Servo Punch Press

Today was build a crossmember day :) (see post above) This is a piece of A36 steel, 6x12x3/4. It needs to get trimmed down to 5.4625 x 11.5 x whatever makes it flat after facing.

So the first operation is to put it in the vice and face both sides to remove the mill scale and flatten it out, most flat bar has some bow in it. It finished at 0.714. I didn't get any pictures of that operation, but you know what a carbide insert face mill looks like :)

Then anchor to the fixture plate and rough cut the OD. Had this been wider than the vice I would have bolted it to the table with the aluminum under. Then the finishing pass on the OD, this pass is taking of the last 0.0005 per side to finish. This is called doing surface grinder work on a mill :grin: ( I have two surface grinders, but neither of them are set up). This again is one of those dimensions that you need to sneak up on, it must be on size. Nice sharp 3/8 solid carbide end mill, ~140 SFM (1420 RPM), 6 IPM feed. It came out on size :).

1538107916129.png

This was bolted down to a 1/2 scrap aluminum piece that was used for a fixture for another project, this allowed working around the entire perimeter.
1538108070646.png

The next step is to machine clearance for the ram since this will cover the entire front of the press. 0.150 deep x 6 inches. The bottom 1/4 inch of this poor old roughing end mill has just about had it. Wasn't sure it was going to finish this cut.
1538108841417.png

But the rest of the end mill is nice and sharp so drop it way down through the hole to cut the bolt head clearance.
1538109044685.png

And the top view
1538109115680.png

And last, I need to cover the bolt head clearance hole because this just created a pinch point, so screw a scrap piece of 3/8 lexan down to the pre-tapped holes and profile the OD. Access to the bolt is required for machine maintenance. This cover actually installs on the other side.

1538109743783.png

Tomorrow the crossmember will be installed on the press, 12 hours is enough for today :)
 
Last edited:
Today I installed the crossmember, it worked as planned. I had to do a bit of prep work on the frame sides to install. so pull the sides one at a time and over to the mill to drill & counterbore the mount holes.

So set on the mill table spaced with some 123 blocks. The dial it in and locate the lower left corner for 0,0. Then do a bit of measuring to confirm that the side plate matches the drawing. Well it didn't :mad: The entire area bounded by the calipers is screwed up. Which explains one of the problems I was having aligning the die. This goes back to what I thought was a backlash problem with the mill. https://www.hobby-machinist.com/threads/another-backlash-problem.72171/#post-606271. The holes and more importantly the frame edge to the opening edge was not located correctly. The good news is the I could fix it by removing metal, needed to take about 0.010 off. The opening is just a bit wider now, but the dimension from the corner to the opening is correct, and now matches the other side.

BTW, those are Harbor Freight 12''calipers, about $20. A bit stiff, but seem to be pretty accurate, had to adjust them to align the jaws, but that only took a minute.
1538283782138.png

I had to move the bolster mount holes a bit also. I just pocketed these since I was already set up. 1/4 inch solid carbide router bit from Home Depot.
1538284175641.png

Here is what the press looks like on the inside with the left side removed. It don't look like much, but a lot of work in there.
1538284545612.png

And the stiffener installed. Works great, holds the press frame square.

1538284608517.png

And the guard over the pinch point................BUT... There is a couple of problems with that guard. https://www.hobby-machinist.com/threads/machine-guarding.73128/
1538284794599.png

More later......
 
A bit more progress, after doing the rework on the side plates I was able to finally get the die aligned with the punches and get it to do so repeatably. Everything is now dowel pinned in place.
1538455854218.png

So now I need to put a 1.375 radius on each end of the die holder to clear the feed rolls. So run a profile routine to do that. Material 4150, 3/4 x 4'' HSS endmill, 340 RPM, 2.5 IPM, 0.060 DOC, 0.050 step over. Total depth about 3.625. Took a while per pass about 82 minutes per, 3 passes. I roughed it out on the band saw first.
1538456657756.png

And done
1538456698244.png
 
I haven't updated this for awhile, been busy trying to get this thing out the door. My customer is getting a bit antsy, I'm way overdue on shipping.

Where I left off above is prepping the the holder to accept the feed rollers. The ends were concave radiused for clearance. The rollers need to stick up above the top of the die holder about 0.030 or so.

Here is a picture with the feed rollers installed and the retractable top pressure rolls just sitting in position.
1539577397856.png

The next step is to build the mounting system for the pressure rolls. The toggle clamp will be used to retract the pressure roll for loading the material into the machine.
1539577942486.png

So first grab an old rusty piece of 3/4'' A36 and clean it up to look a bit presentable and prep it for the next operations...... Building the mounting brackets. So now I have 4 blanks for the mounting brackets and an aluminum fixture in the vice. Since I'm doing 4 identical parts, I'm using dowel pins to locate the part on the fixture while machining, it's held in place with two 1/4'' cap screws, the dowel pins prevent any lateral movement, the cap screws only have to prevent it from lifting. Notice the two screwdriver slots in either end of the fixture, this allows easy removal of the part from the dowel pins.

So first make a light cut of the profile, then over to the band saw to rough them out, actually saves time to do it that way and I don't eat as many endmills.
1539578610349.png

After bandsawing
1539578664502.png

Then rough and finish profile.
1539578747825.png

I didn't get a picture of the slotting operation or the various drilling operations. There is a 1'' wide slot in the bottom of two of the pieces, and a 3/4 wide slot in the in the other two and a 3/8'' hole drilled from the top that intersects with the center of the slot in all of the pieces. After inserting the die springs, I also installed threaded 1/2-13 x 3/8-16 in each piece for the spring adjuster screw. This holds the tension on the spring to press the roller shaft down.

The next operation is the roller shafts. There are turned down to 0.750 on one end to accept the roller bearings that press into the roller bore. Then mill the spring landings and pull attachment point. Also mill the anti-rotation flats on the off roller end, thus the need for the spindex.
1539579496358.png

And installed on the press. I cut the heads off of four 3/8 cap screws and cross drilled them to accept a #40 chain master link. I needed a flexible link and that seemed like the easiest way to do it. The toggle clamp handles are in the normal operating position, push them up against the press frame to raise the rollers. In the bottom of the back mounts is a ball end set screw that sits in a divot in the shaft. This allows leveling adjustment of the roller, without it the pressure rollers would not sit parallel to the feed rollers. They work exactly as planned.:encourage:

1539579860964.png

Another view
1539582714650.png

I have it running now and making product. Have had the normal software glitches and a rather odd control problem. It turns out that the servo drives are so electrically noisy that when turned on they kill my ethernet connection to the motion controller, I have everything grounded properly so this one is a bit puzzling. The work around for that seems to be to use the serial connection on the controller, but that causes some other minor issues in communication speed. I need to find a better workaround. But overall it's a success. I ran out about 1100 feet of product yesterday in just over an hour running at about 1/3 speed, so I can say it works. I should be getting about 12,000 feet of material tomorrow so as soon as I figure out the material handling I'll run that and then ship the machine, and the finished product to the customer.
 
...I have it running now and making product. Have had the normal software glitches and a rather odd control problem. It turns out that the servo drives are so electrically noisy that when turned on they kill my ethernet connection to the motion controller, I have everything grounded properly so this one is a bit puzzling. The work around for that seems to be to use the serial connection on the controller, but that causes some other minor issues in communication speed. I need to find a better workaround.

Excellent! As for Ethernet, it should tolerate hundreds of volts without problems (grounding shouldn't matter), whereas
a serial link only tolerates a few volts. I'd suspect a bad cable, first.
 
I'll just toss out some ideas. Sometimes an inexpensive fix for this sort of noise problem is to fit toroid cores (ferrites) on either the noise source (motor leads) or the susceptible cable (ethernet) or both. Signal and power wiring separation can help too. It is also possible the ethernet cable isn't the way the noise is being picked up. The noise could be conducted into the power supply or control IO of the controllers or computer. A little trial and error with ferrite cores and being sure that noise source cables are separated by some distance from signal cables might help. Also, if some control wiring is not shielded, consider shielded wiring. Be sure shield lead drain lead connections are very short and that the un-shielded sections of the cable near the terminations are as short as possible. From a grounding stand-point, a star type ground is generally better than a daisy-chain ground. One last thing. It is possible to lock the ethernet nic to a slower speed such at 10 Mbps instead of the automatic 10/100 or 10/100/1000 setting, if you don't need that speed. I have used that technique before. It worked once for one problem system I had but did not work on another.
 
This might be a bit of hair splitting but I strongly doubt that the noise is being picked up by the ethernet hardware layer. This is because Ethernet, like RS-485, USB and some other protocols uses a Differential Signal technique . This means that two wires are used for the signal and they are shifted in opposite directions (very simplistic description but you can get copious details online). The bottom line is that this technique is VERY noise resistant. I say that this is a bit of hair splitting because the problem could still be in the Ethernet subsystem like the interface cards or circuits. Still I think that that while adding a toroid to some things will help, I doubt it will have an effect on the Ethernet cable (but I don't see how it can hurt). It could still be helpful on the motor control lines and physical shielding and good grounding of any and all things that have PCBs is a good idea. Don't forget to check the power supply levels to ensure that you are not getting an unexpected supply voltage drop when you start to push the servos (i.e. draw more current).
 
Instead of saying "It is possible the ethernet cable isn't the way.." I probably should have said "It is more likely the ethernet cable isn't the way..". It can still happen though so I would not rule it out completely. Even though the signals are differential, common mode signals can be picked up by the pairs and are presented to the ethernet transformers. These tiny transformers are not perfect and can convert some of the common mode into differential mode on the circuit side. It is easy to get into the minutia on these things but from a practical standpoint, some of the techniques I mentioned and Boswell mentioned above can work to solve the issues. Good point Boswell about diff. signals and checking power supply levels. It made me think about a Phoenix Power Inc. DIN rail power supply. It was extremely noisy and caused lots of problems in a system I helped troubleshoot. Replacing the power supply with one with less ripple and noise solved that case. (P.S. Not all Phoenix power supplies are noisy, just that model I came across was noisy.)
 
Back
Top