(Not a stompbox) Solar power disconnect via transistor

[acehobojoe]

Hey guys! I know this isn't stompbox, but it is DIY!

I am trying to recreate some of the ways we use transistors to power on stompboxes at a much bigger scale.

I'm disconnecting power from a battery bank via a switch. Since the switch will be ontop of a hill, I don't want any voltage drop from sending the main power there. I was thinking a transistor would probably be best. Something to control a higher voltage with a lower voltage.

I found a suitable transistor, that would probably be fine with a 5v supply or so, permitting the full power on it's collector/emitter. http://www.mouser.com/ds/2/308/MJ14001-D-79064.pdf

but then the question is, how do I scale the power down to a miniscule amount to be on the base of the collector?

it's a switch kind of like this setup. I know the theory, but I was wondering if anyone here has worked with similar things.
http://sub.allaboutcircuits.com/images/03078.png

[R.G.]

I used to design switching power supplies for a living, and I designed a solar power controller for a friend last year, so the issues are current with me.

I suggest you use a power MOSFET, not a bipolar transistor, for a couple of reasons.
- bipolar transistors like the MJ series can only saturate down to about half a volt under any significant current, sometimes not that little.
- MOSFETs can be lower "on" voltage if you are careful and take the current levels into account, where bipolars can't.
- MOSFETs take dramatically lower driving power if you don't insist that they switch fast. If you just want to hold one on, the driving power is zero.

I'd need to know the solar panel voltage and currents to suggest a suitable MOSFET.

[acehobojoe]

We are working on different solar battery systems. There is a constant voltage of either 12, 24, or 48v. The 12v one will have a max current of around 40A. There is less current for the higher voltages (20A or 10A).

[acehobojoe]

And about the MOSFETs, we don't really care if it is fast or not. It just needs to hold on and off, so I'm sure it will work great.

[acehobojoe]

 

I was also going to look at some surge protection ideas with some varistors and GDTs. got any ideas?

[PRR]

> control a higher voltage with a lower voltage.

And where are you going to get that lower voltage?

Usually it is convenient to get it from the existing power source. Then it makes some sense to pick a device which can be controlled with that voltage. Especially if the device's current can be traded-off against its voltage. Otherwise you are blowing-off precious energy in a dropping resistor.

I like relays. (Just bought three.) They can be had in 6V, 12V, 24V, 48V (and DC or AC). This eliminates the resistor (and calculation) in your linked plan. The *power* is about the same for any voltage; the current needed drops as the voltage is raised. Offset/loss voltage can be near Zero. Surges and such rarely do harm. Insulation and heatsinking are non-issues.

Go to the car-parts and ask for a "Starter Solenoid Switch, ST81" (fits 1966 Ford Galaxie but also many later trucks)(*). This will switch >200A intermittently in a hot engine bay; should do 40A in the shade all day long. Draws ~~1A (IIRC), 12W loss, -far- smaller than the 112W of the BJT scheme. There is an equivalent part for 24V big-truck systems. In industry, this is one type of "contactor", Mouser or DigiKey will show you an infinite list of contactors with their continuous-duty (not starter) ratings. They may come in 48V; that's not uncommon in larger industrial work. If not, measure current on rated voltage and compute a drop resistor, accept the small extra loss. (Or find a wide-range 20V-50VDC to 12V switching regulator, and run safer 12V on the long line.)

The MOSFET is also a dandy thing. It may take some mA to switch it at high audio frequency, but if you want a few times a day the input current is about Zero. Ideally the "base resistor" can be zero (at least for 12V work), but small charge to over +/-20V blows the gate, also an open-gate invites radio transmissions, so there will be more or less R/C/D network involved.

MJ14002 will "not" pass 40 Amps, well, for long. At that point it will drop about 2.0V. Your 12V becomes 10V. The transistor dissipates 40A*2V= 80 Watts, which is similar to the heat of a 160 Watt audio amplifier, but in just one device (we would use 4 or 6 for an amplifier that big). The base current must be at least 40A/hFE, and while hFE may be high, they only promise 15. This is 2.7 Amps base current! Taken from 12V supply, 32 Watts of heat. About 7 Watts in the transistor B-E junction and 25 Watts in the resistor. The total 112 Watts of waste looms large on a 480 Watt system.

(*)or ST80 (silver contacts, $45!), ST85, ST71, ST67

[R.G.]

I like relays. (Just bought three.) They can be had in 6V, 12V, 24V, 48V (and DC or AC). This eliminates the resistor (and calculation) in your linked plan. The *power* is about the same for any voltage; the current needed drops as the voltage is raised. Offset/loss voltage can be near Zero. Surges and such rarely do harm. Insulation and heatsinking are non-issues.

Coil current and power are significant issues for relays in solar systems, having just been through this.

One way to get MOSFET gate drive is to use PVI MOSFET drivers that work from anything that can drive an LED.  The LED's needs make this something you can't ignore, but it's much smaller than a 50A relay coil.

The MOSFET is also a dandy thing. It may take some mA to switch it at high audio frequency, but if you want a few times a day the input current is about Zero. Ideally the "base resistor" can be zero (at least for 12V work), but small charge to over +/-20V blows the gate, also an open-gate invites radio transmissions, so there will be more or less R/C/D network involved.

I used one R, one C, and the MOSFET had a built-in zener gate protector. If the MOSFET used doesn't have gate zeners, one 12V zener does it.

MJ14002 will "not" pass 40 Amps, well, for long. At that point it will drop about 2.0V. Your 12V becomes 10V. The transistor dissipates 40A*2V= 80 Watts, which is similar to the heat of a 160 Watt audio amplifier, but in just one device (we would use 4 or 6 for an amplifier that big). The base current must be at least 40A/hFE, and while hFE may be high, they only promise 15. This is 2.7 Amps base current! Taken from 12V supply, 32 Watts of heat. About 7 Watts in the transistor B-E junction and 25 Watts in the resistor. The total 112 Watts of waste looms large on a 480 Watt system.

Yep. MOSFETs suffer from the channel resistance causing I^2 R losses, but they parallel well enough to make it a non-issue. Single 60A/60V devices are under $2 each, and you can put several in parallel. The resistive losses do force modest sharing; at least it's non-destructive.

Mouser sells the IRFB7537PBF for $2.06 each in ones. It's rated for 60V, 173A (!) and 230W. This set of specs is highly speculative, as it's in a TO-220 package, and the chip heats up by 0.8C/W (about) and so heat sinking is necessary even though it saturates to 3.5milliohm typical. This is a good place to start for N-channels. P-channels are probably simpler, as you won't be using the solar array unless it's up around 10+ volts, and that's a decent gate enhancement.

P-channel devices will cost about 2-4 times as much for the same nominal power/current/voltage, but will be easier to drive.

[acehobojoe]

I was thinking of using a buck converter to switch on the transistor with 5v.

Let me draw a diagram.

[R.G.]

MOSFETs do not switch on like bipolars. They require a threshold voltage, often 2.5-5V, below which nothing happens. After that, they typically pass something like between one and ten amperes per volt of gate enhancement. The gate takes no DC current at all, of course. So if you're using a MOSFET at high current, you want something like 10-12V on the gate/source.

Another nice thing is you don't have to build a buck converter. 

[acehobojoe]

Great! Here's a picture of what I'm doing :

http://i.imgur.com/294Hx3f.jpg

I'll build a schematic and put it soon. Do you think I should run two mosfetes in parallel just in case as mentioned above?

[R.G.]

Given that, I'd use a MOSFET as a main switch. How big a MOSFET and whether you need paralleling depends on the well pump.

Historically, motor loads are hard to drive, for relays and semiconductors both, because of the highly inductive nature of the motor. MOSFETs and IGBTs have proven themselves more reliable than bipolars, so this is a good fit. However, I have some friends who have gone massively solar, and I get consulted all the time for electronic widgets to do things involving solar panels. There are a host of other considerations in this setup.

The characteristics of the bobber will matter. Typical well-tank bobbers are fairly rugged affairs - and "rugged" often means "not too sensitive to fine points". They are typically intended to switch motor currents all by themselves, so the contacts are not well designed for "dry contact" switching, meaning very low currents and voltages.

A critical issue is whether there are batteries attached to this setup, or just the solar panels. Batteries imply that there can be a minimum voltage available. Solar panels only can feed a very low voltage/current to the whole setup, which means trickier design for those situations where there may be only marginal solar power available for driving the motor. Also, you can get into marginal conditions where the solar voltage starts things, the motor starts and overloads the available power, the voltage drops, the motor/controller turns off, the voltage comes back up... so the motor can sit there being turned on/off/on/off for long periods of time at dawn/dusk and in cloudy conditions.

The MOSFET switching is easy. Using enough smarts to make it reliable under the conditions can get tricky.

That being said, I would design a circuit that
- looks at the solar voltage available and decides when there was enough to run the motor and electronics
- when motor turn-off happens, hold it off for some period of time, at least a few seconds, so you don't get into on/off oscillation so badly
- worked well with the bobber signal; this will take some tinkering unless you just get lucky

The actual drive to the MOSFET is pretty easy - have your controller circuit decide when there's enough voltage to turn it on, and switch a big honking MOSFET to V-. Do NOT forget a big, honking reverse flyback diode on the motor; ideally you'd put some transient suppression right at the motor in the form of some capacitance and TVS or MOV suppressors. Clamp the MOSFET to a decent chunk of metal for some thermal inertia. I don't think it will get all that hot from current if it's driven well, but the startup and shutdown loads on the motor will be brutal. The gate enhancement needs to be probably 12V minimum, but no more than 20V, so the drive voltage is a good match to 12-18V CMOS logic chips to run the controller.

[acehobojoe]

There are batteries for sure. 

Should have drawn them

[acehobojoe]

That being said, I would design a circuit that
- looks at the solar voltage available and decides when there was enough to run the motor and electronics
- when motor turn-off happens, hold it off for some period of time, at least a few seconds, so you don't get into on/off oscillation so badly
- worked well with the bobber signal; this will take some tinkering unless you just get lucky

We really don't have to mess around with the solar supply, because he has charge controllers and many great circuits to make it all smooth. The batteries are what are really driving the motor.

Could I not just use the IRFB7537PBF mosfet connected to the high current path, and then run a separate low current signal into a switch for turning the mosfet on and off. We could use a small regulator since Vin-Vout x output current = heat dissipation, Heat won't be a huge issue with the regulator, because of the low current on the gate. Although, what was the gate enhancement you mentioned? Do we even need to worry about a controller? All we have to do is disconnect the positive side of the power, I believe.

We'll just have to test with long cable runs and make the bobber perform like it does in real life to see if it handles the low voltage "well" at all.

Then we'll add protection with diodes for spikes, and look into some other surge suppression. These solar circuits tend to be loaded up with surge protection because of the possibility of lightning strikes. They also cover the whole board with some sort of protective film. If we end up with a successful design, the boss will want 50, and we can coat all of the pcbs to protect them.

[R.G.]

Ah. Not a hack job for a friend - a commercial deal.

So what's your boss willing to pay for a complete, mature design?   

[acehobojoe]

well, he will probably pay commission for each board that is sold. He does solar installs and well pumps all over the U.S. and it's not hard to convince someone to add this small protection for 50-100 extra, especially when it's for a 10,000 install.

I'm not entirely sure why he needs a low water disconnect, but I told him it was possible, and he wants to invest a couple of hundred into a prototype, and then he's fine with the boards being around 5$ ea. for 200 or so.

I was just spitballing and told him we could probably have the device in an enclosure and fit all of the components for around 20$ each. I'd have to populate them though 

[acehobojoe]

It's also not necessarily about making a little money for him. It's just a pretty specific design he can't find. He's worked with relays and all kinds of other things, and he just wants less hassle.

[acehobojoe]

the 12v should be fine to turn on the Drain to source according to the VGS chart.

Not sure if I'm missing anything though!

[Brisance]

Coil current and power are significant issues for relays in solar systems, having just been through this.

even with latching relays?

[acehobojoe]

I may need to change that mosfet setup to be grounded. Hooking the positive and negative to the drain of the mosfet, while having the source connected to ground. Testing it out today.