CAD CAM EDM DRO - Yahoo Group Archive

Re:My CNC Retrofit Project/Don't cut wires!

Posted by Andrew Werby
on 1999-10-14 04:12:34 UTC
Message: 9
Date: Wed, 13 Oct 1999 16:13:13 +0200
From: "Arne Chr. Jorgensen" <instel@...>
Subject: RE: DON'T CUT WIRES

As you already may have figured out, what I meant was that any
computer will act like a punch tape unit, - just better. In fact it
is what EMC and every others do. So if it has a teletype port, -
many used current loops. ( That is the very best way - noise immune,
easy to work with, will carry signals over large lengths of cable,
etc. )

[I'm still not clear what's the difference between current loop and
RS232-style signaling- or are they different forms of the same thing?]
What I meant about the "old" ISA serial port adapters is:

On the adapter of an original IBM, ( and a bunch of clones ) there
is a socket on the board, with wire wraps. Or an aluminum type of
dual in line pin - device. This one will act as a wire jumper for
the different modes.

( Assuming orig. IBM) - This look a like 16 pin IC has a mark on
the top ( pin 1 ) - If this mark is up, you have a normal RS-232
adapter.

Pull it out, turn it around ( mark facing down ) - you have a
current loop adapter. 0-20mA .

[Poring over my manuals, I did notice something about reversing one of the
RS232 plugs 180 degrees to get that 20 mA current. But what do I have to
change on the other end?]

You have +transmit current - on your DB25 pin 9
-transmit current - ....................... pin
11 , this is the current return path.

You have +receive current - on pin 18
and -receive current - on pin 25

-------------------
here is a little trick:

place a 1N4148 or any diode in series with the wire.

wire ------------>|-------------

Now the current will not be affected by this, so you can measure
across the diode with a milliAmp meter. The internal resistance in
the meter will short circuit the diode, and you will read the
correct current anyway. You don't have to clip any wires.

[How do I put in this diode without clipping the wire?]

Even
during transmission, you will not interrupt the data or disturb
anything by measuring across this control point.

Often you will want to have opto isolation, and this is already
installed on the board. the DB25 pin 18 and 25 is in fact the light
emitting diode in this opto isolator. Note there is no current
limit, that is done on the xmit side. If you would like to wire this
to anything else, like a 5 volt supply, add a resistor of at least
150 ohm to it. ( 220 being very common for 5 volt )

[Why would I want to connect this to a 5v supply?]

The rest of the machine, can be upgraded in small steps, ( I have
not looked at the photos, just read what others have told here ) -
but you could win a lot, by taking it step by step. I am not so
familiar with this kind of machines, but I have seen so many stupid
retrofits. In almost each case, the original system was better.
The reason is that they have spent years to match the various
parameters in a machine, and without all that knowledge - you start
on scratch. Many stupid looking, old and nasty things, showed hidden
attributes of wisdom.

[I hear you, but unfortunately these wise old things tend to die off, like
wise old people...]

By all means, you want something nice and up to date, but I have
rebuilt systems back to it's original status, ( - and some stupid
guy have maybe dumped parts of it ) , - just to find out what was
wrong with the expert system that was put in by very famous makers.
( Siemens, Baily, ABB, GE, - yes, those kind of experts ! )
I don't want to discourage you, but there is so many things you may
learn by just trying out some of it. Even just to get some intuitive
ideas. This can be very helpful later on.

[Since it's going to take me a while to get Linux/EMC going, not to mention
patched into this machine, I'd like to try some incremental approaches
first- like you say, it'll be a learning experience, if nothing else.]

I hope this could be helpful to someone, and I do it just because I
hate to see people disregard old stuff too easy, and making a hell
for them selves.

Best wishes

//ARNE

[Thanks //Arne. I appreciate your input, don't get discouraged and go away.
Heck, I don't even mind when you get drunk and slobber all over your
keyboard...]


Message: 3
Date: Wed, 13 Oct 1999 13:02:13 -0700
From: "Darrell Gehlsen" <darrell@...>
Subject: Re: Re: DON'T CUT THOSE WIRES !! and My CNC Mill retrofit project

<snip>
>
> I looked into one of the motors, and was surprised to see only 4 contacts,
> 2 of which seemed to be the DC current + and - and 2 which came from a
> gizmo on the back which I assume was the encoder. If the things have
> tachometers they must be someplace else- or can speed be calculated
> from the encoder pulses?]
>
<snip/>
You are right about two of the wires being current +/- but I think the other
two are tachometer wires.
Darrell

[It did look like that could be a little generator stuck on the end of the
motor shaft, now that you mention it. So that means I should look elsewhere
for my encoder signals, I guess.]


Message: 21
Date: Wed, 13 Oct 1999 20:18:12 -0500
From: Ray Henry <rehenry@...>
Subject: Re: DON'T CUT THOSE WIRES !!
Andrew wrote

-----om-----
I looked into one of the motors, and was surprised to see only 4 contacts,
2 of which seemed to be the DC current + and - and 2 which came from a
gizmo on the back which I assume was the encoder. If the things have
tachometers they are must be someplace else- or can speed be calculated
from the encoder pulses?]
-----

Yes speed can be calculated from encoder pulses but my guess is that they
were not in 1983 when your machine was designed. My guess is that the wires
that you are seeing coming from a gizmo on the back of the motor are from a
tach. Most encoders have at least 6 to 8 wires to them. Resolvers have
six. You could disconnect a pair of these gizmo wires and put a dc volt
meter across them. Turn the motor a little and you should see a dc
voltage. Reverse the motor direction and the dc should change polarity.

Most likely the encoder or resolver is directly connected to the ball
screw. You may have to crawl under the thing or take off the covers on the
end of one axis.

[I haven't noticed anything like that, but I'll take another look. If there
isn't anything attached to the ball-screws, then should I assume that it's
using a linear encoder? If so, where would I look for the inputs and
outputs? And what's the difference between encoders and resolvers?]

-----om's-----
A while back I prototyped a cable and software
to run g code through the parallel port of a pc to a tape in on an old AB
7300 cnc. The cnc would accept all of the code that it needed and then
pause the parallel port.

[That sounds like a promising approach. Do you have any of these
cable/software kits you'd be willing to sell? I'd like to try this, if I
can coax my machine into booting up all the way. Currently it won't accept
any commands, although I can hear relays working- maybe replacing some (or
all) of them is called for.]
-----

If your machine comes out of the coma, I'll try to rustle up the stuff up
for you.

[That would be great- actually I got the thing running yesterday (thanks
for all the pep-talk, guys) The problem turned out to be that the thing had
run past its limit switches, and wouldn't budge. Repeated rebooting and
gridding finally woke it up. I was so encouraged I started wondering if I
could somehow feed it a short program on mag tape, but it uses a kind of
cassette I've never seen before, so I don't know where I'd find something
that would write to it. (I can give it mdi commands from the keyboard, but
that's awfully tedious.) Really, I like the drip-feed from a parallel port
idea much better, although my programs tend to be so big that I'm afraid
the machine would have a hard time executing them even if they weren't
chopped up into bite-sized pieces and spoon-fed.]

One thing I do for controls that act like your's is to reseat all
of the computer boards. That is why I asked what computer it had in it.
With all power off, you should be able to pull each board out an inch or so
and then push then back firmly into the rack. This might cause problems
depending on how memory is set up and how you reboot the machine exec.

[This computer doesn't seem to be set up that way. Instead of being slid
into slots like a modern computer, its 4 panels are screwed to a back
plate, which mounts on the door of the control cabinet. The only tag on the
computer said "Industrial Information Controls Inc PC-LSP-08." I think they
made the rest of the control stuff too. The only way the machine reboots is
by turning on and off the main switch, which looks like a car door handle
mounted on the back of the control cabinet.]

I saw on your post to Fred that you want to run an indexer rotation around
X. If you do not have this on the current machine that would be an
excellent reason to retrofit. An indexer that will step 1 degree will cost
much more than the hardware costs of a retrofit with emc.

Ray

[Really? I was thinking of a simple rotary table with a chuck, mounted
parallel to the x-axis, with a tailstock like a lathe. I've got a spare
motor and amp that came with the machine, so even if everything had to
match that wouldn't be a big problem. How much do rotary tables cost? I
know Max NC sells a little one for $535 with a stepper and control box- I
figure I should be able to get a larger manual one for not much more and
adapt it to my motor. So what I was mostly worried about was figuring out
how to hook up the extra axis. I was also thinking about running it with a
stepper, since the position and velocity information isn't so crucial, and
the servo motor seems heavy for the job. Would it be possible to have the
program signal the stepper to move a (certain settable) distance at the end
of each linear pass, and accomplish the indexing function that way?]

To get adequate information about the motors and drive amplifiers that you
have on your mill, we will need the numbers from the tag that you found
westamp's name on. There should be model and serial number. Westamp was
recently purchased by Ormec but plan are to keep the companies separate.

If you want you can give me the numbers and I'd be happy to contact them
and see what info we can get.

Contact Information for Westamp, Inc.

For sales information, contact Westamp at 9006 Fullbright Ave., Chatsworth,
CA 91311.

Phone: 818/709-5000;
FAX: 818/709-8395;
E-mail: info@....

Visit Westamp on the Web at http://www.westamp.com

[Okay, I'll get back to you on that- thanks.]

Message: 1
Date: Thu, 14 Oct 1999 00:15:36 -0500
From: Jon Elson <jmelson@...>
Subject: Re: Re: DON'T CUT THOSE WIRES !! and My CNC Mill retrofit project

Andrew Werby wrote:

> You can feed G-codes from any program and computer. These often
> used paper hole tapes, and don't always have a high baud rate, but
> - it is so much faster than the movement of the axis anyway.
>
> [I thought of trying this, but the machine's controller seems to be in such
> a fragile state of health that I figured it would be best put out of its
> misery. Doing this, I can easily see getting into a situation where I
> wouldn't know what was screwing things up- the patched-in drip-feed, the
> old computer, or the associated maze of circuitry.]

I did this with an Allen-Bradley 7320, a 1978 vintage control. I built
up a circuit to simulate the function of the paper tape with a program
that ran under DOS. I could download the executive software that
made the machine into a CNC, and then 'drip-feed' the part program
as the CNC needed it. It would even pretend to back up the tape if
you wanted to re-execute a block or two, and allow you to search
the 'tape' for specific block numbers, etc. It was also at least 4 times
faster than the physical paper tape reader. I had to limit it to about
1000 chars/sec or it could overload the CNC cpu.

[I figured there would be some issues like that, which was one reason I
didn't feel like it was feasible for me to attempt. But I wasn't thinking
about all the geniuses on this list pitching in to help, either...]

Well, it worked, and it got my CNC system making parts, but it had
a lot of problems, mostly loose connections, but also a blown high-voltage
diode in the CRT sweep circuits, and a blown static RAM in the
video generator. Bad connections in the battery back-up power supply
caused it to lose its program memory every once in a while. It became
obvious that it was not going to get any more reliable, at least without
a lot of work. Every time a problem developed, and you pulled a
card out, connectors all over the thing went flakey, and front panel
buttons or lights failed to work, etc.

[Yes, this is the sort of thing I was anticipating happening with my
machine as well. But on its good days, like yesterday, it seems like the
drip-feed idea might work, at least until I get up to speed with EMC and
figure out how to attach it to the parts of this machine I want to keep.]

Fortunately, EMC was getting usable at about that time, and when I was
able to make the switch, I never went back!

Jon

[Thanks Jon, Matt, Darrell, Ray, and Arne- I really appreciate the help
I've been getting here. ]




Andrew Werby - United Artworks
Sculpture, Jewelry, and Other Art Stuff
http://unitedartworks.com

Discussion Thread

Andrew Werby 1999-10-14 04:12:34 UTC Re:My CNC Retrofit Project/Don't cut wires! Ray Henry 1999-10-15 07:58:28 UTC Re: My CNC Retrofit Project/Don't cut wires! Andrew Werby 1999-10-16 04:43:42 UTC Re:My CNC Retrofit Project/Don't cut wires! PTENGIN@x... 1999-10-16 13:54:42 UTC Re: Re:My CNC Retrofit Project/Don't cut wires! William Scalione 1999-10-27 23:44:57 UTC Re: Re:My CNC Retrofit Project/Don't cut wires!