### Length Variance and Encoder Tracking

When someone has a problem with length variance in roll forming applications, the prognosis is almost always encoder tracking.  If you look at length control from the perspective of the computer that controls the process it’s the brains of the machine.  It’s telling the roll former when to go and stop, and it’s telling the shear and punch when to fire.  The encoder and wheel are the eyes and ears of the brain.  Everything the computer knows – distance, speed, direction – all come from a very small contact point between the material and the encoder wheel.

The encoder assembly at the top of this post is an example that I frequently see when someone asks for help with length variance problems on a machine.  Here’s the same assembly viewed from simply leaning over the setup and taking a picture:

Everything that touches the material – the roll tooling and the side-guides – is parallel to the material.  Everything should be perpendicular, parallel, and/or in-line with the strip in order to make a good part.  Why wouldn’t you need to measure it in-line in order to make a good length?

That’s a trick question.  The encoder should always be parallel to the surface it’s measuring.  You generally want the wheel to be perpendicular to the surface, if that’s possible.  If it’s not possible, then the more unforgiving your tolerances on the finished part, the less forgiving the parallel alignment and rigidity become if you want to achieve accurate and repeatable parts.

The less true the encoder bracket assembly is to the strip, the worse the error and variance will be.  Speed will increase the error and variance.  Typically, the result will be long/short pieces, but there’s no guarantee the system will generate a long part immediately followed by a short part.  The amount of mechanical slop in the encoder assembly (including the mounting bracket) can allow you to ride well enough that the line will produce two or three parts that get longer and suddenly you get a short part.  That’s a situation where the slop in the bracket is taken up as the material begins to move, but eventually the friction between the wheel and the steel is overcome by the pressure created when the slop is taken up in one direction and suddenly the wheel slips allowing the bracket to slop back out again.  This process is repeated over-and-over as the machine runs, producing an odd-looking pattern if a 20 part sample is run.

If you grab the encoder bracket assembly and throw your weight against it in different directions, you should see zero deflection.  Again, the more accurate and repeatable your lengths must be, the less deflection is allowed.  If the part tolerance is ±1/16″ then a bit of flex in the steel probably isn’t an issue.  But if you’re trying to hold ±0.015″ then every bit of flex must be eliminated.  In the previous example pictures, the entire encoder assembly is so badly out of alignment, you don’t need a gauge or caliper to tell there’s a problem.

When running a punched or embossed product, you want to make sure the encoder is riding in an area of the product where the wheel won’t ride over holes or embossed features.

As I adjust my angle of view, you can see just how much the edge of this particular hole deflects as the encoder wheel rides over it.  The spring tension on the assembly allows for significant deflection, which will be interpreted as a speed change and small amounts of extra length tracked for every hole the encoder must ride through before the part is cut.  The longer the part and the more holes, the more the error accumulates.

An encoder can ride in virtually any direction, but the more products a given line must run the greater the potential need to move and re-align at tooling changes.  End of coil scrap is generally increased the further from the shear you must mount the encoder assembly, but riding on the strip prior to any forming or punching operations can be ideal for high-changeover machines.