@Felthry Also, fyi, most oscilloscopes are 8 bit

@zetasyanthis Yeah, I know--I think we have some really expensive sixteen-bit one somewhere though

but also we're interested in lower frequency stuff, which means you can increase the ENOB somewhat with averaging...or something? signal processing stuff is complicated

@zetasyanthis current probes almost always use some proprietary connector on top of the BNC though? I know the current probes we have already can let you know when the jaw is open for example

@Felthry "Compatible with any Keysight oscilloscope with a 1 MΩ input BNC" (Which means any. The picture shows a normal BNC cable.)

@Felthry If you click on the link, you'll see. :P 600A, 30Mhz.

@zetasyanthis Ah, didn't realize I had to enable javascript to see the iamge, that's weird

@Felthry Oh, that kinda is. Either way, I think that does what you want? There's a 300A model too. Not sure on the resolution. You'll have to check.

@zetasyanthis Possibly! The price is a bit ehhhh though--we'll focus on designing the buck converter first and then get around to metrology. Will leave a spot for current measurements.

Fair. Keysight's stuff is pricey, but it's good at least.

I'm also a little curious what this is for. I've designed buck regulators before (and was actually working on a few just the other day), but 100A is quite a bit.

@zetasyanthis you managed to lose the @!

Research on three-level buck converters, specifically circulating currents when using multiple in parallel, as well as how to mitigate problems with neutral point drift.

This is based around DC power transmission systems, as used on some ships. The idea is that this small(!) device will have a 10kW max power output

@zetasyanthis The lab is also heavily focused towards EMC (i.e. compliance with UL, FCC, etc standards), and conducted emissions are legally defined as undesired harmonic content up to 30MHz either on the output or back up the input (or through grounding straps, or...). That's why we're looking at 30MHz bandwidth specifically.

@Felthry AHA, that makes perfect sense!

@Felthry Oh, oops! And okay, that's pretty neat! I've dealt with 12V DC distribution in computer racks before (and am dealing with it now a bit), but nothing on the scale of a ship. (I think these power shelves are about 15kVA each. Unsurprisingly, Delta makes them.)

@zetasyanthis The research is being done on a grant from the office of naval research so of course it'd be about shipboard stuff (not a huge fan of working on a thing for the navy, but none of this is classified research and all of it will apply equally to freight ships so....)

@Felthry I feel you 100%. I used to work for the IC (at one of their contractors) and left that forever after they used one of my devices to launch air strikes at people. :(

@Felthry (Technically they gave it to a third party government who then did what they wanted with it, but that's still not what I signed up for, or was okay with.)

@zetasyanthis fortunately I think working on power systems and only power systems is mostly guilt-free as long as it's not classified research. It's a generic enough thing that it's applicable to pretty much _anything_, I guess?

@Felthry Yeah, that's fair. I stayed on for a while doing some work for NASA for similar reasons.

@Felthry And mine research and development definitely was classified, hence not saying exactly who it was for. :/

@Felthry Actually, I suppose I should say "a different organization in the IC," since ONI definitely qualifies.

@zetasyanthis Actually, looking into it, it seems like Tek sells the _exact_ same current probes, just with a tektronix logo instead of a keysight one on them

I assume they both buy from some third-party manufacturer and put their own logos on them.

@Felthry XD

@Felthry I'm not that surprised. Even Lecroy rebrands some of their probes.

@zetasyanthis that rogowski probe you recommended has a minimum detectable current of 250mA, which suggests it's probably got a pretty high noise floor unfortunately. Not sure how good you can get with rogowski coils, though.

@Felthry Ah, dang. They have a few probes with different sensitivities. Might be another can help.

@zetasyanthis I think poor noise performance is kind of baked into the concept of a rogowski coil unfortunately. not having a magnetic core means they don't concentrate flux into the coil, which is what makes cored devices work so well

we need better paramagnetic materials!

@Felthry I mean, there are current transformers too. (I checked digikey, but those go up to about 1 kHz max at the current you're wanting.)

@zetasyanthis Yeah, that's the problem! I'm wondering how hard it would be to roll our own current transformer, at this point, but I don't know the first thing about high frequency magnetics

Some kind of current shunt resistor might be the only option

@Felthry Actually, yeah, that could work. You could A/C couple off the current sense resistor before driving an op-amp with it, potentially. Definitely want some limiting resistors on those lines, too.

@zetasyanthis Problem then is the preamplifier that you'll pretty much definitely need

you need a high bandwidth, extremely low noise op amp, _and_ some way to power it while it's floating at potentially up to 500V (honestly considering a CR2032 for that if we go down that route--it's for research, not a permanent installation, and isolated switching power supplies inject all kinds of noise), and then some way to level-shift that signal down to something we can use as an input

@zetasyanthis In any case, the metrology is probably going to wait. We do have high voltage diff probes so maybe those will be sensitive enough? The problem lies in trying to AC couple a current signal, which means using inductors, and inductors can become very capacitive at high frequencies.

@Felthry Yeah, fair. (And sorry, just thought that's a really cool problem and started digging into it. XD)

@zetasyanthis oh I agree it's a really fun problem!

@zetasyanthis we've learned a lot of interesting things in the process of designing this too, like "actually it turns out you can just buy 2500μF film capacitors" and "instrumentation amplifiers are very low bandwidth" and "holy _fuck_ silicon carbide MOSFET modules get expensive"

@zetasyanthis Also, apparently they make power JFETs designed for switching, which is Definitely A Good Idea to have a system where losing power to your gate drive means you instantly create a shootthrough condition

@Felthry Hahaha. Oh god. IGBTs please! XD

@Felthry I mean, you can do the tried and true "put a nice ADC on the board with the op-amp" and then run the digital data through an optoisolator?

@zetasyanthis Might have to, but _oof_ 30MHz ADCs won't come cheap