High side MOSFET driver for solar panel question

Thread Starter

russpatterson

Joined Feb 1, 2010
353
Hello,

I'm just first experimenting with MOSFET drivers. I bought a Microchip TC4432 High side driver http://ww1.microchip.com/downloads/en/DeviceDoc/21424d.pdf

I'm using an N-Channel MOSFET. The source is connected to a 12V battery positive lead. The Drain is connected to the solar panel positive lead. I can control the current flow (panel to battery) through the MOSFET with the TRC4432. Although I see a little current (400uA or so) getting through even when I have it turned off. Is that normal?

What confused me in the data sheet was the two output pins, source and sink output. They talk about tuning rise and fall times with different resistors on either pin. I just connected the source output (Pin 7) to the MOSFET gate directly (no series resistor). Should I have a pull-up resistor on that output pin? If so how do I determine the value of that?

They also say that supply input, VDD, should be bypassed to ground with a local ceramic capacitor. I left that off too. How should I determine the value of that cap?

I'm using the IRL3714Z http://parts.digikey.com/1/parts/1022752-mosfet-n-ch-20v-36a-220ab-irl3714z.html

I originally bought these for a low side application. They are rated at 20V VDss, most panels go no higher than 18V but I've seen them get up to 22V. So I should probably get a different part for this application?

Thanks in advance.
 
Last edited:

nickelflipper

Joined Jun 2, 2010
280
My limited experience with solar buck converter killed off a couple of high side Pfets due to 20V gate source breakdown specs. The ringing induced by the switching will drive the gate voltage well above the max rating, and pfffffft:(, although you don't actually hear it. A gate resistor would be required to reduce the ringing affect, especially if this a buck converter circuit. A scope would be really handy to see effects of different resistor values.

Tried a feeble attempt to introduce a zener on the gate to source, no luck. Went to a part with a gate source rating of 30V and problem went away. So something like a FDP8876 or higher rated part would seem a better choice?

I thought microchip gave the cap value for their driver, somewhere around 1uf ceramic from memory?, check their smps demo board schematic to see what they use.

For an Nfet high side wouldn't be a pulldown on the gate?, I had a 47k pullup on my Pfet gate.
 

Thread Starter

russpatterson

Joined Feb 1, 2010
353
Fair enough. I had to make the part for the TC4432 this morning. Here's the schematic. I haven't yet connected the MCU IO Pin to TC432 control input yet, just experimented on the bread board.

I updated the schematic. The Supply input was not labeled correctly.
 

Attachments

Last edited:

Thread Starter

russpatterson

Joined Feb 1, 2010
353
For the cap value, they say: "The value of this capacitor should be chosen based on the capacitive load that is being driven". The switching test circuit they show, (attached) a 4.7uF and a 0.1uF cap. So I'm not sure how to know the capacitance of the load being driven. Would that be the battery? How would I learn that value. Also, once I know it how to I get the cap value(s) from that?

So a series resistor between the driver output and the gate? They brag that the driver can source 3A on the the output pin. So wouldn't you need a really small, like 7 Ohm, series resistor? Wouldn't it also need to be big to handle the power?

Thanks for the input on the MOSFET Vdss rating. I thought that setup might be a part cooker. I switched to a NDP6060L, 60V N-Channel MOSFET that I had bought for a "planned" buck circuit. This circuit is just intended to limit current for a 3 stage battery charger.
 

Attachments

wayneh

Joined Sep 9, 2010
17,496
Unless I'm turned around again, current can flow backwards from the battery thru the panel with this configuration, via the body diode of the MOSFET. This will happen even with the gate held low; again, it's a diode. Turning it around would orient the body diode so that current can flow thru that diode if the panel is hotter than the battery, but not the reverse. Something to consider.
 

nickelflipper

Joined Jun 2, 2010
280
russpatterson:
Well should've looked at the TC4432 data sheet first. The terminology seems kind of backwards, but you were correct in wanting to source the output pin7, which they call "Drive Output Pull-Up".

I used VDD supplied directly from the solar panel. If the VDD is several volts lower than the gate source max, then the IRL3714Z may just be fine. That said, using the solar panel for VDD, and the !LOCK DIS pin could be a good strategy to shutdown the circuit when low light conditions exist?

Composing this post when you replied with the new part NDP6060L. My previous comments on the FET related to Vgss not Vdss. The gate source max (+/-16V) is actually lower than the previous part. Otherwise, looks really stout if you use 12V, or the Battery, as VDD. For some reason I choose the panel output as VDD (i.e. gate drive) for my experiments, can't remember why.

I am a bit befuddled also on the resistor and cap values. That 3A drive is with VDD at 30V. Right or wrong, I just used a 1/4watt resistor without much thinking about it.
 

Thread Starter

russpatterson

Joined Feb 1, 2010
353
Thanks for the heads up. The panels I'm using have diodes to prevent current flowing back in. This also allows you to orient multiple panels differently which allows for more complete charging given different weather and seasonal changes.
 

wayneh

Joined Sep 9, 2010
17,496
The panels I'm using have diodes to prevent current flowing back in.
That's a nice convenience but a source of substantial loss at low output. The diode drops ~0.7v across itself and if you're getting only 6v or so out, that 0.7v drop represents more than 10% loss. The loss is proportionally much less if your panels are at higher voltage.
 

Thread Starter

russpatterson

Joined Feb 1, 2010
353
The panels will push 18-22V open circuit. I've done a bunch of testing and two panels positioned for peak morning and afternoon Sun will outperform two panels positioned both for mid-day sun. You get a longer power producing window. Many days out of the year are partially cloudy so expanding the time you can charge/run pays off in the long run.

I think you could rig up a FET based reverse current limiter, but that's more expense.
 

Thread Starter

russpatterson

Joined Feb 1, 2010
353
Thanks for checking into that. I think that's right. You should tie those two outputs together. I think they intend that you put a different series resistor between each output and the MOSFET gate.

Check out the picture. Pin7, Source output drives the gate high, in this case ~14V and Pin6, Sink output pulls it low. When I tie them together you get a perfect square wave.

I still get ~400uA bleeding through the MOSFET when it's off. That bugs me. For the size of battery I'm using I can't see that overcharging it, and it could be my crappy $60 meter reading badly at low currents as well.

So maybe something like a 10 Ohm resistor on the Source Out and a 20 Ohm on the sink out would protect the part.

Overall I got it to limit current flow to the battery very nicely when connected to the PIC PWM.
 

Attachments

I still get ~400uA bleeding through the MOSFET when it's off. That bugs me. For the size of battery I'm using I can't see that overcharging it, and it could be my crappy $60 meter reading badly at low currents as well.
Have you got a pull down resister on the gate drive output, say a 47kΩ, then what do you get? The $60 meter should be O.K. I would think. If there is a low side current sense resistor, then you could double check with the scope.

So for right now this is just a PWM , not a buck converter, charger? How many watts total are the panels, and what type and rating is the battery bank?
 

Thread Starter

russpatterson

Joined Feb 1, 2010
353
I do not have a pull-down on the gate. I have a 150 Ohm series resistor between the two output pins on the driver and the gate. You think I should put a 47K pull-down on the gate? I'll give it a try but the driver says it can sink 1.5A so that's pretty darn pull-down already right?

What exactly do you mean by a current sense resistor to test the meter with? Put in a voltage across a small resistor and measure current to check that your meter reads what you calculate using Ohm's law? That would assume that my volt meter is accurate too. Time to buy a decent Fluke meter I guess. Do you mean check the voltage with the scope and use that as the dependable measurement?

This circuit is for a 3-stage lead acid battery charger. I want it to support panels that produce from 15 watts to 210 watts and charge batteries from 5aH to 120aH. The charge curve will have to be different for the different batteries. I'm not sure how to determine the length of the saturation charge phase, and the exact saturation and float voltage levels.
 
Yes I meant to use the oscilloscope as a cross check to the DVM or meter. And yes I presumed that a current sense resistor was being used as feedback to the micro, so I am going to stop guessing.

That's a wide range of conditions for the charger. There are a fair number of charger ic chips out there. Check them out for charging algorithms, like say bq2031, LT3652, or? A recurring theme on current mode cutoff seems to be about 1/10 of Imax set point.

Looks like a real challenge, and best of luck.
 

Thread Starter

russpatterson

Joined Feb 1, 2010
353
Thanks for the input. For this first charger I was going to leave off the current sensor. If I throttle the current with the PWM on the Panel MOSFET then I can hold the voltage to specific levels. So I don't need to know the current. This is the first time I've built one of these so this may not work, (not knowing the current).

I plan to match the panels to the batteries so at full power the panel won't push more amps than the battery can take. Otherwise you're just wasting money on panels that are too big.

I wasn't aware of the battery charging IC's. Thanks, those are cool but don't handle the amps I'm after. I'll look at the algorithms they use. Good idea.

If I make the device handle 20 amps, then the 1.5 amp setup should work fine. The parts to handle the larger amperage aren't that much more money so I figured just build for the larger current. I have a version testing for the last few months that handles 7 amps charge current and will run 12V pumps at 10 amps.
 
Top