Hi Jon, Thanks. I'll end up trying that or even running separate power.
> Also, a lot of these cheap encoders (CUI and US Digital) > benefit greatly from a small capacitor (0.1 uF) right at the > encoder's power terminals when long cables (more than a few > feet) are used. > > Finally, many of these encoders, even if they have a > totem-pole output, benefit from a pull-up resistor on the > outputs to +5 V at the receiving end. > > Jon Here's a quick update on the CUI AMT112 encoder compared to the US Digital. I received two from Mouser but had ordered the wrong cable. So since I had a Digikey order in the works I added the correct AMT113V-Q encoder from them so that the purchase of the $47 cable from Mouser wouldn't be a total write-off. I then removed the US Digital (which is in a nice aluminium package completely shielding the optical encoder, and installed the CUI AMT. I ran it both with and without the encoder end ground cable connected to the body of the motor. Without it the noise on the quadrature differential outputs is nothing short of amazing. With it connected it's the same as the US Digital. Support from CUI has been outstanding and fast. Seems the Digikey Encoder was made in 2015 and CUI changed the encoder to make it more noise resistant in 2017. So I have to try and return this one to Digikey. In both cases the HP_UHU Atmel processor reported overrun errors (that's where the A or B changes in a way that is illegal for quadrature patterns) suggesting the noise may either still affecting the encoder or it's on the HP_UHU board itself. I changed over to Henrik Olsson's dsPIC replacement module and both encoders appear to work fine. After hours and hours of G-Code moving the motor back and forth for a motion of 5" at F50 the line on the pulley matches the line on the motor. So no lost steps. Since I like the case and the cable better on the US Digital I put it back and have been running again for hours under LinuxCNC which has way less jitter on the stepping pulses. Both systems can generate 12.5kHz for the F50 (50 ipm) on the parallel port. The WIN-XP with MACH3 is just rougher. LinuxCNC is smoother due to less jitter. I have not yet gone through the check list for either OS to improve speed or reliability. I'll give the AMT112 units one more chance since I now have some flat Ethernet shielded cable along with the connectors and crimp pins to create a single cable from encoder to HP_UHU servo drive. And those two encoders from Mouser were made in Week 16, 2019 so are the newer design. John _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
