[EE]: Adapting DIY optoisolated MOSFET driver

[Byron Jeff]

During a search for MOSFET drivers I came across a couple of interesting
tutorials by Julian Ilett using a dual complementary optoisolators to
switch the gate of a high side N-channel MOSFET between the source terminal
and a bootstrap capacitor. If you're interested here are some of them:

Each have an excellent explanation of both the use of a bootstrap capacitor
for the gate and the use of optoisolators in a complementary configuration.

Before moving on, personally I'm not interested in black box gate drivers
at the present time. I'm really interested in how those drivers work and
how they can be duplicated with junkbox discrete parts.

My real interest is how to adapt what is a trivially simple driver into
specifically some step down lithium battery charging projects I have near the top
of my project list. However I'm running into some issues in adapting these.
Here are the main problems:

1. In Julian's examples the load (a lightbulb) is grounded. So when the
mosfet is off, the terminals of the bulb, along with the source terminal of
the mosfet are at ground potential. A battery as a load will never be
grounded at the source. So for example charging a 40V pack, the source of
the high side mosfet is always at 30V or higher.

2. In the same vein, unlike Julian's example where the top voltage in the
circuit is 12V, the main charging voltage will be about 50V. While 12V is
clearly good as Vgs, 50V clearly is not. I have some golf cart DC/DC
converters that will output 12V or so, but getting it switched is going to
be a bit of a challenge.

3. The optoisolators are the third issue. They are 6N139. Their CTR is
excellent, so they can drive the totem pole for the gate, However, they
have a maximum Vce of 18V, running into the same problem as #2.


So given these circumstances, I'm interested in the following:

1. How to switch the bootstrap cap negative terminal between the battery
voltage and ground to charge it?

2. How to protect the optos which may have up to 44V between the the
collector and the emitter?

3. How to limit the gate voltage to a standard Vgs of 15-20V?

Any suggestions welcome.

BAJ
--
Byron A. Jeff
Associate Professor: Department of Computer Science and Information Technology
College of Information and Mathematical Sciences
Clayton State University
http://faculty.clayton.edu/bjeff

[Byron Jeff]

I think that I've found some solutions to the problems I offered below.
Just wanted to come back and share. As I addressed before, Julian Ilett
offered some interesting optoisolated drivers...

Post by Byron Jeff
During a search for MOSFET drivers I came across a couple of interesting
tutorials by Julian Ilett using a dual complementary optoisolators to
switch the gate of a high side N-channel MOSFET between the source terminal
http://youtu.be/tebprQvrqk4
http://youtu.be/uff1nzFo140
http://youtu.be/iNYeww1Sjzk
Each have an excellent explanation of both the use of a bootstrap capacitor
for the gate and the use of optoisolators in a complementary configuration.
Before moving on, personally I'm not interested in black box gate drivers
at the present time. I'm really interested in how those drivers work and
how they can be duplicated with junkbox discrete parts.

I found a solution in a Arduino thread titled "Automotive PWM MOSFET
drivers." One of the posters offers a really simple circuit, which uses a
NPN, diode, and a pullup resistor to create an emitter follower that
actively pulls up the emitter to the upper rail with the lower NPN turns
off, then turns off once the output has reached that point. I believe that
it solves most of my problems.

The second circuit is a form of a Dickson charge pump. The charge pump is a
simple device that ties the top end of a cap to steering diodes, then uses
a clock signal on the lower leg of the cap to alternately charge the cap
through one diode when grounded, then squeezing the charge into a second
cap through a second diode when the clock is raised. A storage cap after
the second diode maintains the higher voltage when the pump cap is dropped
back to recharge.

The adaptations I made was to the charge pump were:

1. Source the pump cap to 12V or so, which is more than enough to raise the
voltage on the gate to turn the N-channel fully on.

2. To clock the charge pump cap between 0V and the top rail of the charger.
At 0V, the pump cap charges to 10.8V. When the clock switches to the top
rail voltage that 10.8V is added to the top rail, squeezing 10.2V + the top
rail voltage into the storage cap.

3. Tie the bottom leg of the storage cap to the top rail voltage. So the
storage cap is really only charged the 10V or so. But the top leg of the
cap will be 10V above the top rail voltage.

So essentially driving the charge pump clock circuit with a PWM signal will
provide a permanent gate voltage for the high side N-channel MOSFET for the
half bridge.

Post by Byron Jeff
My real interest is how to adapt what is a trivially simple driver into
specifically some step down lithium battery charging projects I have near the top
of my project list. However I'm running into some issues in adapting these.
1. In Julian's examples the load (a lightbulb) is grounded. So when the
mosfet is off, the terminals of the bulb, along with the source terminal of
the mosfet are at ground potential. A battery as a load will never be
grounded at the source. So for example charging a 40V pack, the source of
the high side mosfet is always at 30V or higher.

The charge pump solves the problem by independently switching the lower leg
of the pump cap between GND and the top rail using the steering diodes to
isolate the pump action between 12V and ground from the switch action at
the battery voltage at the top.

Post by Byron Jeff
2. In the same vein, unlike Julian's example where the top voltage in the
circuit is 12V, the main charging voltage will be about 50V. While 12V is
clearly good as Vgs, 50V clearly is not. I have some golf cart DC/DC
converters that will output 12V or so, but getting it switched is going to
be a bit of a challenge.

The golf cart converter will supply the steady 12V to the charge pump. The
totem pole switch will switch the charge pump between 0V and the top rail
now.

Post by Byron Jeff
3. The optoisolators are the third issue. They are 6N139. Their CTR is
excellent, so they can drive the totem pole for the gate, However, they
have a maximum Vce of 18V, running into the same problem as #2.

I technically only need the opto in one place: the totem pole switch
between the gate voltage from the storage cap and source terminal of the
high side mosfet. But there is also a low side mosfet that is driven with a
complementary signal. So I could use a DCOI to drive both MOSFETS. But
neither gate is going to be driven more than 10-12 volts away from the
source. So the 18V Vce of the optos won't be exceeded.

Post by Byron Jeff
1. How to switch the bootstrap cap negative terminal between the battery
voltage and ground to charge it?

Totem pole to switch the bottom leg of the charge cap between 0 and the top
rail voltage. 2 NPNs, 2 base resistors, and a diode.

Post by Byron Jeff
2. How to protect the optos which may have up to 44V between the the
collector and the emitter?

By not having 44V between the collector and the emitter. All gate voltages
only swing 12V or so.

Post by Byron Jeff
3. How to limit the gate voltage to a standard Vgs of 15-20V?

By having the charge pump cap only charge to 12V.

Post by Byron Jeff
Any suggestions welcome.

Still interested in seeing if anyone has put together such a driver...


BAJ
--
Byron A. Jeff
Associate Professor: Department of Computer Science and Information Technology
College of Information and Mathematical Sciences
Clayton State University
http://faculty.clayton.edu/bjeff