SparkFun Forums

[Vraz]

I am working on a wireless weather station that uses uses a solar panel and rechargeable battery for power. Typically a diode is used between the solar panel and battery to prevent back current from flowing through the solar panel at night. My plan was to replace the diode with a high-side load switch (P-channel MOSFET) allowing the following:

1- Read solar panel voltage (voltage divider to ADC on uC while solar panel is disconnected)
2- Eliminate the forward voltage drop of a diode (disconnect panel when voltage is below charging range)
3- Use PWM to reduce panel output if it exceeds charging range (since I am not using a regulator between the solar panel and battery).

The high side switch I intended to use was the Fairchild FDC6331 (http://www.fairchildsemi.com/ds/FD/FDC6331L.pdf). It is essentially a P-channel MOSFET with an N-channel driver. It felt like I was being clever by taking advantage of the uC to control the MOSFET and get multiple benefits. That is when it struck me that MOSFETs have body diodes that can conduct current back through them. It would seem like that would still allow the battery to discharge through the body diode to the solar panel at night. Thus, I started looking for similar applications to see if anyone had tried this before. I found the following design note from Linear Technology (http://www.linear.com/pc/downloadDocume ... 142,D14157) which seemed to suggest exactly my proposed solution.

Question-- What prevents the MOSFET body diode from discharging the battery through the solar panel at night?

[felis]

Question-- What prevents the MOSFET body diode from discharging the battery through the solar panel at night?

Reverse bias.

[NleahciM]

If you want to block the body diode you can put two MOSFETs back to back (source to source). Tie their gates together and run it that way. Works quite well.

[felis]

When present, the current flow from solar panel to the battery doesn't need to be blocked, consequently second MOSFET is not needed. Conducting MOSFET here is just an enhancement to its own body diode (which is used just as typical solar panel blocking diode ).

[Vraz]

Just a little clarification that might not have been clear-- I always want to block current flow from the battery to the solar panel. I want to switch current flow on/off from the solar panel to the battery.

Question-- What prevents the MOSFET body diode from discharging the battery through the solar panel at night?

Reverse bias.

Apologies, but I don't understand. In my application, the P-MOSFET is connected Source to panel and Drain to battery. Thus, the MOSFET "switch" (via the Gate) allows/prevents current flow from the panel to the battery. When the MOSFET is off and the battery voltage exceeds the panel voltage (because its dark), it seems like current would flow backwards through the MOSFET through its body diode (and the flow would be forward relative to the body diode). Am I completely missing the point here?

If you want to block the body diode you can put two MOSFETs back to back (source to source). Tie their gates together and run it that way. Works quite well.

I assume you are suggesting two P-MOSFET's back to back? I understand how this blocks the body diode while the the MOSFETs are off. However, when the MOSFETs are on, it seems like this adds the voltage drop of one of the two MOSFET body diodes (depending on the direction of current flow). If so, would I be better off just using a Schottky diode plus a single MOSFET since I don't actually require bi-directional current flow?

When present, the current flow from solar panel to the battery doesn't need to be blocked,

Actually, this is something I wanted. I don't want to overcharge my battery and want to be able to disconnect the solar panel at times. Also, I wanted to be able to monitor panel voltage and battery voltage independently. I assume this can only happen after disconnecting them, reading the voltages and then reconnecting.


My original schematic. (The FDC6331 is just a P-MOSFET with driver.)


From Linear Technology App Note 1012 showing a similar concept (switching the panel with a P-MOSFET based on the panel voltage) but claiming no diode drop and no discharge current-- how does it work?

[leon_heller]

Q3 is turned off!

Leon

[felis]

Just a little clarification that might not have been clear-- I always want to block current flow from the battery to the solar panel. I want to switch current flow on/off from the solar panel to the battery.

It's not going to work with a single power MOSFET; they are not capable of blocking current in both directions. A pair of MOSFETs conected drain to drain will work.

Apologies, but I don't understand. In my application, the P-MOSFET is connected Source to panel and Drain to battery.

It's connected other way around( drain to panel ) on Linear schematic.

I understand how this blocks the body diode while the the MOSFETs are off. However, when the MOSFETs are on, it seems like this adds the voltage drop of one of the two MOSFET body diodes (depending on the direction of current flow).

Diodes will never conduct. Draw the schematic of two MOSFETs with gates and drains( or sources ) connected, you will see two diodes looking at each other.

[felis]

Another option would be to use "four-legged" MOSFET, the one with body terminal, if you can find one with current rating suitable for your application.

[Vraz]

Appreciate the responses and apologizing for seeming dense on this, but something still does not add up for me.

It's not going to work with a single power MOSFET; they are not capable of blocking current in both directions. A pair of MOSFETs conected drain to drain will work.

This makes sense and confirms what I suspected would be the case.

Diodes will never conduct. Draw the schematic of two MOSFETs with gates and drains( or sources ) connected, you will see two diodes looking at each other.

Understand that when >off<, the body diodes being in opposite directions will stop any current flow. What happens when both MOSFETs are >on<? Don't the P-MOSFETs only conduct from source to drain? Thus, since the MOSFETs are back to back, only one of them will see current from source to drain (the other sees current from drain to source). Wont the former conduct with almost no voltage drop and Rds resistance while the latter will experience Vsd voltage drop because of the body diode.

Am I wrong in assuming that when "on", the MOSFET only conducts in one direction but that it fact it conducts equally well in both directions (source to drain and drain to source) with Rds resistance and no voltage drop? If that was the case, then things would suddenly make more sense.

[felis]

Am I wrong in assuming that when "on", the MOSFET only conducts in one direction

The channel is symmetrical, there is no difference between drain and source if body is not connected to one of them. If body is connected, the device behaves very much like a switch with diode across it - it conducts current in both directions when on and blocks current in one direction when off.

[NleahciM]

If you want to block the body diode you can put two MOSFETs back to back (source to source). Tie their gates together and run it that way. Works quite well.

I assume you are suggesting two P-MOSFET's back to back? I understand how this blocks the body diode while the the MOSFETs are off. However, when the MOSFETs are on, it seems like this adds the voltage drop of one of the two MOSFET body diodes (depending on the direction of current flow). If so, would I be better off just using a Schottky diode plus a single MOSFET since I don't actually require bi-directional current flow?

You could use N-FETs too, you'd just have to use fancier drive electronics. What you're missing is that MOSFETs, when on, allow current in either direction, with an essentially identical RDSON. So no, you would not have the voltage drop of a body diode plus the voltage drop across the RDSON of the other FET. You'd have two times the voltage drop across the RDSON of one FET.

However - it sounds like all you're looking for is an ideal diode. You want the reverse current blocking of a diode, without the voltage drop. For this you want an ideal diode controller like you posted about - the LTC4412. What this circuit does is set up the FET so that the body diode conducts only in the forward direction of your ideal diode. Depending on the specific ideal diode controller you use, the controller will watch the voltage drop across the FET (from drain to source). In the circuit you posted from the Linear app note, they are measuring this from pin 1 to pin 6. Typically, the gate pin will be regulated to maintain a constant VDS. If the VDS starts going too low the chip will lower |VGS| and keep on lowering it trying to maintain a positive voltage across the chip. By doing this, it allows current to pass one direction, but not the other, while significantly lowering the forward voltage across your "diode" (mosfet).

[Vraz]

Thanks felis and NleahciM

That a MOSFET conducts in both directions when "on" was the missing piece. While I have some experience using N-channel MOSFETs, I have always used them in applications where the current flow was unidirectional. When I saw the Linear app note, I figured something in my understanding must be wrong as I understood what they were trying to accomplish with their "ideal diode", but it couldn't possibly work if MOSFETs were unidirectional.

For my application, I will look to see if anyone makes a high-side load switch similar to the FDC6331 but with back-to-back high-side MOSFETs. Alternatively, I could use a pair of FDC6331s, one on the solar panel and one on the battery. That would allow disconnecting the battery when its fully charged and running exclusively from the solar panel which might have some advantages.

[metaforest]

Am I missing something here? Solar panels are just huge diode junction arrays, aren't they?

Why would battery voltage leak back into the solar panel? Wouldn't it's reverse breakdown be more than enough to block the battery? Is the reverse leakage current really an issue?

[Vraz]

Why would battery voltage leak back into the solar panel? Wouldn't it's reverse breakdown be more than enough to block the battery? Is the reverse leakage current really an issue?

Just about every solar charger schematic includes a blocking diode of some sort, though I can find no good references detailing how much "dark current" can be lost. The one general reference I have seen is that when you reverse bias the solar cell, it will emit infrared light and that is the cause of the current loss. (And analyzing the infrared light production pattern can actually be used to analyze the cell for defects.) What I have been unable to find (in spite of much searching) is how large the current loss is and what conditions it depends on.

Since my project needs to run 365 days a year and winter days are pretty short in MN, I felt it was safest to include a reverse current block but didn't want to give up the forward drop of the diode. Thanks to the advice here, I redesigned my circuit to use a pair of back-to-back P-MOSFETs. Hopefully that will do the trick.

[leon_heller]

Solar cells aren't built from silicon diode junction arrays! They are too inefficient.

Leon