lekernel changed the topic of #m-labs to: Mixxeo, Migen, MiSoC & other M-Labs projects :: fka #milkymist :: Logs http://irclog.whitequark.org/m-labs
mumptai has quit [Quit: Verlassend]
<rjo> sb0: llvmlite: nice. i'll spin new travis scripts and packages.
<rjo> sb0: re cfl driver. one should not call that an "ion" trap, more of a dust trap..
<rjo> sb0: for paul traps the parameters (voltage, frequency, charge and mass) need to be in a certain range to provide stable trapping.
<rjo> do you know your parameters?
<rjo> i think for macroscopic particles with the usual charge (corn starch) frequencies will be in the 50-5000 Hz range
<rjo> and voltages in the ~kV. but just guessing.
<rjo> and the corn starch needs to be dry to not stick together.
<rjo> usually the particles charge with the same sign. and once these things are charged they will repell.
<sb0> rjo, the designs I've seen use 2-3kV and 50-60Hz. I think I'm pretty much in the same range for the voltages, but some orders of magnitude higher for the frequency
<rjo> higher?
<rjo> ah. you are higher.
<sb0> yeah, those CCFL inverters use a small resonant ferrite transformer, a wide guess for the frequency range would be 5kHz-100kHz
<rjo> yes. if you have a ~20kHz flyback, that will drop the trap strength by almost 6 orders of magnitude ;)
<sb0> ah. maybe that explains it
<sb0> how does the trap strength vary with the frequency, exactly?
<rjo> pseudopotential goes as (voltage/frequency)**2
<sb0> mh. let's see how far I can lower the frequency of this thing...
<sb0> why do you use MHz frequencies for the atoms, by the way?
<sb0> to prevent wide oscillations?
<rjo> Omega_rf >> omega_secular, and we want large omega_secular.
<rjo> your "ion" is 1e16 amu, charge is 1e5 e.
<sb0> ah yes, seems that would need at least 120 million volts at 20kHz (with 1 and 2 cm for r0/z0)
<sb0> even at 1kHz the voltages are ridiculous, and I don't think the ferrite transformer will go lower than that
<sb0> bah. the ccfl inverter is a bad idea.
<rjo> sb0: resonant tesla?
<sb0> it'll flashover
<sb0> seems the microwave oven transformer is the simplest thing you can use as power supply for dust traps
<sb0> oh, or maybe gas stove ignition transformers. more voltage, less current.
<GitHub166> [artiq] sbourdeauducq pushed 1 new commit to master: http://git.io/qZBlIA
<GitHub166> artiq/master 9165031 Sebastien Bourdeauducq: test/py2llvm: support 32-bit machines
sj_mackenzie has joined #m-labs
<GitHub151> [artiq] sbourdeauducq pushed 2 new commits to master: http://git.io/Os44Vw
<GitHub151> artiq/master b830dd5 Sebastien Bourdeauducq: test/py2llvm: pep8
<GitHub151> artiq/master bfe980d Sebastien Bourdeauducq: py2llvm: distinguish between llvmlite Module and ModuleRef
<rjo> even if the secondary is only a single layer?
<sb0> tesla coils have high frequencies too, and according to that page you need a bit less than a million volts at just 1kHz
<sb0> you can do a million-volt tesla coil, but that will flash over any reasonably sized trap. and huge traps need higher voltages. seems any frequency above than a couple dozen Hz is a no-go for dust traps.
<rjo> sb0: i was thinking about someting resonant at 60Hz. but the coils (and caps) might be large
<rjo> sb0: which llvm-or1k do we need for llvmlite? still the same old one? mine does not seem to have AsmParser/Parser.h
<sb0> no, latest one
<rjo> https://github.com/openrisc/llvm-or1k master (as in the install notes)?
<sb0> yes
<sb0> head
<rjo> oh so that part of llvmlite is merely a fancy templating system ;)
<rjo> with text-IR output.
<sb0> yes
<rjo> ah well. i guess that's ok.
<sb0> it's not slower than llvmpy afaict
<sb0> btw if you want to run stuff on the board, the new command is "artiq_run.py <module_name>", e.g. "examples$ ../frontend/artiq_run.py mandelbrot"
<sb0> it goes through the hardware database now
<rjo> 1mH and 1µF cap would give you an 80Hz. i believe you can get kV/µF caps from the trash.
<rjo> sb0: i gave the pdb and ddb a cursory look but have not played with it. maybe i get some time off from fighting with my trap..
<sb0> in sham shui po they sell a lot of vacuum tube audiophile stuff in the streets, including nice HV caps. a good part of it is grossly overpriced and with "features" such as gold-plated leads and hand-rolled dielectric material, but I can have a closer look
<rjo> ha. i would buy hand-rolled cigars any time. but not hand rolled caps!
<rjo> why is clang actually needed?
<sb0> clang is not needed
<rjo> why is it in the installation notes then?
<sb0> I just installed it because I try to build the misoc stuff with it, which doesn't work yet
<sb0> due to missing bits in the MC assembler
<rjo> still there in 3.5?
<sb0> re. installation notes, I'd rather end up skipping GCC/binutils, since LLVM is a necessary evil with ARTIQ
<rjo> agreed.
<sb0> afaik yes
<sb0> I heard about an Italian fork that fixes it, but the modifications are not merged yet, and I have not tried
<rjo> maintaining llvm-or1k looks like herding cats.
<sb0> seems to be the case with maintaining llvm-anything
<rjo> got a link to tha italian job (or fork).
<rjo> damn. llvm 3.5 requires g++ 4.7
<sb0> grmbl. why do inductor calculators online use imperial units.
<sb0> anyway, seems 1mH air core on PVC pipe is very feasible
<rjo> yes. the imperial units.... https://www.jlab.org/accel/inj_group/docs/Coaxial_Resonators.pdf these are the design estimates for our helical resonators.
<rjo> look out for micromicrofarads (sic!) and MC (MegaCycles).
<sb0> but practical LC circuits do not always work well when the inductance/capacitance are imbalanced, and I wonder about this case
<sb0> typical coils in this frequency range are more like hundreds of mH
<sb0> and use ferromagnetic materials, which are another rabbit hole...
<rjo> yeah. parasitics might make it inefficient and shift the frequency but efficiency is not a goal here.
<sb0> the other question is, how would you excite such a resonant coil?
<rjo> if you have a few km of coax around, you can just do a lambda/4 resonator. that will definitely work.
<rjo> sb0: autotransformer style.
<sb0> you typically only want a few turns to get a high transformation ratio. but the impedance will be very low.
<rjo> just tap it at a few turns.
<rjo> sb0: add a capacitor?
<sb0> ...which needs to resonate with the HV circuit. that gets complicated...
<sb0> and that cap will need to be large and cause even more L/C imbalance problems
<rjo> really? all you need is one knob to tune the coupling.
<sb0> I think traditional tesla coil designs solve the problem of low primary impedance by having high-current capacitor discharges into it
<sb0> through a spark gap
<sb0> sounds much easier to gut a microwave oven for a non-resonant low-frequency HV transformer.
<sb0> with a current-limiting resistor, it should be relatively safe
<rjo> well what you really want is a good old neon sign transformer.
<sb0> yes, but they are harder to come by than MOTs, and more expensive
<rjo> back in the old days....
<sb0> and their internal current limit is often above lethal levels anyway
<rjo> nice. looks like a mixture of katana making and how nikon shows of their lens making business.
<rjo> but yes. cool transformers!
<rjo> we need more mercury!
<rjo> travis running fine again.
<sb0> ha, you should visit Alek in Poland someday :)
<sb0> is it mercury? I couldn't say if it was mercury or oil in that tube
<sb0> from the italian llvm: "Initial support to little endian version of OR1K" hmm, why?!
aeris has quit [Quit: en a pas]
aeris has joined #m-labs
mumptai has joined #m-labs
sj_mackenzie has quit [Remote host closed the connection]
bhamilton has joined #m-labs
bhamilton has quit [Client Quit]
Alain has joined #m-labs
kristianpaul has quit [Ping timeout: 264 seconds]
kristianpaul has joined #m-labs
kristianpaul has joined #m-labs
sh[4]rm4 has joined #m-labs
sh4rm4 has quit [Ping timeout: 265 seconds]
sj_mackenzie has joined #m-labs
sj_mackenzie has quit [Remote host closed the connection]
Alain has quit [Remote host closed the connection]
mumptai has quit [Quit: Verlassend]
mumptai has joined #m-labs