Roboclaw motor controller investigation

http://www.racingfr.com/forum/index.php?showtopic=47623

@BlazinH , I to use hall effects, but not use them to the extent @eaorobbie has, but after his initial discovery, I lube mine with CRC spray as well, last thing I need is pots going off. They are mounted on my actuators in this case. I intend to change them to wire wounds later once money is free.
As for wire wounds, I have also used them for years, in fact I sold a sim with over 100 hours use on them and they still run spot on. I cannot notice any "noise" either. They are very accurate and I have no problems with them. I have them on the desk racer and its spot on.
Another member has used hall effects and had the same problem, they tend to seize and require maintenance. No problem, just a spray of CRC lube.
Still its an issue that has been raised and it should be noted as others will use them and they need to know.

@eaorobbie , This is splitting hairs don’t you think? When a bushing is used as a bearing, it is also a bearing. But unlike ball or needle, it is usually self-explanatory without adding bearing to the end of it. The documentation refers to it as bearing bronze bushing.

I not sure what the motor driver has to do with the resolution and accuracy of the hall effect pots. But again, technically these are digital and noiseless. However they have a maximum hysteresis of .2% or .002. At 12bit resolution 4096 * .002 = up to 8 units drift or +-4 units. However, from my experience drift is less than +-2 units. I read mine at full 12bits resolution @ 360 degrees. But since I like to display and use degrees output, I converted it from 4096 to 3600. So I display the output in 1/10ths (.1) of degrees. My drift is never over +- 1/10th of a degree.

True! But they still have 12bit range. It is just done with .2 to 4.8 volts instead of 0 to 5. It’s a small hit though if you are using a 0-5v adc that doesn’t allow for a different reference voltage or you can’t constrain the 0-5v to .2- 4.8.

To my understanding, most noise on a standard pot is caused by the motion of the wiper moving against the wire in this case. So at rest you shouldn’t see much if any noise. But if you are showing hall effect fluctuations of 2% of the total value, it sounds more like the problem is the quality of your motor driver! In fact, this seems to confirm a hunch I’ve had about the Jrk. Since I haven’t used them I can’t confirm this but maybe you can. The documentation says that the Jrk have, “High internal resolution (12 bits) for smooth and flexible calibration to various input and feedback devices.” It doesn’t say that is has 12bit input though does it? The microcontroller used on it only has a 10bit adc and I don’t see another chip on the Jrk that looks like a secondary 12bit adc. So I wonder how it could input at 12bits because it doesn’t seem possible to me with what I know at least. I believe somewhere it refers to input scaling also. So if its input is actually 10bits and scaling it to 12bits, then the drift will also be multiplied by 4 times which would seem to account for a 2% drift reading to some extent at least. Otherwise I don’t know what the issue could be because the hall effect pots definitely don’t drift 2%. That would be equivalent to around 7 degrees drift!

If that was the only issue you were having with them then for about $45usd you could purchase ball bearing hall effects. But, of course, if you are satisfied with what you have already then there is no need to do so!

But I concede that everyone’s situation is not the same! So just because I have been quite satisfied with halls, as you have been with wire wounds, that doesn’t necessarily mean it will be the same for everyone!

Thanks for the link! (I think anyway because I don't read French )

@bsft For full disclosure, the hall effect units I have used for over 3 years and hundreds of hours are ball bearing units. I have just recently started using the BI Technologies cheapo version with the bushing so I don’t have many hours use on them yet but so far so good!

ok then, do you have a link for those pots? because unfortunately the ones we use need constant lubing to keep them fine. Maybe because they are cheap.
Lets not compare that hall effects are good or bad. Obviously you use ones with proper ball bearing sleeves in them. The ones we use are bronze sleeve on steel shaft and have not shown the durability we would like. Still a good unit, just need extra servicing.

Hi
With the BI tech 180 deg Hall Pot, I have these mounted direct with a coupling on the back side of the out put shaft using 160 - 170 travel
On the output window trace I sometimes notice a small jitter other times the trace settles to deadband spot on,
I suspect because the hall pots are so sensitive and accurate they are detecting a small amount of backlash in the motor gearbox, I dont have the jitter with quad encoders but they are mounted on the motor armature so no back lash.

regards Terry

Sure. The supplier I use doesn’t ship internationally but I found a supplier in your country.

http://www.digikey.com.au/product-s...t=0&page=1&quantity=0&ptm=0&fid=0&pageSize=25

The ball bearing “units” I have been using output a pwm signal instead of a voltage though so those wouldn’t work for you. I have not used the BI Tech ball bearing units yet but the front end (hall effect) have the same specs, I just have a different output type. The BI Tech ball bearing units are rated at 100 million shaft turns.

While I was searching for a supplier for you I ran across this product training module for the first time that explains their features and benefits (w/ audio).

http://dkc1.digikey.com/au/en/tod/T...Pot-Rotary-Sensor/MagnePot-Rotary-Sensor.html

To those who like to only post "disagree" without an explanation; it doesn't add anything to discourse! (verbal interchange of ideas; especially : conversation )

Sorry that you don't like that I can't read French! But I'm not sure what the link is actually saying! It is rather ridiculous that you feel compelled to disagree with this posting!

Sorry, I found the new Google website translation feature (it moved to iTools)
http://translate.google.com/translate?hl=en&ie=UTF8&sl=fr&tl=en&u=http://www.racingfr.com/forum/index.php?showtopic=47623&sandbox=0&usg=ALkJrhhnw-Tjgpd5aJg-udPMJ9c8HlUyRQ

Now I understand!

@eaorobbie , A friendly debate on ideas and what the correct facts really are would make it easier to come up with the best solutions for everyone.

But, while you seem to have discarded the hall effect pots already (thanks fine though if you wish) it dawned on me why you are having problems I think. These require a stong 5v power supply capable of more current than a standard pot needs. They recommend a regulated one also. If these are not powered properly and have even a little low voltage or current, the signal will degrade and fluctuate much more rapidly! This seems to be the only answer to me why you are having those wild fluctuations!

@BlazinH , its not voltage, its actual use on the pots and the shaft be twitched up and down rapidly. the pot reads fine, its a mechanical limitation of the pot shaft being stuffed up so it constantly needs lube so they dont stick.
My converted brick spreader actuators shaft, even though its aluminium on a plastic sleeve, still needs a bit of spray lube on it now and then, or they tend to stick a bit.
Same thing we have found with the hall effects we use, need servicing.
If the wire wounds work better for us, we use them, we trialled the hall effects and have reported back our issues.

This discussion is about the roboclaw, so lets get off the pot problem.
As I dont know, nor understand the roboclaw, I am leaving this discussion.

@bsft , I was referring to its voltage requirements due to this statement, “Sitting still and zooming in on the output graph shows them fluxutating about 2% of total value”. This is not normal!

I have yet (knock-on-wood) to have any problems with my cheap magnepots sticking and requiring lube. I'm not sure how many hours I have on them but I estimate around 50 each on three of them. How many hours before you started having issues (I wonder)? They are rated at 500 rpm so I don’t see how being twitched up and down rapidly should make a difference because I doubt they are going faster than that. If your bearings are wearing this quickly, they are probably still getting forces applied to the shaft that causes friction and wear. Do they practically float on the shaft? A standard pot requires a stronger mount because they have more turning resistance. But since the hall effects are non-contacting, they have very little resistance. So all you need is a small flexible wire to mount them with. I use bailing wire which is stiff enough but flexible too. Then mount the wires ends to two points where it keeps the body of the pot from rotating but still allows it to move up and down and sideways with the shaft. This is how I mount mine.



I too stated this thread went south again some time back and needs split again! But as far as I’m concerned, unless someone gets a roboclaw that knows what they’re doing when testing it, the investigation has come to a stall anyway!

This discussion is about the roboclaw, so lets get off the pot problem.
As I dont know, nor understand the roboclaw, I am leaving this discussion.