Carnegie Mellon University > Robotics Institute > Garth Zeglin > Web Log
I am currently engaged in a very frustrating transaction trying to buy a Micron gearhead from Danaher Motion. Our local distributor is Motion Industries, who has contributed nothing to actually getting this order through correctly.
I didn't personally place the original order, but apparently it involved specifying the mating motor as QB02300 (which has a 0.375 inch shaft), getting told it wouldn't fit, speaking directly to the factory, then specifying a different motor QB01701 (which has a 0.250 inch shaft).
Some time later a gearbox arrived on my desk, and I discovered it didn't match either of the two specifications: it seems to be designed for a NEMA 23 face with a 0.312 inch shaft at least 1.1 inches long, which is longer than any NEMA23 motor I have yet seen.
So I call the application engineer Debra Holder and leave a message. A week passes and I call again and get her in person. She remembers the order as NEMA23/0.375, and agrees there was a mistake, promising to call me back, which doesn't happen.
So a week later I call again. Now I am told that the NEMA23/0.375 combination isn't possible, at which point I start to get testy: what exactly did they ship me?
After a long hold, I am then told that the combination does work, and they custom-made the part for me; if they did, I didn't get it.
At that point I also ask about the design spec for the shaft length, and she looks it up and says 1.27 inches. I ask where that number came from, and she says they looked up the data sheet on the web. I ask if they still have the source of the dimension, and she says no. So I download the datasheet again myself and forward it to her; as near as can tell, someone read the 12.7 millimeter reference shaft length dimension off the encoder shaft and used that. It's a three part error: the wrong shaft, the wrong units, and the wrong magnitude. And if you count that I thought the entire order was for a different motor, it is a four part error. And in my mail I suggest the correct solution is just to provide me an entirely different gearhead for the QB01701.
This is old-school engineering sales with all it's waste: a lack of a clear part numbering specification, unclear verbal orders, confusion between the engineering staff and the distributor, a reluctance to use email, and a lack of a design paper trail or system to confirm critical dimensions.
2006-03-30 update: a day later I hadn't heard from Debra Holder so I called again and she told me she would have an entirely new gearhead made for the QB01701 as I wished, and that I should just keep the incorrect one. I then very gingerly asked her to send me a confirming email with a statement of the mating motor and the critical shaft dimensions, and she complied. I am relieved at this outcome, although it doesn't much improve my trust in the reliability of this company; I would describe them at best as "barely cooperative."
2006-04-21 update: the new gearhead arrived, and it is not for the QB01701 as promised. However, it does seem to match the QB02300 spec as was originally requested, so I'm just going to accept it. Still, I find this behavior unacceptable: I regard this as an unauthorized order change, since we did reach agreement on the QB01701.
I am finally joining the horde and putting up some commentary in blog form, albeit very infrequently updated. I'm starting with a few backdated entries from recent gripes.
I envision this mostly will be used as a way of documenting problems with equipment that I purchase. Hopefully that will help someone somewhere avoid some of the same headaches.
The editorial, of course, is that companies have little incentive to advertise the failings of their products, so only by banding together to question them can we pressure them to fix anything...but many people have said this better than I can.
I do feel compelled to say this, however: all information on this page is solely the opinion of Garth Zeglin, and is offered as-is with no express or implied warranty. I make a good-faith effort to provide accurate commentary and analysis, but use at your own risk.
I have a minor gripe about the flanged button head cap screws that 80/20 Inc. ships for use with their aluminum extrusions: when trying to insert them into the inside holes of the standard four or eight hole brackets, the screw flanges prevent installation in adjacent inside corner holes. Here's a photo to show what I mean. You can see that the flange of second inside corner is catching on the top of the other screw.
I did send a message to 80/20 customer service which did not yield any response.
I actually purchased the screws kitted with T-nuts from McMaster-Carr as part 47065T142, so I'm not sure of the exact 80/20 part number.
Also, the overpriced 80/20 T-nuts look pretty painted black, but seem functionally identical to a much cheaper 1/4-20 plain steel tab-base weld nut. I purchased McMaster-Carr part 98001A130, and while not as nice-looking, they seem to work the same.
I recently purchased an Omegadyne TQM202-28 reaction torque sensor, distributed by Omega Engineering, for a test setup. But I have a word of warning to anybody who might be researching this product: it has an unusual shaft design with three symmetric shaft flats, and the company does not seem to appreciate the naivete of this approach. I sent a query asking for the preferred coupling technique, and engineer John Menke sent me a sketch involving three symmetric set screws. By my analysis, this is a flawed technique. The engineer did not respond to my critical commentary and analysis.
I'd welcome any corrections to my analysis, although I'm confident I have the correct basic argument: symmetric setscrews do not provide the strong frictional coupling of a single setscrew, and introduce substantial stress concentrations. I'm not putting up the sketch here just to avoid any potential copyright violation claims, although I made no agreement as to confidentiality, so I reserve the right to do so.
I'm still in the process of building the setup, so I don't know yet how well the sensor works or how well my own coupling design will hold up, so this is still a theoretical design argument.
If you try using a Mac G5 running 10.3.x with a Hitachi CM721F, you will have problems with the completely broken color profile. I figured this one out, but I guess this was the penalty for trying to use an old-fashioned and non-Apple CRT with a new Apple computer.
Another minor annoyance is that to connect two analog CRT VGA monitors, I had to buy a third party cable for about $35; Apple supplies one DVI to VGA adapter, but I also needed an ADC to VGA adapter.
But don't get me wrong, once past this hassle I've been quite satisfied.
Page revision: 2006/09/17 15:39:23.
Garth Zeglin, Robotics Institute, Carnegie Mellon University.