Hello everyone,

I was building a circuit that includes a High Side Driver which can be summarized from Capture2. I noticed after construction that the boot strap has a hard time charging because of high load impedance to ground and the long Vgs on times.

I found a way to use an astable oscillation with a charge pump to keep the boot strap continuously charge as seen in Catpure3.

I was playing around and found an alternative as seen in Capture1 that prevents more expensive timers and uses extra IO ports from the main IC. The 200k R1 is for the low side driver leakage, C1 is the normal boot strap, D3/D2/C2 is the charge pump injecting a 12VAC waveform at 20kHz.

I noticed that during initial charge from the boot strap when Vc1 ~ 0, instead of charging through the load path to ground, there is a much smaller impedance through D3->C2->GND when Q3 is on and thus makes the original boot strap time constant much less.

Any reason why I shouldn't try this?

Thanks,
BSECE OSU

 

Well since nobody wants to add in on the subject I just thought I'd say it works nicely.

Started out with a Darlington pair but eventually brought the gain down to keep from pushing too far into saturation with a high hfe. For the tinkerers that aren't in the industry like I am, if the release time of the BJT is too long (in my case higher frequencies -> lower impedance -> higher injected current ) and the circuit impedances don't add up to the circuits poles dynamics, pull the gain back. The BJT has a longer time recovery from saturation and gets worse with temperature (use to have issues with manufactured ballast Dali when temps were high for just this reason).

Feel free to use my design (as shown is not recommended for obvious reasons).

Cheers,
BSECE OSU

 

Just in case anybody is interested, I have attached my next attempt for all 3 high side drivers. I have a buck converter circuit to bring voltages 12.6V - 38V down to 9V then a linear regulated 5V for most of the circuit (not shown) and then a double charge pump to get to 15V to charge the boot strap.

The BJT Push Pulls minimize input impedance for the charge pumps and losses when switched by the open collector.

All of this is switched by one IC output. I kept the normal boot strap diode in for quicker initial startup and if there is a failure in the switching components, the driver will still get 9V so it won't be catastrophic.

 

Here's a look at the actual driver, its a (5-40)V sensorless DC brushless controller I have been builder for fun. Once I get a finished design, I already have a 300A board design in the mix

 

Just to keep this thread up to date, I found how to center my circuit within Vgs limits throughout the range of Vgs duty (see images with equations and scope captures of high side Vgs with circuit put together "no load")

Note: Iboot is average per cycle

 

Just curious. Why not a low side switch?

 

Hey RONV, the circuit is for a DC Brushless Motor. It uses a three phase bridge to create a rotating magnetic field induced in the wound stator poles. you like it?

 

It's all good. Now I know why you needed the charge pump.