Posts

This profile is from a federated server and may be incomplete. Browse more on the original instance.

jeffpc, to random
@jeffpc@mastodon.radio avatar

Oooh, the new defines 4 new uuid versions!
https://www.rfc-editor.org/rfc/rfc9562.html

jeffpc, to Electronics
@jeffpc@mastodon.radio avatar

The project detour number...I lost count...is progressing quite nicely. The goal is to line up the edges. Instead of the brain-dead-simple (too high? -1 too low? +1) logic, I need to write a PID controller (or something like it) to avoid this sort of amusing to watch hunting.

Screen recording of Siglent oscilloscope. 100ns/ 1V/ Shows channel one 1Hz positive edge (and the source of trigger), and channel two a falling edge that's oscillating near the channel 1 edge.

jeffpc,
@jeffpc@mastodon.radio avatar

@jaseg A simple PD controller is doing better and worse at the same time.

It could be bad constants or bad fixed point math :D

jaseg,
@jaseg@chaos.social avatar

@jeffpc If the bang-bang controller is already oscillating like that, I suspect you might need an integral term to dampen that oscillation.

jeffpc, to random
@jeffpc@mastodon.radio avatar

I just spent the afternoon reading about PCI bus signaling. This is normal, right?

jeffpc,
@jeffpc@mastodon.radio avatar

@gsuberland Weird but really clever.

gsuberland,
@gsuberland@chaos.social avatar

@jeffpc yup. it's kinda funny how close to the limit of single-ended signalling PCI was on regular FR4 boards. the insertion loss is pretty massive so you dump a ton of power trying to drive the bus over the distance of an ATX board. reflected wave was just enough to make it workable.

jeffpc, to FreeBSD
@jeffpc@mastodon.radio avatar

Perhaps a silly question... what's the easiest way to use in a jail? All the instructions I found just install FreeBSD-* which seems...excessive.

jeffpc,
@jeffpc@mastodon.radio avatar

@gsuberland well, I'd use pkgbase to make a template that I'd zfs clone for each jail, but that's (IMO) not relevant.

The "problem" is that I can't seem to find a list of packages that correspond to base.txz or more accurately a list of packages that are required in a jail. I don't need to waste 100s of MB on what amounts to kernel*.txz, etc.

gsuberland,
@gsuberland@chaos.social avatar

@jeffpc not sure exactly the best approach for that. if the goal is to have some default set of packages when you create a jail, it might make more sense to have an installer script that you run in the jail, so you don't end up with weird complexities when you're doing jail upgrades.

jeffpc, to Electronics
@jeffpc@mastodon.radio avatar

layout question for anyone who knows more than I do about the subject (essentially everyone)...

Is it ok to put decoupling caps under the package (like the 3 red ones in the image) or am I better off keeping them on the same side as the IC (like the blue ones)? I only have 2 layers to play with and lots of routing on one (still to do on blue), so plan to use the other as (mostly) ground plane and power anyway.

I'm asking before I do the tedious routing ;)

jeffpc,
@jeffpc@mastodon.radio avatar

I suppose the options are:

  1. keep caps on the blue side
  2. move caps to red, but outside of the IC footprint
  3. move caps to red, but it's ok for them to be inside the IC footprint
jaseg,
@jaseg@chaos.social avatar

@jeffpc Having them on the other side looks fine to me, as long as they don't split up your ground plane too badly and you keep ground loops in mind. You add a bit of inductance with those vias, but since that chip is in a massive LQFP with long (~10mm) internal leads anyway, that 1.5mm should hardly matter. To keep inductance as low as possible, I'd maybe consider using wider traces where possible (e.g. from via to cap), and making the vias wider where space is available.

jeffpc, to FreeBSD
@jeffpc@mastodon.radio avatar

Any recommendations for a graphics card for a desktop? I've been using a ~10 year old nvidia card, but the drivers haven't been updated in a while and have incomplete implementation of libEGL (see bug 263347) which makes some 3D accelerated software (like kicad) crash.

So, even though the current card still works, the closed source nature of the "driver" is making me consider replacing it with something that works and is open source.

jeffpc,
@jeffpc@mastodon.radio avatar

@gsuberland I've only seen kicad get tripped up.

It is my understanding that EGL is not X11 specific, and so I assume Wayland would be affected as well. I haven't paid attention to the whole X11 vs. Wayland thing... aside from occasionally dreading the inevitable breakage as software switches over.

gsuberland,
@gsuberland@chaos.social avatar

@jeffpc from what I've heard, the switch to Wayland is pretty smooth, and a lot of apps have already gotten past the initial hurdle of bypassing a lot of the X-specific hacks they had to make, but on the flipside there are some major regressions in accessibility features in Wayland that the blind and disabled communities are less than happy about.

jeffpc, to random
@jeffpc@mastodon.radio avatar

Strolling through I spotted a box of mice. So, here Iโ€™m sharing the photo for

adamsdesk,
@adamsdesk@fosstodon.org avatar

@jeffpc Wow Sun Microsystems, an amazing system from the past.

jeffpc, to RaspberryPi
@jeffpc@mastodon.radio avatar

After I replicated @BatchDrake I2C over coax setup (https://mastodon.radio/@jeffpc/109425277089482499), I wondered if the 's outputs are weak. Tonight, I finally got a few minutes to try out a similar thing on a . I fed the SPI clock output through the same 30.5m coax. It looks quite good!

This time, the setup was a bit different, the connections were:

AVR -> ~1m long coax -> T adapter to channel 3 (cyan) -> long coax -> T adapter to channel 4 (green) -> 30dB attenuator as a 50ohm load

gsuberland,
@gsuberland@chaos.social avatar

@jeffpc @BatchDrake wow, that's a huge difference in rise time. seems like the drive current is better suited to the termination impedance, too.

jeffpc, to random
@jeffpc@mastodon.radio avatar

Hm... why does my 10.6MHz pierce oscillator oscillate at 44.25kHz? ๐Ÿค”

jeffpc,
@jeffpc@mastodon.radio avatar

@gsuberland I tried replacing both 47p with 30p and I got ~600kHz.

I definitely agree with breadboard capacitance being hard to quantify. Doing 10MHz on one is definitely asking for a bit more than is healthy :)

The datasheet (LFXTAL012945Bulk by IQD) lists 30pF as the "load capacitance". ๐Ÿคท

gsuberland,
@gsuberland@chaos.social avatar

@jeffpc ha, datasheets useful as always! ๐Ÿค”

I guess maybe try 27pF or 22pF if you have them - could be that the capacitance they spec is the nominal parallel, in which case 30pF would be a smidge over given the parasitics. All just a bit of a crapshoot.

jeffpc, to Electronics
@jeffpc@mastodon.radio avatar

My first PCB is being manufactured! I really hope I didn't screw it up too badly :)

gsuberland,
@gsuberland@chaos.social avatar

@jeffpc nice! it's a liberating step once you get past the first hurdles. so many more designs within reach!

jeffpc, to hamradio
@jeffpc@mastodon.radio avatar

I though Iโ€™d learn code and programming at the same time. I started with an ATmega48P gating a 555 generating a tone. But thatโ€™s boring. So, I extended my code to generate messages toggling D#6 on and off.

A left hand in the lower left corner is seen holding iambic paddles. The rest of the frame is close to the display area of a mid- to late-90's consumer MIDI keyboard. The hand is seen alternating between actuating the two paddles changing the program on the keyboard.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • โ€ข
  • JUstTest
  • kavyap
  • DreamBathrooms
  • thenastyranch
  • magazineikmin
  • tacticalgear
  • cubers
  • Youngstown
  • mdbf
  • slotface
  • rosin
  • osvaldo12
  • ngwrru68w68
  • GTA5RPClips
  • provamag3
  • InstantRegret
  • everett
  • Durango
  • cisconetworking
  • khanakhh
  • ethstaker
  • tester
  • anitta
  • Leos
  • normalnudes
  • modclub
  • megavids
  • lostlight
  • All magazines