A really great MOSFET to use with the Propeller... except don't forget your the

ElectricAye · 2009-09-02 20:30

Hi all,

PhiPi recently suggested I use an IRF3708 MOSFET with the Propeller, and I must say it works really well for switching current (Thank you, PhiPi!). But I noticed that its drain is connected to the metal tab that I use for screwing the MOSFET to a heat sink, and that means the heat sink is sitting at +15 volts in the circuit I am using (see attachment). So the configuration I'm using sorta freaks me out. Is it normal to have the heat sink at voltage? I always thought things like heat sinks, etc. should be kept at ground.

Am I doing something totally stupid here or is a charged-up heat sink just something people live with under these circumstances?

thanks,
Mark

Post Edited (ElectricAye) : 9/4/2009 3:20:49 AM GMT

Ale · 2009-09-02 20:49

There are insulators for well isolation of heat sinks. Some are made of mica others are some kind of polymer. And of course plastic screws or insulating rings. 15 V are ok, the problem starts at 50 or so. But if you are worried... just insulate it.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Visit some of my articles at Propeller Wiki:
MATH on the propeller propeller.wikispaces.com/MATH
pPropQL: propeller.wikispaces.com/pPropQL
pPropQL020: propeller.wikispaces.com/pPropQL020
OMU for the pPropQL/020 propeller.wikispaces.com/OMU

Leon · 2009-09-02 20:49

The drain is connected to the tab with the MOSFETs I've used. Power BJTs are usually similar, with the collector connected to the case or tab.

Leon

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle

Phil Pilgrim (PhiPi) · 2009-09-02 20:51

In cases like this you want to use an insulating thermal pad and a flanged nylon washer to attach the MOSFET to, and insulate it from, the heatsink. DigiKey carries both. Check out #BER102-ND and #3102K-ND.

-Phil

ElectricAye · 2009-09-02 22:07

To Ale, Leon, Phil,

wow, what a relief! That's interesting. I never heard of those insulating pads. I thought for sure I had done something that rendered my design unsalvageable. Thanks, you guys, for saving my neck!

Leon · 2009-09-02 22:24

You can insulate the whole heatsink, as well, by mounting it on plastic spacers. It's OK if it is inside an enclosure so that it can't get shorted accidentally.

Leon

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle

Peter Jakacki · 2009-09-04 02:32

Mark, do you really really need a heatsink at all? The mosfet is a 0.012ohm (max) device and unless you are doing some high-frequency or very high currents with it the thing will probably never even start to warm up. Anyway, unless your heatsink is connected to the chassis or a common heatsink there is never any need to insulate it in low-voltage circuits (<50V).

*Peter*

ElectricAye · 2009-09-04 03:05

Leon said...
You can insulate the whole heatsink, as well, by mounting it on plastic spacers. It's OK if it is inside an enclosure so that it can't get shorted accidentally.

Thanks, Leon, it's nice to know it's not an outrageously uncommon practice.

Peter Jakacki said...
Mark, do you really really need a heatsink at all? The mosfet is a 0.012ohm (max) device and unless you are doing some high-frequency or very high currents with it the thing will probably never even start to warm up. Anyway, unless your heatsink is connected to the chassis or a common heatsink there is never any need to insulate it in low-voltage circuits (<50V).

*Peter*

That's a good question. I'm using this MOSFET to power a Peltier device whose performance peaks at 15 volts at 10 amps. Since Power = I ²R, I guess that means it could be "generating" about 1.2 watts of heat. That doesn't sound like much, but the device isn't very big, so I'm somewhat concerned that with my luck it might approach incandescence if I don't heat sink it somehow. In any case, I'm not familiar with MOSFETs, so I was caught off guard when I found my heat sink was able to charge my electric toothbrush. :-D

By the way, I found out the hard way that you can't use a 1k ohm or 2 k ohm resistor for R₁ in my above drawing. Though it's common practice to protect the Propeller pin with such a resistor, it seems that a much lower valued resistor is needed otherwise the gate-to-source signal from the Propeller output drops too much. In fact right now I'm driving the gate without a resistor. I'm using only R₂ valued at 10k (to drain the gate capacitance) and so far things have survived.

Peter Jakacki · 2009-09-04 03:42

1.2W is starting to get hot but it's quite ok for the TO-220 package without a heatsink I've found. The trouble is that you really need more drive and 3.3V is a little on the low side whereas Ron is usually spec'd with 10V gate drive.

At 3.3V drive you may need a heatsink but certainly it doesn't need to be insulated.

I only just looked at your drawing but you certainly don't want a gate to source resistor at all. If you are worried about what happens when the pin floats then put the pull-down on the prop side only. If you want to protect your prop from becoming a pyrotechnic device then use a series resistor (R1) to the gate. Even 1K will stop the prop from burning even if it is zapped somehow.

Those "pull-down" resistors you see in many transistor circuits aren't necessary for your common switching circuit.

*Peter*

Phil Pilgrim (PhiPi) · 2009-09-04 04:29

A gate pull-down resistor is always desirable on a MOSFET fed from a micro because, when the micro powers up, that pin will be an input with an indeterminant voltage level — possibly high enough to turn the MOSFET "on". The sereis resistor is only necessary to limit peak gate-charge current flow to levels that won't harm the micro. If that results in too-slow switching, a MOSFET driver chip can always be interposed.

R_ON for the MOSFET he's using is spec'd for a low gate drive voltage, which is what makes it such a nice complement to the Prop.

-Phil

Clock Loop · 2009-09-04 04:37

What you want is a SIL PAD

www.orionind.com/sil-pad.html

If you open up a switching power supply, you will notice that many HIGH-WATTAGE IC's use their TABS without a sil pad, and actually use it as a way to FEED the IC with electricity.

I have had a few devices specifically tell me to NOT use the heatsink as a way to feed power to the IC, but usually they state this.

Peter Jakacki · 2009-09-04 04:52

@Phil - my point about the resistor was that it needs to be on the prop side to pull-down *without* affecting the drive voltage, not on the transistor side. I had a closer look at the datasheet and it is indeed low at 3.3V and so it seems from previous discussions that we agree it's a nice 3.3V drive mosfet. *If* there was more drive voltage the Ron would be lower still and so would the heat dissipation too.

But the insulator on the heatsink is a definite redundancy.

*Peter*

Phil Pilgrim (PhiPi) · 2009-09-04 04:53

I've used power ICs (UDN2998W) that required that the tab be the main feed and would self-destruct if you ignored that and used the pin instead!

-Phil

Phil Pilgrim (PhiPi) · 2009-09-04 04:55

Peter,

Got it about the pull-down resistor. Thanks for the clarification. And you're right: put it on the Prop side to get the most drive voltage you can.

-Phil

ElectricAye · 2009-09-04 13:27

Peter Jakacki said...
@Phil - my point about the resistor was that it needs to be on the prop side to pull-down *without* affecting the drive voltage, not on the transistor side....

Okay, I think I understand what this means and why my Prop-to-gate voltage took a dive when the +15 volt supply was turned on. It's nice to know I don't need to leave my Propeller totally exposed to the inevitable screw-ups I am destined to inflict upon it. And these thermal pads, etc. are now on my To Buy list. I've learned a lot thanks to you guys!

I'm just soaking up the love here,
Mark

Toby Seckshund · 2009-09-04 16:03

Just remember that the insulator adds one more interface surface and its own thermal resistance and so the heatsinking will not be quite so good. At low volts it is not a problem to have live metal and if the volts get higher put a cage around it, but then you will require forced air.·A lot of·mains SMPSUs have live (350 V here) heatsinks.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Style and grace : Nil point

Agent420 · 2009-09-04 16:14

I've used various mosfets for switching and pwm applications, and never really had any thermal issues...· Really the only time heat is generated is during the transition phases;· simple on/off and pwm circuits typically spend very little time in partial conductive states, so heat should not be an issue.· In fact, that is pretty much the whole point of pwm.

For example, I've had mosfets controlling multi amp steppers, and never even needed a heatsink at all.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔

localroger · 2009-09-04 18:06

I can top this. I regularly work with an industrial terminal that has an exposed power supply PCB mounted to the back plate, which you must remove to get inside the device. This PCB has a forest of heat sinks, and one of them is hot with line voltage. Everyone I know who works with the thing has accidentally touched it at least once.

Toby Seckshund · 2009-09-04 18:59

Back when I was a lad ...

The senior engineers would give us a "training course" on measuring the incoming 11KV rails. This involved lowering down the 250KVA 3 phase Switch fuses, racking them out and making sure the guard plates were locked back into the open position. With the EHT probe the rails could the be got at, if you sat cross legged in the bay and looked strait up. The volts would oblige by fizzing and crackling a bit as the probe neared.

On the HF sets the 11KV was 3 phase rectified smoothed a bit, and stuck onto the anodes. A lot of the door interlocks could be over ridden

I don't think there was ever a need of this but the seniors probably had to go through it too and now it was the fresh lot of trainees turn. There are probably a whole raft of laws against it, now. Tested the underwear, not thermally but for absorbancy.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Style and grace : Nil point

localroger · 2009-09-05 02:06

@Toby, I once met a guy whose job had been doing exactly that kind of test on 10KV+ rails. One day he encountered a new rack that had been wired in reverse. It was OK when he hooked the test lead to the rail because the meter was sitting on a rubber pad, but when he hooked up the ground he turned the whole test apparattus into a fuse, right in front of his face. He had third degree burns over 70% of his body and spent a year in the hospital. His new job, by which I met him, was giving safety instruction.

Peter Jakacki · 2009-09-05 03:01

See Mark, what can I say? Here you are worried about low voltages and insulating them etc and yet there are the 350V hot tabs on commercial PSUs (caution! do not open). I'm not quite sure how all this relates to the 11KV bus bars but it does help to put your situation into perspective.

As I said, I wouldn't bother or want to insulate these low voltage mosfets if they had separate heatsinks so it's like a lot of overkill but your choice. It's good to know about the pros and cons of insulating devices so this thread does contribute positively to the forum. If you do heastsink, use sil pads and bushes (or nylon screws), way easier.

*Peter*

ElectricAye · 2009-09-05 03:12

Peter Jakacki said...
See Mark, what can I say? Here you are worried about low voltages and insulating them etc and yet there are the 350V hot tabs on commercial PSUs (caution! do not open).....

I must admit all these electrified heat sink tabs leave me a bit baffled and stunned. I would have never guessed such things would be legal. I don't know enough about semiconductors to understand why such tabs must be allowed to harbor any kind of voltage. But I'm sure glad I saw it on the data sheet and didn't learn the hard way. I know it's only 15 volts, but the lesson here is to always read the all of the data sheet like your life depends on it, because one day maybe it will.

thanks for helping out,
Mark

localroger · 2009-09-05 13:06

@ElectricAye:

 I would have never guessed such things would be legal.

Well you have to watch out, because there is a tax of USD$3.50 per exposed volt per unit, and if you don't pay it the Heat Sink Police will show up and confiscate your circuit board.· If the exposed voltage is over 40V you also have to pay a 20% surcharge to fund Medicare Part E, which is the universal electric hazard coverage.· It's easy enough to pay the tax by phone, the Heat Sink Police are a division of the Phone Police and they know where you live.

Also, you know that tag on your mattress that says DO NOT REMOVE?· Don't remove it unless you feel like spending the rest of your life on the Mattress Offender's List and sleeping on the grass.

Toby Seckshund · 2009-09-05 13:18

Probably the reasons that the devices are not proofed against voltage being connected to the HS is that it would lower the Wattage figures. Then they can sell a hi volts version to you as an added bonus.

There will always be some nasty volts is a mains psu. As long as there is adiquate guards, and a label "Nicht der finger gerpoken" it's all your problem.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Style and grace : Nil point

PJAllen · 2009-09-05 13:25

A GND (or V_ss, Return, et al.) in an electrical circuit is no less "live", no more or less 'safe' for the purpose of this discussion,·than·those at·a positive or negative voltage.· There are two subjective issues here: reference and expectation.· Your UL/CE-Mark/TUV consultant will be glad to assist.

Toby Seckshund · 2009-09-05 13:41

Very true. TV Tx Klystons had their anodes at GROUND and their Cathodes at -22KV. Don't assume anything.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Style and grace : Nil point

A really great MOSFET to use with the Propeller... except don't forget your the

Comments