Re: Chopper max voltage?

On 4 Oct, CAD_CAM_EDM_DRO@egroups.com wrote:

> 20Khz or there abouts is commonly quoted on Application notes, I've used the
> App note values (3.3nf and 22k) for the R/C as well as 3.3nf and 10k which
> gave about 42khz I seem to remember... The higher frequency gives higher
> switching loses in the 298 (gets hotter) but works fine. I can't remember
> the reason for the higher frequency now..

The optimum switching frequency is related to the inductance of the
motors. With a VERY low inductance motor, a very fast switching
frequency is required to prevent too much current running through the
windings for too long.

Likewise, for very high inductance motors ( unipolar motors run in
bipolar mode, for example ), a lower switching frequency is required,
although this is somewhat less critical.

( An Ericcson application note covers this to some extent, for those
REALLY interested. )

20KHz is a nice frequency for use with conventional bipolar switching
transistors, as too much higher and switching losses become very
significant. The L298 datasheet is a very interesting document from
the standpoint of power consumption.

There are two primary sources of heat from the L298. The first is
steady-state heat generation caused by the inability of the output
totem-pole drivers to turn on fully. The worst case condition shown
here ( and it's ALWAYS best to look at worst case ) is a total drop
across both transistors of 4.9 volts. At 2 amps continuous current and
both windings driven, that's almost 20 watts of power to get rid of.
Note that this loss is independent of switching frequency ( when used in
fast current decay mode ), because when a top transistor is turned off,
a bottom transistor is turned on. There are still two Vce drops to
create heat, the current being supplied by the inductor. Note too that
this power loss is relatively independent of supply voltage.

Now, 20 watts is a lot of heat to begin with, but switching losses also
start to add up at higher and higher frequencies. While we like to think
of the transistors as "switches", going from off to on and back off
again, they do so by going through a region where they are only partly
on.

In between the time the transistors are off, and generating no heat,
and the time they are "fully" on, and generating 5 watts or so of heat,
they are partly on. Worst case figures are NOT given here, but typical
figures are, to the tune of ~2.5us. For a single 20 KHz ( 50 us )
cycle, 10% of the time ( 2.5us on, 2.5us off ), the transistors are in
linear mode. In this mode, an average of just under 1/2 the power
supply voltage is being disippated by the transistor pair. At 12
volts, 2 amps and 20 KHz, an additional 1.2 watts per side ( roughly ),
or 2.4 watts total is being generated inside the device. At 36 volts
and 50 khz, we're up to six times that number, or almost 15 watts,
which is added to the 20 watts shown above.

In fact, the 20 watt figure does include some of the switching losses,
so the true number would be more like 30 watts, but that's still a LOT
of power to get rid of in a small package. I've got 30 watt soldering
irons, and their surface area ( and hence their temperature ) is not
THAT much different than that of the MultiWatt package.

Let's just sum up by saying that there's a good reason why all the
modern designs use FETs on the outputs.

Alan



--

Alan Rothenbush | The Spartans do not ask the number of the
Academic Computing Services | enemy, only where they are.
Simon Fraser University |
Burnaby, B.C., Canada | Agix of Sparta