#homecmos IRC log for Wednesday, 2011-07-27

azonenbergTaking a brief break from my MEMS work to try a slightly different microscale project01:17
azonenbergMounting of a 5-ball CSP chip (200um ball size, insanely small pitch) using my projection rig to make pads for it01:18
azonenbergi cant do a PCB with that kind of pitch using contact litho :P01:18
azonenbergi'd need like 75 micron traces01:18
kristianpaulholegrid looks defiant02:45
kristianpaulazonenberg: http://en.qi-hardware.com/wiki/File:Cat_gds_imported_in_toped.png02:49
kristianpaultoo dark :/02:52
kristianpaulis up to wolfspraul make it colored ;)02:52
bart416azonenberg, http://www.coultersmithing.com/forums/viewtopic.php?f=10&t=17218:49
Action: azonenberg_work looks up18:54
azonenberg_workBuilding a profilometer... doable?18:57
azonenberg_workThinking of piezo scanning18:58
azonenberg_workOoh, interesting18:59
azonenberg_workI plan to go and make a serious home SEM at some point18:59
azonenberg_workAnd wow, these guys made homebrew FESEMs?19:03
azonenberg_workI think i'd prefer a simple tungsten filament source19:04
azonenberg_workAlso, very very interesting19:06
azonenberg_workLooks like FIBs are possible to homebrew o_O19:06
azonenberg_worknow THAT would be awesome19:06
azonenberg_workDo imaging and alignment at low mag19:08
azonenberg_worklow voltage*19:08
azonenberg_workthen bump up the beam current to do cutting19:08
azonenberg_workimagine patterning metal layers with that :)19:08
bart416I like the idea of hijacking a vidicon tube to do it19:16
azonenberg_workMy design will be using a homebrew tungsten source and an everhart-thornley detector19:26
azonenberg_workrun through a fast ADC, an FPGA or MCU to do USB, then stream to a PC19:26
bart416For measurement instruments you really need fpgas sadly :(19:30
bart416I tried using microcontrollers19:30
bart416The only thing that gets the job done is something like an arm cortex a819:31
azonenberg_workIf not an FPGA i'd use a PIC3219:41
azonenberg_work80 mhz fully pipelined mips3219:41
nathan7Hi azo19:45
lekernelwhat's wrong with fpga's?19:48
bart416meh, not fast enough19:52
bart41680 MHz is too slow19:52
bart416To do DSP on a signal sampled at roughly 100 MSa/s (what you want for this sort of thing at the very least) you need at least 1 GHz to process it and throw it somewhere19:53
azonenberg_worklekernel: Nothing wrong with them, except i dont have it yet19:54
azonenberg_workbart416: For slow SEM scanning, actually, i was envisioning long integration times19:54
azonenberg_workso maybe tens to hundreds of KHz19:54
bart416You want over sampling to avoid noise19:55
bart416Serious over sampling19:55
bart416If you have enough data you can use a good algorithm to remove or at least reduce the noise19:55
bart416If you only have the required samples it becomes harder19:55
lekernelwhoa, this CRT-based SEM is awesome!!19:56
bart416Even averaging 10 images is better than anything else19:56
azonenberg_workbart416: I actually wanted to use a long sampling time on the ADC as a first-level filter20:03
azonenberg_workAnd then do a 5-10 sample line average20:03
bart416Increasing sample time often reduces precision of the ADC20:04
azonenberg_workI would be using a 10-bit 1 MHz one if i used the PIC3220:04
bart416Keep in mind the S/H mechanism is only reliable for a small timeframe20:04
bart416The less you relly on S/H the better20:04
azonenberg_workBuffer a single scanline in the PIC20:04
azonenberg_workAverage five or ten of them20:04
azonenberg_workthen DMA to the USB controller while repeating the process20:05
azonenberg_workAssuming i used 1 MHz, 1920 samples wide, i'd need 3840 bytes of temp storage20:06
azonenberg_work260 scanlines per second20:06
azonenberg_worktimes 1080 is 4.15 seconds per frame20:07
bart416Your system also does not allow you to reduce artifacts over several scanlines20:07
azonenberg_workI've gotten good results with line averaging on real SEMs20:07
azonenberg_workI could do lower resolution scanning and upsample initially for aiming and focusing, then high-res capture20:08
bart416anyway 'nite20:08
azonenberg_workNormally on a real SEM i do several lines averaged for maybe 30-60 seconds per frame when doing an image for analysis20:08
azonenberg_worki think 10 lines average, which actually puts it around the same sampling rate as i would be using20:08
azonenberg_worki.e. 1 mhz horizontal scan frequency20:08
bart416Yeah, but that's fine equipment with good noise prevention methods put in place20:09
bart416Good luck doing that at home20:09
azonenberg_workSo i'll get a noisier image20:09
bart416The 50 or 60 Hz noise alone will already be considerable20:09
azonenberg_workbut i dont need *higher* sampling frequencies than them20:09
azonenberg_workAnd yes, it will be20:09
azonenberg_workThere are a couple of options for that20:09
azonenberg_workOne is to try to eliminate it by sampling only during peaks20:10
azonenberg_work*zero crossings20:10
azonenberg_workphase lock onto the AC signal and turn off the sampling during the, say, 1/2 of the period that the signal is strong20:10
azonenberg_workThere are several implementations possible, easiest is probably to run the AC through a step-down transformer (possibly the same one that my PSU runs off) and sample that20:11
bart416anyway 'nite20:12
azonenberg_workrun into a comparator and check if abs(x) is above some K20:12
azonenberg_workif so, pause sampling20:12
azonenberg_workand ttyl20:12
--- Thu Jul 28 201100:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!