Prop and Mosfets - 48V DC 7A
GeeksGoneBad
Posts: 100
I need to drive four solenoids with 48V DC - is there a good mosfet that would work with the 3.3V?
I have a bunch of IRF511 (80V N Channel mosfet) but it needs 5V to turn it on (at least in my experiments - I am not really good at reading the datasheets yet)
I obviously want to keep my part count down and complexity somewhat simple
I WAS trying to use a 5V relay and ULN2803 but the relay is only good to 28V
ANY ideas are uber appreciated
I have a bunch of IRF511 (80V N Channel mosfet) but it needs 5V to turn it on (at least in my experiments - I am not really good at reading the datasheets yet)
I obviously want to keep my part count down and complexity somewhat simple
I WAS trying to use a 5V relay and ULN2803 but the relay is only good to 28V
ANY ideas are uber appreciated
Comments
For that amount of current and voltage, optoisolation will save you a boatload of headaches. In the long run, you won't regret it.
-Phil
55V N-channel mosfet
Gate threshold voltage 1.0-2.0 V
I_DS 49 amps
less than $2.00
The biggest missing feature, compared to typical relays, is the lack of multipole contacts.
-Phil
+1024
It is not easy to find High Drain Voltage, and very low thresholds specified.
If it is not for volume production, you may be OK with a more typical spec, and selecting a over-rated part.
Or, you could use a Gate Driver, or Level Shifter, and choose a cheaper, less over-rated FET.
If you want Common GND, Hi Side drive, companies like Allegro (A3942) and Linear (LT1161) make Quad High SIde drivers, for Generic N-FETS.
Or, if you want to isolate, and roll your own SSR, then the IR PVI50xx series are interesting devices - Mini Photo-Cells that bolt straight onto a FET Gate - fine for relay replacement use.
DO you have a part# for the solenoids as you mentioned 7A but is that pull-in current or continuous? Must be pull-in as otherwise the solenoid would be glowing.
But I characterized them myself. See:
IRLZ44N
See the graph.
This should do just fine at 3V.
If you want to see the raw data see:
IRLZ44N Excel Data
This is the data for 1 but I then check at least 25 more.
Duane J
The MOSFET Cascode amplifier
The bottom MOSFET, IRF3708, can be a low threshold voltage type and the top MOSFET, IRL540N, can be a high voltage type.
Also, the CasCode generally switches much faster than the normal single MOSFET circuit mainly due to the low Drain to Source voltage.
Duane J
Hey, that's interesting. Might such a circuit, or one similar to it, be fast enough to handle pulse widths of about 10 microseconds at about 10 amps and, say, 12-ish volts?
In your cascode circuit, where is the discharge path for Q2's gate when Q1 turns off?
-Phil
However, whether using CasCode or conventional MOSFET circuit one must use one of the high drive current MOSFET gate drivers.
The main advantage of the CasCode is there is relatively low Drain to Gate charge transfer due to the low drain voltage.
10uS is not to fast. Good MOSFET gate drivers can switch a MOSFET in 50nS.
The OP was looking for a relay driver driven directly from the Prop.
If you have another purpose that requires 1uS rise times the naked Prop just cant do it. A driver clearly is required.
Duane J
If Q1 is OFF, it doesn't matter if Q2 is off or not... but in this case it does turn off.... From a layout perspective, the Source and Drain can be swapped, however from a discrete component perspective there is a bulk diode to deal with. In any case it still works out in this configuration where the bulk diode still remains reverse biased. When Q1 is OFF, then the reference voltage to Q2's Source (acting as a Drain) is higher than the Gate voltage of 5V so Q2 turns 'OFF'
-Phil
This isn't for production, it's a one off project - just for myself
the solenoids are actually 24V 10A but my power supply is 48V 7A and I am really just "clicking" the solenoid quickly - not a constant push - and also not a very high frequency - more like one click every 30 seconds maybe
OK so I'm going to study the thread now and I'll ask more n00bish questions in a while
Thanks!
15k 10k 48V >----/\/\----o----/\/\----< GND | o------------> To Gate of Q2Well yes, you could use a voltage divider or zener diode to develop a higher voltage on the gate of slave Q2. One thing to keep in mind is there should be a low impedance connection between the gate of slave Q2 and the source of master Q1. This is best done with a stiff power supply but a resister divider and capacitor, a relatively large one of say 10uF to 100uF in parallel with at least a 100nF ceramic capacitor will do nicely. Remember, there will be a fairly high current in and out of the gate of slave Q2 for a few uS or so.
My schematic showed an IRL540N MOSFET which has a guaranteed on resistance of 53mO at a gate voltage of 5V. This is considered a "Logic" gate voltage MOSFET. Yes, you could increase the gate voltage but this would be at the expense of slower switching times. For really fast switching times a MOSFET gate driver is required on master Q1.
OK, back to relay drivers. And in this case a relay that has a pretty high coil current. Speed is not really a big factor here. So, I guess a 10V gate slave MOSFET will do.
39k 10k 48V >----/\/\----o----/\/\----o----< GND | 10uF | 10V o-----||-----o | o------------> To Gate of 10V gate voltage slave Q239k 5.1k 48V >----/\/\----o----/\/\----o----< GND | 10uF | 5V o-----||-----o | o------------> To Gate of 5V gate voltage slave Q2Duane J
Mickster
BTW, your average photoFET based DC SSR is not exactly the fastest device around ... I once did a prototype job (I think I've mentioned it here before) that required some reasonably precise switching for a rotary impulse sealer, well, within 10usec (both turn-on and turn-off) at least. And had to deliver a decent punch of 100+ amps per impulse. It only had to achieve pulse lengths of around 100-200 msec every couple of seconds, so, I could take advantage of that and use both an optocoupled FET driver and parasitic supply source when turned off ... which meant it could be turned on hard and not need any heat sinking.
The parts were cheap enough, primarily the PCB, HCPL3140 and IRFP2907 and built to fit on DIN-Rail. But, given my labour to hand assemble them, work charged the customer $100, on top of the development costs, a piece to build them.
does anyone see any major problems? this works and so far has not fried anything and it switches plenty fast enough from my application
Please don't take my word on this, but I thought that any kind of load, like your K1.1, must be placed between the +V and the mosfet (your IRF) so that the "gate-to-source" voltage is properly maintained. In other words, I've always thought about a mosfet as a kind of "trap door" and so no components must be between the mosfet and the ground. I hope somebody with either confirm or deny that for me.
Oh, yes, good point. You've got your mosfet gate tied directly to ground.
but you guys are also saying that my solenoid should be on the other side of the mosfet too, right?
Yes. That's how I wire my solenoid valves. Otherwise the gate-to-ground voltage is not fixed on a known number. Beware that "source" in mosfet language actually refers to the ground - don't ask me why.
See for example what PhiPi shows here:
http://forums.parallax.com/showthread.php?137035-MOSFETs-and-the-Propeller&p=1063401&viewfull=1#post1063401
The resistor you see above PhiPi's mosfet is the load, which in your case would be the relay.
so I should put the load on the other side of the mosfet so the relay is using up the load before it goes through the mosfet - if it's on the other side then the mosfet has to endure the load as well?
to add - the IRF511 can handle 5.6A so it should work and not blow up - but why make it handle that if it doesn't have to - right?
thanks for the help guys
oh and what simulation software are you using?
That's probably not the best way to think about it. Let me suggest this mental picture:
The mosfet is a trap door. When you open that trap door, you want your current to rush through the opening and dump straight to the ground without anything getting in its way. If there's something below the trap door, then current will get splashed around by that something and make a mess. Sometimes the splashing can happen and your circuit will seem to work, but over time this splashing can damage components, generate more heat than you want, or cause irregular and maybe unpredictable operation, etc. When the current dumps straight to ground, then no splashing happens, and everybody stays high and dry.
What's really happening inside the mosfet is that an electric field must be set up to allow the mosfet to conduct the current. For that electric field to be just right, it needs a reference point to the ground, otherwise the gate does not really "know" where it stands in the circuit, so weird things can start to happen. Imagine an astronaut floating around in space with nothing else to hold on to but a valve, and he's trying to turn that valve, which is somewhat stuck. Without having his feet planted on the ground, or holding onto something, he's just going to spin around. Under these ungrounded conditions, the astronaut might be able to get the valve to crack open and closed, but he's going to be spinning all over the place as he does so. On the other hand, if he's got his feet fixed to the ground, he'll have no problem.
BTW, I'm clueless as to the other part of your circuit. I personally just don't know anything about how those types of components are supposed to work.
I hope that helps a little.