Hi all,

I am building (trying to build) a mini underwater ROV for a bit of fun. I have everything working except, I have overlooked the charging of the battery, and more specifically the electrolysis of the charging contacts when in salt water. I am mainly concerned with:
- Maintaining exposed charge contacts (so that the ROV can sit on a charging cradle without the hassle of unscrewing ports)
- Ensuring that the battery is fully and properly charged by the external charger (I do not wish to integrate the charger into the ROV, due to heat dissipation issues)
- Charging a single cell Li-ion at 2.5A (max 5A if possible)

My initial thought was a diode between port and battery, i.e. between battery and charger. This has many issues the most significant being that the voltage drop across the diode would change the charging characteristics.

My next thought was a MOSFET which was 'switched' on when the charger was connected, which then lead me to an ideal diode design such as the LTC4352.

My question is, how would you guys stop the reveres current from the battery and how would you ensure the battery charger was charging the battery correctly, considering there is probably going to be a voltage drop across any switch or diode used?

It is worth noting: I have a microcontroller monitoring battery voltage and so there are a few I/O pins free which could be used to implement some sort of switching, and the charger I am using can have the voltage parameters altered i.e. if the was going to be a constant voltage drop independent of current I could increase the charger CC/CV voltage characteristics. I have included two high-level schematics, the first 'basic' is what I hope to achieve bare minimum, where as the second also includes a rectifier so the polarity of the charger contacts is interchangeable.

Any help confirming my direction would be much appreciated!

James

 

Yes, I would think you could use a low on-resistance MOSFET that would be externally controlled by the uC for your "basic" circuit.
You can readily buy a MOSFET with few mΩ of on-resistance which will have negligible effect on the charging characteristics.

The "ultimate" design would be more complex.
You can't use a diode bridge due to the voltage drops, so you would have to use something like a MOSFET bridge controlled by the external voltage polarity.

 

Thanks Crutschow,

Attached is the kind of thing I am thinking. I went for a P-Type MOSFET as the voltage at S and D, under charge conditions, is higher that the logic level of my μC. Obviously the parts need specifying, but does the principal of the design seem reasonable?

Thanks

 

One possible problem with that circuit is, if the uC is powered from the battery, and the battery is very flat (or Li-ion when the protection circuit kicks in, there will no supply voltage to the uC so it will not be able to switch on the charger.

 

Thanks for the feed back Albert, In principal you are right. In practice the protection is provided by the uC and so output to the load is terminated before full discharge occurs. There should always be just enough juice left to activate charging. If the battery voltage is such that it has dropped below that needed for the uC, it may well be a good thing that the battery cannot be recharged, as it is most likely damaged and therefore not really the kind of thing you want to be charging in a sealed container! (I have a pressure relief built into the design but it is catastrophic...)

 

You apparently have Q2 shorted to ground, which I don't think you want.
Also the source and drain must be interchanged (drain goes to the charger with R3 connected to the source).

But your circuit may not work, since the battery voltage will feed back through Q2 (MOSFETs are bidirectional) when you remove the charger, making the uC think the charger is still attached.
For it to work properly you would have to detect the difference between the charge voltage and the battery voltage when the charger is removed which may be quite small.

Alternately you could add a separate connector terminal from the charger to the uC sense line to detect when the charger is attached.
Actually with that connection you don't even need the uC in the loop, you can just connect the added charger terminal directly to R1.

But the simplest solution my be to use the LTC4352 ideal diode circuit.

 

You apparently have Q2 shorted to ground, which I don't think you want.
Also the source and drain must be interchanged (drain goes to the charger with R3 connected to the source).

I don't understand what you mean by Q2 shorted to ground.

The charger+ connection must be positive of the battery+ connection (when the charger is connected). It is a P-channel FET so the source should be connected to the charger and the drain to the battery so it is the right way round.

I agree with the bit about the body diode.

PS I think R3 should be connected between source and gate of Q2.

 

As Albert says, as it is a P-channel FET and as the +v supply of the charger is 'more +v' than the battery I assumed it should be the way I presented it?

I suppose if R3 was small then I would effectively be shorting Q2 to ground with Q1 turned on. It's my fault as I should have put values on the components (although I am still undecided). I have placed R3 as a pull up resistor to keep Q2 turned off untill Q1 brings the gate down to ground to turn it on. I was thinking around 1MOhm for R3. Without R3, what is to bring Q2's gate up to turn it off once Q1 is turned off?

I was also concerned about the body diode, do you think I could put a pull down resistor from the +ve charge pin to ground, there by having a significant change in voltage when the charger is connected? As I understand it the body diodes are far from ideal and only pass very small amounts of current?

 

I have placed R3 as a pull up resistor to keep Q2 turned off untill Q1 brings the gate down to ground to turn it on.

But if the charger is connected R3 will connect the gate to a more negative voltage than the source thus risking actually turning it on when it should be off. Connect R3 between gate and source of Q2 then when Q1 is not conducting it will keep the gate at the same voltage as the source.

the body diodes are far from ideal and only pass very small amounts of current?

No. They are generally rated for the same current as the MOSFET.

 

Hi all,

I am building (trying to build) a mini underwater ROV for a bit of fun. I have everything working except, I have overlooked the charging of the battery, and more specifically the electrolysis of the charging contacts when in salt water. I am mainly concerned with:
- Maintaining exposed charge contacts (so that the ROV can sit on a charging cradle without the hassle of unscrewing ports)
- Ensuring that the battery is fully and properly charged by the external charger (I do not wish to integrate the charger into the ROV, due to heat dissipation issues)
- Charging a single cell Li-ion at 2.5A (max 5A if possible)

My initial thought was a diode between port and battery, i.e. between battery and charger. This has many issues the most significant being that the voltage drop across the diode would change the charging characteristics.

I routinely use a series diode on lithium batteries for my E-cig, firstly it bypasses the button so the threaded connector can double as a charging point. Secondly - I use a shunt regulator, if that should happen to fail; the diode protects the battery from the short. You just have to measure Vf while charge current flows and calibrate the regulator accordingly.

Lead chemistry is fairly critical voltage points - but without the dangers of lithium.

Nickel chemistry is usually constant current charging, so a Vf is of little consequence as long as the charger can meet the voltage requirement.

A Shottky-barrier diode has Vf around 0.2V compared to about 0.7V for regular silicon - but SB has some leakage, so you'd have to resistor shunt the charging contacts to eliminate electrolysis.

 

But if the charger is connected R3 will connect the gate to a more negative voltage than the source thus risking actually turning it on when it should be off. Connect R3 between gate and source of Q2 then when Q1 is not conducting it will keep the gate at the same voltage as the source.

Got ya, yes I will have a think about that one.

No. They are generally rated for the same current as the MOSFET.

Does this mean that the whole exercise is futile, as the diode will effectively create a path by which current can pass from the battery to the +v charge pin, which is what I was trying to avoid in the first place?

 

I routinely use a series diode on lithium batteries for my E-cig, firstly it bypasses the button so the threaded connector can double as a charging point. Secondly - I use a shunt regulator, if that should happen to fail; the diode protects the battery from the short. You just have to measure Vf while charge current flows and calibrate the regulator accordingly.

A diode was my first thought, but as Vf varies with current I was not confident that it was a safe solution for charging Li-ion? At what point to do you measure the Vf of your diode? In CV mode as the forward current drops so does Vf. if you have calibrated your charger for Vf @ max If (in CC mode), would you not be over charging your cell by Vf(max)-Vf(min) at the end of the cycle?

 

Does this mean that the whole exercise is futile, as the diode will effectively create a path by which current can pass from the battery to the +v charge pin, which is what I was trying to avoid in the first place?

Yes it would be a problem.

But the simplest solution my be to use the LTC4352 ideal diode circuit.

Agree.

 

So with all this said, how does the LTC4352 overcome the fact that all MOSFETs have body diodes and perform as an ideal diode using a standard MOSFET? Based on what has been said here, it should not work should it?

 

It uses the FET in the 'reverse' direction. MOSFETs are bi-directional so when the gate-source has voltage to turn on the fet it will conduct in either direction. In your case the fet is connected so that the conduction direction of the body diode is to pass current from the charger to the battery, but then the chip turns on the fet to conduct and so reduce the voltage drop to a few tens of millivots. When the charger voltage is below the battery voltage (or disconnected) the body diode is reverse biased and the fet is switchd off. Thus it acts like you have a 'perfect' diode.

 

It uses the FET in the 'reverse' direction. MOSFETs are bi-directional so when the gate-source has voltage to turn on the fet it will conduct in either direction. In your case the fet is connected so that the conduction direction of the body diode is to pass current from the charger to the battery, but then the chip turns on the fet to conduct and so reduce the voltage drop to a few tens of millivots. When the charger voltage is below the battery voltage (or disconnected) the body diode is reverse biased and the fet is switchd off. Thus it acts like you have a 'perfect' diode.

Thanks Albert, with this in mid I have edited my design and revered the FET. This should work now should it not?

 

I have edited my design and revered the FET. This should work now should it not?

No it won't. If the fet was fully conducting then the Charge+ and battery+ will be at the same voltage but you need somewhere positive of that to get a suitable voltage for the gate to turn on the fet. The 'ideal' diode uses a charge pump to achieve that.

 

No it won't. If the fet was fully conducting then the Charge+ and battery+ will be at the same voltage but you need somewhere positive of that to get a suitable voltage for the gate to turn on the fet. The 'ideal' diode uses a charge pump to achieve that.

As it is a P-Channel FET, surly I just need the gate to be < source - threshold. So if the source of Q2 is at +V (3V min), the gate is at ~0V and the FET has a min threshold of -1.6V, then it should be on?

 

Regarding the circuit in post #16.
I have difficulty seeing how T2 can disconnect again?

When the charging circuit is connected then both T1 and T2 is on.
When the charging circuit is disconnected after the charging, then T2 still provide battery voltage back into the "+ Ve Charge" terminal.

 

Have you considered inductive charging? Making reliable electrical contacts in salt water will be problematic.