Show more

Free sunday tip: if your refuses any flash upload or erasure, check you don't have a serial monitor running in a random terminal...

At @logilab we have deployed 5 @peertube instances over the years :

1. intranet from source
2. intranet using @yunohost
3. extranet using yunohost again
4. public for @semwebpro conference using yunohost
5. public for logilab videos using kubernetes and s3-type storage

With this new kubernetes + s3 deployment we're trying to contribute and have feedback from other users or potential users. You can review forge.extranet.logilab.fr/open and get in touch ! Thanks ! Boosts welcome.

The ref voltage is a bit off (6.0v) but well within the required range (+/-0.3v). After a quick adjustment, it's starting to look better

Show thread

But it was pretty unstable, mainly because the 2DW233 was drived too low with current schematic: at 1.5 mA instead of the expected 7.5mA (for the 233). So I played a bit with ltspice to find better values for R61 and R63, and I also replaced the opamp with a 741 I found while cleaning a bit my bench :-)

Show thread

So here is current situation: the 5.9v zener reference has been replaced by a 2DW233. First quick check was a bit dodgy: this ref is normally 6.2v, but actually consists In 2 zener head to head, so in practice a 5.5-ish v zener in series with a forward one (so .6-ish v). So I've used only one of the zener of the 2DW233 in series with a germanium diode. I also used a random opamp I had but had not the correct pinout (it's a dual opamp). The result is a bit dodgy as I said, but it worked: 5.87v

Show thread

now I need to find a replacement for this 5.9v reference voltage. Maybe it's time to start playing with these 2DW233 I have...

Show thread

Yes, this poor little optocoupler is not straight, but it does work. Now I have decent voltage readings (here fed with 0.5v)

But it was still not working. After fighting with the poor quality (but available) schematics from the service manual still available on Keysight web page, I could identify the culprit: one of the optocouplers was dead (on the LED side). Funny enough, it is listed as a TIL111, but it's a 8-DIP package (the til111 is a 6-DIP). So I had to improvise...

After replacing it with what I had, a simple 6.2V zener and replacing the opamp with a random one (but with matching pinout), I finally got the voltages in the correct ballpark.

Then I checked the power supplies, and sure enought, voltages on the inguard board were completely wrong. Turned out the 5.9v reference voltage zener is cooked, as well as the opamp driving it. This 8 pins can is actually a zener diode (obviously made of unobtainium)

The unit did power up, but the displayed voltage was all over the place, and a couple of keys had no spring left, so I started there with my plastice made replacement spring trick

Show thread

Hey I have a new toy I'm working on, a glorious HP3437A "High-speed" system voltmeter (from late 70s) . Got it for a cheap as non-working, as usual. TBH, I've picked mostly for the fun amd because I love these "bubble" 7 segment LED display.

Pour ceux qui ne l'ont pas regardé, le webinar organisé par l'ENS sur le co2 et covid est vachement bien savoirs.ens.fr/expose.php?id=3

Connaissez-vous Toobib?

toobib.org/

"Toobib est un outil opensource / libre et décentralisé de prise de rendez-vous médical. Toobib présente les informations (numéro de téléphone, courriel, etc) d'un praticien de santé, sans compromettre les données des utilisateurs. "

Créer et réserver des plages de rendez-vous,

Fédérer des instances Toobib (utilisation du protocole ActivityPub),

Au revoir Doctolib?
😀

#médecine #privacy #santé #données

Pfff 🤯 j'ai résolu mon premier Rubick's Cube, tu sais, le truc qui traîne sur la table du salon, où tu arrives à peu près à faire une face depuis quelques années...

C'est peut-être d'un banal à pleurer, mais le sentiment que ça donne est chouette.

J'ai suivi youtube.com/watch?v=R-R0KrXvWb

fr.wikipedia.org/wiki/Rubik%27

#rubickscube #rubick

Thrilled that our paper (w A. Pietri & T. Allançon) «The Software Heritage Filesystem (SwhFS): Integrating Source Code Archival with Development» has been accepted to #ICSE with great reviews. It presents a #FUSE layer on top of #SoftwareHeritage. Preprint in the working! #ICSE2021

Pour en savoir plus sur Geotrek,
la suite logicielle libre pour gérer et valoriser les sentiers

(Ré)écoutez le #podcast de « Libre à vous ! » (50min), ou lire sa transcription

Avec Camille Monchicourt du Parc national des Écrins
et Amandine Sahl du Parc national des Cévennes

april.org/86-geotrek

Gave an updated #Guix-#Jupyter talk this morning at the UST4HPC workshop, before an audience of sysadmins.

That was a good opportunity to raise the issue of how hosted notebooks deprive scientists from their autonomy and undermine #reproducibility. "There is no cloud".

git.savannah.gnu.org/cgit/guix

#ReproducibleResearch #OpenScience

La discussion a mené au dépot de dossiers de financement et à un site : fedeproxy.eu/ et son forum : forum.fedeproxy.eu/

... à suivre

@loic @ChatonsOrg

Show thread
Show more
Logilab's Mastodon

The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!