Losing counts on one axis

I posted this here because I’m using Mesa hardware. 5i20 and 7i33, lcnc 2.7.14. Encoders are mounted on the screws. 

I have an extremely odd condition on a machine that has been working flawlessly for about 10 years.  It’s a lathe and on the z axis it moves toward the spindle at aprox .0005 increments each time the axis is reversed . 
   After the machine is referenced by the GUI, it will return to that point only by using the GUI , if G28 is issued  or G53 z0, it will be off by .0005 each time the commands are given and will accumulate. Telling the machine from mdi the following will result it the z axis being off position by . 0025   G53 z-1G0 then G53 Z0 repeated 5 times.  Commanding less then the lead , or 1 turn of the encoder does not produce an error. I have done the following:

 Replaced the thrust bearings, the was no backlash and is none now.
 replaced the motor
 replaced the drive
 tried 3 other encoders which all had the same but different then the above issue. They tripped the following error and lunged forward .02 in the direction of a jog after approx .075 of movement. 
  Tried 3 different 7i33 boards.
   Swapped X for Z on the 7i33, issue follows the Z axis .

    Initially I suspected backlash in the thrust bearings or a loose screw to encoder connection.
  Only a bad carpenter blames his hammer, I’m out of ideas as to what this could be. It seems like 4 new encoders being no good is unlikely.  It worked with out issues for almost a decade. Suggestions?
     
  
   



 

What about encoder cabling?

It sounds like you have not successfully swapped the encoder
so the Z encoder is still suspect

Is it possible that the new encoders are scaled differently or backwards?
(these would both cause following errors)

  A small electrolytic cap smoked and puked electrolyte all over the glass. Same issue with 2 of the encoders. I cleaned the goo off and it works. I’m trying to get some info on the cap and it’s place in the circuit to repair. Who woulda thunk?

 

I did check the 5v power supply and it was at about 5.1 with no ripple . The encoder is rated 5-28 volts input voltage. I should mention, these are new as in new old stock, they are over 15 yrs old. My understanding is electrolytic caps have a 5-8 year shelf life. In retrospect, I should have opened up them up sooner, it just seemed really unlikely that I would have 4 new encoders with issues. Lesson learned!!

If it's the blocky component that's still there that's vented, I think it is a tantalum cap rather than an electrolytic one. Tantalum caps are famous for exploding when they get old... at least certain types/series.