matthieuvigne / STS_servos

Arduino library for Feetech's STS smart servo
MIT License
12 stars 6 forks source link

Ask Accuracy #5

Open reb1302 opened 3 weeks ago

reb1302 commented 3 weeks ago

I have a project that needs a replacement for the AX-12A servo. I would like to ask if the STS3215 has comparable accuracy in feedback and torque, and if it can replace the AX-12A. My project requires relatively good accuracy. Thank you if you can share your experience with me.

matthieuvigne commented 3 weeks ago

Hi !

Unfortunately no, I would say the STS are slightly less accurate than the AX12.

In terms of position, their only drawback is backlash, which is quite visible (I would say at least 2°, but I never measured it precisely), and I don't recall the AX12 having such backlash. If you can live with it, the encoder has a high resolution (4096 ticks), and the result is quite repeatable: that means that for instance, for a robot arm always working against gravity, because the servo is always on the low-side of the backlash, the results are quite accurate. But if the load changes sign, or if you do very dynamic motion, this could become a problem.

In terms of torque, the measurement is extremely poor - in fact, I haven't even managed to use it for a simple binary information 'am I holding an object or not'. I think this is linked to the design itself, and it may be the same on an AX12 (I don't remember ever testing that when I worked with these servos): there is no torque measurement, only motor current is measured, and these servos have a very high reduction ratio (close to 300 I think). So everything is lost in friction: basically you see torque (current) when the motor is moving and almost no torque when the motor stops - but you can't use it to measure a weight precisely.

Nevertheless, I think the STS3215 (and other Feetech servos) are great for their price - they are after all quite cheap and make a very affordable replacement for the AX12 in my own projects. I don't think Feetech specifically has 'high accuracy servos', but I could be wrong. I know however that they recently launched a new product which they call 'constant force servo' - the name seems to say they have improved on force sensing, but I don't know what technology really is behind it (I suppose it's still current sensing and not true torque sensing), I haven't tested them: https://www.feetechrc.com/hl%E6%81%92%E5%8A%9B%E7%B3%BB%E5%88%97%E8%88%B5%E6%9C%BA.html

reb1302 commented 3 weeks ago

Thank you very much! The article is very detailed, and I have gained a lot of information about the STS3215.