Press Consistency

Unless you’re using a servo, press consistency is crucial to length accuracy in flying punch and cut applications.  Without servo positioning, timing variances in the press will directly relate to length tolerance error, and the faster you run the line, the worse even small timing variances will affect length.

For instance, if you’re running a roll former at 400 fpm, that’s 80 ips.  At that speed a timing variance as small as ±0.003 s will cause a total length variance of ±0.24″!

Even the best 24 VDC relays have a switching time of ± 0.003 s.  Thus, running a press fire output through an electro-mechanical relay could easily induce a total length variance of almost 0.5″.  That assumes you have no encoder tracking issues.

Assuming a constant line speed, it’s easy to calculate the variance by using the formula for time, speed, and distance.  Let’s assume we’re running a roll former at 100 fpm, which is 20 ips.  If we’re measuring a variance of ± 0.25″, then we can divide that distance by the line speed to find the timing variance:

d / s = t

0.5″ / 20 ips = 0.025 s

Now, find the process on the machine that takes 0.025 s to do its job.  You’ve probably found the source of your variance!

Press timing variances are very difficult to track down if you don’t have the right tools.  If you have those tools, then there is a procedure you can follow to isolate and troubleshoot the issue to prove whether or not the press is the problem.  I’ve used it many times.

I was helping install a line at a manufacturer of semi-truck trailers when the Maintenance Manager asked if I could break away for a while and help troubleshoot a length issue on another production line.  The problem line had an open loop flying punch and open loop flying cutoff.  The overall length of the parts was holding to within ±1/32″, but the holes were ±1/2″.

As we were looking at the issue, the Plant Manager walked up and wanted to focus on the machine controller that ran the line.  I explained that the computer wasn’t the problem, because it’s simply executing its code – doing the same thing, over-and-over, really fast.  That means consistent errors could come from a setting or a problem with the controller, but variance is almost always a problem caused in the real world.

I knew the problem wasn’t encoder-related, otherwise both the overall length and the punch holes would have the same error.  Also, the dies were not boosted.  The tooling contacting the material would drag the die forward, and springs pulled them back to their home positions.  The only options were – 1. something in the press fire circuit, 2. a problem with the solenoid, or 3. a problem with the press, itself.

The press fire circuit was simple.  It was a hard wire from the solid-state output of the controller, through a solenoid driver module.  These modules take the 24 VDC output from the control system and boost it to something closer to 60 VDC to force the solenoid to react quicker and more repeatably than it normally would.  These are also solid-state devices, so it was very unlikely to be the issue.  Normally, if a solid-state device fails, it simply stops working.  Tracing the rest of the  wires, I could see there were no electro-mechanical devices that could cause an obvious variance.  That left the press and/or the solenoid.

At that point, I pulled out an oscilloscope, a high-speed prox. sensor, and a magnetic base.  With this setup, I could directly measure the time between when the controller turned on its output to the point in time when the tooling reached bottom-of-stroke.  The O’scope I had was capable of measuring time down to the microsecond range, and the prox. sensor had a switching capability of tens of microseconds, which was plenty of resolution to accurately measure the issue to the millisecond range.

I walked the PM and MM through the math – the machine was running at 100 fpm, which is 20 ips.  Their variance was roughly ±0.5″.  Distance divided by speed equals time, therefore 0.5/20 = 0.025, or ±25 milliseconds of time.

The great thing about this setup is that you can manually fire the press from the controller without needing to run the line, so you don’t have to make scrap or try to perform these measurements on-the-fly.  We connected the setup, and then I manually fired the press over-and-over.  I setup the O’scope to give us a persistent trace on the screen, and I would move the sliders (vertical lines to help with measuring time differences) to the left or right of the signal coming from the prox. sensor each time it fell outside of the original signal location on the screen.

After roughly 50 firings, we could see that the signal no longer drifted any further to the left or right of where it showed up the first time, and at that point we could all see the time displayed on the O’scope (the scope shows the time between the sliders) – it was 0.025 s!

The Plant Manager was amazed, “That’s the exact time you just calculated!”

“Yes,” I said.

“So…what’s that mean?” he asked.

“It means you have a problem with the press.  It could be a sticky solenoid, or a problem with the valve itself, or maybe even a blown seal in the press.  You need to bring in a press expert to go through it and maybe even do a rebuild.”

Two days later, a local hydraulics expert had replaced a sticky solenoid and replaced a blown seal on the cylinder.  The punch was holding ±1/32″ just like the cutoff.

At the time I was working for AMS Controls, and I had asked our on-staff machinist to mill out a small aluminum bracket that I could fit onto my magnetic base using the same dial indicator mount that came with the base.

I purchased a high-speed prox. sensor from SICK.  I don’t recall the sensing range, but the crucial thing is the switching speed and consistency.  The sensor had a switching speed of 0.000005 s with a guaranteed consistency of ±0.00001 s.  This is more than sufficient to measure time down to 0.001 s, which is usually good enough for most roll forming applications.

All those years ago, I carried a 100M Tekscope, which was overkill for most applications in our industry.  For almost the last decade I have carried a 10M Picoscope.  The specific model I carry is their 2204A.  It has 2 channel capability with an external trigger, and supports invert-and-add functions, which makes it a fantastic scope for everything from measuring timing issues to checking for electrical interference.  I am unaffiliated with Picotech, but I have used this little scope dozens of times to troubleshoot ink jet printer issues, noise issues on Ethernet lines, to measure timing problems on presses and to troubleshoot software bugs in control systems.  The best thing about these new scopes is that your laptop acts as the front-end for the scope, which means there’s no delicate display to worry about and they can data-log directly to the hard drive of your laptop.  Have a weird, intermittent problem?  Put the scope on it and data log over the next several hours.  You’ll catch it happening, and then you can peruse the information at your leisure.

Setting up to test for open loop press consistency is simple.  Wire your prox. sensor into the system.  These sensors come in 24 VDC models, and this is a voltage typically used to fire the press.  You might need to wire-nut longer wires to the sensor cable, but it’s a temporary test setup, anyway.

Mount your prox. sensor to your magnetic base, and lock the base to the bottom die platen or base.  Adjust the articulated arm of the magnetic base so that the prox. turns on only when the tooling reaches bottom of stroke.

Connect both probes to your O’scope, and clip  one probe to the output of the control system and the other to the output of the prox. sensor.  You can now see both signals on the scope.  You can directly measure the reaction time of the press between when the control system tells it to fire to when the tooling physically reaches the material.  This, by the way, is a great way to nail down Press Reaction Time, if you’re using a control system that accounts for it.

Perform a standing press operation (Manual Punch or Manual Shear).  Drag the scope slider bars to either side of the prox. output signal on the screen.  Fire the output manually over-and-over, and each time the prox. signal falls outside of the slider bars, drag the appropriate slider bar to the outside of the signal line.  Repeat the process until the output signal no longer falls outside the sliders.

If your signal goes through a driver board or a PLC, you can even move the probe from the controller output to different points down the circuit to verify there’s no timing variance coming from, for instance, the PLC scan rate.


Leave a Reply

Your email address will not be published. Required fields are marked *