[Date Prev][Date Next][Date Index]

RE: step-motor controller/driver connection


Re controller <--> driver cabling

% Gerd Rosenbaum suggests DB-25 instead of DB-15

I don't have a problem with this if Leo Bric can cram eight of these things
onto his Step Pack interface module.  (I think he can.)

% Mark Rivers suggests we avoid those dinky little screws usually found
% on the floor near a nest of DB connectors.  There are two possibilities
% that seem reasonable to me:

1) get backshells with thumbscrews (e.g., Newark stock #93F8861)
2) use spring latches (e.g., Newark stock #93F8863/93F8865)

I'd prefer (2) with DB-9 connectors, since these latches can probably
be attached (retrofitted, even) to anybody's cable and panel-mount DB
connectors.  However, I'm not sure there's enough room to put eight
spring-latched DB-25's on even a double-width VME transition board.

Gerd suggests the following signals:

> 1. amplifier fault
> 2. overtravel
> 3. CW limit N/O (in addition to the CW limit N/C signal)
> 4. CCW limit N/O (in addition to the CCW limit N/C signal)
> 5. a second home signal or second index pulse

How about the following DB-25 pinout:

==========================================================
VME58 <--> (New Step Pack OR isolated motor drivers)
----------------------------------------------------------
Pin	Dir Function		Pin	Dir Function
----------------------------------------------------------
1	Out step +		14	Out step -
2	Out direction +		15	Out direction -
3	In  high limit N.C.	16	In  high limit N.O.
4	In  low limit N.C.	17	In  low limit N.O.
5	In  encoder A +		18	In  encoder A -
6	In  encoder B +		19	In  encoder B -
7	In  index +		20	In  index -
8	In  home 1		21	In  home 2 (coarse index)
9	In  driver fault	22	Out auxiliary
10	In  overtravel		23	    uncommitted
11	    uncommitted		24	    uncommitted
12	    uncommitted		25	    uncommitted
13	    ground/shield

By the way, the 'auxiliary' signal is intended to tell the driver to switch
from drive current to hold current, and/or to drive a clutch.
Also, I don't believe the VME58 will look at any of the newly defined signals.

> Function of signals:
> 
> overtravel:
> 
> shuts off the power to the stepper driver (or the output to a DC-motor) if the 
> motor travels past the limits i.e. controller does not receive or disregards 
> limit signals or other malfunction.

I think this is overkill, but I see no point in excluding it.
It's the driver's responsibility to stop on an open limit switch.
Sending the controller an unsolicited message that the driver screwed
up is not a robust way of getting the motor stopped.  (If you don't
trust the driver hardware to respond to the limit switch, why would you
trust it to tell you it didn't?)

> CW (CCW) limit N/O:
> 
> At X9 NSLS we found that when you drive very slowly into a limit switch it is 
> possible to get a transient limit signal (=switch opened). As a response the 
> controller stopped the motion. But by that time the limit signal was gone (= 
> switch was closed again) and no limit indication light was on. You scratch your 
> head and ask why did the motion stop? 
> 
> To cure this little problem we used a R/S-latch and a DTSP switch to create a 
> stable limit signal plus get some hysteresis. The R/S-latch goes up (= limit) 
> when the N/C contact of the limit switch closes for the first time. It stays 
> this way until one drives back enough so that the N/C contact closes again. 
> Then the latch goes down (= no limit). Spurious openeing of either contact does 
> not change the state of the latch.
> 
> At X9 we had the R/S-latch together with a buffer driver for the encoder 
> signals on a small PC-board on the device. Then we had only 2 limit signals to 
> transmit. But if someone want to do the latching on the distribution panel you 
> may need 4 limit signals.

Limit switches need debouncing one way or another.  Those who prefer to debounce
with one signal and an RC filter should agree on which signal is to be guaranteed
present.  I suggest the N.C. signal.

Also, I'd like to draw a clear distinction between these pairs of
complementary TTL signals, and truly differential signal pairs, such as
'step +'/'step -'.  All signals marked with +/- are differential for
noise immunity, and handled by differential line-driver electronics.
(E.g., the OMS VME44 uses MC3486 line receivers for the encoder A+/A-
pair, although they allow you to artificially bias the negative line
with a DIP jumper, so you can use the positive line as a single-ended
signal.)

> second home signal / index pulse:
> 
> The encoder index comes once per turn. So your home signal has to be precise to 
> about 1/2 turn in order to identify 'home' in combination with the index pulse. 
> In a highly reduced drive or a high resolution drive one encoder turn may only 
> be 20 micrometer drive travel which is too small for a reliable home switch. 
> You may need another signal, once per turn, at a less reduced place of your 
> drive and have to combine 'home', 'index2' and 'index1' (from encoder) for 
> homing.

This signal will probably not be delivered by name from the Step Pack
interface, since it gets only one home signal from the motor
connector.  However, Leo Bric plans to use DB-9 connectors for the
encoder inputs, and since only eight signals are currently specified,

phase A +/-
phase B+/-
index +/-
+V (currently derived from the same +V as appears on the motor connector)
gnd/shield

we could use the ninth conductor as the 'second home' or 'index2'
(coarse index) signal.  Or we could leave the line unspecified and
simply require Leo to bus it.

> Responding to Mark Rivers' concern about the tiny, flaky fasteners for DB-
> connectors:
> 
> You can get spring clip latches for DB-connector shells. Also, make sure you 
> get shells with sturdy metal strain reliefs for cables. Not the shells with the 
> flimsy plastic wedges or the one screw which is driven into the mantel of the 
> cable from the side. Those don't work, they always rip out.

Tim