiting.  I don't see how that affects the ability to drive it, though..?

Hypothetically, the solenoids would be running at 12v, but have 20-30 ohms resistance.  My calculations put that around 0.4 to 0.55A total current.  Within safe limits, especially assuming 20 ohms is a temporary worst-case.

However, I haven't taken dissipation into consideration up til now.  That presents a problem:  It looks like there's a duty cycle for high-current applications.  At 25c ambient, 100% duty cycle use is limited, linearly, from 12v @ 50mA to 2v @ 400mA.  OK, so not so well suited to this application.

Again, I haven't learned all there is to know, so this is purely a what-if in case my assumptions and datasheet reading skills aren't quite up to snuff.  Hope this is relevant to someone besides me.

[Reply #19 on:]

Yes, I would definitely reiterate the need for heat sinking. These chips run fairly hot if your solenoids draw a lot of current.

If you are making this on a PCB (doesn't sound like it) definitely use a large copper plane for heat sinking, and ideally bolt the board to a piece of metal (if you can avoid shorting out vias). Otherwise, consider something like this:

http://search.digikey.com/us/en/products/580200B00000G/HS179-ND/228497

It is a $0.74 heat sink designed to press fit directly on top of the 16-DIP package of a chip like the SN754410. Having used these chips before (on a PCB with well heat sunk ground pins) it is probably worth the investment. That heat sink will dissipate 2.8W with a ~55C temperature rise, which is definitely still pushing it at full power. Assuming an ambient of 20C, that's an operating temperature of 75C.

The chip itself claims to be good at up to 85C, but you'd definitely be pushing it if you ran all four simultaneously at full power, even with a decent heat sink. And 75C is hot enough to potentially burn you anyway. You might be better off splitting the outputs into two chips. Of course, you may well end up dissipating substantially less heat than that, so you might be okay. If you had a way to flow air over the chip that would also substantially improve the heat sink performance.

[Reply #20 on:]

I tried the SN754410 over the weekend and it wouldn't drive the solenoids (Common 12V and using the SN754410 to give a low signal to each of the 4 solenoids). I measured around 20 ohms with a multimeter. But regardless running two at a time wasn't working with this chip. It powered the leds just fine(had leds connected to the output of the SN754410 so I could visualize that it was working).

But no go. I definitely got VERY hot.

[Reply #21 on:]

But no go. I definitely got VERY hot.

This is why I recommended using mosfets. However, a high-gain low-saturation medium-current BJT such as the ZTX851 would also be suitable for switching 0.55A.

[Reply #22 on:]

I'm a fan of MOSFETS at this point!

[Reply #23 on:]

Are these DC solenoids or AC? For AC you need Traic's.

[Reply #24 on:]

DC I believe.

[Reply #25 on:]

Edit the chip actually does work. The LEDs were not getting juice because I had a 1.5k resistor in series (12V) and the solenoids were taking all the power so they weren't lighting up. But the solenoids are working just fine.

The chip does get very hot though I should add.