Battery PWM loading

[Mosaic]

Need some advice here.
How to derive a 10V to 18V gate drive from this situation?

 

[crutschow]

Can you connect the negative of the battery and the source terminal of M1 to the negative side of for V1 while opening the plus connection to V2?
Alternately you can get a logic level MOSFET for M1 (fully turns on at 5V) and use V2 to drive M1's gate.

 

[Mosaic]

Disconnecting V1 is difficult as it will require a P Channel FET additional to switch and additional Gate voltage to be generated.
A logic FET gate drive may not hold up as the 0.65 ohm load can cause a Vdrop on the battery which is a D.U.T. and can be undercharged.

 

[ronsimpson]

 

[crutschow]

Disconnecting V1 is difficult as it will require a P Channel FET additional to switch and additional Gate voltage to be generated.
A logic FET gate drive may not hold up as the 0.65 ohm load can cause a Vdrop on the battery which is a D.U.T. and can be undercharged.

You don't need to generate a gate voltage for a P-MOSFET in series V1 and the battery. A resistor from source to gate shuts it off, and grounding the gate turns it on.

If the battery voltage is too low to drive the MOSFET gate then you will need to generate that voltage in some other fashion.

 

[dr pepper]

No problem, a gate drive transformer, with the appropriate step up turns ratio from 6v, you can add a turn off transistor on the gate as well if you want.

 

[ronsimpson]

Question:
What is the PWM frequency?
Will you ever need to be on 100% of the time? or more than 90% of the time?

 

[crutschow]

Here's an LTspice sim of a circuit that should do what you want.
I changed the transistor to a P-MOSFET to simplify the drive requirements and allow for DC drive if needed.

 

[dr pepper]

Rons quite right there I hadnt thought of that a transformer needs a certain time to reset, so max pulse width will be around 90%.

Nice circuit cruts.

 

[ronsimpson]

This circuit will work only if (from V1- to V2- has less than 0.5 volts of noise)
C1*R2 RC time constant should be 10x of the switching frequency.
Diode makes the turn off voltage on M1 to be -0.6 volts. The gate will go a little negative.
The positive voltage swing will be on the gate.
The duty cycle can not be 100%.
If the duty cycle goes to 0% R2 will turn off M1.

 

[Mosaic]

Yes the duty cycle can be 100%. Swapping to a P channel isn't ideal for me.... B.O.M limitations. If cruts can get the N channel to work with a similar circuit, that would be swell!

I m actually thinking of a transformer less isolated supply via switched capacitive balancing. Like how its done to balance individual cells in a mono block battery.

 

[crutschow]

Yes the duty cycle can be 100%. Swapping to a P channel isn't ideal for me.... B.O.M limitations. If cruts can get the N channel to work with a similar circuit, that would be swell!

I m actually thinking of a transformer less isolated supply via switched capacitive balancing. Like how its done to balance individual cells in a mono block battery.

Sorry, the simple biasing doesn't work with an N channel.

You have B.O.M. limitations on the MOSFET type but you can add some type of switched capacitor circuit?
I don't see how a switched cap scheme would work for a 100% duty-cycle (DC).

 

[Mosaic]

I have a selection of caps in the BOM already.
No Pchannels though. Also a Pchannel will have a high-ish RdsOn for this app .

I'll try to rig up a smoothed charge pump type circuit to supply the Vcc to run the gate.

The idea is based on this device...note fig 12
**broken link removed**

 

[crutschow]

I have a selection of caps in the BOM already.
No Pchannels though. Also a Pchannel will have a high-ish RdsOn for this app .

I'll try to rig up a smoothed charge pump type circuit to supply the Vcc to run the gate.
..................

The P-MOSFET I showed in my simulation has on ON resistance of 12mΩ.

A charge pump seems like a lot of extra parts and complexity just to avoid using a P-MOSFET.

 

[Mosaic]

Well...perhaps. I have access to a 1Khz pulse already so no osc. required.
There is another Nfet handling polarity protection should the battery D.U.T be connected in reverse. Using a Pchannel requires rerouting that circuit as well.

 

[crutschow]

Well, I think the charge pump design will involve more complexity then you might think to get it working properly, but I can see you are determined to pursue that approach so, good luck.

 

[Mosaic]

Since the Tl494 is available in my BOM...it looks to be the way to go with the dual alternate NPN outputs with guaranteed dead time.

here is the sim> the isolated voltage is achieved in about 12 mS

 

[crutschow]

You need to reference that voltage to the minus side of the battery.
Below is the conceptual LTspice simulation of that.
V3 generates the switched-cap voltage referenced to the minus side of the battery (SCom).
V4 is the PWM signal driving the MOSFET gate from that voltage using an opto isolator.

 

[Mosaic]

I think that's the best solution cruts. Lower parts count. I have a free OPA in a quad 324 nearby in the cct running on regulated 12V that can produce a 10.5V V3 signal. I'll prototype today!

 

[crutschow]

You can pick up about 0.7V in the DC output if you use Schottky's for the diodes.