Filter cap at dc motor input

[pimpom]

I'm rebuilding an old machine for a local company by replacing
all the electronics with my own design. It uses a 24V DC brushed
motor with an integral reduction gear. The original drive circuit
used a couple of SCRs for speed control and current was limited
to 3.5A. My circuit uses high frequency PWM (~22kHz). (The motor
control is only a small part of the overall design).

My design worked nicely at home, but lost power drastically as
soon as it's fitted on the machine. I traced this to a shunt
capacitor at the motor input. I opened the motor and sure enough,
there's an LC filter inside the gearbox housing. Like this:


+ --------L-------
| |
C |
|
--GND Motor
|
C |
| |
- --------L-------

My circuit uses a MOSFET output stage with the motor floating
between the drain and Vdd. The two caps are 1uF each. Should I
reduce them to something like .01uF each or just omit them
altogether?

 

[pimpom]

On a sunny day (Tue, 20 Jul 2010 18:39:27 +0530) it happened
"pimpom"


Those filters are there to reduce RFI originating from the
brushes.
I have a very similar setup here, and the RFI even causes bit
errors
on my sat receiver,
and that is a 600mA motor (RFI up into the GHz range, sparks
are BAD,
as from those brushes). Simple caps help only very little.
So removing the filter may cause problems.
The right way would be perhaps to use a different kind of
switcher,
with a series L and filter capacitor, if you want to use the
existing
motor.
So the motor sees pure DC, and the filter can be left intact.

Thanks. Yes, an L-C filter is probably the best solution, and it
may not even need changing the switcher design. The problem must
have been caused by the cap between the negative terminal of the
motor and ground as it essentially parallels the output
transistor except for the (small) lead inductance.

 

[E]

I'm rebuilding an old machine for a local company by replacing all the
electronics with my own design. It uses a 24V DC brushed motor with an
integral reduction gear. The original drive circuit used a couple of SCRs
for speed control and current was limited to 3.5A. My circuit uses high
frequency PWM (~22kHz). (The motor control is only a small part of the
overall design).

Try lower PWM frequency, like 50 Hz or whatever the original SCR
circuit used.

-ek

 

[Cydrome Leader]

Try lower PWM frequency, like 50 Hz or whatever the original SCR
circuit used.

-ek

This sounds like the best idea. Even at 2 or 3kHz I ran into weirdness
with motors that had capacitors across the leads. They would race instead
of running slow though.

 

[pimpom]

This sounds like the best idea. Even at 2 or 3kHz I ran into
weirdness
with motors that had capacitors across the leads. They would
race
instead of running slow though.

It's not the cap across the motor terminals per se that's causing
trouble. As I said earlier, it runs fine and speed control works
as expected as long as the motor frame is not grounded. (I use a
slotted wheel optical tachometer for feedback and, with a
counter, for duration control).

It's when the motor is mounted on the steel frame of the machine,
connecting it to electrical ground, that things go wrong. The cap
from the negative terminal (also MOSFET drain) to ground must be
playing havoc with the switching output.

 

[pimpom]

Try lower PWM frequency, like 50 Hz or whatever the original
SCR
circuit used.

I'll consider that option. Thanks.

 

[Grant]

It's not the cap across the motor terminals per se that's causing
trouble. As I said earlier, it runs fine and speed control works
as expected as long as the motor frame is not grounded. (I use a
slotted wheel optical tachometer for feedback and, with a
counter, for duration control).

It's when the motor is mounted on the steel frame of the machine,
connecting it to electrical ground, that things go wrong. The cap
from the negative terminal (also MOSFET drain) to ground must be
playing havoc with the switching output.

So tip your circuit upside down and pop in a P-channel MOSFET?

Grant.

 

[pimpom]

So tip your circuit upside down and pop in a P-channel MOSFET?

I've thought of that as a possible solution, but for various
reasons not a practical one in this case: 1) I don't have a
P-channel MOSFET in stock (I can never remember to include it in
my orders). 2) It's not locally available in my small town.
Placing small orders from another city is not as easy in India as
it is in places like the US. 3) Unless it's absolutely necessary,
I want to avoid having to start the pcb design, construction and
assembly all over again.

 

[Grant]

I've thought of that as a possible solution, but for various
reasons not a practical one in this case: 1) I don't have a
P-channel MOSFET in stock (I can never remember to include it in
my orders). 2) It's not locally available in my small town.

I live in a biggish small town, can get overnight deliveries from
Sydney, about 1000km away

Placing small orders from another city is not as easy in India as
it is in places like the US.
Unless it's absolutely necessary,
I want to avoid having to start the pcb design, construction and
assembly all over again.

That's a really good reason. So you add inductor to isolate the
cap, or rip cap out of motor?

I prefer leaving the motors alone, because a replacement down the
track will cause same problem if 'they' forget to take out the cap.

Grant.

 

[pimpom]

I live in a biggish small town, can get overnight deliveries
from
Sydney, about 1000km away

Judged by population, my town is not truly small with about
200,000 souls. But it's the capital of one of the less developed
states of India in a remote corner of the country, with poor
facilities and resources. (We have little in common with the rest
of the nation). The problem of ordering from other parts of the
country has less to do with our location than with the mentality
of Indian business people.

That's a really good reason. So you add inductor to isolate
the
cap, or rip cap out of motor?

For the time being, I've gone the route of reducing the caps from
1uF to 0.01uF. This was the easiest solution, esp. as I'd already
opened the motor. It worked fine and the owners are enthusiastic
about being able to use their machine again. I'm going back
tomorrow for extended trials and final tweaking. I'll also try to
find some way of estimating the amount of RFI (I don't have
dedicated instruments for that).

I prefer leaving the motors alone, because a replacement down
the
track will cause same problem if 'they' forget to take out the
cap.

There's that, yes. I have not yet completely ruled out inserting
a series inductor. OTOH, it's very unlikely that anyone else will
work on the machine during its lifetime.

 

[Grant]

Judged by population, my town is not truly small with about
200,000 souls. But it's the capital of one of the less developed
states of India in a remote corner of the country, with poor
facilities and resources. (We have little in common with the rest
of the nation). The problem of ordering from other parts of the
country has less to do with our location than with the mentality
of Indian business people.

Ah. My country town is large (for Victoria, it's fourth largest
in state) at around 100k population, but we're only 160km from
Melbourne, so overnight from Sydney to Melbourne, then a two hour
drive up to Bendigo, delivery here around lunchtime It's good.

For the time being, I've gone the route of reducing the caps from
1uF to 0.01uF. This was the easiest solution, esp. as I'd already
opened the motor. It worked fine and the owners are enthusiastic
about being able to use their machine again. I'm going back
tomorrow for extended trials and final tweaking. I'll also try to
find some way of estimating the amount of RFI (I don't have
dedicated instruments for that).

A normal AM radio tuned off-station would tell you something,
perhaps TV or shortwave Rx for higher frequencies? Compare with
RF noise from other devices.

There's that, yes. I have not yet completely ruled out inserting
a series inductor. OTOH, it's very unlikely that anyone else will
work on the machine during its lifetime.

Then you're in control

Cheers,
Grant.