Thursday, 28 January 2010

Air in the rams and sloppy bearings.

We have had trouble with air somehow getting into our rams, even after bleeding the system several times. It seems as if air is somehow getting sucked into the rod seal on our low friction rams. We have had the sales rep and ram tech rep from parker take a look at the system, and they have suggested that we change the seals to standard type.

There is alos a fair amount of play in the spherical bearing on the rams rod eyes. This is not helping with the control system, and the combined play and air in the system is making tuning very difficult. We are hoping we can swap the bearing for higher tolerance types to reduce the play.

Once we have eliminated these issues, we will carry out further tests, but I fear it part of the problem is having a rotary encoder attached to the mechanism, instead of using transducer built into the rams, unfortunately we couldn't afford the latter. The only other thing we are going to try is to pre-load the ram with a spring acting against the push stroke, this should eliminate any play in the mechanism and bearings.

Wednesday, 27 January 2010

MDF & old tyre foot..

We have attached a basic test foot to the bottom of the ball joint, primarily to protect the ball joint from an undesired plunge into the concrete, as well as testing the operation of the tyre as a shock absorber. It may not look pretty, but it does the job and was cheap! The tyre came from a local John Deer tyre skip :)

Monday, 25 January 2010

New Encoder Mount

The new encoder mount have been implemented on the Coxa and Femur axis, the new design is much better! Essentially this design places the bellows on the opposite side to the encoder, which allows the encoder to sit 23mm closer to the mounting plate. It all looks allot neater and better protected.

The Tibia has yet to have the modification, you can see the bellows in the picture below, these have been moved to the opposite side.

Thursday, 21 January 2010

Hose Rollers

The leg is going back together, with hose rollers installed and a larger gap in the hose pass through plate in the back of the Coxa. Clearances look much more comfortable now. Also have been taking a look at the use of Flexor 7 cables, they are alot lighter, a couple of mm smaller diameter, and a smaller bend radius of 75mm, however, they will be more expensive!

Josh has also replaced the square steel section coxa support, this was always supposed to be round section, but we didn't have an at the time. The round version is less abrasive and gives more room for the hose bundle.

Tuesday, 19 January 2010

HexEngine Linux Real Time Patch

I decided I had to take a look at patching the X-Linux system with a real time patch which should be more suitable for the control system. The system is currently running very well, very little jitter, but it would be nice to remove possible kernel interruptions! I have been putting this off, as I have no idea where to start, however, I followed the how-to on this site: and after half a day I was able to patch and re-compile the kernel and install it on the HexEngine target :) well pleased! This also means that I am also able to add and remove features to the X-Linux system such as wireless internet drivers, FTDI serial drivers etc and remove unwanted stuff such as audio driver that may be slowing boot time. This has ticked off a major issue for me that has been lurking in the back of my mind, I'm sure this is childs play to a linux guru, but to me it is good progress.

One problem with the patch so far is that the Vortex86DX does not seem to support high resolution timers, so my jiffy time is still 4ms (250Hz). However the real time patch has seemed to improve jitter.

Friday, 15 January 2010

Initial tests complete

Having completed some initial test of the FACU PID control box, its time to strip the coxa down and make some changes to the design. This is to facilitate better hose movement through the coxa section, including three rollers to guide the hoses. Also we are going to change the encoder mounting design, which will bring the encoder at least 20mm closer to the body, keeping it out of harms way.

I am experiencing some issues with the position control of the rams. First off, having bled the system it seems that the Femur and Tibia are gathering air in the rams, even after running the system for some time the air does not disappear, its as if the air is being sucked in, yet we have no oil leaks? The next issue is that the rams which are not under load, such as the coxa while it swings freely on the coxa pivot, are very difficult to tune, I think this is due to the momentum of the leg and slight play in the mechanism causing an error on the encoder, which in turn is translated back through the system as a correction on the cylinder, this then causes oscillation. In order to tune this oscillation out, the drive has to be reduced to a value that is undesirable for normal use. We have a plan to pre laod the leg with a spring or gas strut, this may help.

Josh makes the first adjustment to the coxa.. with an alligator saw!!

Tuesday, 12 January 2010

Complete 4 axis leg

Just started testing the complete 4 axis leg, great to see it together, apart from the foot. I will send the next couple of days testing the PID control algorithms and possibly hook it up to the HexEngine for a quick test.

Monday, 11 January 2010

Hydraulic Commissioning

Commissioning our hydraulic rams became a little easier today following a visit to the pound shop, we picked up this clear container, which we attached two ball valves to, through holes in the bottom. Now to commission cylinders and pipes, we attach each end of the cylinder to one of the ball valves, adn drive the cylinder back and forth until all the air is removed. Obviously some air gets back in when re-connecting to the manifold, but its a pretty good place to start.

Snow Joke!

Our first week back ended up being two days! Then the snow came.. which is great fun, but has set us back a bit. Would have been great to test the finished machine in such conditions, lets hope it snows like this next year :)