Useful Tidbits

From OrbSWARM
Revision as of 20:29, 2 August 2011 by Lee (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Put useful emails and (as yet) unfiled tidbits on this page


2008 Afterburn transportation report from Michael

Just for those of you who may be interested in logistics in the future-

The final charge from Penske for the truck was $912.30. This is a couple hundred less than my initial estimate because I did not realize they have a lower weekly rental rate- i.e., they charged for one week and 4 days. I estimated 700miles, and we came in at about 750miles. We sprayed out the interiors with compressed air, swept and mopped the floor of the box, and then went to the power washer/taqueria place off of bayshore to spray it down. Four of us spent about 90 minutes spraying it down, and then I spent about 1/2 hour back at the shop detailing the cab with shop towels. We returned it Monday night via the key drop at Penske Oakland on the Oakland/San Leandro border.

The truck had both a lift and a ramp. We used the lift for un/loading crates and mothership on the playa, and a forklift for crates at the shop. It was easier to use the truck lift for the mothership than the forklift, and we required no DPW support.

I spent about $300 on diesel fuel, which I will give to the project as my rental fee for living in the truck for 9 days. It was a fine home, and left room for a good workshop in the back.

Penske didn't sweat the cleanliness, nor did they scrutinize preexisting damage which went unrecorded when they rented it to me. When I reserved it, they asked what I was doing, and I told them "going to a remote site in Nevada", and they left it at that. There was a huge difference between different Penske offices I talked to; east bay was better than SF/Peninsula.

The overall experience was great, and significantly cheaper than I predicted going in. As, according to legend, Native Americans use every part of the Buffalo, so did SWARM use every feature of the truck. I would say, if you need less than a 40 foot truck for the playa, go and rent the thing, you will be happy in the end.

-mike

How to Keep Parts Rust Free

The Box Shop is just a hundred yards from the San Francisco Bay. Salt air is a problem on the equipment. Here are some hints on keeping steel tools from rusting

  • Coat parts with Vaseline petroleum jelly. Coat it as thick as a penny.
  • I keep all my machine tool surfaces coated with Mobil DTE lite machine oil, it wont evaporate off, it only wipes off with use, so re-oil after each use. If it is a higher load-bearing way surface use Mobil Vectra #2 or #4 way oil. You can buy these in a gallon at McMaster. After several years in my moist SF garage, all my bare metal surfaces still look new!
  • Do not coat parts in WD40, because it will promote rusting; WD40 is corrosive over the long term.
  • Some recommend Boeshield T-9 [1], [2]
  • LPS3 was recommended

How to make motors run quieter and with less vibration

from Carl Pisaturo on dorkbotsf-blabber
Sent: Thursday, June 24, 2010 11:15 AM

I've recently confronted annoying gearmotor acoustic noise, and aside from spending big bucks on maxons, there are these options for whine abatement...

A) look for (or possibly mod yourself) a gearmotor where the first pinion gear is plastic. This is the one attached to the actual motor shaft and goes the fastest, thus is the biggest noise contributor when metal-on-metal.

B) fill the gearbox with thick synthetic grease (may be able to do it with a syringe via a front face mounting hole). This trick costs power - the motor will draw far more current for a given output speed, but the grease damps out a lot of the sound.

C) soft-mount the motor to whatever you're mounting it to. Rubber washers and grommets (check mcmaster carr) can prevent much of the gearmotor's vibration from getting into the main structure and using IT as a speaker.

D) running-in: gears smooth themselves a bit and get quieter with hours of vigorous use. Then disassemble, clean well and re-grease. I've resorted to run-in with abrasive lapping paste in the gears - worked but harder cleaning op.

E) use a much bigger gearmotor than needed, and run it slower.

f) consider not using a gearmotor at all... can a straight motor and belts achieve the required speeds?


How to Find a Short on a PCB

Mark Lakata writes on SFMicrocontrollers July 29 2011 in response to Adam Glass' question, "What process or resources do people use for finding shorts on a board?"

  1. turn up the supply voltage slowly, if you get significant current < 0.6V, then you have a dead short somewhere. If it kicks in above 0.6V, then you have bad silicon
  2. If you have a microvolt DVM, crank the current up as high as you feel comfortable, and probe around the supply nets and look for microvolt differences. the lower the voltage, the closer you are to the short
  3. crank the current up to about an amp or so and find the hottest component. remove it and look for more shorts.
  4. crank the current up as high as you can go and let the short blow itself up. (This is a technique that works with EL wire. If you have a short with EL wire somewhere, plug it into 110 VAC wall socket and blow out the short. then cut out the bad section and patch it)

Rolfvw adds

  • if you made PCBs for this project, look them over

carefully for defects. you might also poke around with a meter checking for continuity where there shouldn't be.

  • is it possible to remove the 7805? if you can do that, then

you can apply 5v directly and measure the current. see if your circuit (less the 7805) is consuming more than expected.


zsradding adds
I recently had a regulator (not a 7805) draw more current then it was supposed to. Turned out that the wrong sized caps were soldered on the input and output sides of the regulator. The wrong cap values made the regulator unstable (it was oscillating) and drawing about twice the normal amount current. Probably not what's going on here, but it doesn't hurt to check.


Jesse Lackey adds
Another approach when the easy stuff doesn't help ... cutting power traces. How easy/useful this is depends on how the +5V (or whatever) is routed. If you cut the +5V to a chip/section of pcb and your current draw is suddenly reasonable, you've got your answer: the problem is on the other side of where you cut...

I'd look at the current draw as it stands now before doing anything. If it seems actually reasonable, you can compute how much heat the 7805 needs to dissipate, and maybe it is failing or maybe everything is OK but you need a heatsink. I always have a good ammeter in series with my main normal bench power supply (which also indicates current, but only to 10mA increments) and I watch it as much as anything else (scope and circuit outputs or whatever) when developing/debugging. Unexpected current draw (or lack thereof) indicates that something is wrong, either with the circuit or one's understanding... but I digress.

For the cutting approach: a combination of divide and conquer, educated guesses (a dc/dc that is a new design is more likely to have a short/serious problem than simpler circuitry that I've used before), and choosing cuts that are easy to patch drives the decision of where to cut. Easy to patch would be if you can do a cut on a trace that has two vias, so the cut can be patched later by a jumper wire (wirewrap wire is great) between the two vias. etc.

Another angle ... with SMT, using hot air to remove parts is a breeze, even chips with a lotta pins. Start removing components. If you've got a +5V that then goes thru a voltage regulator to make +3.3V for a subsection of the design, pull the 3.3V regulator, and now that whole subsection is out of the game. For designs that are pretty complex I will sometimes put ferrite beads (0603 SMT) on the power feed to major subsections so I've got something very convenient to pull to find power problems. (plus filtering on the power feed to each section).

I rarely let the current go above 500mA if there is a problem (i.e. I expect 100mA, but have a short/near short, so with the current-limited bench supplies it only goes to 500mA b/c that's what I set the max to).

I may go to 750mA or 1A relatively briefly (15 seconds), only to heat up the problematic component so that I can find it with a (wet) finger (sizzle ouch!!! ok found the problem is how this goes.)

One caveat: unless the various subsections that are getting their power cut are truly independent, you can get strange interactions between the still-powered part and the supposedly off part. Example: the still-powered part has a microcontroller sending a digital high to an unpowered chip. Well, that chip is now powered via that input; the digital high goes thru the unpowered chip's input protection diodes and it maybe runs/maybe not. This normally doesn't cause damage but definitely flaky behavior. But it shouldn't lead you astray on the where-is-the-short quest.

Good luck!