In article ,
Ignoramus26424 wrote:

On 2010-12-29, Joseph Gwinn wrote:
In article ,
Ignoramus24647 wrote:

On 2010-12-29, DoN. Nichols wrote:
On 2010-12-28, Ignoramus24647
wrote:
A couple of days ago, I started working on getting my 4th axis rotary
table to work.

http://igor.chudov.com/projects/Brid...act-2-CNC-Mill
/24
-Troyke-Rotary-Table-U12PNC/

I mounted it on the milling table a while ago and started integrating
it on Sunday or thereabouts. Constantly being interrupted by the
family, of course.

For feedback, it has a separate tachometer and a "resolver". This
resolver has six leads, two for rotor excitation and two for stator.

http://igor.chudov.com/manuals/Harow...300-F10-10.pdf

I would use Jon's resolver to encoder converter board:

http://pico-systems.com/resolver.html

Hmm ... the resolver has specs at 400 Hz (aircraft use) and at
2.5 KHz -- but the board seems to want to talk at 10 KHz. Any certainty
that this one will handle the 10 KHz frequency?

I have NO idea, this is a good question. I wondered about the same thing.

Yesterday I had a revelation and changed my mind. I decided that an
extra week spent on doing things the right way would be time well
spent. I decided to replace the terminal strip on the rotary table
to accommodate completely isolated stators, add a connection for home
switch, etc.

So I ripped everything apart again and will use a 12 lead cable, with
every lead pair shielded individually.

Hmm ... I would instead put each of the secondary windings from
the resolver in a separate shielded twisted pair instead of individual
leads shielded. Better common mode rejection.

Yes, this is what I tried to say, every pair is shielded.

1) 6 wires (3 pairs) for the resolver
2) 2 wires for the tachometer
3) 2 wires for home switch
4) (possibly) 2 wires for limit switch.

Best of luck,
DoN.


thanks.

Any thoughts on this excitation frequency?

Whatever the resolver wants. The datasheet provided above says 400 Hz
through 2500 Hz. I bet it is not critical within that range, but 10 KHz
is too high. I would ask Pico. Much more important to accuracy than
excitation frequency is good shielding.

Joe Gwinn

I use a multiconductor cable where each pair is shielded.

That's good and usually necessary. Pico may have an opinion on how best
to connect the shields and pairs so the capacitance from pair to shield
has little effect. The point of the shielding is to prevent leakage of
excitation drive signals into the signals coming from the resolver,
which will cause phase shifts and thus angle errors.

The resolver angular accuracy is +/- 10 minutes of arc, according to the
datasheet. If we assume leakage in quadrature with the true signal,
then leakage ratio exceeding Tan[10' of arc]= 0.0029:1 voltage, or about
-51 dB will cause that magnitude of error. Now, it is not hard to
achieve this much shielding effectiveness, if one does a reasonable job.

Joe Gwinn