DPRG
DPRG List  



[DPRG] thermal glue

Subject: [DPRG] thermal glue
From: David Peterson robodave1 at attbi.com
Date: Sun Dec 22 02:21:01 CST 2002

Actually, 50 Hz PWM can very well be the problem. The motor is indeed a
power conversion device, but it is also an inductor. Higher current
requirements from low frequencies heat both the motor and the h-bridge.
Here's a quote from an email about reasons behind choices in PWM frequency,
explaining it quite nicely :
***********
   Universally, the motor inductance is used to smooth the DC motor
current. The concept of producing a smooth DC voltage across the motor is
not
applicable here - only the motor current waveform is important. The net
result
is that there is very little point operating a PWM motor controller above 25
kHz
or so, and you will find that the vast majority of commercial motor
controllers
do indeed operate below 25 kHz.
   OK. So why not operate (as was suggested) at 500 Hz, or even 50 Hz or 5
Hz? The
answer is that at frequencies below typically a few kHz, the motor
inductance is
insufficient to "smooth" the motor current. So what, you may say, for in
fact
the controller would still "work", even at a few Hz. Actually, long ago when
all
transistors switched very slowly, motor controllers had no choice but to
operate
at 1 kHz or less, and the result was (and still is) that the motor runs much
hotter and less efficiently at partial speed than if the PWM frequency was
higher. Why?
   An example is easiest. Let's say the supply voltage is 24V, and the motor
is
operating at 25% of  full speed, at full rated torque, corresponding to an
average current of 10 amps. Let the motor DC resistance be 1.0 ohms. With a
high
frequency PWM controller (or a variable DC power supply) the average motor
voltage is 6V, the current is a smooth and constant 10 amps, and the
resistive
motor dissipation is IxIxR = 100 watts.
   Now compare with a PWM controller operating at say, 100Hz, so there is
negligible current "smoothing", and the motor current is effectively either
"on"
or "off". To obtain 25% of full motor speed the contoller will be "on" for
25%
of the time, just as with the high frequency controller. During the "on"
period,
the motor current will be 40 amps, so the average motor current is still 10
amps, as before. The resistive dissipation during the "on" time is 40x40x1 =
1600 watts, so the average dissiaption is 25% of this, or 400 watts. Hmmm.
Motor
dissipation is 4 times higher than for the high frequency controller, and
the
motor will run damn hot, or quite likely burn out.
   The bottom line here is that the PWM frequency should be chosen high
enough for
the motor current waveform to be relatively smooth, and generally this means
an
operating frequency of 5kHz or above. Add the fact that audible motor whine
can
be quite objectionable, and you can see why commercial motor controllers
usually
operate in the 10 to 20 kHz range.
********
   Also, the 754410 has a maximum rating of 2075 mW continuous free air
power dissipation, though perhaps derating this to a third of it's value as
Chuck has done might be an adequate rule of thumb. Both using ground planes
connected to the 4 grounds of the chip and using heatsinks, even possibly
wired to these grounds help significantly in getting heat off the chip.

Dave

----- Original Message -----
>From: "Chuck McManis" <cmcmanis at mcmanis.com>
To: "Daryl Gallatin" <dgall723 at mail.dal.devry.edu>; <DPRGlist at dprg.org>
Sent: Saturday, December 21, 2002 5:05 PM
Subject: Re: [DPRG] thermal glue


> At 03:26 PM 12/20/2002 -0600, Daryl Gallatin wrote:
> >... the TI SN754410 H-Bridge driver ... overheats in about 30
> >secs with only small 6V DC motors and I'm pulsing the motors at 50hz
>
> Don't be confused by motor size, it isn't a good metric. A DC motor that
is
> 6V can draw 175 amps at stall if it is wired that way. Not surprisingly
> some of the R/C car motors do exactly that because they are expecting to
be
> hooked into NiCd battery packs with FET based speed controllers. As the
> magnetic field a motor produces is proportional to the current it is
> drawing, the higher the current the more "torque" the motor can deliver
and
> thus the easier it is to use it directly rather than through a gear
system.
>
> Remember that motors are fundamentally power conversion devices, the
> convert electrical power into mechanical power. You can rate them in watts
> just like you do electrical circuits. So a 10 watt DC motor running at 6V
> will be drawing about 1.8 - 2.0 amps in order to deliver that power
> (efficiency is power out / power in and a 75 - 85% efficient DC motor is
> not uncommon.) so 10 watts out needs 10/.8 or 12.5 Watts input.
>
> The TI H-bridge is limited to a steady state dissipation of about 750
> milliwatts of power without a heat sink. Slightly more more if you use the
> recommended ground plane/heat sink pattern underneath the chip, slightly
> less if you use it in a socket (it conducts its heat through the ground
> leads and the socket reduces efficiency there.
>
> If you solder it to the board, have the heat sink pattern in 1 oz copper
or
> better, and have one of the glue on heatsinks on the back. You can run 1
> amp motors pretty much all day. The trick is making sure your motors are
in
> fact 1 amp when they are running.
>
> I don't believe the 50hz PWM is an issue.
>
> --Chuck
>


More information about the DPRG mailing list