Gas station without pumps

2017 October 24

Shaving my head (and maybe my beard)

Filed under: Uncategorized — gasstationwithoutpumps @ 14:06
Tags: , , ,

In Shaving my head, I announced that I would be having my head shaved tomorrow, but not my beard.

On Wednesday 25 October 2017, I’m going to get my head shaved (just the top, not my beard) as part of St. Baldrick’s fundraiser for pediatric cancer.  I missed the head shaving last year, as I was at the iGEM Jamboree in Boston, or I would have done it then.

The head-shaving event is being organized locally by the Treehouse Childhood Cancer Initiative at UCSC.  Cancer genomics has become a major research topic at UCSC, and pediatric cancers are a particular focus, so this fund raiser has particular significance for UCSC.  Last year they raised $29,488 with 28 shavees.  I don’t expect to raise anywhere close to $1000, personally, because I’m not willing to badger my friends and family to make donations, but I hope that a few people will be willing to contribute token amounts.

If you feel like giving to a good cause (or just want to see a picture of me with no hair on top, to be posted after the event), you can donate at https://www.stbaldricks.org/participants/Karplus2017.

Last night, I got permission from my wife to have my beard shaved, but only if $1000 is raised by 1pm Pacific time, Wed 2017 Sept 25. (I’m scheduled to be shaved at 1:15).  Because I’d only raised $60 so far, I figured my beard was safe, but some of my colleagues have decided they want to see my chin, so an extra $400 was raised this morning.  There is a chance now that I’ll be beardless after tomorrow (something I haven’t done for about 25 years—the last time I did it, for a Halloween costume, my wife threatened to divorce me if I shaved it again, which is why I needed her permission for the $1000 offer).

More recently, I’ve colored my beard for St. Patrick’s Day, but not removed it.

The link for donating again:

https://www.stbaldricks.org/participants/Karplus2017

Update: we hit the $1000.  Someone asked how much for the eyebrows—if we hit $2000, I’ll let the eyebrows go also.

2017 October 22

A beacon detector for mechatronics

Filed under: Robotics — gasstationwithoutpumps @ 22:04
Tags: ,

I’ve not been blogging much lately, because so much of my time has been taken up with taking the Mechatronics course at UCSC.  I’d love to blog about what I’m doing in the course, but they use the same assignments every year, and I don’t want my solutions to replace student ones.

I will post a picture of the beacon-detector board that I soldered up for Lab 2, as reverse-engineering the design from the photo would be harder than doing the design from scratch:

This is a two-chip solution, using eight op amps. It is definitely overkill for the problem of detecting an IR LED flashing at 2kHz, while rejecting ones flashing at 1.5kHz or 2.5kHz. The Molex connector for power is a bit of a pain, as its pitch is wider than 0.1″. I ended up having to drill extra holes in the PermaProtoboard to accommodate it.

I did this design with very conventional approaches (though not the ones that the course was encouraging) and got a filter whose measured characteristics closely matched the theoretical ones over a wide frequency range:

The active filter does a good job of responding to 2 kHz while rejecting other frequencies. I initially had made the Q higher (that is, made the peak narrower), but the filter started oscillating.
The increasing response at frequencies above 10kHz is expected from an active filter—but the problem is not significant here.

I also characterized the filter including the infrared emission and reception:

The infrared signal here was strong enough to cause clipping around the central peak, which reduced somewhat the discrimination between the desired 2kHz and the undesired 1.5kHz and 2.5kHz. This plot is from an earlier version of the design—the one that had some oscillation.

After building this conventional solution, which takes 8 op amps, I also built (just on a breadboard) a 3-op-amp solution, which I have not yet tested in the lab, only at home with different beacon sources. I don’t plan to solder it up, though, as I plan to use the digital beacon detector that I built over the summer, which has been tested and seems more reliable than the analog ones (as well as giving 8 channels of detection, rather than just one).

2017 October 14

Shaving my head

Filed under: Uncategorized — gasstationwithoutpumps @ 12:51
Tags: , ,

On Wednesday 25 October 2017, I’m going to get my head shaved (just the top, not my beard) as part of St. Baldrick’s fundraiser for pediatric cancer.  I missed the head shaving last year, as I was at the iGEM Jamboree in Boston, or I would have done it then.

The head-shaving event is being organized locally by the Treehouse Childhood Cancer Initiative at UCSC.  Cancer genomics has become a major research topic at UCSC, and pediatric cancers are a particular focus, so this fund raiser has particular significance for UCSC.  Last year they raised $29,488 with 28 shavees.  I don’t expect to raise anywhere close to $1000, personally, because I’m not willing to badger my friends and family to make donations, but I hope that a few people will be willing to contribute token amounts.

If you feel like giving to a good cause (or just want to see a picture of me with no hair on top, to be posted after the event), you can donate at https://www.stbaldricks.org/participants/Karplus2017.

If you want to know why you should give to this charity, rather than the 100s of others asking you for money, St. Baldrick’s has a good list of reasons at https://www.stbaldricks.org/blog/post/your-donation-to-a-shavee-matters-and-heres-why.  The reasons I found most compelling is that St. Baldrick’s is the biggest funder of pediatric cancer research other than the US government, and they are a pure research play: funding nothing but pediatric cancer research. This focus makes St. Baldrick’s a long-term social investment, rather than addressing immediate emergencies.

Charity Navigator rates St. Baldrick’s moderately—their transparency is excellent and their administrative expenses low, but their fundraising expenses are moderately high and they have already committed to spending most of what they have collected, so they have low reserves.  St. Baldrick’s explains that the financial thing is an accounting difference—their commitment to funding grants appears as a liability, even though there is no external debt involved and they bank the money before they make the commitment: https://www.stbaldricks.org/faq/#sbfinfo-answer10

The link for donating again:

https://www.stbaldricks.org/participants/Karplus2017

2017 October 10

Why the blog has been quiet lately

Filed under: Robotics — gasstationwithoutpumps @ 21:49
Tags:

I’ve not been writing much for the blog lately, because my time has been consumed by two projects:

  • rewriting my book—I have to get at least all the parts needed for BME 51A done by the end of December.
  • Gabriel Elkaim’s Mechatronics class, which I’m sitting in on this quarter.

The Mechatronics course has the well-deserved reputation of being the most time-consuming course on campus.  I’m trying to do all the labs for the course without a partner, as it would not be fair for me to join a student team (both because I already know a lot of the electronics and debugging parts of the course, and because I may not have time at the end of the quarter to spend all my time in the lab as the students do).

I got the first lab (Lab 0) done without too much difficulty—most of the problems I had were with learning MPLABX and the Events and Services Framework that they use for programming the “roach” robots.  The students may have hand an advantage over me, as they had all (or almost all) used MPLABX in prior courses—I’ve never programmed a Pic32 processor before.  One irritant for me is that they were using a different bootloader, so software had to be downloaded with ds30, for which they only had a Windows version available.  That meant that I had to use the lab computers at least for the downloads.

I ended up with very different solutions to the “roach” programming than most students did, because I started with the idea of simulating bacterial swim-tumble chemotaxis to do the required negative phototaxis. My design was inherently probabilistic—a Markov model rather than deterministic state machine.  The flat state machine worked fairly well, but was needlessly complicated with four states, despite which it could sometimes get confused about whether a bumper was still pressed and continue pushing into a wall for a while. For my hierarchical state machine, I managed to separate the task into 3 different small state machines: one for left-motor reflexes, one for the right-motor reflexes, and one for the swim-tumble phototaxis.

One problem I had with Lab 0 (other than a general distaste for the MPLABX IDE) was that the roaches lacked any mechanical feedback other than the bumpers, but different roaches ran at very different speeds.  I had tuned the parameters on the first roach I worked with, which turned out to be one of the fastest ones in the lab.  When I went to get checked off, I ended up with one of the slowest roaches, and the mean tumble time was too low—the roach did not turn enough to make a big enough change in direction.  I had to change one time constant and retry checking off later—again with a very slow roach, but this time the random search did eventually manage to crawl into the dark area.

I’m considering trying to port their Events and Services Framework to Teensy boards, so that I can do almost all my work at home—I’m not that happy with the Uno32 boards.

Lab 1 is more straight-forward than Lab 0: analog design for a track-wire detector (inductive pickup of 25kHz signals) and a beacon detector (which I looked at this summer) to pick up 2kHz IR signals.  In class today, Gabriel challenged the students with the claim that a one-chip solution was possible.  This afternoon I did a 3-op-amp version of the track-wire detector (and I could reduce that to one op-amp with some small changes to the requirements).  I think I can use a similar approach for the IR beacon, but that will probably still take 4 op amps, as the initial signals are much smaller.  I might try to get that done tomorrow, if I get enough done on my book—I’m trying to alternate days of working on the book and playing with the mechatronics.

I’m not going to share my solutions on the blog, much as I’d like to, because the Mechatronics course uses the same design exercises each year, and I don’t want students copying what I did without understanding.  (There is enough of a problem with that in my own course—I don’t want to mess things up for someone else’s course.)

2017 October 8

Mail delivery problems

Filed under: Uncategorized — gasstationwithoutpumps @ 23:19
Tags: ,

For decades we had excellent service from the USPS, with prompt deliveries to our door 6 days a week.  USPS has the lowest prices for small items (I don’t know any commercial service that will deliver anything for under 50¢, as USPS still does for first-class mail), and the service used to be great.

In the past two years (since our long-time mailman retired), delivery has gotten a bit unreliable.  I hear from neighbors that the problem is occurring all over Santa Cruz—the USPS is understaffed locally, and they can’t cover all the routes, so they drop routes on a rotating basis, providing approximately 5 day a week service.

This is mildly annoying, but what irks me is not the reduced service so much as the way that they report it on their tracking service “Informed Delivery”.  Instead of something honest like “understaffed—route skipped”, they report the non-delivery as “receptacle blocked”, blatantly lying about the problem and blaming the customer for their failure.

Even in the past, when delivery was routinely good, I noticed a tendency of the Santa Cruz Post Office to push the boundaries of honesty.  For example, when you paid for 2-day delivery, they were very careful not to scan the packages when they were handed over the counter at the post office, but waited until the post office closed at the end of the day to scan the package, giving themselves an extra 24 hours to make their delivery promises.

I imagine that the USPS pay scales are set based on cost of living in some cheap part of the country and are not really living wage in Santa Cruz.  They have job ads out for “City Carrier Assistant” at $16.41 an hour, which just meets the City of Santa Cruz definition of a living wage (they require $16.21 an hour with benefits for anyone hired by a city contractor, or $17.68/hour without benefits).   These aren’t even permanent jobs: “This is a non-career position through which employees are hired to serve a 360-day term, with the possibility of reappointment to additional 360-day terms.”

The pay is a little better than entry-level jobs in retail (which everyone is having a hard time filling, because the pay scales are so low), but less than shift supervisors or assistant managers.  Anyone with skills (apprentice electrician, plumbing installer, automotive technician, …) can expect higher entry-level pay.

Incidentally, I don’t see $16.41/hour on the postal-workers’ union pay-scale chart—the closest I can find is Pay grade 4, Pay step GG at $16.4688/hour.  Does the less-than-a-year contract circumvent union protections also?  I’m beginning to see why there is a shortage of postal workers locally.

« Previous Page

%d bloggers like this: