Question 6DOF HeXpod: DIY or COTS

Hi all,
I have a quick question: I am really interreseted in the design of the 6DOF called HeXpod.

It has a very small footprint and seems nethertheless pretty powerful. I which I could build something similare but there isn't many information on it.
So my question is the following: Do you guys have more info about it ? I mean, do you know if this is a commercial product we can buy or is it a DIY rig ? If it is a DIY, I would assume there is a thread on that forum but I couldn't find anything related to it.
Thanks in advance for your help.

@hexpod
By chance, are you the owner / builder of the 6DOF motion platform mentioned above?
Thanks mate

heXpod 6dof all-in-one motion simulator is a DIY Stewart Platform based on "bare bone" AMC1280usb controller equipped with Teknic BLDC servo-motors with external feedback positioning sensors.

Small footprint (90cm/140 once stored) was a prerequisite to reach the famous WAF* factor.

Back than I was searching for servo-drives with absolute positioning capability but unfortunately the interface to the internal encoders was not possible due to manufacturer limitations. (That's why we switched to feedback sensors)

heXpod 6-dof software is a thir-party IK plugin for X-Sim and SimTools ajustable for any kind of Stewart platforms based on AMC controllers.

More info about the plugin here:
Http://heXpod.xyz

@hexpod - I was looking at the Teknic website and was curious which 3432S motor you used. I’m guessing MC series? If so, did you use the ASU-FR510 Analog Send Unit to convert from the AMC 1280USB signals?

Also, the Newstart IB90 looks good. Does it take the 0.5" motor shaft directly? I’ve been looking for drawings but haven’t found them. I also don’t find a vendor that sells them. I have only found the specs. Where did you get those?

And from the two big transformers, I bet you built your own power supply for the motors? Did you go 75 volts or a different value?

It looks like you changed from tooth belt feedback to a drum and pull string. I like both setups. Is there a reason you switched?

I really like the outriggers in your design. Much simpler and more compact than a full hexagon.

Sorry about all the questions.

@Zed

Exactly, I am also really interrested in this design and would like to build something similar. Hope you will get your answers

@hexpod
Thanks for your answers. Is there a build thread somewhere for people like me who really like the design for its compactness and would like to design something similar ? (Or at least inspired by your design)

@FoxONe42 - There are these specs from Hex's videos. The BLDC motors have some advantages over AC for me because I can use batteries to get past the average current draw issue on a 120 volt mains. I also like the planetary gearboxes over worm since it appears wear is less localized with more gear surface carrying the load. I think I have enough to get started except for a gearbox source but Teknic says contact them for compatible gearbox sources.

Definitely a more expensive rig but Hex did a beautiful job with the design and it looks very capable and no biger than my current 2 dof when stowed.

ASU-FR510 Analog Send Unit to convert from the AMC 1280USB signals?

No, the AMC is directly interfaced with the servos in PWM velocity mode

Does it take the 0.5" motor shaft directly? I’ve been looking for drawings but haven’t found them.

Yes, The IB series from NewStart are directly compatible. btw they changed the labeling since I bought them. I think it's AB now. You have to deal with the application engineer to get all infos you need.

I bet you built your own power supply for the motors?

No, I use Teknic power supplies.

It looks like you changed from tooth belt feedback to a drum and pull string. I like both setups. Is there a reason you switched?

No, it's the opposite. The timing belt is more reliable I think.

Teknic has lounched the new AC servo. I am sure Newstart can provide a good box for it.

You can go with higher speed and more load.
Ultra simple wireing. Remember, heXpod max speed is 16rpm

The only pain with Teknic is that you can't do the absolute positioning. If it would be possible that would be a perfect engine in my opinion.

Maybe you can find on the market another brand with similar specs which allows the absolute positioning.

For the interface possibilities, you have to talk to Thanos.

Cheers

@hexpod - Thank you so much! That’s perfect information!

@hexpod Thanks man for all the useful information.
I actually have a couple more questions if you don't mind, since I am at the early stage of designing and building my own rig largely inspired by yours to be fair.
Unless these info are somewhere else and I miss them, otherwise, can you please let me know the exact reference you used for:
-- The motors (Teknic 3432S ? but this reference only seems uncomplet) and input voltage ?
-- Gearbox (Newstart IB90 ?)
-- Power supply (Teknic IPC5 ?)
-- Positioning sensor ?

I already selected the basic design but now want to go for the motor selection and it is quite an important choice since everything that follow will depend on that choice. I build a scale model based on your hexpod design to run some test with Simtools and I should confess I really like it.


One last quick question for now, I realise your rods are bend in the latest videos I have seen from your rig, can you tell me what is the reason for that ?

Thank you so much in advance mate

@Zed
Where are you with your build mate ? Any build thread where I can follow your progress ?

-- The motors (Teknic 3432S ? but this reference only seems uncomplet) and input voltage ?
I am using MCPV series but as most propably you will not use the internal encoders, maybe you can go with cheaper MCVC series. CPM-MCVC-3432S-RLN

-- Gearbox (Newstart IB90 ?)
With the motor from above I use the IB90 with ratio 50 which gives you 16rpm. you could maybe risk a ratio 40 depending on the weight and lever length you want to deal with.

-- Power supply (Teknic IPC5 ?)
No. I use two "Teknic SSt-EMF75". It can be that only one could do the job for all the motors.

-- Positioning sensor ?
AEAT-6012 digital SSI


realise your rods are bend in the latest videos
Yes I bend them for two reasons:

1.) the design with motors inside and a platform ratio 0.82 that I use, gives you to much collisions while the arms are operating in the lower range.
2.)bending the legs corrects the joints starting point bank angle. see the picture below:

@FoxONe42 - no build yet but am more and more convinced this is essentially the direction I’ll be going for a 6 dof. The outriggers are what do it for me by allowing compact stowage but that compact base limts other actuation methods. I also like the planetary gear outputs better than worm. Worm puts all the load in a fairly small gear surface for more stress and wear I believe. Planetary, at least for the gearboxes Hex used, looks like lots of gear surface area. Just no unpowered static hold but the droop to park is even better. @hexpod has a really innovative and well-done design. It’s brilliant.

It’s going to be a bit longer before I start on one of these. Still sorting out how I will approach the design. I don’t have a shop or the kind of tools it takes for metal work so it’s either get them or sort using strut like my 2 dof. I am leaning to strut. More expensive but I just order the parts cut to size, ready for assembly.

@Zed
I hear you and I am in the same situation, I don't have the proper tools or even knowledge to cut or weld the frame altogether so this is a bit concerning for me. Besides, the costs of all 6 actuactors + gearboxes is a bit salty I should confess. But hey, it is gonna be worth it I hope I am also at the stage of designing the rig using SOlidworks, very much inspired by hexpod design which is indeed brillant. I think I am pretty much done with the scale model I built for test purposes, everything work as it should (expect DCS plugin but I have to figure out why). Let me know when you start your rig, we could share some tips

@hexpod
Thanks againg mate for all the info you share, I am leaning more and more towards starting the full size project thanks to you.
A few more questions however:
--Regarding the rods, I understand why you did that, it makes sense. Now, I came up with the same concern when designing the RC scale model, and I made sure both joins were aligned so that at the start position, there is indeed no angle. See picture below. Any reason you didn't go that way initially ? (I mean, I just want to make sure there is nothing wrong with that if I go full scale)

--Regarding the motor sizing, do you have somewhere the calculation you make ? Or this was done with an application engineer from Technik and newstart ? I am asking cause my knoweldge regarding this is very limited so I an trying to gather as much info as possible before contacting those guys.
--Can you tell me what are the rods made of ? (I mean the orange part)
--Did you use high missalignement spacers for your rod ends ?
--Just to sumerize briefly, and to be able to budget more accurately the project, what I will need is (main actuators component only): 6 servo motors, 6 planetary gearboxes, 6 position sensors, 1 or 2 power supply, 1 Thanos AMC 1280 USB. And that;s it, no need more Arduino, Sabertooth, .. if I understand correctly ?
--That's it... for now
Thanks in advance

--Regarding the rods, (…) Any reason you didn't go that way initially ?

yes, platform / base ratio. basically speaking: bigger the ratio, bigger the rotational amplitude. (you can easily simulate it with heXpod software). The rotational ability is pretty narrow in your design.

--Regarding the motor sizing, do you have somewhere the calculation you make ? Or this was done with an application engineer from Technik and newstart ?

definietly not Teknic. Newstart guy should give you all the info you ask for.
Your biggest enemy is the motor RMS (continus power). I would advise to keep your torque below 50 percent in static calculations if you want to go with different reductions.

ex: (RMS 2.7 Nm) * (reduction 40) * (lever length 16cm) = 675 Nm = (68kg on each leg * 6) / 50 percent = 204 kg.
Now, you will accelerate this 204kg lets say up to 0.5g. which means that your reducer should be able to take 4.05 Nm of torque.
As you can see in the doc., the AB90 is rated for 120Nm nominal 180Nm max and 360Nm emergency stop so you should be fine:
http://www.newstartmotion.com/d/file/PRODUCTS/IBIBR Series/2017-07-04/1c039bae8c46a023fc2f0f82b7ec897e.pdf

Usually, they are able to produce the ratios you dont see in the documentation.

If you want deal with more load or more accel or longer lever or more speed, than choose bigger motor and sronger gearbox. Or reduce your lever or your accel or load

--Can you tell me what are the rods made of ? (I mean the orange part)

normal plain alu.

--Did you use high missalignement spacers for your rod ends ?

no. the metal is pressed into the joints with 1-2 cent of millimeter tolerance. I advise the high missalignement spacers. (easier to replace)

IMPORTANT:
you may want to limit your teknic motor max torque to fit the max gearbox torque otherwise you will most probably grind your boxes quiet quickly.

Good luck

@hexpod
Those are invaluable information really

--The rotational ability is pretty narrow in your design.
Yeh I do agree, my previous scale model platform had a different platform to base ratio, and a better rotational ability. However, it was very easy to reach the limit angle of ball joins since the original position already started with a non nul angle. Anyway I guess it is a matter of compromise, as always. You say it can be simulated with heXpod software. Can this software be used as q stand alone, or it necessarly has to be interfaced with AMC 1280 USB control device ? Righ now I am using an arduino Uno to control my 6 RC servos and I am wondering wheter or not I could be using heXpod in that case.

Can this software be used as q stand alone, or it necessarly has to be interfaced with AMC 1280 USB control device ?

The only compatible hardware interfaces are all Thanos AMC series but the heXpod software works and connects to X-Sim or SimTools also in "software mode" without any hardware pluged in.

Actually you can also work on your profiles for games in the "software mode". Not everything is modelised but heXpod 3d simulator gives you for example a good idea of the workspace according to the geometry you may want to choose.

"it was very easy to reach the limit angle of ball joins since the original position already started with a non nul angle."

Heim joints are a very good crossover solution but I believe one could get a better results with a combination of rotational universal joints commonly used in platforms equipped with linear actuators.

I would open a dedicated thread to cover this topic.

I get it thanks.
The more I dig into your rig, the more questions I have
Now, regarding the position sensor, it looks like it is your own 3D printed design, including pulleys, am I correct ? It looks like the position sensor shaft is directly connected to the pulley, isn't there any risk to damage it due to the belt tension ?
Can you share with me the gearbox support thickness and the lever arm thickness as well ? last for tonight, what is the size of your heim joint ? M14 ?
As usual, thanks in advance mate

isn't there any risk to damage it due to the belt tension ?

No, I've printed a socket which incorporates two bearings in order to deal with relatively strong radial loads due to the belt tention.





I use the ratio 1:1 but in order to get even more resolution you could use a smaller one like 1:0.7.

Theoretically even Something like 1:0.2 could be used as @Thanos can convert those sensors to multiturn by reading them multiple times. No idea if it worth the hassle to go this way. Talk with him if you are interested.



I theory again, one could imagine even a homig function being implemented.

Steel levers with thickness 15mm and joints 20mm

The holding plates has to be really steady, (you will deal with fast load changes 0-100kg) talk with the gearbox manufacturer, maybe they can make them for you.

Teknic has updated the firmware on their MCPV series and apparently allows now the "follow position" commands.

That seems to be a highly promising feature which could be used with a barebone AMC12800 controller.

A simple one cable from the AMC into the MCPV servo could make out of this solution the simplest interface ever.