Jack Crossfire's blog View Details
Archive for May, 2018
Posted by Jack Crossfire | May 30, 2018 @ 11:58 PM | 5,664 Views
Drilling the holes in a space shuttle injector plate, in 1977. It was 1 of the 1st high quality photos a lion ever saw on a computer & the moment it became clear how laborious spaceships were to create. Decades later, lions realized after all that work drilling the holes, they had to plug the holes back up when they cracked.
Posted by Jack Crossfire | May 29, 2018 @ 10:00 PM | 5,542 Views
Finally cracked open the ice budget & attempted the 1st refrigerated transportation. This requires going both ways with ice blocks, significantly reducing range. 2 ice blocks were quite heavy, but it didn't show a significant increase in tire wear after 10.7 miles. Power consumption increased to 328mAh/mile. The weather started hot & got cold by the end. Sandwiched the salad between the ice blocks. The top ice block finished more melted than the bottom. The salad arrived still cold, which isn't very conclusive unless another test is done without ice, in the same weather. Ice cream definitely needs to be tested.

Whether ice on top or ice on the bottom is more effective is unclear. Most heat should get in from under the robot, but cold should travel down.
Posted by Jack Crossfire | May 26, 2018 @ 01:29 AM | 5,939 Views
Fixing the camber spread the tire wear over a larger area in the center but introduced bigger problems. It significantly reduced the range because of the higher rolling resistance & made the servo skip cogs because of the increased rolling resistance. Went back to the stock camber which wore down the inner edges & searched for a coating to resist the wear.

The solution was to glue fabric straps to the inner edges with E6000 adhesive. After hundreds of miles, they didn't wear off & the power consumption dropped to 280mAh/mile. Initially, the straps were common shoelaces, but kevlar would be the ultimate strap. The key is getting the strap to cover the entire wear pattern. If wear hits a strap edge, it'll peel off.

Earlier coatings were white electrical tape & CA glue, which instantly wore off. E6000 adhesive without fabric seemed to wear off.
Posted by Jack Crossfire | May 12, 2018 @ 05:44 PM | 5,355 Views
NASA Mars Helicopter Technology Demonstration (1 min 23 sec)

The speech:


More details on kiwipedia:


After decades of rumors about such a thing, which no-one took seriously, it was finally included on the next Mars rover. The finer details had a few shockers.  It uses solar charged lithium batteries, coaxial blades, has no blade shroud, & no obvious way to right itself.  It weighs 4 pounds & looks quite top heavy.  It's basically the exact opposite of a copter you'd expect to be sent to Mars.

The equivalent atmospheric density on Mars is 100,000ft on Earth or 3x higher than the highest a copter has ever flown on Earth.  A quad copter would not be efficient enough to do the job.  It's inevitably going to crash & roll over, so apparently this simple use of long legs is able to recover without human intervention.  Maybe the blades can kick it upright long enough to get going or it can tumble down a hillside until it reaches a flat spot.  There are still many ways it can get stuck, so it needs a major investment in autonomous programming which can avoid getting stuck.

The dark, blurry footage from JPL shows the solar panel on top of the blades blocking airflow, no obvious use of a flybar, & lots of vicon balls for tracking position.  You'd think a solar panel below the blades would allow more lift & get dusted off by the airflow.

A copter that could travel great distances on solar power & recover from crashes is what we all wanted, 40 years ago.  The solution seems so simple, it makes you wonder why China didn't already give hobbyists such a thing.  
Posted by Jack Crossfire | May 10, 2018 @ 10:28 PM | 5,480 Views
Mane engines stepped up to 190,000lbs thrust with higher ISP.
Merlin vac increased to 220,000lbs.
The merlin vac is still throttled down to 210,000lbs to study vibration.
The octaweb is stronger aluminum 7000 instead of aluminum 2000.
The rocket base is now titanium with water cooling.
The black sections are a proprietary material, probably derived from shuttle tiles, but without the porous sections which absorbed water.
The avionics & IMU were redone.
Current 2nd stage recovery efforts are only to determine the mass impact of recovering it without impacting BFR development.
Falcon 9 launches would cost under $6 million if they could reuse the 2nd stage.
He wants to launch the same booster twice in 24 hours, in a demo mission next year.
The new COPVs are the most advanced & tested ever made.
Posted by Jack Crossfire | May 10, 2018 @ 02:36 AM | 6,164 Views
Got a somewhat acceptable wing by creating 4 profile sketches, then lofting them. Freecad can't create a solid from a surface, so you have to create 2D slices for the lofting operation, with matching curve & line segments. Freecad did manage to interpolate the sketches into a smooth curve. Suspect this process would be much easier to tweek & give smoother results if it was built up from equations rather than paw sketching, but the reference design is going to change. The automated version would have to use a constant number of line segments for each profile sketch instead of curves.

The only way to create windows was writing the basic equations in a python script. There are no basic cutting, pasting, copying functions in the sketcher. The lion kingdom wanted see through windows instead of textures.

The script created a window creation tool. Then the tool cut out windows you can see through. OpenCAD has a few rounding errors, causing some windows not to open. Offsetting by 1mm in random places seemed to work. There's the issue of a tinted, reflective texture. The whole model may end up being from equations. Tweeking is a buster because it takes a long time to recompute the model.

The parts -> thickness tool doesn't work at all with booleans or constantly locks up. The only way to make hollow solids is sketching a cross section & revolving/lofting with solid mode enabled.
Posted by Jack Crossfire | May 09, 2018 @ 02:01 PM | 6,240 Views
Long thought to be unfixable, the camber is determined by a bend in the wishbones. It turns out the wishbones can be flipped over, giving a closer to ideal camber position. Further refinement is within the limits of changing the suspension height. The result was much closer to ideal, but increased the rolling resistance, put more stress on the steering servo, & decreased the range. The steering horn jumped cogs much more often.

There are servo savers & a return to sloppy steering or going back to stock camber & tire wear. Steering with the direct servo link was so solid in this spherecam video, it didn't need any stabilization.

3DRRrrrrrun (3 min 18 sec)