Encoded motor recommendations

[robof]

I'm looking for recommendations on 12-24v dc brushed motors and encoders, preferably with the encoders pre-installed, that have been tested with roboclaw and can be accurately positioned via the encoder. A pre-installed gearbox is a plus as that's one less thing I have to mess with.

I would like to avoid anything that uses a cheap hall effect quadrature encoder like the actobotics/servocity/robotzone motors as I've been having issues with the noise on their encoder circuits and do not want to modify the circuits.

I currently have 2x7A, 2x15A, and 2x30A roboclaws so any motor that is drivable by any those would be fine. I'm mostly trying to do a proof of concept so no real load requirements at this point; the only major requirement is that I need to be able to position the motor with some degree of accuracy and repeatability.

Thanks!

[Basicmicro Support]

Your best option is to get motors with dual shafts(front and back) and the US Digital E4P encoders(press fit on the back shaft). I've also been happy with AMT encoders and they come with multiple shaft size support attachments.

In general you are not going to find motors with optical encoders that compare on price with the Hall sensor motors. For something in the same price category as the hall sensor motors you will need to look on Ebay and hope there is some left over stock from a custom motor if you want something with the encoders already installed.

As for the modification of the motor, all you need to do is remove the snubber cap(some times it is one, and sometimes it is 2 smaller caps). The cap(s) will be the only ones with continuity to the main motor leads.

[robof]

Thanks, I will get both of these and try. Are both/either of these plug and play, or would they need a pull-up resistor?

I may try the modification to the snubber cap mentioned on one of the motors, it's about a $50 test, but assuming it worked perfectly I guess it would save me the time of using after market encoders.

I wasn't necessarily expecting anything as cheap as the hall effect encoders; I know servocity charges about $20 extra to include the encoders, but I wouldn't mind paying $50+ for an after market encoder if I knew it would work as expected.

[Basicmicro Support]

I doubt US Digital has changed there designs since I bought the EP4 encoders(though they may have a new model number) so you should not need any pullups with them. I'm also pretty sure the low cost AMT encoders dont need them either(but it has been a while since I used it). Last time I bought either encoder they were in the $25 price range. But they do require the rear axle on the motor which isnt always easy to find. The US Digital is probably the simplest to install though. You just tape(part is already double sided taped) the bottom of the encoder on the motor using a special positioning tool(included with the encoder) and then press the encoder disc on the axle using the same tool and then just snap the cover on and your done.

IIRC the AMT mounting was the reverse. You push on the axle sizer(comes with several sizes to choose from) and then just push the encoder on over it for positioning. I think it also came with double sided tape already installed but its been a while so Im not sure.

[robof]

I went ahead and purchased both to try. I will report back with my results for anyone else that might be interested.
- US digital E4T
-- https://www.usdigital.com/products/enco ... ry/kit/E4T
- CUI AMT10
-- http://www.cui.com/product/components/e ... mt10-v-kit

Looking forward to testing these and thanks again for the help.

[Basicmicro Support]

E4T looks a little different from my E4P but the AMT looks exactly the same. You shouldnt have any trouble.

[jwatte]

Another option, if you want gearbox, motor, and encoder all integrated into one nice package, for a reasonable price, is the Pololu D37 gearmotors.
They're not super powerful, so if you need something that's both torque-ey and fast,you might do better with something you assemble yourself.

https://www.pololu.com/category/116/37d ... gearmotors

[Basicmicro Support]

Unfortunately these motors also have onboard snubber capacitor, and are suseptible to motor noise on the encoder signals.

The noise is being generated because the snubber caps pass A/C current even though they block D/C current. PWM generates both A/C and D/C current. The cap effectively removes the D/C offset of the voltage so at the snubber cap the PCB is seeing +-Main Battery voltage swings at 20khz when using Locked Antiphase PWM and +- 1/2 Main Battery voltage swings when using Sign Magnitude. So Sign Magnitude should have a significantly lower amount of noise but there should still be noise.

Frequency of the PWM may also play a roll since the snubber cap is effectively a LC circuit(combined with the motor inductance) and has a natural resonnant frquency.

Currently removing the snubber cap(s) is the best option if you feel comfortable doing it but if not please check back next week.

[jwatte]

That's interesting!

When I run those motors and hook a scope to the encoders, the signal actually looks clean, but I can't run them under load with the scope connected (don't have a dyno or anything like that.)

Are you suggesting that the cap on the board would pass enough HF that it would make the RoboClaw miss counts?

[Basicmicro Support]

It does seem to be hit or miss. Based on the number of people that have used them its probably a small percentage that see the problem. But I have personally seen the problem on several motors. May be there are different revisions of the boards. Different values of the snubber caps are used etc...

Its fairly clear on the motors I've used. Even under no load. You will see noise that lines up with the rising and falling edges of the motor PWM(eg 20khz) on top of the encoder pulses. It may also show up on the 5v power line when the encoder is plugged in.

Removing the snubber cap removes the noise so its definitely related to the cap. My theory is that the A/C component of the PWM 20khz passes through the cap(completes the circuit). On these motors that cap is very close to the hall sensors. Theoretically the noise could be getting on the signal line through the hall sensors. Once the cap is removed there is no longer an A/C component passing right next to the hall sensors so the noise is gone. However this doesnt completely explain how the noise gets on the 5v rail. Possibly because through the hall sensors as well. To test my theory I would need to remove the hall sensors to see what happens.

The symptom people see that dont have a scope to look for the noise, is missed counts or extra counts on the encoder readings from the Roboclaw. I've seen the noise reach +-3v on the signal line. I always assumed all the motors had the noise problem and most people just didn't notice a few missing counts(it wouldn't be obvious in speed control mode unless it was very bad). But I get the feeling it may be luck of the draw which motors show it and which dont.

The motors I have used(several from different resellers) all had the noise though not all had it as extreme as +-3v. Only one had it that bad and its why I found the problem in the first place. It was obviously getting wrong encoder counts when moving a linear slider with one of Servo City's encoder motors, specifically worse in one direction than the other. At the time(this was about 2 years ago) I thought I had a bad Roboclaw so I started digging and couldn't find anything wrong with the controller so I finally thought maybe there is something wring with the encoder. The noise was obviously 20khz related so I looked at what was on the main motor leads of the encoder and saw they had 2 caps(on this particular model), another unpopulated cap and what looked like a test pad. So I removed the two caps to see what happened and the noise disappeared.