H-Bridge Controller

When you say h-bridge circuit, this can mean just about anything, chip,
commerical device, or something you designed yourself. Hard to know
what is going on.

However, if you are using any avaialble off-the-shelf h-bridge chip for
this, it is doubtful it can handle a 1-ohm load. At 12v, a 1-ohm load
means 12-Amp, and few of the h-b chips can handle this.


- dan michaels
www.oricomtech.com
======================

I am assuming that you want to re-create the bi-polar input signal
across your resistive load, so that when the input signal is above
ground, one side of the H bridge pulls high and when the input signal is
at ground or below ground, the other side of the H bridge pulls high.

 From looking at Blanchard's H Bridge, jumper the low side inputs so
that they are driven off of the other leg of the bridge as Blanchard has
shown. Now you only have two high side control inputs to worry about.
One of the high side control inputs should be tied directly to the input
signal. The other leg of the H bridge gets it's control input inverted
with this circuit. If the Ascii art gets killed, it is just an NPN
transistor with it's emitter grounded and a 10K resistor from the
collector to +12V. A 4.7K  resistor is used to tie the signal generator
output to the base of the transistor. The output signal is at the
junction of the 10K resistor and the collector of the transistor. The
resistor values I supplied may need some fooling with. No additional
diode stuff should be needed for separating the parts of the square wave.

                      +12V
                      -----
                        |
                        |
                        \
                        /    10K or so
                        \
                        /
                        |-------------------  To other leg high side
                        |                      input
                        /    C
                      |/
              4.7K    |
    sig in --/\/\-----|      NPN
                    B |
                      |\
                       \/    E
                         \
                          |
                          |
                          GND


 From looking at the drive circuits on that bridge, I don't think that
bridge will work very well, especially not at high frequencies. I would
expect problems to start showing up at frequencies above a few KHz.

Good Luck,
Bob

Thanks for all the replies! For Dan, the schematic of the bridge can be
found here. http://www.geocities.com/fet_h_bridge/np-s.pdf . I made the
circuit using a printed circuit board and some standard parts, the
MOSFETS I am using are IRF4905, and IRF3205. For Bob, I'm going to try
the circuit you have suggested this afternoon, and hopefully it works.
I've been trying to replace the PNP transistor on the high side, which
didn't work. As far as problems when the frequencies are < 1KHz, I dont
expect to be using this circuit any higher than probably 100 Hz. I
think also I'll get rid of the diodes I have on there.

- JJ

In the previous message, where I said PNP, I meant NPN

The inverter circuit you suggested worked like a charm. Thanks for the
idea! Anyway, am I correct in assuming that this circuit setup would
not need any negative component in the f.g. output to the bridge?

- JJ

I'm glad it worked for you. Yes, you are correct that it does not need
the function generator to go negative, only near to ground. An LM555
timer would drive this just fine instead of a function generator.

Bob

The NPN is an OK choice for that location, it will switch the FET on as
quickly as possible. The switch off is likely to be slow and dirty
without changing over to a "totem pole" driver configuration (NPN and a
PNP). It does make dead time (the time between turn off of the high side
and turn on of the low side FETs or vice versa) harder to control
externally.

Because the high side and low side FETS are controlled separately, I
would suggest doing your debugging with a current limited power supply.
If the current limiting is set to to a value that the FETS can safely
carry and there is not a lot of capacitance on the DC bus, errors and
problems will result in the supply limiting rather than the FETs
catching fire.

Once the circuit is working the way you want it, adding capacitance on
the DC bus is a good thing.

Good Luck,
Bob

Looking at the schematic shows that's a rather marginal electronic
design. Here's a few problems I see:

1. the 1N4001 diodes are probably too small and too slow for the
capability of the bridge. The mosfets can handle much larger currents
than 1 amp.

2. the ckt will not switch very fast. Q2+Q6 use passive pullups to turn
off, and the large value 47K base resistor in Q1+Q5 will probably cause
slow turn on of those devices.

3. if you dirve this with external signals coming in at C+D, they have
to go fully 0-12v. or else Q4+Q8 will probably never turn off.

4. if you're not careful with the signals on A+B, you can get all 4
mosfets on simultaneously, and short the battery and blow the mosfets.

5. I haven't looked at it closely, but these sorts of cross-coupled
ckts can have timing issues regards turn-on and turn-off race
conditions.

6. another disadvantage of such cross-coupled ckts is that you
generally cannot get active motor braking.

Remeber, you're driving a 1ohm load, which means 12A and 12W and a lot
of heat.
So, have fun ... :)

Actually, I should have said 144W, and a ==lot== of heat.

I = V/R = 12v / 1ohm = 12A
P = V^2/R = 12*12 / 1 = 144W
or V*I = 12V * 12A = 144W

Can you point a noob like me towards a better design ?
I'm a mechanical engineer turned programmer, so this EE stuff hurts my
head : (

Thanks,
-JSM

dan michaels wrote:

Motor controller design is more difficult than most other electronic
design. There are many problems involved in dealing with large
inductive loads, the more so the more amps are being switched. Most
digital design is a breeze in comparison. I've designed a number of
motor controllers, and am not really happy with any of them, for
various reasons. I've literally blown the ends off of large 50W
extruded load-resistors. They fire like a bullet :).

For a non-electrical engineer, I would suggest buying a commercial
controller.

[oops, another addendum]

For a non-electrical engineer, I would suggest buying a commercial
controller .... if doing anything more than simply hooking up a
standard
h-bridge chip [typically good to only 2-3 Amps of current].

You might take a look at the Open Source Motor Control project. It is
now a Yahoo group. The original project was the power section for a
robot drive, as I remember it was in the 50+ amp range and I believe
that they got a good working board. These guys sell bare boards from the
project and have some of their own stuff as well:
     http://www.robotpower.com/products/osmc_info.html

The project has wandered off into a brushless multiphase project that
isn't relevant to what you want but the original is probably a decent
starting point.

If you want to learn it and blow up FET's on your own, International
Rectifier has a couple of good app notes and sells driver chips and
FET's. I took this path: it was interesting. It was an easy way to burn
a year of spare time and a pot full of FET's and I am an EE.

Good Luck,
Bob

I did find that this circuit attenuates an 8V at a 2 Ohm load cycling
at 100Hz after a couple of minutes. The real disadvantage that I am
having with this thing is that since the voltage to the motor activates
the MOSFETs as well, when you lower the motor voltage less than the
gate voltage (~4V), the mosfets don't fully open, so they heat up and
there is serious motor voltage attenuation. Hopefully I won't have to
drive it at a lower voltage than 5V. There isn't an appreciable amount
of inductance in the load that I have this thing hooked up to, and
braking isn't really necessary for the operation. What really would be
nice is some sort of IC that would send voltage to the 4 MOSFETs on a
trigger I guess. I'm going to have to use the existing circuit until I
can find a way to do this. Somebody on another forum suggested I use a
PWM IC. The reason I don't use one is that there is an internal
oscillator in the chip, where the frequency is regulated by a RC setup.
Is there some sort of IC that can use an external trigger to pulse
voltage to the MOSFETs? Or even A and B?

I don't have a solution to your problem, but there are a couple of
things you should be aware of.

First off, 4v is near the threshold level of any MOSFET, and you need
to have at least 5v gate-source voltage to turn them on adequately, so
you'll have problems unless you use a special charge-pump circuit to
generate the gate voltages, along with a "separate" supply to drive the
pump ckt. Logic-level MOSFETS will have a lower Rdson at Vgs = 5v than
regular MOSFETs.

Secondly, you need to check the minimum Vdd for any off-the-shelf
h-bridge chip. Not many will work at Vdd as low as 5v. Many require at
least 10v minimum, I assume many due to the Vgs levels required.


- dan michaels
www.oricomtech.com
=====================