Serial Encoder General Information

Serial Encoder General Information

Serial encoders are encoders which provide the position count and/or other information (such as error detection bits, overheat bits, etc.) as a serial data stream to the drive or other external electronics. This serial data is somewhat similar to data from your PC RS232 port, your PC network, etc. The encoder normally maintains the count and other information internally, and transmits the resulting data in a serial format which is typically proprietary to the encoder manufacturer. Some encoders provide both serial dataoutput and quadrature pulse output. For thorough testing, these encoders should be tested as both serialand quadrature pulse incremental encoders. Examples of this would be the Sumtak, Sanyo Denki, and Heidenhain Endat encoders. Please refer Section to 2.6 Manufacturer Specific Encoder Details for more information on this topic.

Data Display

Data Display is selected to read an input pulse signal and display the count and other information. The Data Display format will vary depending upon the encoder manufacturer. Some manufacturers include more data in the serial data stream than do others. The display format for Fanuc Alpha I serial encoders is shown below as an example.

Encoder Selection

The Encoder Selection frame displays the current selection which is Fanuc Serial Pulsecoder Alpha I64 (8 pole), with 65,536 counts per revolution. Normally either a TI-5005 (or TI-5005W, depending on the keying) or TI-5006 cable would be used with this encoder depending upon whether it used a 17 pin circular or DB15 connector respectively.

Commutation

The commutation frame provides information that can be used to set and check encoder commutation alignment for permanent magnet brushless servo motors. The C1, C2, C4 and C8 boxes show the commutation gray code pulses that are commonly used to indicate commutation position for Fanuc encoders. However the next two boxes indicate the position in terms of mechanical and electrical degrees, and this is usually a more accurate measurement to apply to commutation.

To the right of the Electrical Angle box is the setting for Poles. It is disabled and set to 8 poles for this Fanuc motor. Because this motor has 8 poles (4 pole pairs), the commutation pattern will repeat 4 times in one revolution of the motor shaft. This means that the commutation gray code bits will go through the same sequence 4 times per revolution. Likewise, the electrical angle reading will go through 360 degrees 4 times in one revolution. Checking the commutation by static lockup of the rotor (applying a DC test voltage to the armature winding to lock the rotor in a specified position) will result in the same commutation gray code or electrical angle reading at all 4 lockup positions (within a close tolerance). Notes are available from Mitchell Electronics, Inc. with procedures that describe checking and setting commutation for many kinds of brushless PM servo motors.

The information described above is shown in the Data Display for a Fanuc Alpha A64 which is communicating properly as shown in the figure below.

Position Count

The Count frame shows the position count in Decimal and HEX (hexadecimal). An initial check to make is to verify that the encoder count changes by approximately the correct number of counts per revolution when the shaft is rotated approximately one revolution. This count display can be used for that purpose.

Encoder Status

The Encoder Status frame shows information that helps determine whether the encoder is working properly. 

The first box is labeled INDEX. It will show OK if the encoder has been turned to the position where it will index the count to zero, and it will show ALARM it is has not. This Alpha A64 does need to be indexed, but many other serial encoders do not need to be indexed. They produce the correct position count immediately upon power up with no need to index. For these encoders, this box will be disabled.

The next box labeled DATA will indicate RECEIVING if the encoder is communicating properly with the TI-5000JX, and it will indicate NONE if it is not (as seen in the first figure above). An encoder not communicating may be the result of a bad cable or possibly a bad encoder. Nothing can be done with an encoder until the problem has been corrected, if indeed it can be corrected.

The next column, labeled Internal Error should read OK indicating no internal error indication in the data. If it reads ALARM, then the encoder is indicating that it has detected an internal error, and normally the resulting data is not dependable. An internal error would cause a drive fault in most cases. This usually requires an encoder repair or replacement to correct. This field may not be in effect for all serial encoders.

The next column, with the heading BATTERY ALARM, will show ALARM if there is a battery error alarm and
OK if there is not. It is often possible to alternately connect and disconnect a battery voltage to the encoder to verify that this bit is working properly. This field may not be in effect for all serial encoders.

The box labeled OVERHEAT, will show ALARM if there is an overheat error alarm and OK if there is not. It is often possible to alternately connect and disconnect thermal lines to the encoder to verify that this bit is working properly This field may not be in effect for all serial encoders.

The box labeled ENCODER ID will show the type of encoder detected by the tester for brands of encoders that provide encoder identification information in the serial data. For encoders that do not indicate identification (or identification is not supported by the TI-5000JX software), this field will be disabled.

Count Test

The count test will verify that the encoder is incrementing the correct number of counts per revolution. It works basically the same way for incremental encoders as it does for serial encoders as described in a previous section. The Count Test must be run by turning the encoder slowly by hand.

In addition to checking for the correct number of counts/rev, this test checks for “stuck bits”. For testing to be as complete as possible, some of the fundamental differences in quadrature pulse incremental encoders and serial absolute encoders require the testing to be done differently. Quadrature pulse encoders do not actually count. They simply produce pulses, and external electronic circuitry does the counting correctly if the encoder produces the pulses correctly. Testing for incremental encoders relies heavily on testing for the correct number of pulse edges per revolution. Absolute encoders produce a bit pattern to represent the angle. This bit pattern may be produced by internal counting, by absolute tracks on the code wheel, by interpolation of those tracks, or some combination of these things. The possibility exists for an absolute encoder to produce the correct count bit pattern at the end of each revolution but not show some intermediate angles correctly due to bits not changing properly. One or more bits never changing during a revolution would represent such a problem. The Stuck Bit test checks to make sure that each bit used to represent the angle in one revolution does actually change states during the revolution.

The Count Test is performed by slowly rotating the encoder by hand through 4 or more revolutions until the box at the bottoms indicates Test Completed. The data in the results boxes will indicate how close to the exact number of counts per revolution were registered, whether any bits failed to change state during a particular revolution (stuck bits), and whether any internal errors were detected during a particular revolution. The data boxes are described in detail below.

Revolution

Each row on the display contains the results of the test for one revolution. The revolution box shows which revolution the test is on at any time during the test. After 4 revolutions, the test is complete. It is a good practice to re-run the test turning the encoder in the opposite direction, but the encoder must be turned in only one direction in each individual test. Changing directions during a test will cause the test to fail. Running the test several times may uncover a problem if the problem is intermittent.

Count

The actual count at the end of each revolution is displayed in the Count box. Ideally this number should increase each time by the number of counts per revolution for the encoder. Due to time delays in receiving the data packet from the encoder, the error is seldom zero. The higher the encoder resolution and the faster it is turned, the higher the error will be. Turning the encoder as slowly as possible gives the best results. Serial encoders that are counting incorrectly normally show large errors, so any error below 500 is considered good. See the section below on the ERROR field for more information.

Target

The Target box provides the target count which would ideally be registered at the end of each revolution. It is against this target that the actual count is compared. 

Error

The Error box displays the difference in the data in the Count box minus the Data in the Target box. Since the target count is what we expect and the actual count is what we got, the difference is going to be the error. Due to differences in how that Count Test is performed for incremental encoders and serial encoders, the error is quite different. For incremental encoders, the error for a good encoder should not be more than ±2.

For serial encoders, there is a time lag for the data to be transmitted from the encoder and received and interpreted by the tester. This time lag causes a measurement error for serial encoders. This error will be larger for high resolution encoders that are turned quickly, and it will be smallest for low resolution encoders that are turned very slowly. Normally serial encoders that have counting problems will have very large errors, so the error limit to indicate faulty counting is set at 500 counts/rev.

Count/Rev

The box called Counts/Rev is the difference between the count at the end of the current revolution and the count at the end of the last revolution. Ideally this should be equal to the number of counts per revolution of the encoder. For incremental encoders, it will usually be within ±2. For incremental encoders, it is useful in determining the number of counts/rev of an encoder for which you have no specifications.

Stuck Bits

During the count test, the tester checks to make sure that all the bits that are active in one revolution (or at least the first 16 bits for encoders with more than 16 bit resolution) change state at least once during the revolution. If they do, the data in the box shows OK. If a bit does not change states, it is determined to be a “stuck bit”. Stuck bits are identified by a ‘1’ in the position of the bit that was stuck, while bits which checked out OK are designated by a ‘0’ in those positions. The bits are displayed in groups of 4 with bits 15 to 12 on the left and 3 to 0 on the right. For instance, the following bit pattern would indicate bit 13 as a stuck bit: 0010 0000 0000 0000

This error was generated by selecting a Mitsubishi OSA14 (14 bit encoder) and actually using an OBA13 (13 bit encoder) during the test. The test expected bit13 to change, but only bits 0-12 change for an OBA13. This means that you should check your setup to make sure you have selected the correct encoder if you encounter stuck bits.

Internal Error

As described in the section on Data Display, many serial encoders include an internal error bit in the serial
stream that they transmit to the receiving electronics. Typically a HI error bit will indicate that some self test within the encoder has uncovered an error. The exact meaning of this event depends entirely on the specific encoder and you would need to read information on that encoder to understand the significance.

In the Data Display window, the Internal Error box will show ALARM whenever the tester receives data with the error bit HI, but it will go back to displaying OK if the error bit is received LO. It is not uncommon for this bit to be HI only part of the time when there is a problem. This means that it may be difficult to actually catch the bit HI in Data Display. For that reason, during the Count Test the internal error bit is latched HI for a particular revolution if it ever appears HI during that revolution. That means that it will show up with ALARM in the box if it ever went HI during that revolution even if it cleared itself during the revolution. This latching of data keeps you from missing the fact that the internal error alarmed.