Slant Six Forum
https://www.slantsix.org/forum/

N-channel MOSFET vs mech relay for lights & electric fan
https://www.slantsix.org/forum/viewtopic.php?t=19985		 Page 1 of 2
Author:  emsvitil [ Mon Oct 02, 2006 5:24 pm ]
Post subject:  N-channel MOSFET vs mech relay for lights & electric fan
Any downsides to using a N-channel MOSFET instead of a relay.

You can get N-MOSFETs cheaper than relays and there's no points to wear out or corrode. You can even have redundancy built in by MOSFETs in parallel (still have a cost advantage).


I usually make little black boxes when I add things like relays and wiring for neatness, so a MOSFET has a size advantage.

You could even gut out a broken relay and install a MOSFET if you wanted plug & play..........
Author:  Joshie225 [ Mon Oct 02, 2006 5:28 pm ]
Post subject: 
What's the heat dissipation if you're pulling 20 amps for an electric fan?
Author:  emsvitil [ Mon Oct 02, 2006 6:36 pm ]
Post subject: 
Quote:
What's the heat dissipation if you're pulling 20 amps for an electric fan?
Should be less than a relay.

I just measured the amp draw of a 30amp relay for the coil at 140ma.

At 14 volts, the relay eats up and needs to dissipate (14v*.14A)= 1.96 watts of heat no matter how much current it carries. Plus any heat due to the voltage loss through the point contacts and how much current.


I'm no expert on MOSFETs, but the heat gain should be from the resistance in the MOSFET and the Amps flowing though it. Shouldn't be any heat from the on/off gate as MOSFETs are voltage driven not current driven.

Here's an example MOSFET datasheet:

http://www.irf.com/product-info/datashe ... z44pbf.pdf


Since we're not switching it constantly on and off (where I think there would be heat gain as it transitions from on to off to on....) the worst case voltage loss is .028v (page 2 datasheet)

So it would have to dissipate .028v * 20A = .56 watts of heat.
Author:  Joshie225 [ Mon Oct 02, 2006 7:48 pm ]
Post subject: 
I think figure 4, normalized on resistance, is the specification we would be concerned with. The chart specification you cite is for very short duration low duty cycle. Pulse width ≤300 µs ; duty cycle ≤2%

If we use Fig. 4 we have at 60°C about 1.2Ω resistance. Since power equals I²xR we get (20A)²x1.2Ω= 480 watts. Of course the reistance will drop the current some, but 18 amps is still 389 watts.
Author:  emsvitil [ Mon Oct 02, 2006 9:28 pm ]
Post subject: 
Figure 4 is at 51 amps.


I think resistance goes up with higher amp draws......
At 20 amps it may be less...


Anybody out there work with MOSFETs on a regular basis????????
Author:  emsvitil [ Mon Oct 02, 2006 9:44 pm ]
Post subject: 
Figure 4 says normalized... whatever that means.

I looked at other data sheets, and there's no mention of pulse width and duty cycle. And they're around the .03 ohm resistance level


BTW, 2% of 1.2 ohms is .024, which is very close to the .028 figure. Might have something to do with figure 4??????????


lookup mosfet at www.digikey.com for other datasheets
Author:  Joshie225 [ Mon Oct 02, 2006 10:48 pm ]
Post subject: 
I'll ask the station engineer tomorrow about MOSFET characteristics. Doesn't the gate voltage have to be above the source voltage? That might preclude their use instead of relays.
Author:  emsvitil [ Mon Oct 02, 2006 11:16 pm ]
Post subject: 
Quote:
I'll ask the station engineer tomorrow about MOSFET characteristics. Doesn't the gate voltage have to be above the source voltage? That might preclude their use instead of relays.

Yes, to get the device on. Once on, it stays on until gate voltage goes to 0.

(Definitions for n-channel MOSFET, drain is + input, source is output, gate is control (for anyone that doesn't know))

Heres another datasheet from another company:

http://www.st.com/stonline/products/lit ... /11325.pdf

This one has a resistance of .003 (order of magnitude less)

figure 8 here is equivalent to figure 4.

Notice that this device has a higher amp rating, and that the resistance drops. (MOSFETs in general have this characteristic, so BIGGER is better)
Author:  KBB_of_TMC [ Thu Oct 05, 2006 10:45 am ]
Post subject: 
The new MOSFETS are now used in some high current applications; their very low resistance means that heat isn't a problem, provided they are turned all the way on or off. Running them partially turned on or rapidly switching is where you must consider the thermal load.

I'm sure it will work fine if your allow for the turn-of spike of an inductive load (like a big fan motor); it can hit 100's of volts and arc through a MOSFET. (Also, a motor can act as a generator when it's moving, but the voltage will be low.) A capacitor or reversed bias diode or zener would be good protection.

A MOSFET also has a very high gate impedence, so I'd have a small load on it to prevent damage there too.

Please let us know how it goes and the cost w/ connectors vs the cost of relay & socket.
Author:  emsvitil [ Thu Oct 05, 2006 1:17 pm ]
Post subject: 
Quote:
I'm sure it will work fine if your allow for the turn-of spike of an inductive load (like a big fan motor); it can hit 100's of volts and arc through a MOSFET. (Also, a motor can act as a generator when it's moving, but the voltage will be low.) A capacitor or reversed bias diode or zener would be good protection.

A MOSFET also has a very high gate impedence, so I'd have a small load on it to prevent damage there too.

Please let us know how it goes and the cost w/ connectors vs the cost of relay & socket.

Might be awhile........

I'm setting up an order from digikey, and wanted to check the feasability of a MOSFET. I figured it it was feasible, I'd get a couple while I was ordering.

For the gate, I was planning on a high resistance resistor to ground for protection. I also figured that a snubber circuit would be a good idea for a fan application.
Author:  SlantSixDan [ Thu Oct 05, 2006 4:58 pm ]
Post subject: 
I'll be interested to learn your results, but my initial reaction is that your relay-elimination project is a solution in search of a problem! :shock:
Author:  emsvitil [ Thu Oct 05, 2006 6:46 pm ]
Post subject: 
Quote:
I'll be interested to learn your results, but my initial reaction is that your relay-elimination project is a solution in search of a problem! :shock:

I've pitted a few relays, so I don't think it would be any worse...........

And it would make my control boxes I make smaller :wink:
Author:  Pierre [ Thu Oct 05, 2006 7:29 pm ]
Post subject: 
May want to look into solid state relays - mosfet based relays with input protection and everything all rolled into one. Most you may have to add is external diode on the output for inductive loads. Some even come in packages with screw terminals, completely sealed with screw holes for mounting as well. No box needed, mount against metal to dissipate heat and use ring/spade terminals for wires.
Author:  emsvitil [ Thu Oct 05, 2006 7:42 pm ]
Post subject: 
Quote:
May want to look into solid state relays - mosfet based relays with input protection and everything all rolled into one. Most you may have to add is external diode on the output for inductive loads. Some even come in packages with screw terminals, completely sealed with screw holes for mounting as well. No box needed, mount against metal to dissipate heat and use ring/spade terminals for wires.
Have you seen how much solid state relays cost?


This is a frugal DIY.............. :wink:
Author:  Pierre [ Thu Oct 05, 2006 8:26 pm ]
Post subject: 
Depends on current desired - you can sometimes nab em cheap from surplus stores.

Careful with mosfets as they are static sensitive devices. Some you can trigger even with your finger depending on input sensitivity.

FETS are popular with the robot crowd - granted most of the use here is in full H bridge configuration but still good for info gathering purposes. Look into the OSMC - open source motor controller - 4 fet per leg, no heatsink just forced air, rated for 160A continious, more with sinks.

No matter how good your heatsinks, protection circuits, etc you'll be limited to ~75A on a to-220 package.
Page 1 of 2 All times are UTC-08:00
Powered by phpBB® Forum Software © phpBB Limited
https://www.phpbb.com/