General Grievous

[Pengatom]

Its been around 10 years since I purchased my Robonova-1, after playing around with it for while before a house purchase took over all my spare time. Back in summer of 2015 I changed my job and I felt it was time to get back into robotics. After a lot of searching and "window-shopping" on the internet, I decided to give the old Robonova a new approach to see if the interest and time was there without investing to much money. I did some mechanical changes to the bot to make it more similar to the DARwIn-OP. After a while I figured out that time had run out for the Hitech servos and abandoned the whole project. I had my eye on the HR-OS5, but at a price tag of $20 000 it was a pretty hefty investment, and I spent half a year playing around in SolidWorks and ROS before I did my first purchase, which was six MX-64AT and frames for the upper body:



I had an accident with the Arbotix Pro controller, breaking of the micro usb connector. It's not pretty, but I fixed it with some soldering and using the Molex connector instead.


Got around and ordering the rest of the parts arriving mid February 2017


My idea is to make this a ROS "powered" bot, and in addition to the HR-OS5 I have added two extra DOF in each arm, including a gripper.

I'm not 100% satisfied with the add-on design, and I want to use the available HR-OS5 framework by interbotix, I'll take it of until the base configuration is ready.

I've always been a StarWars fan, so I also want to try to make the bot look like General Grievous. This is not a main priority at the moment, but I'll at least stick with the name!

[jwatte]

You're writing all that, and we don't get to see a picture? That counts as teasing :-)
It sounds like you're making progress, and by all means, I look forward to seeing updates!

[Pengatom]

You're writing all that, and we don't get to see a picture? That counts as teasing :-)
It sounds like you're making progress, and by all means, I look forward to seeing updates!

I've added some pictures and updated the first post. Planning on future updates to the first post and some comments (hopefully from someone else than myself) and progress in future posts.

I have put most of my effort in assembling the robot and updating the assembly document while building. The main bot is now completed, except for the head and the grippers. I forgot to order the idle bearings for the MX-28 when I put in the big order back in January, and now I've been travelling with work for the last month first Houston and these days in Singapore. I order all parts needed to complete the head and grippers back in March, so when I come back home I have to get all designed parts 3D printed and I'll be able to complete the whole bot.

Mechanically that is, I still need to finish up everything related to power supply. For now the servos are running off the 12v10a SMPS power supply and the NUC is using its original power supply, so I need to look into batteries and wiring everything together. I'd really like to have some schematics on how this is done, which input is used for the NUC for instance. I'm playing around an idea to add a On-Off-On switch between the NUC and the main power so I have the option of powering it either from batteries or the power supply, but I would really like to have possibility to power it from a external power supply while the batteries are charging without the need of powering down either planned or unexpectedly...

[jwatte]

Love the pictures! Looks like a fun project.
Personally I'd prefer to read updates in additional comments, rather than an updated first post. (Although adding the pictures was swell :-)

You can make the power supply powering the robot be automatic by adding a diode. When both battery and power supply is connected, if battery voltage is less than (12.0V - Vf_diode) then the battery will be charged by the power supply. For a Schottky power diode, you'll see about 0.6V drop, so it'll be about 11.4V battery float voltage.
I find that keeping the power supply in parallel with the battery is the best of both worlds, because the battery is there for the large current spikes when things start up or move strenuously, whereas the power supply makes sure I never run low on battery and can keep it on indefinitely while developing.

If you use a power supply with a current limit (rather than hiccup mode) and with adjustable voltage, you don't even need the diode, as long as you unplug the power supply when you turn if off to avoid draining the battery through the PS output. Typical "bench" power supplies work well for this.

[Zenta]

Good to see you having some progress with your project!

[KevinO]

On Golem i use a voltage up-converter to 19v to power the NUC. This allowed me to stick with 3cell Lipos. Works fine the past two year without problems.

[Pengatom]

Appreciate the comments!

Was thinking of updating the first post to make it more of a overall presentation of the bot, duplicate/addition to whatever happens throughout this thread.

Regarding power, the HR-OS5 is designed with dual 2cell LiPo for better balance and I think I'll keep this feature. This will give me 14.8V to work with and according to the specs the NUC can be powered either with 12-24v to the internal connector or 12-19 on the external connector and the Arbotix Pro takes 12-24v input so I don't really think I need to step up the voltage to 19, unless I'm missing something?

Now, for parallell powering, since I'm gonna use LiPo batteries, wouldn't that cause an issue if I try to charge two 2s batteries connected in series with my 12v power supply? My impression is that LiPo's are somewhat unstable and should be removed and charged separately under close supervision. This is a issue that I've been very unsure on how to deal with in the best way, so any input here is highly appreciated

[tician]

Very strongly recommend you use 3S LiPo until you have a working gait as a fully charged 4S LiPo will be almost 17V, which will cause overvoltage alarms on the MX servos until the battery voltage drops below 14.8V. While the specs state that MX servos can handle up to 14.8V, they will experience higher rates of damage/failure at any voltage over 12V during any collision, stalling, overloading, etc. Using a 3S LiPo will also permit using a 12V power supply, which would not be otherwise practical as 12.8V is dangerously drained for a 4S LiPo.

If you must have the higher voltage battery, then 4S LiFePO4 is a safer choice than 4S LiPo given the fully charged voltage of ~14.4V and nominal voltage of ~13V. LiFePO4 is also much more tolerant of abuse during both charge and discharge. The big problem is that the series connected cells tend to become unbalanced much faster than LiPo, so will have to perform a proper balance charge much more often if normally performing a quick charge.

A 3S LiPo has a nominal voltage of 11.1V which will likely cause undervoltage alarms on a NUC without a 19V boost converter. The higher performance consumer systems I own are even less tolerant regarding the input voltage and absolutely require a 18~20V supply or they will beep incessantly and refuse to boot.

[Pengatom]

Ok, so its better to use a lower voltage source and step up the voltage for the NUC than to use a higher voltage and limit it for the servos? I'm pretty new to these kinds of batteries, only batteries I have is NI-MH, so its a different ballgame altogether.

Also, this is the setup which is sold by Trossen, wouldn't this be things they thought of? I' just curious, and want to know as much as possible to do the right choice here before spending money on batteries and electronics and possibly frying some of my servos. I almost haven't had time to play with them at all yet!

[tician]

4S LiPo works if you have a well-functioning gait, but not great for initial development or any other time when collisions or falls are likely. Even when the gait is working well, it does not hurt to have proportional torque limits (decrease maximum torque limit at higher voltage to get a consistent 'maximum' at all battery voltages) and fall/collision protections (e.g. if the IMU detects a fall, immediately disable torque to all servos to protect them from damage).