- Contents in this wiki are for entertainment purposes only
Bro and the Ph.D in Earth Field: Difference between revisions
Jump to navigation
Jump to search
XenoEngineer (talk | contribs) (Created page with "Category:electrical engineering Category:Earth field Category:Schumann resonance Category:QFT <div style="background-color:azure; border:1px outset azure; padding:0 20px; max-width:860px; margin:0 auto; "> [Same picnic table, one year later. The epoxy “cuddle-blob” is now under a machined-aluminium heat-spreader anodised matte-black. Bro’s coffee cup reads “∂V/∂t keeps me awake.” Ph.D’s cup still warns about δT/δt, but the font i...") |
XenoEngineer (talk | contribs) mNo edit summary |
||
| Line 6: | Line 6: | ||
<div style="background-color:azure; border:1px outset azure; padding:0 20px; max-width:860px; margin:0 auto; "> | <div style="background-color:azure; border:1px outset azure; padding:0 20px; max-width:860px; margin:0 auto; "> | ||
---- | <big>'''Same picnic table, one year later. '''</big> | ||
The epoxy “cuddle-blob” is now under a machined-aluminium heat-spreader anodised matte-black. | |||
| |||
Bro’s coffee cup reads “<big>'''∂V/∂t'''</big> keeps me awake.” | |||
| |||
Ph.D’s cup still warns about <big>'''δT/δt'''</big>, but the font is now Comic Sans—an inside joke about peer-review formality. | |||
| |||
They open their laptops side-by-side; the RP2040 has been promoted to a dual-core STM32H7, yet the $20 budget somehow still balances.] | |||
BRO ( | '''BRO (M.S. EE, still wearing skate shoes): | ||
I’ve been modelling the auto-zero loop as a discrete-time Lyapunov drift. | |||
| |||
If I treat the digi-pot wiper position as state xₖ, then ∆xₖ = −µ·sign(Vos,ₖ) gives global asymptotic stability for 0 < µ ≤ 1 LSB. | |||
| |||
But the residual dithers between ±½ LSB. I’m thinking… inject a shaped PRBS so the error spectrum spreads like spread-spectrum CDMA. | |||
| |||
That way we gain 3 dB effective resolution. Legit? | |||
'''EE Ph.D (raising eyebrow, pleased): | |||
Legit. You’re basically dithering a 1-bit ΣΔ inside a multi-bit wrapper. | |||
| |||
Keep the PRBS amplitude at ¼ LSB so the peak error stays inside the bound, and high-pass above the 1/f corner—your 10 Hz Schumann line stays untouched. | |||
| |||
Nice bridge between stochastic control and classic mixed-signal lore. | |||
'''BRO (grinning, but calm): | |||
Cool. I’ll call it “whisper-noise dither” so my homies don’t fall asleep. | |||
| |||
Skate-video soundtrack meets Shannon—perfect. | |||
'''BRO: | |||
BRO | While I was at it, I re-did the tail-current source. | ||
| |||
Instead of a lone 33 kΩ, I stacked two 0.1 % thin-films with a 2.5 V reference and a BJT current mirror. | |||
| |||
Temperature coefficient dropped to 8 ppm/°C. | |||
| |||
We’re now burning only 80 µA, yet output impedance hit 2 MΩ at 10 Hz. | |||
| |||
That’s… spa-day quiet. | |||
EE: | '''EE: | ||
You just described a Brokaw-bandgap-less reference topology, minus the bandgap. | |||
| |||
Clever—by letting the mirror run at < 0.5 V VCE you avoid the curvature term. | |||
| |||
You realise you’ve entered the “sub-threshold, sub-ppm” club? Membership includes log amplifiers and seismic front-ends. | |||
BRO | '''BRO: | ||
Membership card can double as a guitar pick—multifunctional, like us. | |||
BRO ( | '''BRO (pulling up LTspice): | ||
Long-tailed pair AC simulation shows CMRR peaking at 94 dB @ 100 Hz, then rolling 20 dB/dec above 3 kHz because of the 2 pF protection diodes. | |||
| |||
I’m eyeing bootstrapping the guard ring with a miniature op-amp to keep the diodes at zero bias. | |||
| |||
Think we can push CMRR > 120 dB out to 10 kHz? | |||
'''EE: | |||
Theoretically, yes—bootstrap cancels the junction capacitance, but watch the phase-margin. | |||
| |||
You’re introducing a feed-forward path; loop gain may kiss unity at 2 MHz and sing. | |||
| |||
Add a 10 Ω damping resistor inside the boot loop, keeps Q < 0.5. | |||
| |||
Also, mind the guard-ring leakage: 1 GΩ || 1 pF creates a zero at 160 Hz that’ll eat your lunch. | |||
'''BRO (nodding): | |||
Trade-offs, always trade-offs. I’ll 3-D print a Teflon shield so the guard never sees sweat—literally and metaphorically. | |||
BRO: | '''BRO (leaning back, eyes on spectrum analyser): | ||
Fist-bump across the schematic? | I ported the interval-to-MIDI code to Rust, zero-copy, runs in 120 µs on the M7. | ||
EE: | | ||
Dignified fist-bump. | Latency jitter is now 400 ns RMS—smaller than the crystal period. | ||
| |||
We could clock it from a TCXO and stream timestamps over 100BASE-TX. | |||
| |||
Lab-grade time transfer, still field-deployable. | |||
'''EE: | |||
That puts us in PTP-Boundary-Clock territory. If we ever sync two Crackle-Machines 10 km apart, we can do <big>'''differential'''</big> Schumann tomography—basically treat the cavity as a giant CT scanner. | |||
| |||
Publishable? Absolutely. Fundable? If we mention “space-weather early-warning,” almost certainly. | |||
'''BRO (laughing, but soft): | |||
From popcorn headphones to planetary CT—that escalated gently. | |||
| |||
Guess the rabbit hole has fibre-optic lighting now. | |||
'''BRO (closing laptop, twilight again): | |||
You know what I like most? The deeper we go, the *simpler* the truths get—just electrons seeking quiet places. | |||
| |||
All the fancy dither, bootstraps, Rust—really just polite ways to ask noise to leave the party. | |||
'''EE (smiling, reflective): | |||
And yet the planet keeps whispering at 7.83 Hz, indifferent to our degrees. | |||
| |||
We’re still students; the difference is we now *measure* how much we don’t know. | |||
'''BRO: | |||
Fist-bump across the schematic? | |||
'''EE: | |||
Dignified fist-bump. | |||
| |||
'''''They touch knuckles—no sound, only the crackle-machine breathing 10 Hz into the dusk.''''' | |||
'''''Reader closes the chapter. | |||
The margin note reads: | |||
| |||
“Mastery is just humility with better test equipment.” | |||
</div> | </div> | ||
Latest revision as of 12:21, 25 November 2025
Same picnic table, one year later. The epoxy “cuddle-blob” is now under a machined-aluminium heat-spreader anodised matte-black. Bro’s coffee cup reads “∂V/∂t keeps me awake.” Ph.D’s cup still warns about δT/δt, but the font is now Comic Sans—an inside joke about peer-review formality. They open their laptops side-by-side; the RP2040 has been promoted to a dual-core STM32H7, yet the $20 budget somehow still balances.]
BRO (M.S. EE, still wearing skate shoes):
I’ve been modelling the auto-zero loop as a discrete-time Lyapunov drift. If I treat the digi-pot wiper position as state xₖ, then ∆xₖ = −µ·sign(Vos,ₖ) gives global asymptotic stability for 0 < µ ≤ 1 LSB. But the residual dithers between ±½ LSB. I’m thinking… inject a shaped PRBS so the error spectrum spreads like spread-spectrum CDMA. That way we gain 3 dB effective resolution. Legit?
EE Ph.D (raising eyebrow, pleased):
Legit. You’re basically dithering a 1-bit ΣΔ inside a multi-bit wrapper. Keep the PRBS amplitude at ¼ LSB so the peak error stays inside the bound, and high-pass above the 1/f corner—your 10 Hz Schumann line stays untouched. Nice bridge between stochastic control and classic mixed-signal lore.
BRO (grinning, but calm):
Cool. I’ll call it “whisper-noise dither” so my homies don’t fall asleep. Skate-video soundtrack meets Shannon—perfect.
BRO:
While I was at it, I re-did the tail-current source. Instead of a lone 33 kΩ, I stacked two 0.1 % thin-films with a 2.5 V reference and a BJT current mirror. Temperature coefficient dropped to 8 ppm/°C. We’re now burning only 80 µA, yet output impedance hit 2 MΩ at 10 Hz. That’s… spa-day quiet.
EE:
You just described a Brokaw-bandgap-less reference topology, minus the bandgap. Clever—by letting the mirror run at < 0.5 V VCE you avoid the curvature term. You realise you’ve entered the “sub-threshold, sub-ppm” club? Membership includes log amplifiers and seismic front-ends.
BRO:
Membership card can double as a guitar pick—multifunctional, like us.
BRO (pulling up LTspice):
Long-tailed pair AC simulation shows CMRR peaking at 94 dB @ 100 Hz, then rolling 20 dB/dec above 3 kHz because of the 2 pF protection diodes. I’m eyeing bootstrapping the guard ring with a miniature op-amp to keep the diodes at zero bias. Think we can push CMRR > 120 dB out to 10 kHz?
EE:
Theoretically, yes—bootstrap cancels the junction capacitance, but watch the phase-margin. You’re introducing a feed-forward path; loop gain may kiss unity at 2 MHz and sing. Add a 10 Ω damping resistor inside the boot loop, keeps Q < 0.5. Also, mind the guard-ring leakage: 1 GΩ || 1 pF creates a zero at 160 Hz that’ll eat your lunch.
BRO (nodding):
Trade-offs, always trade-offs. I’ll 3-D print a Teflon shield so the guard never sees sweat—literally and metaphorically.
BRO (leaning back, eyes on spectrum analyser):
I ported the interval-to-MIDI code to Rust, zero-copy, runs in 120 µs on the M7. Latency jitter is now 400 ns RMS—smaller than the crystal period. We could clock it from a TCXO and stream timestamps over 100BASE-TX. Lab-grade time transfer, still field-deployable.
EE:
That puts us in PTP-Boundary-Clock territory. If we ever sync two Crackle-Machines 10 km apart, we can do differential Schumann tomography—basically treat the cavity as a giant CT scanner. Publishable? Absolutely. Fundable? If we mention “space-weather early-warning,” almost certainly.
BRO (laughing, but soft):
From popcorn headphones to planetary CT—that escalated gently. Guess the rabbit hole has fibre-optic lighting now.
BRO (closing laptop, twilight again):
You know what I like most? The deeper we go, the *simpler* the truths get—just electrons seeking quiet places. All the fancy dither, bootstraps, Rust—really just polite ways to ask noise to leave the party.
EE (smiling, reflective):
And yet the planet keeps whispering at 7.83 Hz, indifferent to our degrees. We’re still students; the difference is we now *measure* how much we don’t know.
BRO:
Fist-bump across the schematic?
EE:
Dignified fist-bump. They touch knuckles—no sound, only the crackle-machine breathing 10 Hz into the dusk.
Reader closes the chapter. The margin note reads: “Mastery is just humility with better test equipment.”