Grant,
That's going to be the same filter you would see on the output of a SMPS like
in a computer, adjusted for higher voltage/current. You need to pass 16 kHz
but attenuate everything above 1 MHz. If I understand that this is a DC motor
driving a ring rotator and you are switching 180V off and on at 16 kHz to
control speed. Outputs are ground and switched DC.
Look at the square wave output with a scope. You might be able to filter it
down some if the square wave output has lots of ringing or overshoot/undershoot
on the square wave. They want to switch those transistors very fast so that
they don't heat up and overshoot is the result. Not to mention that the rise
time of that square wave is going to determine how high in frequency those
harmonics go. However, rolling that off will increase heating in the
transistors.
I think the big problem is that you probably have a 100'+ of cable going from
the PWM box to the ring rotor and that the twisted pair might not be enough.
It might require shielded twisted pair, especially since you note that you can
hear it on another tower a distance away. Throw a big chunk of ferrite on the
pair at the power supply end. If you notice a big change, especially on the
higher bands if you use the standard type 43, 850u, 20-200 MHz, ferrite then
this is a common mode problem and not purely differential. Put a separate
chunk of ferrite on each lead if you think it is differential. You want to
make sure to have enough there so that it doesn't saturate under high current.
I have been seeing 5000u and 10000u ferrite showing up with ferrite vendors. I
have some samples of it that I am experimenting with for baluns. Looks too
lossy for RF power applications and the supplier is selling it for EMC
suppression. Wound some 9:1 beverage transformers up and need to get around to
testing them. Nice to be able to use 4-5 turns instead of 60 to make the
transformer. Have to see how lossy they will be. The problem with these cores
in your application I think is going to be the high current and saturation of
the core making it useless as a suppressor, unless you have a big enough core.
Earl
N8SS
------------------------------
Message: 3
Date: Wed, 26 Aug 2015 19:11:59 -0700
From: Grant Saviers <grants2@pacbell.net>
To: towertalk@contesting.com
Subject: [TowerTalk] filter for rotator DC motor PWM amplifier
Message-ID: <55DE71EF.0@pacbell.net>
Content-Type: text/plain; charset=utf-8; format=flowed
I'm seeking some guidance for the design of a PWM amplifier noise filter
for a 180v 1/2 hp DC motor that drives a ring rotator. My googling
turns up a lot of information for 3 phase VFD filters but little for DC
motor applications, so I'm hoping there is some experience/expertise
among towertalkians.
The PWM amp is a commercial Minarik 16Khz switch and drives the motor
via twisted pair. The amp is in a RFI gasketed Hoffman steel enclosure
and I have a Corcom 20a/240v line filter on the AC source. However, I
have hash everywhere and S9 noise on a separate tower 20m antenna. The
overwhelming source is the PWM amp, not motor brush hash. The box is
grounded to the tower master ground plate.
I think a DC filter design is similar that of AC line filters, a equal
current forcing/common mode balun followed by capacitors followed by an
L-C filter in each leg. My major question is about the practical
parameters of a balun in this application. A high perm toroid seems a
likely choice for a parallel L1-L2 balanced winding. I have some
FT-193-JA cores, u=5000 and the question is what assumptions/parameters
govern the balun winding design? For the L in the L-C section, high
current toroid inductors are widely available (e.g 50uH @ 10amps).
Any links to references would be appreciated.
Grant KZ1W
------------------------------
_______________________________________________
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
|