@lily you could get sillier and give your cpu cycles in terms of clock wavelengths. this is a 7.5cm CPU connected to 22cm RAM. actually i imagine a variation of this is somewhat useful for circuit design
@astrid@lily ok but bloody web designers actually do this shit
they say shit like "our website is designed around a minor third interval" to mean that headers are 20% bigger than body text and just please shut the hell up, no, first of all that's a preposterous thing to do and second of all you haven't defined if you mean an just intonation minor third or an equal temprement minor third because if you've designed the whole thing around the fourth root of two then your devs are going to at best ignore it and at worst murder you
@andrewt@lily see this would only somewhat work if 2x the font size had some significant meaning. like 2x the frequency does, but 2x the font size doesn't really? and it would work better if there were harmonics and beats between specific intervals but there aren't either? like it's not even a good analogy
@astrid@lily it really isn't. Not least because they'll never get to the ×2 interval because their minor third isn't ⁴√2, it's 1.2, so stacking four of them will give you ×2.0736. And one of their favourite intervals is ϕ — the whole point of musical intervals is to be (or approximate) neat, simple ratios and the whole point of ϕ is to be as far away from any of those as it can. It's arguably the most violent possible discord you can produce with two notes. It's not just a pointless analogy, it's actually fighting them.
@astrid@lily also "minor seventh" is also the name of a chord so I assumed we were using the ratios of all the notes in that chord and it sounded very silly but no, they're doing a perfectly reasonable thing and just calling it something silly
@andrewt@lily I hate everyone's obsession with the golden ratio so much. It's an interesting number mathematically but for literally none of the reasons anyone thinks it's interesting. They'll go impose a spiral or ratio onto literally anything, it won't even be aligned, there's enough tolerance that they can get away with it, and they're like "wow cool universal constant" when no, you're just doing designerist numerology
@astrid@lily the amount of people that think A4 paper is in the golden ratio is maddening. there is more than one interesting number! A4 is in √2, not ϕ! Although whatever you do don't tell the designers that or they'll be calling it "tritone paper" and insisting it's cursed
I remember seeing a clip of a Grace Hopper lecture in which she held up a length of copper wire (somewhere around 20 cm) and said, "This is a nanosecond." For pretty much exactly this purpose.
@astrid@lily yeah the wavelength is used to constrain chip sizes based on planned operating frequency, and therefore is an important part of very high frequency ASIC layout
@spinach@astrid@lily it's also extremely useful for knowing when you need to switch from a lumped circuit model to a transmission line model. essentially figuring out whether the signal voltage will be near constant across the length of the trace (lumped, modelling a simplistic conductor) or if there will be considerable variation in the voltage across the length of the trace (transmission line, modelling a waveguide).
and for digital it's the rise/fall time that matters, not the fundamental.
@spinach@astrid@lily also fun fact: characteristic impedance and transmission line effects in circuits are the exact same physical phenomena as refractive indices and refraction/reflection of light. literally the same thing, just different frequencies!
I am glad I happened to be learning signal integrity and laser physics at the same time because noticing that the two sets of equations were identical by pure coincidence was extremely satisfying. super cool eureka moment.
@gsuberland@spinach@astrid@lily In the early days of radio, wavelength was the preferred unit of measurement because signals were measured by finding the physical distance using a ruler between two standing wave nodes on a Lecher line or slotted line without any electronics. The appearance of practical high-speed counters is the only reason why modern electronics shifted to frequency instead. Optics still uses wavelength for the same reason, most measurements are still from counting fringes from interference patterns, cycle counting is far less practical.
@astrid@lily speed of light in copper is like 0.99c, however, the energy (and therefore signals) on a PCB aren't propagating through the copper! the copper is there for the light to [extremely over simplified] bounce off
the signals are propagating through the insulator. basically all of the energy on the PCB is stored in the insulators (whether the PCB dielectric or the capacitor dielectric)
@astrid@lily you can figure this out from first principles if you consider that copper basically cannot contain an electric field within it because if it did, the free electron gas would almost instantly shift to neutralize it
without a potential difference in a field, there is no energy stored (since that's what potential is)
across the dielectric, you can have as much as the breakdown voltage allows
@astrid@lily did you know that the anisotropic woven nature of the fiberglass [FR4 is woven glass fabric filled with epoxy] changes the speed of light for the transmission line enough that for especially high speed signals (like PCIe Gen4 and higher) it pays off to make weird traces that aren't going vertically or horizontally because then the changes average out and it distorts the signal less
@astrid@lily because like, if one half of your differential pair is consistently routed over the bit with more glass fiber and the other half is consistently routed over the bit with more epoxy, the differential signal will get smoothed out, and the eye will close a bit. this can appreciably eat into your link budget!
so you do it like the pic below. you might have seen this on server motherboards
"Buildings have walls and halls. People travel in the halls, not the walls. Circuits have traces and spaces. Energy and signals travel in the spaces, not the traces."
Ralph Morrison, Grounding and Shielding: Circuits and Interference
Add comment