Archive for category Projects

ORD Camp Photos and Panoramas

The Whisky fest

The Pig fest

Ord Camp 1

Ord Camp 2

Ord Camp 3

Ord Camp 4

Ord Camp Shop

, , , , ,

1 Comment

Filtering EMI from Switching Power Supplies

For some time now we’ve been experiencing strange problems with the USB system inside the cube.  Somewhere along the USB path something drops the connection, either the Arduino or the Hub or the Linux box. We first tried swapping out USB cables or Hubs. We suspected the Arduinos were drawing more power then the hubs could provide so we separately power them with power supplies. Finally we started suspecting the Dimduino boxes them selves were at fault.

IMG_0196_small

Nothing seemed to provide any particular relief. Oddly the frequency of the disconnections was not isotropic and certain boards tended to cause disconnections much more frequently (every few minutes) then others. Swapping out the dimduino boxes or the arduinos didn’t affect this though, which was suspicious. Finally it dawned on me that the PowerSupplies that power the Lights must be at fault because that was the only element we hadn’t yet tried to change. Sure enough, analysis of the freuquency of the dropped connections with whether or not the lightbox was powered by a new style power supply (an IP67 compliant, waterproof switching powersupply) or an ATX powersupply was conclusive. The new switching powersupplies were in fact at fault. Removing them all and powering everything off ATX PSUs fixed all the disconnection problems overnight.

So far so good. Now, however the cube was running on a single PSU per box, i.e. our redundancy was compromised (we only just had enough ATXes to power each box with one PSU). Thus I’ve been trying to figure out what the issue is with these new PSUs and if there’s a cheap fix.

I hooked them up to my oscilloscope and it was pretty clear that these new power supplies are switching power supplies but it appears of poor quality. Unlike an ATX which gives perfectly smooth power there’s a lot of high frequency noise on the output.

IMG_0192

Measuring Vout and V1 (Fig A) shows some interesting patterns: B) Vout has a 83KHz beat of about 2V magnitude and a faster superimposed frequency in the Mhz range decaying exponentially like a damped oscillator after each 83Khz beat.

C) V1, between 0V output and Ground, is clearly carrying a 60Hz mains hum and a faster ripple on top of that, which resembles the ripple in B). The 60Hz oscillation has a large magnitude (like 60V or so) but clearly very high impedance (you dont get zapped when you touch it and barely any current flows if you connect 0V to Ground, but the 60Hz oscillation on both 0V and 12V disappears completely).

These oscillations make the whole cube circuitry swing like hell, and sends out a ton of EMI and RF everywhere. I can pick up the Mhz signal using just a piece of wire in the air. No wonder the USB was struggling.

diagram1 The two solutions I arrived at, empirically are shown in D) and E).
One involves a resistor and a capacitor and the other has an additional direct resistor.
Both smooth out Vout ( B) ) quite a lot, though not completely and V1 also looks quite a bit better but also not perfect. E) i think works a weeny bit better, but its all comparable. Some spike voltages remain on V1 that I just can’t get rid of it seems. Vout is now pretty smooth. Adding another cap between Out+ and Out- doesnt help to remove any residual noise. Nor does adding Y caps for the +12V line (from +12V to GND ).

emifilter
I found a nice article on Switching PSUs and EMI which suggests a more elaborate but straightforward circuit with additional caps and particularly in-series inductors. Makes sense, the inductors will further resists the AC going forward. I’m tempted to try it but maybe what I cooked up is sufficient (it certainly appears to be, empirically speaking).

Also I talked to a knowledgeable person at ORDcamp this weekend who suggested using higher voltage caps should not only make things safer but also help absorb more of the 60Hz hum and the RF.

If all of this doesn’t help the next step is clearly to add in series inductors but this is a slightly bigger undertaking and begs the question if it wouldn’t be better to just move to a different PSU. Would be a shame though to toss the 14  $70 PSUs that in all other ways are absolutely perfect.

, ,

1 Comment

Groovik’s Cube Panoramas

Using Photosynth I made some panoramas of the cube, on the inside and the outside :)

, ,

No Comments

Groovik’s Cube in the News

Seattle Met Magazine (Nov Issue): http://www.seattlemet.com/groovikscube

On KING5: http://www.king5.com/news/local/Giant-Grooviks-Cube-featured-in-new-Pacific-Science-Center-exhibit-130875678.html

Groovik’s Cube on Q13 FoxNews: http://www.q13fox.com/videogallery/65109801/News/Kaci-Pac-Sci-Center

Seattle Times: http://seattletimes.nwsource.com/html/photogalleries/localnews2016222326/2.html

IO9: http://io9.com/5840738/watch-people-constructing-a-26+foot+high-fully-playable-rubiks-cube

KOMO News: http://queenanne.komonews.com/news/arts-culture/668165-giant-rubiks-cube-arrives-burning-man-frustrate-you

PCWorld: https://www.pcworld.com/article/240168/26foottall_rubiks_cube_is_even_harder_to_solve.html

Seattle PI: http://blog.seattlepi.com/thebigblog/2011/09/13/seattles-giant-rubiks-cube-where-to-see-it/

Seattlest (Blog): http://seattlest.com/2011/09/20/attention_stoners_a_giant_rubiks_cu.php

Opodo: http://news.opodo.co.uk/NewsDetails/2011-09-22/Seattle_s_Pacific_Science_Center_launching_two_new_exhibits

Associated Press: Associated Press (Yahoo news)

TechFlash: http://www.techflash.com/seattle/2011/09/a-giant-grooviks-cube-in-seattle.html

SLOG (Seattle Stranger Blog): http://slog.thestranger.com/slog/archives/2011/09/13/grooviks-cubism-at-pacific-science-center

No Comments

Grooviks Timelapse

YouTube:

Local AVI:

http://www.miketyka.com/data/grooviks_timelapse_highres.mov

,

No Comments

We have lift off!

DSC_1590_sm

,

1 Comment

Day 8: Groovik’s cube is nearly done!

Its sunday night and a bunch of hackers are sitting at the door of the cube fixing bugs and figuring out the last few hick-ups for the cube to conclude a phenomenal week of cube building. This is day 8 of the installation and things went a lot faster then we first anticipated. The main structure was done after 300 man hours of work. The new frame worked out great and despite somewhat larger flexibility then we thought its doing its job perfectly. The new weight estimate for the cube is in the 1700 pound range which is *a lot* lighter then the 4500 or so pound weight on the playa (we know this accurately since the crane operator told us). The new cube has a nearly 2/3rd weight reduction. 1150lbs coming from the new aluminum frame.
Installation time is also significantly down (about 40-50% less) but i think there’s a lot of mileage here still. Many things were learned and figured out during this isntallation, especially about the new frame and I firmly believe installation time could be reduced to 200 or even 150 man hours with an experienced team and a number of possible and easy hardware upgrades. Watch this space.

We lift it into place tomorrow. :)

Below: The Software team engrossed in the code next to the finished cube ont he floor.
Photo on 2011-09-18 at 17.52

, ,

No Comments

2nd and 3rd Cube installation days timelapse

Work is going really smoothly and fast. Much fast in fact then we had anticipated. Videos speak a thousand words (25times a second).

, ,

No Comments

First Build Day

Today we started our installation of the Groovik’s cube at the Pacific Science Center. After a smooth load and unload session we finished 2 hours early and started assembly. Unfortunately I managed to overlook one box containing all of the bolts and nuts needed for the frame assembly so there was some hectic running around trying to scavenge, buy and retrieve bolts in order not stall the build. Once we figured that out though everything went very quickly and we erected the entire frame.  After putting in the internal cable bracing we test lifted the cube by two opposite corners and found that the structure is actually fairly elastic and will flex somewhat under this unnatural load vector. Tomorrow morning we plan to do another full test lift hanging the cube from all three corners and seeing if the problem persists in the hanging orientation it was actually designed for. My guess is that once under the right load direction things will look a lot better but who knows. Alternatively we will be figuring out how to stiffen out the frame while we press forward with the installation of the top layers of fabric, lights and reflectors. I’m sure many unknowns stil lay ahead but for now things are going forward ok. Right now we’re ironing out the last few bugs in the software at ALTSpace and hopefully we’ll have a fully running, mappable and color calibratable interface in the next few days. Just in time to be tested on the whole object – i expect instllation will go smoothly and quickly.

, ,

No Comments

Successful end-to-end test

We hooked up the Grooviks hardware and the software simulator last night and with only a few code changes we had the simulator drive the lights. Very very satisfying.

simulator_running_the_hardware

,

No Comments