balrog changed the topic of #homecmos to: Homebrew CMOS and MEMS foundry design | Wiki: https://github.com/homecmos/homecmos-wiki | Repositories: https://github.com/homecmos/ | Logs: http://en.qi-hardware.com/homecmos-logs/
kristianpaul has quit [Read error: Connection reset by peer]
kristianpaul has joined #homecmos
Degi_ has joined #homecmos
Degi has quit [Ping timeout: 256 seconds]
Degi_ is now known as Degi
Degi has quit [Ping timeout: 258 seconds]
Degi has joined #homecmos
kristianpaul has quit [Read error: Connection reset by peer]
kristianpaul has joined #homecmos
_whitelogger has joined #homecmos
kristianpaul has quit [Read error: Connection reset by peer]
kristianpaul has joined #homecmos
berndj has quit [Quit: ZNC - http://znc.in]
berndj has joined #homecmos
berndj has quit [Quit: ZNC - http://znc.in]
berndj has joined #homecmos
<fltrz> anyone ever tried to use the piezo rods from a piezo igniter from a lighter as a transducer? (context is for making low cost piezo inchworm motor)
<Degi> No, but you can get these relatively cheaply https://www.ebay.com/itm/4Pcs-lot-PZT-Piezoelectric-Ceramic-Actuator-Longitudinal-Polarization-Electronic/113732056951 (and they don't need as high voltages as piezo igniters)
<fltrz> Degi, do these alternate poling internally to trade high voltage to "high" currents?
<fltrz> thats really cheap indeed
<Degi> I think so
<Degi> They work at 90 V I think, for 3 µm
<fltrz> neat
<Degi> Datasheet says 3.7 µm at 90 V
<fltrz> anyone built an inchworm with these?
<Degi> Yes, that is like 3.25 $ per piece
<Degi> Idk, I just found them when I wanted a few piezo actuators.
<Degi> I wonder how to connect the ends
<Degi> Can you just epoxy two of them to a tiny metal cube?
<fltrz> the piezo rods in lighters are a lot cheaper, but this thing does deprioritize sourcing igniter piezos as a source material
<Degi> Hm, I think igniter piezos would have less distance per voltage
<fltrz> Degi, yes I believe they are typically either epoxied, or clamped
<Degi> And these are in a practical format
<Degi> I wonder what "Self locking force:1200-78000 N" means
<fltrz> Degi, right, I will definitely follow your suggestion, its just thinking longterm for reprap style open source micro EDM, once first iteration is out, people can design custom inchworms using igniter piezo's for cheapness
<Degi> Hm yes, I'm just saying that the driver electronics would be more complicated if you have to drive it with a few kV
<Degi> How many piezos do you need? 6 per inchworm, right? Do you need 4 inchworms for a 2D stage? For a total of 24 actuators / 78 $
<Degi> How would you do position sensing?
fltrz has quit [Ping timeout: 260 seconds]
fltrz has joined #homecmos
<fltrz> sorry, ebay / gmail typically cause some memory leak or perhaps just hoard in my firefox, 3GB ram
<fltrz> Degi, so last I'm aware you spoke was: <Degi> How would you do position sensing?
<Degi> Yes I didnt write anything after that
<fltrz> yeah I think 6 per inchworm, or perhaps 3 if pushing against air bearing
<Degi> Oh yes, maybe a spring on the other side.
<fltrz> I should probably look up some comprehensive piezo textbook for the self locking force
<fltrz> it sounds like a parameter of an actual motor rather than a single piezo element
<fltrz> perhaps 4 so I don't have to mount precisely halfway a piezo, nor worry about this middle piezo getting tilted when only one foot is pushing
<fltrz> (which would bring the lifted foot down)
john_cephalopoda has joined #homecmos
<john_cephalopoda> Hey
<fltrz> hi
<john_cephalopoda> Haven't been here for ages.
<fltrz> Degi, for position sensing I was planning on using interferometers
<Degi> How would you do that? Won't you need relatively high phase stability, like with a stabilized HeNe laser?
<fltrz> i think wavelength stability is enough?
<Degi> (My suggestion would be a laser modulated with a few GHz and then a photodiode. Then the photodiode signal and the laser input signal get mixed with a second frequency that is a few kHz off and by correlating the two mixed results you can get the relative phase and the precision is limited by the ADC sample rate / resolution and the accuracy by the clock source used (for example an OCXO with < 1 ppm error)
<Degi> )
<fltrz> yeah, do you think I could use GDD as mixer?
<Degi> GDD?
<fltrz> like a neon bulb?
<fltrz> miniature indicator lampp
<Degi> Hm, you wanna go real cheap?
<fltrz> :), its just that if self-reproducing micro EMD kicks off, you can quickly parallelize if cheap elements are used, like a box of 10 x 10 x 10 = 1000 devices
<fltrz> *EDM
<fltrz> material costs are low, but outsourcing costs are high
<Degi> You mean building like 1000 EDM devices to parallelize?
<fltrz> sure
<Degi> You can get mixers for like ~1.5 €, if you want to go cheaper, you can use some RF filters and a diode
<Degi> Though the laser diodes would be pricy (old SFP modules?)
<fltrz> I understand how theoretically to use a diode, but what are the RF filters for?
<fltrz> Degi, when Ben Krasnow did his self-interfering laser diodes demonstration, you could see that you didn't need too special electronics
<Degi> I think you only need a lowpass filter, if you put like 2 GHz and 2 GHz - 10 kHz into it and put a 15 kHz LPF on it, you should get a sine with 10 kHz out of it
<Degi> But how accurate are they?
<Degi> Like if your produced devices are all weirdly scaled and some parts bigger than others because of thermal effects etc. on the wavelength of the LDs, that wont be so nice...
<fltrz> sometimes I think of using optical indexing or alternatively the comb like capacitive sensing like in micrometers, but micromachined
<Degi> optical indexing?
<fltrz> like some kind of grey code engraved on the moving stage
<Degi> Ah yes
<Degi> Maybe just a piece of plastic with lines on it and then quadrature encoding on that
<fltrz> yeah, something like that perhaps
<Degi> I think some inkjet printers use that
<Degi> That might be cheaper to produce than the RF modulated laser thingie
<fltrz> still I think the self-interfering diode might work sufficiently well, but its still outsourcing of course
<fltrz> if it can work by engraving or similar, that its just own work, I'd prefer that for scaling cheaply
<Degi> outsourcing?
<fltrz> like with working EDM one can engrave micro patterns for gray code or whatever, which just costs time and electricity, vs, buying lasers or RF filters etc
<Degi> I wonder if you could use a diffraction grating with big lines (100 lines/mm?) to stamp encoders out of
<fltrz> because x 1000
<Degi> Hmh
<fltrz> ooh, that should work
<Degi> It should be enough to engrave one as a stamp
<fltrz> perhaps just make the diffraction gratings with analog film
<Degi> Somebody on youtube made a diffraction grating chocolate by pouring chocolate onto a diffraction grating
<fltrz> that was Ben Krasnow again
<fltrz> from Applied Science YT channel
<fltrz> I guess best approach is still start with functional device by buying more traditional components, then use the device to replace the expensive parts one by one
<fltrz> best to start making 2 in parallel so you can always use one working to improve or emcheapen the other
<john_cephalopoda> How would you read out the encoder with such a dense encoder strip?
<Degi> Two photodiodes and maybe a slit in front of them?
<Degi> Usually these strips have some kinda encoder which is 2 combined PDs
<john_cephalopoda> Building that sounds complicated...
<john_cephalopoda> If you were just working with one axis, using a scanner directly (with some beam device mounted on the moving part) could give you relatively good accuracy…
<Degi> scanner?
<Degi> And how accurate would EDM be here?
<fltrz> I was also thinking for say a horizontal moving stage, just having vertical stripes, with the inter stripe distance sinusoidally modulated, then shine a laser perpendicularly, and inspect the 1st diffraction order
<Degi> How much precision do we need and what is the budget
<Degi> And why would you modulate it sinusoidally, wont that be hard to fabricate
<fltrz> since we bootstrap with more conventional components the bootstrap budget is allowed to be bigger, its just for ultimate scaling that I'm considering alternatives, in case some of them provide a shortcut
<Degi> What precision do we need and how precise is EDM anyway?
<fltrz> its micro EDM
<fltrz> so on the order of gap width
<fltrz> say a micron or perhaps a few
<Degi> So like 1-10 µm
<fltrz> sure
<fltrz> I mean i've seen such resolution through a conventional microscope
<Degi> Maybe capacitive sensors are accurate enough?
<fltrz> I also think so
<fltrz> I guess optical is still desirable for fast response rate?
<Degi> Eh, how fast?
<fltrz> harder to say, you don't want the tool electrode to bump into the workpiece creating a short or even welding itself to the workpiece
<fltrz> I think micro EDM will be crucial for home cmos
<Degi> Hm why?
<fltrz> if photolithography is Santa Claus, then micro EDM and similar are like Santa's little helpers, they get less attention but all the mechanical jigs, wirebonders, ... need to be precision machined
<Degi> Oh right, the wire bonders
<Degi> Hmh
<Degi> What about flip chip and then just soldering the IC to the PCB?
<fltrz> I think its also just desirable in general and conductive to improvement to have micro EDM equivalent to 3d printers RepRap
<fltrz> also micro manipulators etc to test circuits without necessarily soldering to PCB
<Degi> Hm yes
<fltrz> once we get open source micro EDM, there's going to be plenty of non homecmos users using it for their own goals and purposes, so that we can freeride on their improvements and emcheapening
<fltrz> this should help homecmos massively
<fltrz> lets help users of micro EDM so that their ecosystem automatically helps us
<fltrz> if every student, PhD, ... who uses micro EDM for microfabrication is exposed to a cheap micro EDM at home, they will participate in it, some students drop out but are still interested, some PhD's don't get faculty position, and they have no access to the fantastic toys, ...
<fltrz> the threat of a functioning ecosystem for micro EDM will force either academia to hire all of them, or industry to increase their wages so the ideologues don't sit at home building the ecosystem
<fltrz> functioning *open* ecosystem...
<fltrz> I think capacitive might even have faster response, because cheap photodiodes have to large area for high bandwidth
<Degi> Hm, it all depends on the price
<fltrz> what kind of printer is that? 42 inch encoder strip?
<Degi> What is the max working area supposed to be
<Degi> Printer?
<Degi> Its an encoder strip for inkjet printers I think
<fltrz> dimensions will probably be on the order of 5 to 10 cm linearly (perhaps unnecessarily big
<Degi> I wonder if we could use something like the TCD1707
<fltrz> but 42 inch, sounds like a huge printer
<Degi> Linear CCD with 7450 pixels and 4.7 µm pitch
<Degi> I wondered that too
<Degi> Maybe its rolledu p?
<Degi> Thats like more than a meter...
<Degi> Maybe for A0 sheets
<fltrz> yeah definitely huge, do we know the encoding on the strip, and if its capacitive or something else?
<fltrz> the linear CCD's have a linerate of 1kHz, a bit too low I fear
<Degi> The encoding strips are optical
<Degi> I think its just stripes
<fltrz> most micro EDM setups appear to use coarse positioning with precision leadscrew and backlash free nut, probably lapped against each other or smth
<fltrz> then piezos for fast response
<Degi> What do we need the fast response for?
<Degi> If we go slow enough, it shouldnt bump into the workpiece...
<fltrz> to bring the electrode back in case of short
<Degi> Cant you just sense the current / resistance?
<fltrz> can still happen with the swarf in the gap
<fltrz> you might be able to detect the short but if you just wait the molten metal might cool down and tool and electrode are welded sometimes
<Degi> As soon as the current exceeds a limit, move the tip away?
<fltrz> yeah essentially
<fltrz> but you are absolutely right that slowing down helps, but you also want to finish at some point
<fltrz> typically you machine at higher spark energies and faster motions, then do the finishing at lower energies lower speeds
<Degi> Hm
<Degi> Maybe just increase the current
<fltrz> once its welded I think the lack of resistance between electrode and workpiece would mean that the whole electrode is the resistor instead of the plasma in the gap
<Degi> Hm yes it might explode the electrode instead
<Degi> But you can prevent the plasma from condensing xD
<fltrz> yes, so keep the arc and then slowly move away
<Degi> Maybe have an oscillatory motion on the work tip to prevent welding
<Degi> Hmh, its not really an arc, is it?
<fltrz> I guess its fine during the rough machining, but if it happens during the finishing you are f*cked
<fltrz> Degi, the shorting is plasma arc
<Degi> Hm yes but its not continuous, more of a spark I mean
<fltrz> suppose we are machining slower than we are advancing, then the gap for each spark will get narrower and narrower, which means the energy in the initial spark (which is very selective for closest points between tool and workpiece) is lower and more of the energy will be in the subsequent arcing which is not as selective for shortest distance (i.e if the spark already eroded the bump, now we are just melting surrounding metal))
<fltrz> I'm sorry for oversimplifying earlier
<fltrz> you are asking the right questions
<Degi> Hm, so you want low spark energy for precision
<fltrz> right
<fltrz> a typical order of magnitude is micro joule
<Degi> Maybe the distance between tool and workpiece can be measured from the waveforms
<fltrz> Degi, thing is both tool and workpiecee erode
<fltrz> but tool slower than workpiece usually
<Degi> Hopefully
<Degi> I think by measuring the impedance of the spark you can measure its length, longer spark should have a higher impedance
<fltrz> the tool is charged negatively and the workpiece positively
<fltrz> Degi, theres a paper about measuring the spark/arc impedance during the pulses, so yes its a useful process monitor to decide if we should slow down, advance, or quickly retract and get back to mix the swarf away
<fltrz> theres a bunch of parameters that have been tried to monitor the process conditions for controlling CNC etc
<fltrz> another paper used RF spectrometer, so perhaps RTL dongle or similar could be used
<Degi> Maybe a photodiode on the tip to measure the light
<fltrz> I'm under the impression that the process parameters are mostly a solved problem, now its mostly bad luck, like a bal of molten metal flying in the wrong direction, or too high a density, so you need better flushing, or tool motion to circulate the dielectric to get rid of all the metal swarf
<fltrz> I'm more concerned with the piezo steppers and position readout
<fltrz> thanks for that piezo ebay link
<Degi> np
<fltrz> if you have ideas regarding inchworm motor or position readout, or motion stages (I sometimes think I should go for parallel motion platforms instead of serial XYZ
<Degi> Maybe as an encoder you can have a wiggly line on a transparent sheet and then a line CCD sense it heh (but low refresh rate)
<Degi> parallel / serial?
<fltrz> like delta printer vs XYZ
<Degi> Are they sufficiently precise?
<fltrz> I don't mean using an actual 3d printer, but the type of motion platform
<Degi> Hm yes, can you get that precise enough?
<fltrz> like once could also use delta platform
<fltrz> iirc positioning errors accumulate with serial chaining, but not with parallel actuation
<Degi> Hmh
<fltrz> but I feel a bit unsure
<Degi> Maybe a linear resistive encoder can be used for position sensing
<fltrz> huh
<fltrz> didn't even cross my mind
<Degi> Im not sure how precise they can get
<Degi> And how they change with usage, hysteresis etc
<fltrz> I wonder if the friction could be a problem, perhaps have a long channel / bath with conductive salty water and a moving electrode in it?
<Degi> Dont they have some kinda electrode drop or os
<Degi> And waves
<fltrz> yeah, bad idea
<fltrz> oooh I know, I forgot the idea
<Degi> OK maybe capacitive thingie os good actually
<fltrz> use pattern engraved on metal, and use a static STM
<Degi> Not sure if you can do that good enough to not crash the hdad
<fltrz> Degi, well have 2 coaxial STM tips separated a bit wider than thickness and use the difference in current to keep it "centralized" on flat piece of "encoder" then an extra STM tip to read the actual dimples
<fltrz> so the 2 coaxial are reading top and bottom of the encoder
<fltrz> the 2 tips are solidly connected
<fltrz> I should draw a picture
<fltrz> consider a horizontal plate, and along and over and under one edge a C shaped piece of metal, on a vertical actuator that prevents the heads from crashing, and on both bottom and top there is an stm tip, oen upright the other upside down
<Degi> Huh
<Degi> Idk the printer strip seems easier / less tip-crashing prone...
<fltrz> if the tunnel current is equal its central, if unequal actuate to keep the tunnel current equal
<fltrz> right
<Degi> Hm yes
<Degi> But what about thermal expansion etc
<fltrz> yeah the metal plate would have to be same material as bracket and STM tips
kristianpaul has quit [Read error: Connection reset by peer]
<Degi> And be kept at the same temperature
<fltrz> but it would be cheap since reproducible
<Degi> We could make them from quartz lol
<fltrz> lol
<Degi> A strip like from printers should be easy to produce with a micromachined wheel or plate
kristianpaul has joined #homecmos
<fltrz> yeah, I wonder what the encoding and readout setup looks like?
<fltrz> do they use similar strip in scanners
<fltrz> ?
<Degi> Probably
<Degi> I think the strip is like a grating but with less resolution, like just black and transparent stripes
<Degi> The readout is two photodiodes offset by half the pitch on the strip
<fltrz> what you said about sinusoidal variation: dont just have a reflective line, but let the area under curve be reflective
<fltrz> then you don't care about position but intensity
<Degi> Hmh
<fltrz> intensity is then immediately cosine or sine
<fltrz> 2 such curves (since and cosine) and you have position
<Degi> I think you can measure the density of the two photodiodes, maybe they don't give perfect sin/cos but still you get I/Q signals
<fltrz> we can probably callibrate imperfections in software
<fltrz> liek it knows 14.7 pi radians since start is actually off by x radians
<fltrz> with gray codes its not sin/cos but a linear intensity change
<Degi> Hm, I think a strip with just black / transparent would be easier to construct and still give exact position if the photodiodes are read with an ADC
<Degi> (Assuming the photodiodes arent perfectly on the track but like a few dozen µm away from it)
<fltrz> right, but you'd use photographic film? or off the shelf strip?
<Degi> Probably off the shelf strip
<fltrz> did you ever see the self-interference laser diode YT clip?
kristianpaul has quit [Ping timeout: 256 seconds]
<Degi> Hm yes
<Degi> Though it doesnt have IQ output
<Degi> And probably a lot of PVT variation
<fltrz> well, they no longer make the laser diodes with built in monitor PD
<fltrz> so you have to build interferometer anyway, then you have both I,Q
<Degi> They dont?
<fltrz> I was also surprised but he said they don't and had to find new old stock
kristianpaul has joined #homecmos
<fltrz> I think reliability and durability increased to the point they could have product engineers blindly trust datasheet specs
<Degi> And you can see the PVT variations
<fltrz> thanks!
<fltrz> I might try the selfinterference thing anyway then
<Degi> See page 2 the graph of wavelength and temperature
<fltrz> yeah
<fltrz> self interference depends on this effect, it actively changes wavelength by modulating laser power
<fltrz> I think modulating distance between interference fringes on film is not that hard: make a sinusoidal groove in something, insert a piece of photographic film, use 2 expanded parallel plane wave beams from a beamsplitter so that the planes of constructive interference intersects the photographic film at varying angles
<fltrz> but yeah thats not really helpful because for readout you need a lens and a position sensor, although theres analog position sensors...
<fltrz> position sensing photodiodes, with either a shade baffle or multiple electrodes on the same PIN photodiode
<fltrz> probably a pain in the ass to make sure the film is bent precisely like a sinusoid though...
<Degi> Hm what would the downsides of the printer strip thingieb e
<fltrz> wait, cant we just read out the phase of normal constant spatial frequency grating with lens to fourier plane? have a or pinhole at the right spatial frequency, then interfere this light with light directly from the laser, should give phase of the spatial frequency
<fltrz> Degi, probably none
<Degi> Idk needs a laser
<fltrz> Degi, do you know where I can find high resolution scans or pictures of what the printer strips contain?
<Degi> The strip idea only needs two PDs and a LED and some kind of housing
<Degi> Hm no idea
<Degi> Afaik they're just on/off modulated, basically black strips with constant spacing
<Degi> At least thats what they look like with bare eye
<fltrz> oh really, wow
<fltrz> that would be very simple then
<fltrz> how often have you seen one up close?
<Degi> I have one here
<fltrz> pehaps theres diferent periods on different heights?
<Degi> I'd assume they're all similar, just quadrature encoders
<Degi> heights?
<fltrz> ir just simple vertical lines?
<Degi> I think so
<fltrz> Degi, like multi track signals
<Degi> Why would it be different, they only need to sense relative position and direction
<fltrz> Degi, so theres no quadrature then even?
<Degi> No, I think its single track and has two photodiodes offset by (n+0.5)*line distance
<Degi> It has quadrature from two photodiodes
<fltrz> oh quadrature is by placing photodiode?
<Degi> yes
<fltrz> yes, I can see that working
<fltrz> and eventually micromachine that from say aluminum foil
<fltrz> for scaling up
<Degi> I'd suggest micromachining a roller and then hot-pressing that into some tape or foil
<fltrz> like plastic tape?
<Degi> yes
<Degi> And then put some ink on the ridges
<Degi> Or maybe not, might work without
<fltrz> so I assume you'd use those of the shelf LD/PD encoder combos?
<Degi> Only ordinary LED
<fltrz> or is it LED/PD
<Degi> Yes
<fltrz> yes that will probably be it
<fltrz> as I can imagine machining physical tabs / slots, and as I can imagine cheap poor mans photodiode from cuprous oxide
<Degi> Why would you need to go that low in cost
<Degi> Pretty sure the oxide costs money to make
<fltrz> nah its just ordinary copper heated above a certain temperature in air
<fltrz> sppontaneously oxiidizes
<fltrz> sorry for my sticky keys btw
<Degi> I mean photodiodes cost like 30-50 cents
john_cephalopoda has left #homecmos [""A tree works like your brain." ~ The Org Manual"]
<Degi> Making it yourself always takes time, compared to buying COTS stuff
<fltrz> of course it will take time, 1000 x 30c is $300
<Degi> No then you get discount
<Degi> Like 80 € for 1000
<fltrz> its ridiculously easy to make cuprous oxide photodiode, we don't need high performance
<fltrz> and it will encourage others to optimize their manufacture
<Degi> How do you wanna control the thingie
<Degi> You still need ICs etc for that
<fltrz> yes
<fltrz> what we can't make yet we can't make yet
<fltrz> so CNC control, the LED, the computer, ... will be commercial devices
<Degi> Well we could make LEDs
<fltrz> I'd be very interested
<fltrz> btw with micro EDM you can easily make optical quality mirrors
<Degi> I mean you just need a semiconductor junction
<fltrz> with short polishing step at the end
<fltrz> hmm, but how prevent thermal emission instead of optical?
<fltrz> or you talking one time use LED :)
<Degi> Well it surely wont be efficien
<fltrz> do you happen to have references for DIY LEDs?
<fltrz> in physics bachelor there was different labs for bach project, one of them was making LEDs but I did other project
<Degi> Hm, we have a project once in our physics bachelor
<Degi> Hmh, I have no references for DIY LED / PD
<fltrz> ^ cuprous oxide photodiode
<Degi> Hmh
<Degi> Cant you just make an incandescent lamp
<Degi> Or a neon lamp
<fltrz> quite hard, need to evacuate, have good hermetic seal
<fltrz> but yeah, perhaps I'm getting unnecessarily getting carried away to much by self fabrication, first make 2 functioning prototypes, then look at improvements and substitutions,
<fltrz> how many stripes per inch are the encoders would you guess?
<fltrz> the strip you have I meant
<fltrz> or per cm ;)
<Degi> I think the one I have here is 3-10 per mm
<Degi> I think its 5 or 10 per mm
<Degi> Hm, maybe you can use an analog camera to photograph a sheet with stripes on it to make a stripe foil lol
<fltrz> yeah, was thinking the same, but with a lense and a monitor
<fltrz> I do think I might use some other photodiode than the one in the quadrature sensors (I think its often phototransistors) for higher bandwidth
<Degi> Hm yes
<Degi> With like 50 ohm lines xD
<Degi> Not sure how green cadmium and selenium are lol
<Degi> Sounds complicated compared to COTS photodiodes
<Degi> At that point you could probably make a silicon PDF
<Degi> *PD
<fltrz> the cuprous oxide photodiode does not use cadmium and selenium
<Degi> Hm yes
kristianpaul has quit [Read error: Connection reset by peer]
<fltrz> hm if its just vertical lines, why do they make the strip of lines so wide?
<fltrz> they might angle the lines for quadrature?
<Degi> I think they just offset the PDs, they dont look particularly angled
<Degi> Wide?
kristianpaul has joined #homecmos
<Degi> I think because then it needs to have the PDs less precise
<Degi> I mean its for a printerl ol
<fltrz> but it seems so much wider than the spacing between the lines?
<Degi> The what?
<Degi> Photodiode?
<fltrz> sorry, the strip seems so much wider than the distance between the grid lines
<Degi> You mean the grid it prints out?
<fltrz> no, I mean the strip is like a long transparency with lines right? why are the lines so long (width of the strip) compared to the spacing between the lines
<Degi> Oh
<Degi> I think cause the strip is that wide for mechanical reasons? Also I think the LEDs are on top of each other with maybe 1-2 mm distance and then slightly shifted sideways by I think half a linewidth / quarter period
<fltrz> suppose that the 2 photodiodes are in fact positioned vertically then the quad would result in a quarter shift say 1/4th from the top and 3/4th from the top, that would be an imperceptible angle to naked eye
<Degi> Hm yyes
<fltrz> so both seem equally feasible, but with slanted lines you could calibrate say with a screw the top photodiode up and down position until its nice quadrature
<Degi> Hm yes and not as sensitive as sideways motion
<fltrz> while horizontal calibration would require very precise screw control
<fltrz> right
<fltrz> its amazing what we can deduce from peculiar dimensions lol
<fltrz> its like a lesson in learning to build this
<fltrz> note that on the ebay picture the lines actually go right up till the edge, so they might start out with perfectly perpendicular lines, and cut the strips at the desired angle from the larger roll?
<Degi> Hm could be
<Degi> I presume they're cut from a big roll
<Degi> Like toilet paper
<fltrz> :)
<fltrz> becausee of the slant they lose some film at the edges
<fltrz> but this way the manufacturer of encoder TP doesn't need to care how sensitive and at what PD spacing the product designer wants for sensitivity of calibration, and produce just a few types of rolls
<fltrz> yes this will probably be the method
<fltrz> (for me to use in the micro EDM)
<fltrz> I was also just thinking, what about temperature stabilization (a bit above temperature) and optionally a diffraction grating for selecting a precise wavelength?
<fltrz> *above ambient temperature
<fltrz> for the interferometer option
<Degi> Hmh I dont know
<Degi> Maybe if you make the grating reflect part of the beam back into the laser that could stabilize it
<fltrz> longer wavelength laser (telecom IR?) would be easier I think
<Degi> Idk
<Degi> I mean you probably need a high coherence length
<fltrz> yes! was thinking the same
<Degi> Like with stabilized HeNe
<Degi> The grating reflecting approach might be worth a try xD
<fltrz> wait wait, theres this paper about coherence length of random red laser
<Degi> Lol
<Degi> Nice
<fltrz> where scientists in 90s were surprised after hearing some rumor about the coherence length of cheapo red lasers, tried building simple interferometer with pane of glass as beamsplitter and observed the fringes easily
<fltrz> like coherence length of meter(s)
<fltrz> let me find it if it interests you
<Degi> yes
<fltrz> closest I find is of course http://www.repairfaq.org/sam/laserdio.htm#diobcc4
<fltrz> but it was an actual paper from the 80's or 90's , I think a south american university
<fltrz> they literally made a cheap ass interferometer with glass pane and red laser diode, and couldn't believe its coherence length
<fltrz> for first surface mirrors, I typically use platters from broken HDD's
<Degi> Lol yes, had that idea too
<fltrz> I always worry about cutting it, so better outside, because I'm not sure of the materials used. probably just paranoia similar to that stuff you can find in a microwave
<fltrz> so I don't know and have zero indication its a poisonous metal, but it just popped in my head someday that I can't know
<Degi> Welll its mostly aluminium and a thin layer of potentially toxic stuff
<Degi> Whats inside a microwave
<fltrz> great
<Degi> Are those optics just soldered onto a PCB
<fltrz> they are on heating elements I think, considering the electrical paths going to them, for temperature stabiilization
<Degi> Hm yes
<fltrz> if you go above ambient you are in control of temp, and not the environment (oversimplified due to currents), just box it in
<Degi> It has a TEC though
<fltrz> oh yeah could be cooling too
<fltrz> we should contact this dude just for tips and tricks
<fltrz> or advice
<fltrz> whats the original html pointing to the image?
<fltrz> in the paper above they concluded the coherenence length was at least a meter since the fringe visibility did not change over the 50cm path length difference their diy michelson interferometer permitted
<fltrz> so that seems more than long enough for micro EDM
<fltrz> I think I'll go red laser diode way after all
<fltrz> I mean they use random float glass as a beamsplitter
<fltrz> if you at least use a microscope slide or smth, you'd be fine I guess
<Degi> the image?
<Degi> Its from the page you sent on repairfaq
<fltrz> was also thinking of taking part of the laser beam and using a static michelson morley interferometer without changing relative path length and use changes in I,Q to regulate power hence temperature to put wavelength back and prevent the crystal from modehopping
<fltrz> oh
<Degi> Oh yes a static interferometer to keep it constant by changing the voltage
<Degi> That sounds good
<Degi> I think you can get IQ by building an interferometer and placing two PDs slightly besides eachother?
<fltrz> imagine the first 4 mirrors in LHS are the only ones used, and upper left one and lower right one being beamsplitters, ignore right half, ignore the cells in the path
<fltrz> then laser splits top left, each goes to full mirror, then they join on bottom right beamsplitter, and place one PD below and one PD to right of beamsplitter
<fltrz> I think that gives IQ
<fltrz> but yeah possibly slight offset also, would have to calculate that, because not familiar with such a slant setup
<fltrz> but could be genius trick
<fltrz> so thanks for the suggestion of slight offsetting
<fltrz> anyway theres bound to be simple IQ interferometer setup if I look it up
<Degi> I mean just radially offset the laser diodes
<Degi> One at like the center and the other a few mm from it
<Degi> Then when the fringes move outward / inward you have IQ
<fltrz> yeah thats how I interpreted your comment, like when you project the interferometer image on a screen you see the central disk, and then fringes radially
<fltrz> I think a lot of these high outsourced prices are for mental / conceptual simplicity of the experiment / product designer
<fltrz> I don't fault the experimenter, because you really don't want device imperfections polluting science
<fltrz> and I don't mind the product designer for the very first prototypes before cutting down and simplifying for price
<fltrz> but there should be like a simple handbook to learn these pragmatic tricks
<fltrz> so that low budget science / engineering can happen
<fltrz> I think quarter wave plate or similar could be used for IQ in orthogonal setups
<fltrz> but yeah, in static setup who cares if one PD is to the side
<fltrz> I think control would be best in FPGA
<fltrz> back in university I had to do a lab for a plasma generator with a liquid electrode (salty water). this liquid electrode doesn't wear
<fltrz> so Im thinking of using super fine pipette tips with salty water as the tool electrode
<fltrz> if furthermore I set the pressure in the chamber such that the water is a little below boiling point, then every spark / short arc will also heat the salty water which should expands into a large volume of steam (miniature steam explosion), this should blow away the swarf from the electrode / tool gap, and allow higher pulse repetition rates
<fltrz> so at low pressure
<fltrz> I didn't see that yet, so its not a proven method
<fltrz> this would also quench the heat affected zone (HAZ), not sure if that is good or bad, probably depends on the metal and the desired surface finish
<fltrz> ^low cost wavelength stabilizing interferommeter for diode lasers
<fltrz> that is exactly the paper I need to read, with great references :)