2-axis Sharp knee mill revival - Now with 50% more axes!!

[martyscncgarage]

I don't know the history of the machine, except that the control was not working. You're right: for all I know, they put mains powers into the encoders and blew them both out, and the subsequent runaway is why they decided to get rid of the machine.

Thanks for the encoder supplier suggestions.

One bad move with power and you can take that stuff out. Might be why they downed the machine. Never know.

Marty

[JoeG]

Just for grins put a couple 2.2k pull down resistors on the outputs from the encoder......

JoeG

[FlySox]

Good suggestion. I was hoping this was going to do it, but all it seemed to do was pull down the voltage ( I used a single 5.1KOhm resistor; didn't have any 2.2Ks handy). Still no signal change when turning the encoder :

[FlySox]

Where we last left this story, I had decided the encoders were the cause of issues and had begun the search for replacements. Marty had offered some suggestions, but I had already been looking at US Digital. One of their mounting options of their inexpensive E5 “kit” line jumped out at me, the addition of 1.812 spaced “ears”.

That’s an odd measurement, where have I seen that before?

Looks like it’ll fit. I ordered two 2000cpr kits to fit my 6mm shafts with the differential option and the ear mounts for $85 each. The connector wires added a rather steep $13 per unit, so let’s call it $100 per encoder. Not too bad. Interestingly, the 5000cpr option (their max in this model line) is the same price. I opted for the 2K since I had read my drives can get finicky when paired with high resolution encoders (although that may have been due the the lag introduced by the capacitive encoders that were used by the person reporting this issue, but I digress).

While there is no online shop for US Digital, you use their configurator and send the model number to sales@ and they send you back a link which enables you to pay online via payfabric.com. They were quick to respond and since I ordered before 11:00am Pacific, they shipped that same day!

A few days later, I got my kits, the contents of which are shown here:

Assembly is trivial. Here’s a quick walkthrough:

Mount the base:

Install the spacer. This ensures the right height for the optical disk.

…which you tighten down via set screw.

The differential line driver is a module you press onto the standard 5 pin connector of the optical module:

Install the optical module:

Install the cover, and you’re done. Whole process takes five minutes.

Now, some of you have already guessed at the issue I faced. You’ll notice that the differential module adds almost a quarter of an inch to the length of the encoder, a fact that I did *not* pick up on in ordering (although I now see it clearly in the data sheet; ugh). This means that when I plug in the connector, it’s actually too long to fit within the rim that defines the encoder protector. More than that, because the connector sits on that rim while the encoder is lower, the connector “finger” doesn’t engage its slot, so it doesn’t lock into place.

Not sure how I’m going to solve this one. I’ll probably crimp female pins onto wire and solder each of them to the differential pins. A bit hacky, but it should be secure and allow me to use the existing encoder protectors without modification. So much for the expensive connectors!

[FlySox]

There is some good news, though. It does indeed seem that the encoders were the issue. I wired up the motor with the new encoder to my drives and was able to read them and begin the PID tuning process. I’m not going to put a lot of effort into that until I remount the motors on the machine and have them running off the 99V power supply (vs the 24V bench supply I’ve been using for testing), though. Looks slightly underdamped for now (?)

[martyscncgarage]

I bet US Digital had an encoder where it exited at a right angle to the encoder....

Let's see how you get along with the tuning process.

[Muzzer]

Looks like a nice project. I've got the DG4S-16035 drives on my big machine, along with SEM brushed DC motors. Works very nicely indeed. The encoders in my motors are British clones of a Heidenhain model that looks rather like the original ones on yours.

IIRC, Marty reviewed a Chinese 1000PPR encoder for use as an Acorn spindle encoder (Marty - didn't you do a YT video on this?). I bought and used one of those myself and they are functionally similar to the motor encoders, so you could perhaps use those if you want a better fit for the space available. They look as if they may even be a drop-in part.

The tuning software is pretty good (better then the DMM equivalent for instance, with transient response "scope" etc). There's no real alternative to iterative adjustment of the P, I and D terms until you are convinced no further improvement is possible. Calculated starting values only get you to a very rough starting point.

I notice you have large hand wheels on your X and Y axes. May be worth comparing the response with those removed. If you sit down and calculate the moments of inertia seen by the motor as I have, you will see that they are significantly affected by the handles, certainly if they are metal like mine. I've now removed the handles on my smaller (Bridgeport) machine and the transient response is much better.

[martyscncgarage]

Yes, I did a video on the Chinese Omron knock off as did Franco CNC.
Some have used it with good success. I've not heard of anyone saying they have died, other than because of misuse/mounting so far.

[FlySox]

Funny you should mention Chinese encoders. On Friday, my backup plan arrived: Three encoders ordered in late January from AliExpress. They go under the brand name Calt. You can see their offerings here: https://caltsensor.aliexpress.com/store ... 01496.html. I paid $53 each, plus a total of $10.50 shipping. The downside, of course, is you don’t get them for 3-4 weeks (that and the unknown quality).

They appear to be well made, and are more of the industrial form-factor than the US Digital encoders, plus they’ll fit! OK, let’s give them a whirl…

I want to make them as close to a drop-in replacement for the original encoders as possible, which means putting the same connector on them as the original (pictured here). I ordered the same Molex connector on DigiKey.

Cut the new encoder cord to size:

Debraiding the shield. I found that using the point of a compass made this really easy:

Twist up the braid and prep the wires for stripping

I’m not a fan of crimping Molex connectors. My cheap crimper is probably a big part of that. It doesn’t crimp them so much as make a suggestion for the subsequent manual plier squeeze. If I did more than a few of these every 5 years, I’d upgrade.

All pins connected, now slip them into the housing

After heat shrink, it looks pretty good:

I haven’t had a chance to test them yet, and probably won’t until next weekend, but I’ll post an update when I do.

[FlySox]

Not a lot of progress to report, but I was able to test my newly wired encoders with my oscilloscope. The A/B signals are clear and appropriately offset. I’m slightly worried that these encoders have a B signal that leads the A signal, but I think in the worst case, I might have to use the Acorn option that swaps motor direction.

On another note, I’ve been thinking about how to fit my components into the rather small control panel I’ve got. One of the ways I decided to deal with this is by making a riser for the Acorn and to mount the relay board on top of it to save some footprint.

Spent some time in Fusion 360 and modeled a board with the footprint and mounting holes of the Acorn, along with the mounting holes of the relay board:

A lot of the time was spent figuring out how to use the F360 post-processor for my Glowforge laser. Got it working and am sure this new-found knowledge will come in handy for this and other projects.

Here’s the model brought into the Glowforge interface:

…and the 1/4” piece of acrylic right out of the machine (with protective sheets still on):

With the relay board mounted:

And the final product: