bodges are inevitable 😖

I wonder how reliable this measurement is

so not labelling the parts was a mistake, and making a test board as dense as the final product is always a mistake, so I added more test points! (boards from @aisler.net, NNNI26 for €10 off year-round, multiple times!)

two D flip-flops give me I/Q signals at any frequency (assuming the FF is fast enough, I'm using an 'ACT74)

routing using planes is something I enjoy, especially for power stuff

it's been a few years since I ordered something halfway complicated, but thanks to @aisler.net's rapid registration stencil, it should be easy (NNNI26 for €10 off throughout the year! 😎)

always remember to order a stencil with your PCBs! I tried smearing paste on the board here, but the results were blobby 🤰 (if you get your PCBs from @aisler.net, use NNNI26 for €10 off year-round! 😳)

one of the best moves I've made in the recent past is blocking the uni's "AI center" nonsense, I don't need that slop in my uni inbox every day, please crash and burn

@aisler.net to the rescue with some TMT SOIC-8 adapters! The year 2026 still has 7 months where you can use my discount code NNNI26 to get €10 off multiple times!

modern oscilloscopes are universally pretty meehhh, but the @rohdeandschwarz.bsky.social MXO38 is an exception - compact, dense (I love density), the display is amazing, the UI is ACTUALLY USABLE

hello followers, I would like to announce my exhibition at the Maker Faire Hannover 2026 this year in Hannover, Germany! if you can make it, I'd be glad to meet you and bore you with analog thingies

my long-awaited follow-up article is finally out! nnnilabs.net/trash/2026/0...

attoamperes?

today I'm testing the dielectric absorption test stand

populated some more breakout boards from @aisler.net , bonus picture taken with a microscope (psst: NNNI26 gets you €10 off throughout the year, multiple times!)

I just realized that my new signal generator has adjustable rise and fall times, so here's a demonstration of I/C = dV/dt using a 10 nF / 1 kΩ highpass filter

the HMC8043 completes my all-LAN all-SCPI setup! to celebrate, I wrote a script that loads a few useful voltage/current configurations without having to fiddle with the front panel controls

phase one of my LAN upgrade is complete - scope, sig gen and dmm are now all connected to LAN 🎉

being able to RELIABLY read back 50,000 data points is amazing

I played with the 34410A for the entire day yesterday and scripting for it is so painless here, I measured the RMS noise at different NPLC settings and compared it to what it should "ideally" be (decrease with sqrt of integration time, normalized to lowest NPLC)

thanks to LAN and sockets, I can run the exact same script on both my PC and RPi 4B with 0 changes

😤

I finally got myself a 34410A! I impulse-purchased this because: 1. it has a multislope IV ADC which I've always wanted 2. it has LAN, so the first step towards converting my bench to LAN + SCPI

okay so now I have like a very expensive clock

JFETs are underrated

fun little project, thanks @aisler.net for making this happen

feels like a good time to do some pcbposting, so here's another run from @aisler.net that I assembled using stencil-and-paste, and then a few minutes in the ultrasonic bath to get them super clean

I swear to god, no ulterior motives!

the restorative power of good anime is highly underrated

tested the 3458A track-and-hold circuit with various input voltages, droop rate (after what looks like dielectric absorption settling) settles to -50 μV/s, which corresponds to -2.35 pA flowing out of the hold capacitor... the good old TL072 is pretty good in this regard

the averaging mode of oscilloscopes is highly underrated

I replicated the track-and-hold circuit from the HP3458A (as described in the April 1989 HP Journal), and it's pretty cool - you can see the hold step, which stays constant because of the clever way they bootstrapped the sample JFET

I love cheap Chinese gear but I am coming to terms with the fact that SCPI is kinda iffy... does not work as flawlessly as stuff from Keysight for example (who are waaaay out of my budget)

I've spent four days trying to get something out of this chip but it's like 🤐 its capabilities are amazing but it just does not want to get configured and talk

yep... 4 gigasamples per second with "6.5 digit" resolution... it took 33 minutes to capture 1.2 μs worth of waveform (timebase not calibrated yet)

it's this circuit: www.radiolocman.com/shem/schemat... but much faster! delay is adjustable between 0 and 1 μs and the output pulse is 10 ns (available in both polarities) a 12-bit DAC controls the delay, so the delay should be adjustable in steps of 250 ps

populated board, the device under test will be connected "through-hole" to the machined pin headers so it's up in the air to prevent leakage currents hopefully

fresh set of PCBs from @aisler.net more to come

getting somewhere

now that's more like it

yeah not very impressive, but I think I know why - the OPA145 I'm using as a pre-front-end to the LM311 has a very low bandwidth (5.5 MHz) compared to a straight fast comparator like the LT1016 (10 GHz)

things are being measured again after months of "hobbyist's block"

populated this TPS7A39 board yesterday (±5 V output)

@aisler.net do you have a tutorial for castellated PCB edges? not just pads, but around the entire board edge

@aisler.net bump!

fortune.com/2026/02/24/s... oh my god, why is this stuff not dead yet? stop talking about "holistic AI education" and burn this down already

I nerdsniped @dmytroengineering.com into helping me out with something and here are the results (a friend in need is a friend indeed!) github.com/diminDDL/pic...

adapters!!! thanks @aisler.net! (NNNI26 for €10 off, works multiple times!)

I like this

ready for the next prototype github.com/macaba/TheMa...