DPRG
DPRG List  



[DPRG] voltage multiplier for NMOS Hbridge

Subject: [DPRG] voltage multiplier for NMOS Hbridge
From: Chuck McManis cmcmanis at mcmanis.com
Date: Wed Jun 25 18:08:00 CDT 2003

At 03:17 PM 6/25/03 -0700, Ted Huntington wrote:
>That is definately happening, an entire side will fail (although the other 
>side
>is ok), but could that also be from the MOSFETs wearing down from getting too
>hot (many times the heat of the FET will easily go over spec and that will
>definately cause damage) over time?

Typically when a FET exceeds its junction temperature there are a mS or so 
where if you get back into the legal thermal range they won't 
decrystallize. However given that most FETs are arrays of FETs in parallel 
(in the silicon sense) its conceivable that the next time they will heat 
faster the next time. FETs are also susceptible to a thermal runaway 
solution where as their temperature rises so does their resistance, if you 
don't back off the current they will dissipate more heat and eventually melt.

> > Over voltage failures (where the input
> > voltage exceeds the breakdown voltage of the FET) are characterized by
> > failure of the high side FETs.
>
>Sometime I do only lose the high side FET.  Since I am within the 10V gate to
>source voltage, perhaps that could be from voltage spikes when the motor 
>stops?

Yes, to "fix" this put a capacitor between V+ and Gnd near the FETs in the 
bridge. Also be aware that you will get a "ripple" current that can be 
fairly severe. Use a low ESR capacitor (typically used in switch mode power 
supplies). This problem gets worse when there is a lot of inductance in the 
wiring between the batteries and the bridge.

>I think the 555 to the transformer is a cool idea (or perhaps even going 
>without
>the transformer) and using capacitors and diodes.  Why two supplies for each
>side, because of low current?

No, you really want to "float" the supply across the gate/source of the 
high side FETs. This is because the voltage on the source of the high side 
fets starts out near ground as the motor is stalled/starting then rises to 
near V+ as the motor spins up. If you connect the ground of the floating 
PSU to this leg, and the output is 10V relative to this ground then the 
gate sees a nice continuous 10V Vgs. If you don't do this and instead 
simply use a voltage that it relative to system "ground" then when the 
motor starts (or is stalled) there is a very high Vgs level, but then as 
the motor spins up the Vgs level seen by the gate shrinks to the delta of 
the input voltage minus the drop across the motor/lower fet combination. 
Designing a gate driver that doesn't exceed the gate input during this 
swing can be challenging. The most common "hack" is to put the zener on the 
gate so that at start time the zener is conducting and keeping the voltage 
within range and then not conducting once the motor spins up. However this 
is both inefficient and it cannot respond quickly (zeners take a while to 
cascade) so sudden spikes on the motor can still cause over voltage on the 
gate.

>I am not sure what latch-up is.

Every CMOS device has an internally created parasitic diode that conducts 
when the input voltage is reverse biased relative to the topology (so 
that's below ground in the n-channel case or above ground in the p-channel 
case). When that diode becomes forward biased it can dump charge into the 
device which turns on all the gates causing all of the CMOS transistors to 
get into an "on" state. The device gets stuck in that state and is said to 
be "latched up."


> > Overshoot? Etc.
>
>Again, not sure what means.

Overshoot occurs on a gate because the gate looks like a capacitor to the 
driver circuit. If you have a resistor on the gate or even just the wiring 
there can be sufficient inductance to create an RLC tank circuit. If your 
gate frequency gets near the resonant frequency of the tank circuit you can 
get "ringing" on the input that exceeds the voltage you are putting into 
the gate. This ringing overshoots the gate voltage.

 >I look forward to seeing the h-bridge page.  There should be a good video on
> > basic electronics and robot electronics available in public libraries, 
> but none!

I don't know about videos but there are good books. I'd highly recommend 
"The Art of Electronics" by Horowitz and Hill. Also many community colleges 
offer basic electronics classes and Amateur Radio (or HAM) clubs often have 
a number of classes on basic electronic theory and practice (needed to pass 
the radio license tests)

>I think you meant to type 30V?  Linear Tech makes voltage doubler chips?

Linear Tech makes a whole slew of chips that convert one voltage to 
another, either fixed (switching power supply chips), multiples (voltage 
doublers, triplers, quadruplers), and inverters.

--Chuck


More information about the DPRG mailing list