[rbcmetalwork]

I have a 3-axis retrofit knee mill with G320s. The X and Y axes operate fine. They are accurate and repeatable. The Z axis is the problem. It homes okay but movements are not accurate nor repeatable. Troubleshooting (swapping drives, wiring, step/direction signal) narrowed the problem down to the G320. Gecko told me that indeed the drive had a problem. They sent me a new one. The newly installed G320 is better (positioning/repeatability error is less than before) but the problem persists. Please help.

Randy
Honolulu, Hawaii

[mc-motorsports]

Maybe a problem with the ballscrew?

What encodor are you using? Actual brand and line count?

[rbcmetalwork]

The servo motor/encoder is by Baldor. 1000 cpr. The ballscrews are 5 tpi.

In my troubleshooting, I swapped motors, X for Z, and the X-axis worked fine. Conclusion: Z-axis motor is okay.

I also used the Z-axis step/direction to move the X-axis accurately and repeatably. Conclusion: the breakout board signal is okay.

I was able to get the Z-axis working accurately and repeatably by using the X-axis G320. Conclusion: the Z-axis G320 was the problem.

Have I made an error somewhere?

[mc-motorsports]

Are all 3 servo's the same manufacture? And all encodors are the same manufacture?

The Z axis driver is not repeatable, but if you use another existing known good driver, everything is ok... But you already replaced the Z axis driver and have the same problems with a new drive? Wow. If you had a faulty drive, I would say it falls under the "stuff happens" catagory. If Gecko sent you another drive and that one is bad too??? I would think there is something causing the drive to fail. Gecko normally has excellent quality, but like I said, "stuff happens" and it's not impossible that a bad drive slipped out the door. 2 in a row???

I had a similar problem. All 3 drives failed, it was because of mis-matched encodors. The encoders were drawing too much amperage (renco?) and caused the drives to fail and or fault. I sent them back to Gecko for testing, 1 was beyond repair, 1 was repairable and 1 was fine. All 3 servo's were the same, all 3 encoders were the same. Gecko called me, through conversation, diagnosed the problem, sent me 1 new driver, 1 repaired driver and returned the good driver. They also told me were to get the proper encoders to work with the Gecko Drives, I think US Encoder, everything has worked great since with absolutly no issue and I can't say enough good things about Gecko and thier customer service. The engineer over there really knows his stuff!

Next question, are you dealing directly with Gecko or with a distributor/company who supplies retrofit components?

[rbcmetalwork]

All three motor/encoders are of the same manufacturer and model.

The (3) G320s were purchased directly from Geckodrive. They told me that their technicians found that the G320 was inconsistent and that they were stumped as to why.

[rbcmetalwork]

After an educational phone conversation with Mariss, the culprit seems to be the encoders which are drawing too much current for the G320s.

[mc-motorsports]

I'm sure Mariss told you where to get the proper encodors and such. I'm sure once you get everything up and running, you'll be very happy. Thanks to Mariss, I got the right encodors and have been running with absolutley no problems for 2 years.
Good stuff!

[mc-motorsports]

BTW, my problem was the guy who sold me the complete package of a control unit using Gecko drives, 3 servo's and the 3 encodors that were installed drew too much current. The system failed immediatly. I called the guy who sold me the stuff, he was no help. Called Gecko, they got me fixed right up! Marriss told me everything that I needed to do to clean up the other guys mess. I called the guy who sold me the complete system and he avoided me like the plauge... Never did make right for his mistakes which cost me more money, I think he went out of business, wonder why...

[rbcmetalwork]

I purchased and installed 3 new U.S. Digital single ended encoders. They passed the encoder test fine. However, the motor won't lock. Same problem on all three axes. More phone calls to Mariss. Seems like the encoder signal is getting disturbed when the motors are connected. Keep in mind that these same G320s worked fine with the old encoders drawing 160mA. I'm going to change to differential encoders and see if that works.

[rbcmetalwork]

I removed the single ended encoder and installed the differential encoder on the X-axis motor. The motor now locks and rotates fine. Y and Z axes next.

[mc-motorsports]

That's new to me, haven't heard that one before. What encoders are you using now? Did you figure out what was wrong with the old ones?

[rbcmetalwork]

I coupled the X-axis motor to the ball screw and did some accuracy tests. No joy. I'm still losing position. Not accurate nor repeatable. I use the glass scales from my DRO as a reference for positioning.

My next plan is to purchase and try the twisted pair cables and differential receivers. That should hopefully rule out any noise issues with the encoders. My X-axis cable is about 6-feet long which just so happens to be the length at which they don't recommend a single ended encoder.

The differential encoders on there now are from U.S. Digital, part number H15-500-N-D. The original Sumtak encoders worked (on the X and Y axes but not on Z), although the 1000 CPR was a little high and they draw 160mA.

[BobWarfield]

What level of inaccuracy over what distance does your DRO tell you?

i.e. I jogged 12" and it was off 0.003"

What backlash have you measured on the axes?

Just curious what the magnitude of the error is.

Also, is the error the same every time? Try not to reverse directions. Measure a couple of jogs in one direction of the same length. What's the error look like for each?

Now reverse direction and do the same. Zero after reversing and move at least 1" to start the next 2 jogs.

[Mariss Freimanis]

It seems the USdigital encoder CH_A and CH_B single-ended line drivers spec out at 8mA sink current and 40uA source current. 40uA! You can't drive a cable with 40uA.

Mariss

[mc-motorsports]

It seems the USdigital encoder CH_A and CH_B single-ended line drivers spec out at 8mA sink current and 40uA source current. 40uA! You can't drive a cable with 40uA.

Mariss

is 40uA something like .0040 amp? Or .00040 amp?

[rbcmetalwork]

The inaccuracies vary. Sometimes as little as .002" in a 4" move, sometimes more than 0.3".

The backlash on the X-axis (which I'm currently working on) is 0.0003". The motor on this axis is directly coupled to the ball screw.

[rbcmetalwork]

40uA is 40 micro amps which is = 0.000004 amps

[Mariss Freimanis]

It's like 40 millionths of an amp (as in 0.00004 A). If you order from them, MAKE SURE you order the 'EM1' optical head and not the 'HEDS' one. They skin you for an extra $3.15 but at least it provides a +/-8mA sink/source capability. It can drive a short cable (<6') while I wouldn't trust the HEDS spec to drive a 1" cable.

It used to be their encoders were rock-solid. Now we have been dealing with stuff like "The G320 passed 'encoder test' but goes nuts when I connect the motor!".

Begin rant:

Something must have changed at USdigital. 40uA is not something in good conscience I would ever use to drive a cable! I would even think twice to even drive a 0.1" long pcb trace with that current. 8mA is barely acceptable but not 40uA.

Figure on 50pF per foot of cable; a 6' foot (2 meter) cable has 300pF of capacitance. How long does it take to charge 300pF to 5V at 40uA? 37.5uS is how long and it's also the rise time. It also means if CH_A goes from 5V to 0V it injects a 38uS 'notch' in the CH_B wire if it is misfortunate enough to be at 5V at the time (50% chance). That kind of 'notch' guarantees malfunction of the encoder signals. Malfunction on the encoder lines means malfunction of the drive.

They offer "Line Drivers" at $15 and change. The 'line driver' is a $0.12 IC that should have been included inside of their encoder. They ship a cable with their encoder, the expectation is the encoder can drive that cable so why isn't the miserable $0.12 driver IC inside the encoder where it belongs? Why are they charging more than 100 times what it costs when it should have been for free inside the encoder?

I can no longer recommend USdigital encoders until they get their act together, stop the cost-cutting and include a proper driver for their cables they supply.

If you want a really superior encoder for just $29, go to www.digikey.com and tap in 'AMT102-V'. It has a serviceable line driver, has nearly perfect 90-degrees quadrature and get this, it is 16 encoders in one. It has a DIP switch that sets the encoder line-count resolution from 48-lines to 2,048-lines with 14 settings in between. It's a capacitive-type encoder so there is no dirt sensitive optical code-wheel. Finally, the pinout and cable connector exactly matches what's used on the USdigital encoders.

Why the rant? It is because USdigital is costing us time and money dealing with application support calls that devolve into the short-comings of their encoder cable drivers. These calls wouldn't come up if their encoders could actually drive the supplied cables.

End of rant.

Mariss

[rbcmetalwork]

Apologies. Mariss is right of course. I was thinking of 4uA. It was a rough week.

[zapmk]

Mariss,

Thank you for the rant, was just about ready to purchase my encoders from
USdigital until I read this, I don't need the head ache on my first time build.
Will be purchasing the one's from digikey!