User talk:Arjs

From OLPC
Revision as of 21:16, 11 March 2008 by Wvbailey (talk | contribs) (→‎Calls: I don't really believe the $2.00 is accurate --)
Jump to navigation Jump to search

re: mediawiki script

Thanks for the reminder! Turns out that you need not just admin accounts on the wiki but access to the webserver, so I passed it on to SJ; when I see him in the office tomorrow afternoon I'll try to catch him to sit down and actually work through it (if it hasn't already been installed by then). In the meantime, try linking to the youtube video... Mchua 01:00, 22 July 2007 (EDT)

It works now; good luck. Sj talk 21:05, 22 July 2007 (EDT)

measure jam

Emailed the olpc.tv folks about the youtube-on-mediawiki coolness, thanks for the suggest! Also passed on the Measure project ideas to my classmates who were looking for projects; hopefully some of them will ping you soon.

The Jams page has a short paragraph describing Jams that you can use to send out to your classmates and friends in new Delhi - you can hook up with any of the existing/upcoming Jams or plan your own. I'm filling in the How To Run A Jam book as the Curriculum Jam Taipei gets underway, so hopefully that might be useful to you soon... and I think the idea of running an engineering Jam around Measure is awesome, you should go for it! (actually, it might work well if the leaders of one of the IEEE student chapters organized the Jam, and you were the resident OLPC/XO/Measure expert on hand.) Keep me posted, Mchua 21:31, 23 July 2007 (EDT)

Sweet, Measure Jam! Are you thinking of doing a tech-flavored Jam (develop software/periphs for Measure) or a more education-ish Jam (here, teachers, take stuff we've made and teach things with it)? There's also the option to do both; see Journalism Jam New York for an example of a Jam that combines toolmaking with tooltesting. 65.78.19.90 21:32, 13 August 2007 (EDT)

the pictures on your userpage

...are absolutely wonderful. Thanks for sharing them! You should email Walter so he can mention them in the weekly update. :) Mchua 18:22, 15 August 2007 (EDT)

Re: Picture positioning trouble

Hi could You show me an example of what your tying to do? --Chief Mike 09:36, 17 September 2007 (EDT)

picture tutorial

how's this? Mchua 13:15, 17 September 2007 (EDT)

RE: Status of Measure in OBX template

follow-up to User talk:Xavi#Status of Measure in OBX template

Oops! yes you're right. And yes, 'bundled' seems fine. The categorization is still not totally defined (that's why they are red/missing), so any suggestions & feedback are more than welcome.
I haven't removed the old {{Status box}} as I want to make sure there are no objections, and that the OBXes are helpful. Comments? :) --Xavi 11:20, 22 September 2007 (EDT)
I love the OBXes! They concisely convey all the most important things a new user might want to know about any activity. Although, I cannot think of any now, I would let you know of any suggestions for additional fields in the OBXes. Great work on the OBXes! --Arjs 11:53, 22 September 2007 (EDT)

USB measuring sensors for the XO

We are working on the USB measuring sensors for the XO. Here are some technical details. We can make the hardware work but we need an application for the XO that will allow kids to work with the sensors. If you are interested we can work together. ZdenekBroz 16:18, 14 October 2007 (EDT)

I would be glad to collaborate. I am a little hard pressed for time these days, so might not be able to undertake a lot of code-development work (I also have to complete feature requests and bug fixes within Measure ACtivity), however I would be happy to answer any of your questions and help you use or develop Measure Activity to suit your purpose. --Arjs 16:23, 14 October 2007 (EDT)
Feel free to email me at arjun at laptop dot org

polyphase filtering

Hope this helps. I checked the algorithm on paper (the math works out) and ran through a couple of test cases where it seemed to produce the correct output, but you may want to compare it to your current code and see if you get (1) the same answers, and (2) any performance gains. But in any case, I tried to comment the code enough that it's easy to follow the algorithm - polyphase filtering is really just doing (and adding) only the convolutions you won't be downsampling out. Mchua 01:21, 28 October 2007 (EDT) 

def polyphase(signal, filt, samplerate):
        """Takes in a signal and a filter to convolve and a rate to sample at. Implements a polyphase filter that returns a downsampled version of the convolution."""

	# load convolve function
	from Numeric import convolve

	# Pad signal with zeroes so it's a multiple of samplerate
	# When the signal is split into polysignals, this should
	# pad the last N signals with zeroes at the beginning
	# (where N is -len(filt)%samplerate)
	# so that all polysignals are the same length
	signalpad = (-len(signal)%samplerate)
	signal = signal[:signalpad] + [0]*signalpad + signal[signalpad:]

	# Pad filter with zeroes so it's a multiple of samplerate
	# unlike the signals, the filters that need to be padded
	# should be padded at the end
	filt += [0]*(-len(filt)%samplerate)

        # Split original signal into polyphase signals.
	polysignals = []
	polysignals += [signal[offset::samplerate] for offset in range(samplerate)]
##	print "Polysignals", polysignals # debug

        # Split original filter into polyphase filters
	polyfilters = []
	polyfilters += [filt[offset::samplerate] for offset in range(samplerate)]
##	print "Polyfilters", polyfilters # debug

        # Filter each polyphase signal by corresponding polyphase filter
        # Add the results together
	ans = convolve(polysignals[0],polyfilters[0])
##	print "Ans:", ans # debug
	for count in range(1,samplerate):
		ans += convolve(polysignals[count],polyfilters[count])
##		print count, "Ans:", ans # debug
		
        # Done
	return ans

if __name__ == "__main__":
	from Numeric import convolve
	signal = range(2,16)
	filt = range(7)
	samplerate = 4
	print "The signal", signal
	print "Filtered with", filt
	print "And sampled at a rate of", samplerate
	print "Yields:", polyphase(signal, filt, samplerate)

Reverting a page to an earlier version

I saw your question elsewhere on the wiki; if you haven't already figured it out, here is a page describing how to revert a MediaWiki page to an earlier version. —Joe 13:35, 31 October 2007 (EDT)

image sharing

Images and files can't be shared across wikis in a simple way. You can link directly to images on the public site from the teamwiki page, though. Sj talk 09:19, 2 November 2007 (EDT)

volunteers for your projects

Back from three months of roaming the world and getting geared up on OLPC stuff again. Sounds like you've been busy! Some stuff you might want to know:

  • Re: Help needed wrangling young testers at OLPC office - forwarded to the Olin university chapter folks, might be able to get some to help when they're back from winter break (January 20-something).
  • Measure: The IMSA chapter, which you've already seen (and messaged User:ScottSwanson about, is making...
    • sensor input modules (starting with an adapted version of the differential amplifier and going forward; they're learning electronics as they do this so progress may be a little slow at first.
    • a collection of experiment demos (with videos) using Measure, among other Activities, so students and teachers have a few examples of the kinds of things they can try out.
    • More might be coming once the sensor input modules work. There was interest in continuing work on Measure-related things as a full-time research project over the summer, if we can wrangle some engineers to mentor via email. Let Scott know if you run across anyone; we'll keep our eyes open as well.
  • Power supplies: Apparently one of the top engineers at Fermilab is very interested in OLPC, IMSA, and sustainable energy. He and one of the IMSA parents are building a small sustainable energy lab on campus (solar panels, wind power, and other common variants) and wants to donate to form an OLPC lab as well. So there's a chance we may be able to get some collaboration on alternative power sources. We're getting in touch with him (will mention Cow power) and will let you know if anything happens.

See you soon in Boston, I hope - will you be around for part of this winter?

Mchua 10:17, 11 January 2008 (EST)

Temperature measurements

I'm in the process of ordering some parts, so I'll order what I need to build your temp-probe. It'll probably be a week before I get the parts...

The photos you've added are good and helpful.

I worry that people won't have access to the tools they need, or can't afford them. So here's some research:

  • Soldering iron (or a wood-burning iron): $8.99 at Radio Shack is the best one, has a nickel-plated tip at $1.99 each. It looks like small amounts of rosin-core solder is about $2.00-$4.00 per oz i.e. in small quantities. ( ~$30 per 16 oz with lead or ~$60.00 per 16 oz lead-free from Digi-key).
  • Platoshear diagonal cutter (p. 1915 in Mouser, PN 577-170): $4.91. This is the lowest-cost "professional" diagonal cutter I've found.
  • Eagle Plastic Products Wire Stripper and Cutter (p. 1914 in Mouser, PN 5876-100): $5.05
  • Eagle Plastic Products Wire Stripper and Cutter with spring (p. 1914 in Mouser, PN 5876-101S): $7.50
  • Nothing of note in Digi-key
  • Nothing affordable in Granger
  • Techni-tool Wire Stripper and Cutter (page 260, PN 758TT0010): $6.80 [seems to be the same thing as the Eagle Plastic stripper]
  • Techni-tool Wire Stripper and Cutter with spring (page 260, PN 758TT0012): $7.68 [seems the same thing as the Eagle Plastic stripper with spring]
  • Sears Craftsman has a three-piece small-plier set: 4 inch diagonal cutter, end nipper, and 4 1/2 long-nose plier for $7.99 (Three-piece Model 45698 usually sold for $9.99). They sell also a needlenose (I own one of these, a very good tool) for $9.99. They sell the good-quality three-piece set for $20.00 including a flush-cutter, the needlenose that I have, and a long-nose (Three-piece Model 45673).
  • The best deal? -- Home Depot sells a Husky 5-piece plier set Model 165250 for $9.99. This is similar to the Craftsman set plus a needlenose and a second long-nose.

Maybe you have some thoughts here. Bill Wvbailey 18:57, 14 January 2008 (EST)

A quick look at the spec indicates the LM35 has a pretty small (low) output at 10 mV/deg C, or 0.25 volt at 25 C. This is at the low (i.e. sensitive) end of the 50:1 attenuator I've been working on (I've fiddled with fixed attentuations of 20:1, 50:1, and 100:1). The 50:1 attentuator yields about 0.025 volt per major division when the "gain slider" is "full up" (most sensitive). At this maximum sensitivity, I've seen a lot of warmup drift from the laptop; it drifts for about 10-15 minutes before things settle down. I suspect that as the laptop's heatsink heats up the AD1888 heats up and its onboard reference drifts.

I definitely will have to order the part and see what it does (Radio Shack doesn't carry it). I'm a bit hesitant to download the new activity version -- I don't trust myself to do it right. Maybe there's a simple way to get the trace at center-screen at 25 C. But when I get the part I will pursue this. I've left a few notes also at Talk:Measure. Bill Wvbailey 00:01, 15 January 2008 (EST)

Hi, Bill here. I got my temp IC's but haven't had time to work with them. And now I'll be out of town for a week. But when I get back I'll wire them up. And I want to see how it works with the "differential probe", too... to see what I can learn from it. By the time I get back I hope my magnets will be here for the "gravity experiment". Have you done any more thinking about a "scope trigger?" That will be necessary, plus some kind of time-axis calibration. To start off I'll try wrapping wire around a soda straw and dropping the magnets down the straw.
Temperature probe: With regards to the photo of how to wire the audio plug, there may be confusion about a stereo plug, i.e. which of the two terminals to solder to. The photo is a bit ambiguous. A person could solder to both, I believe, without any harm done. One of the two terminals appears to be open, thus the input is essentially monaural. If they intend to strain relieve it, the assembler should be reminded to put the strain relief "boot" on before they solder, also if they intend to crimp the ground/common "tabs" of the plug, I suggest that they put some heatshrink tubing over both wires first, then crimp the connector's ground over the tubing with both wires inside it, then screw on the "boot". This appears to make a fairly rugged connector. Also, I'd suggest that the audio wires be twisted at about 3/4 twist per inch, and ditto for the power wires. Perhaps the whole probe -- LM35 but without the tape -- can be 'potted' in something (even wax will do the trick but won't be good for high temps... sealing wax might work, I dunno ), 2-part epoxy?, model glue? RTV would be best but i don't know if regular RTV that outgasses vinegar/acetic acid will hurt the probe.... Longish heatshrink tubing slipped over each wire of the IC and then shrunk (carefully not to overheat the sensor), this assembly then potted (i.e. with the heatshrink inside) would make a fairly rugged, quasi-water-proof probe. Back further certainly regular RTV will make a good flexible "boot" without risk of causing corrosion.
Attentuator/differential probe: I finally found a decent potentiometer (one from Bourns at ~$5.00 dropping to ~$2.50 in 100's, one from Honeywell at about $3.00 in singles, pin-for-pin compatible...) for the "differential probe" and I've mocked up the circuit on two on pc boards and put into the little boxes. (I will build a few more to get more hands-on experience and make little revisions). In the process I ran into an interesting problem. I built both on raw double-sided 0.062 pc board material that I got from Radio Shack. (Clearly not a 3rd-world friendly process: To do the work you need a dial caliper to 'scribe' lines on the copper, a Dremel tool, a drill press and a tiny 0.035-.040 drill(s), various other drills (e.g. .075, .093), a tiny Dremel 'burr' to carve away the copper, a stainless-steel ruler with a cork backing to guide your hand as you carve away the copper, a saw to cut out the board, a file, wire stripper, cutters, etc, etc. ) The first board I wired basically "point to point". When I assembled it in the box, I found that there are issues with 60 cycle capacitive pickup ... my hand holding the box ... so the second one I built as a pretty-good mock-up of a double-sided pc board (again with Dremel tool and all the other stuff) with ground plane all over the bottom except where the component leads come through to their "pads". This second mock-up works much better -- the ground plane acts as a shield. I dunno ... this project may require more skill and tools than most folks can muster, no matter where they live.
The raw double-sided .062 laminate is about $4.50/6 = $0.75 a board. But the drills cost about $1.00 each, and the Dremel burrs are ~$2.50 each. Point-to-point wiring on unched board (0.10 inch centers, no copper) has a lot of issues and will require shielding in the bottom of the box (the easiest is single-sided .032 raw pc board with a wire soldered to it and connected to the circuit common). I found a pc board fast-prototyping service (www.expresspcb.com) that gives you three boards for $55 or
$55 + ($0.65*NumberofBoards*BoardAreaInSquareInches) + $1.00*NumberofBoards) + shipping.
At 1.5 inch wide x 3.0 inch long there's 4.5 in^2. So this is $2.93 + $1.00 per board, plus the $55 setup charge. This means that a 100 would cost $448, divided by 100 = $4.48. There may be cheaper places at this volume. The boards would be plated-through but without solder masks and component locators.
Hope all's well ... Bill Wvbailey 12:55, 26 January 2008 (EST)

call sunday

Yes, I'll be there with you. Rest easy, --Sj leave me a message 01:29, 27 January 2008 (EST)

using google spreadsheets

That's not a bad idea. That makes cutting-and-pasting easier, at least. Anyone can set up a spreadsheet, publish it, and use that plugin... Just use the "publish" tab along the right-hand side of a spreadsheet once you're inside it. --Sj leave me a message 13:42, 29 January 2008 (EST)

reviewing

A review process should be something like this: a public place to list new projects, and another for finished projects. Guidelines for making a project useful and educational. And a few community members who regularly visit that page and update it / discuss new projects there every day.

We have a place for proejct descriptions now, on the contributors page; but could use something for health projects (say, just transcluding those project pages). The discussion can take place right along with the project description:

  • who is interested,
  • how it could be improved,
  • how it applies to OLPC,
  • what formats and languages it is in,
  • who its audience is.

--Sj talk 13:00, 3 February 2008 (EST)

Editing the sidebar

Leave your thoughts and inspirations here: MediaWiki talk:Sidebar! --Sj talk 13:22, 26 February 2008 (EST)

Mozambique OLPC

Hi Arjun--I look forward to the next Health call and I hope I will be available. I've added some information on the Mozambique intervention I'm working on, so if you have any feedback I would love to hear any comments/suggestions. Thanks! --Jumpbean 17:00, 8 March 2008 (EST)

Calls

Sorry we keep missing each other! We should definitely talk soon. I'll keep pinging you on irssi and hopefully I'll catch you soon... Mika 19:04, 9 March 2008 (EDT)

Pricing issues re the USB sensor. Update on oscilloscope probe/isolator

I've come to a similar conclusion as the $2.00 USB sensor -- a pc board that plugs into the USB port. But it costs about $5.00 - $6.00 in qty. It will protect the unit against 120 VAC (tested) and should also against 240 VAC. The significant cost items are the potentiometer and the pc board.

To be realistic and fair to your readers, your designer should quote against 100 and 1000 piece prices for the pc board. If you quote at www.apcircuits.com you will find that an approx 2 inch-square board (1.15 inch x 2.125 inch with 50 holes) in 100 pieces is $3.31 and 1000 pieces is $2.25.

If you guys know of something that I don't re pricing ... please lemme know.

Bill Wvbailey 17:16, 11 March 2008 (EDT)

Retrieved from "http://wiki.laptop.org/mediawiki/index.php?title=User_talk:Arjs&oldid=116020"