Oops, my mistake -- I just found them via a Google search:
I don't know why I couldn't find them before.
So this does offer a potential solution -- if your frame had all the
joint hinges built into it, with a gear on each joint shaft, then you
could use these to drive the joint from a servo, without the servo
experiencing any lateral force at all.
I have what they call a "threaded stud," which sounds like the makings
for a good joke, but it's actually a useful hardware item. It's molded
plastic with a flat flange on one end, and a threaded post off the
other. It uses a very high contact adhesive (3M VBH) on the flange, and
once placed on the bottom of the servo and allowed to set, is extremely
difficult to remove. You then use a nylon or steel nut to secure the
threads through whatever support you're mounting to on the other end.
The biggest problem is that the dimensions of the flange are larger than
the base area of the typical servo, so you have to cut it down a little.
Not hard, but it makes for an amateur-looking setup.
Let me know if this sounds like something you'd like to try, and as a
Budget Robotics customer I'll send you out a sample.
Joe Strout wrote:
I have also milled new bottom plates out of Delrin (a very durable
plastic), adding the missing stud an allowing for the idler. It is alos
easy to add a channel and cutout for the wires, so they come out at the
This works great, but it is a lot of work and an new (or modified) CAD
file needs to be created for every servo type. For my particular
machine, I also need to do four manual tool changes, so at 20 bottom
plates, I am busty for two hours and I mill away 80% of the Delrin which
then becomes toxic waste.
The obvious way would be to mill a negative instead and have these
plates injection molded, but again, that would be different for every
servos type. Something I would highly recomend to a servo manufacturer.
Finally, there is a project on the web, OpenServo, that gives
instructions, pcb layout and the firmaware to build drop-in servo
electronics that are controlled via i2c (a high speed two wire serial
bus, making the controller obsolete) and knows everything and anything
about the servo (power consumption, own ID, position to 12 bit
resolution, temperature inside the servo, etc., etc.).
To sum this up, I would generally be willing to organize injection
molding and pcb production and soldering, but we would need to agree on
a single servo type, and we would need to be able to organize the sale
of 1000 or more servos.
Looks like you're gonna have a lot of binding and/or foot skidding in
the feet, as the knee servos rotate to lift and lower the frame, and
also as the hip servos rotate the flegs front-back. You should be able
to see this if you put the bot up on blocks and exercise all the
If you rotate the knee servos by 90-degrees, the linear distance
between the front feet will change by about 8-cm. And as you rotate the
hip servos +/- 45-degrees, the distance from the feet to the side of
the frame will change by about 9-cm.
Yes, that is correct. Thanks for even estimating the amounts. The goal
was to make it as simple as possible and use only eight servos. Skidding
is allowed by using pegs under the feet. If the terrain is flat, I don't
have to lift the feet too much and I can do short steps to avoid most of
The next iteration will be a 3DOF per leg quadruped from wood, basically
the continuation of this guy from 1999 (before CNC).
That's very cool; I hadn't seen that before. I really like your
approach of cutting 2D parts and assembling them into 3D structures. It
lends itself very well to laser cutting, available on the net from
services like Pololu.
I also like the way you released the drawings and assembly instructions.
These days, rather than JPEGs for pasting to wood, I might suggest
vector art of the sort that can be sent directly to someplace like
Pololu. Then, anybody could use your drawings, as-is or modified, to
replicate what you've done -- and hopefully feed back some improvements
to the community. A true open-source model, applied to hardware gizmos.
This is something I've been struggling with myself: should I spend my
limited budget on Lynxmotion "servo erector set" parts, or instead learn
to cut and assemble my own parts, as you've done? Ultimately I'll
probably do some of both. Lynxmotion's prices are really very good; I
suspect that one of these brackets, if cut commercially from (say)
lexan, would end up costing more than the Lynxmotion equivalent. On the
other hand, you can make custom parts shaped almost any way you need,
whereas with the SES you may end up cobbling together a shape that is
Thanks. Yes, it was an inner fight, but in the end, I decided to put
image files up. That's how I built tha quad myself: I glued printouts on
with Elmer glue, cut it with a small saw, and later peeled the paper off
again. I had no CAD experience back then either.
I am involved in a few OpenSource software projects, and feedback has
been very mixed. I would not expect too many valuable additions from the
outside (never mind if the *do* come, of course). I have seen many
commercial copy and past jobs of my stuff though... .
I will probably do that with Squish as soon as I find a way to take
contributions, so I can easily add sensor rigs, camera towers, etc. to
the web page.
The Lynxmotion stuff seems pretty cool. It's probaly the rigght to not
limit yourself, but buy whatever was already invented and fits your
budget, and add those parts that are too expensive or too custom. I
myself would say that I am some silly purist, having to build everything
myself, but that's just me ;-)
Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here.
All logos and trade names are the property of their respective owners.