What is “Minimum Edge Separation” and Why Should I Care?

When it comes to magnetic incremental linear encoders, sometimes the configurable performance parameters can be a tad obscure.  One of the most puzzling is the feature called “Minimum Edge Separation”.

Just for review, a magnetic incremental linear encoder system consists of a precision-encoded magnetic tape and a precision magnetic reader head.  It is a non-contact device and is an incremental measurement system.  Incremental means the position is given as a series of pulses which must be counted by the controller.  If the count is lost or corrupted for any reason (such as a power-down), the system must be re-zero’d to a known home position before controlled positioning/measuring can resume.

In an incremental encoder, the position output signal is encoded as a quadrature pulse train.  An increment of position resolution is defined as the distance between two adjacent edges of the quadrature pulse train.  For example, the leading edge of Signal A followed by the trailing edge of Signal B equals one increment of resolution.  A unit of resolution is a unit of length, for example 5 micrometers.  Every appearance of a new edge equals another increment, which increases the position count by one.   For example:  5 counts * 5 μm/count = 25 μm of movement.

But, there’s a fly in the ointment.  In addition to a physical displacement or unit of length, the “distance” between edges can also be measured as a unit of time.   The time between edges is variable, depending primarily on the speed of relative motion between the encoder head and the encoded magnetic tape.  In general, the speed of the counter must be fast enough to catch every incoming edge of the pulse train at the highest velocity expected in the application.  If the counter is not faster than the application velocity (taking resolution into account), some pulses could be missed and position errors will occur.

But wait, there’s more!  Even when the encoder is motionless, it’s still possible that the counter could be over-run by edges coming in too fast.  How’s that you say?

At its core, a magnetic incremental encoder consists of two precision analog magnetic field sensors that provide the inputs to an internal ASIC that generates the quadrature digital output.  As in all analog circuits, there is always a minute amount of residual electrical noise present in the system.  Even if it is only thermally generated electrical noise…it is still there.  This noise is seen at the analog inputs of the ASIC and is indistinguishable from changes in the analog signal level due to actual physical movement.  So, the ASIC responds to the level changes by outputting a few pulses.  When properly decoded by the counter, these pulses always sum to zero.  For example +1, +1, -1, -1, +1, -1 = 0.

Trouble is, the noise is typically very fast and the resultant output pulses can have very short pulse widths.  The time value from pulse edge to pulse edge can be extremely short.  So short, in fact, that the counter may not catch all of them.  This will show up in the application as an apparent position drift even when the machine is standing perfectly still.

How can we combat this situation, other than installing the world’s fastest and most expensive counter?  That’s where “Minimum Edge Separation” comes into play.  The encoder manufacturer can provide an internal low-pass filter to remove the high frequency noise before it even gets to the ASIC.  This filter needs to be configured so that it passes the frequency of the actual position changes at the maximum application velocity, but blocks any false signals coming from internal noise.   The faster the application, the higher the cut-off frequency of the filter needs to be.

Rather than speak in terms of internal cut-off frequencies, though, it makes more sense to talk about the impact on the actual speed of the quadrature output signal going into the controller.  And that’s how we finally come down to the real meaning of “Minimum Edge Separation.”

Minimum Edge Separation means the smallest possible time between two adjacent edges of the quadrature signal.  It is given as a unit of time.  Typical selections are 0.12 μs through 24 μs.  When specifying a magnetic incremental linear encoder, you want to the select a Minimum Edge Separation value that is slower than the speed of your counter but faster than the highest velocity of your application, when taking resolution into account.

2 Replies to “What is “Minimum Edge Separation” and Why Should I Care?”

  1. Good article. And I thought I knew it all before reading this. 🙂

Leave a Reply