libera/#neo900/ Tuesday, 2019-08-13

houkimethere are two variations in this family of sensors.  TMD26721 like  TMD26711 has i2c voltage = vdd.  TMD26723 has i2c at fixed 1.8v00:31
houkimethe part used in neo900 schematics is  TMD26713 and it is powered from 2.7v00:31
houkime(ie a part with fixed 1.8v i2c from a previous family)00:33
houkimeso the logical move is to employ TMD2672300:33
DocScrutinizer05:nod:00:34
houkimehowever availability of TMD26723 seems to be quite low. For example, on Digikey while the part is "active" there are 0 available immediately.00:35
DocScrutinizer05our daily business for almost a year, just we rarely did discuss it publicly00:37
DocScrutinizer05there's a lot more work in design than an occasional visitor would expect00:38
houkimemost of neo900 parts are ok in this regard except for oldies that are needed for compat and that are sourced from original n900.00:40
houkimeI am rechecking them as i go through courtyards and layout.00:41
Joerg-Neo900lis302 is EOL afaik00:41
Joerg-Neo900a few more as well00:41
Joerg-Neo900bq2720000:42
Joerg-Neo900wpwrak  had a script to check *all* components00:42
houkimebq27200 will need to be sourced because it is a compatibility part((00:43
houkimewe have already discussed this one earlier00:44
Joerg-Neo900well, leste isn't the 100% compatibility approach Neo900 was designed for. So it's questionable if this reqzurement is worth anything at all, adter all00:50
Joerg-Neo900unless you use genuine maemo dremantle, it won't matter00:51
Joerg-Neo900fremantle even00:51
houkimeI checked ams website. Pretty much everything optical measuring which uses 1.8 i2c is out of stock.00:51
houkimehowever there was one more manufacturer that sdoes similar parts, called TAOS00:52
Joerg-Neo900We git some 3V3 I2C chips00:56
Joerg-Neo900thus a 3V3 bus, behind level shifter00:56
houkimeah, taos is just one of the ams's properties it seems.00:57
Joerg-Neo900those some other chips also were not available 1V800:57
Joerg-Neo900sucks since it's just a logic level issue00:59
Joerg-Neo900I2C is OC00:59
Joerg-Neo900worts case use two FETs to levelshift I2C just for this chip01:00
Joerg-Neo900BOM +2 :-/01:00
Joerg-Neo900avctually +4, you need 2 pullup R too01:01
Joerg-Neo900see hackerbus whitepaper for discussion of FET levelshifters01:01
houkime3.3v should be ok01:05
Joerg-Neo900houkime: if you vould finf a simplr singlewire 1/0 sensor without any programming, this would be great. But seems they are not fasgionabke anymore01:07
Joerg-Neo900fashiomable*01:07
Joerg-Neo900whatever, afk cu01:07
houkimeJoerg-Neo900, see ya.01:08
houkimeJoerg-Neo900, I couldn't find any details on proximity sensor for stylus detection in whitepapers.13:30
houkimeJoerg-Neo900, and in stuff also for that matter13:30
houkimeJoerg-Neo900,  what i want to know is allowed wavelength range, power consumption and calculations for aperture diameter in LOWER.13:31
houkimethe position of the stylus bay i can calculate from the wpwrak's scans myself.13:32
houkimeJoerg-Neo900, i have little idea now about bay wall reflection spectrum.13:34
DocScrutinizer51houkime: I'm sorry for missing details in whitepaper. Wavelength is irrelevant, Appertures (plural) size depends on sensor design, the idea I tried to13:34
houkimecan't one have only one aperture (via hanging sensor from the lower side of the UPPER?)13:35
DocScrutinizer51sketch yesterday is: sensor sends throuch one apperture and receives through a second. Both appertures get "shu" by stylus13:35
DocScrutinizer51shut13:36
houkimeok, so two apertures are for led and detector itself. got it13:36
DocScrutinizer51purpose of proto_v2 is (among other) to verify functionality of this design13:38
houkimeso what we are looking for here is very narrow angle for emitting LED, almost a laser-like thing, Otherwise it will illuminate LOWER and the integrated reflection power from LOWER will be much greater than from a tiny aperture.13:42
houkimeor rather, one should look at detection and LED emission cones, make sure that they intersect well at the depth of the bay13:55
houkimeand then make an artistic orifice based on these cones in LOWER13:55
houkimewhich might turn out to be either 2 separate orifices or one merged.13:56
houkimeTMD49033 looks nice. It is multi-wavelength and can detect color. Same footprint as other TMDs which we already have. 1.8v all across.15:02
houkimebut it doesn't have a datasheet. though knowing ams it is probably included in TMD49031 datasheet or sth15:03
houkimechecking15:03
Joerg-Neo900houkime: the sensor was planned to get mounted onto S2 looking *down*ward through apperture un PCB S1-S215:03
Joerg-Neo900the idea is to have receiver optically separated from LED when some object sits on surface S115:04
houkimeJoerg-Neo900, separation is not necessary since this kind of sensors can measure yield and compare it with reference15:04
houkimebasically it measures distance by comparing integrated reflection15:05
Joerg-Neo900go ahead as you think works best15:05
Joerg-Neo900just keep in mind this sensor is always-on and thus will have massive impact on battery standby time15:06
Joerg-Neo900it should have active supply current <1mA15:06
Joerg-Neo900ideally <<1mA15:06
Joerg-Neo900average15:07
Joerg-Neo900if it doesn't "out of the box", you might need to define a sort of "polling" scheme driven by CPU that enables sensor every 5s for 0.01s or whatever15:08
houkimeJoerg-Neo900, the number of pulses per measuring act is programmable and less pulses are needed for smaller distances15:08
houkimeJoerg-Neo900, they don't need polling because they have a configurable interrupt pin15:08
houkimeJoerg-Neo900, though to configure it one probably needs to calibrate it and then write needed values into i2c registers15:09
Joerg-Neo900as I said: >>if it doesn't "out of the box", you might need ...<15:09
houkimeJoerg-Neo900, well, it technically doesn't work out of the box)) it needs calibration ad experiments.15:10
Joerg-Neo900initializing is "OOTB"15:10
houkimeok15:10
Joerg-Neo900just needs to run autonomously after that, with a supply current average <<1mA15:10
Joerg-Neo900funny little detail that even works for N900: insering a lamba/4 antenna into stylus bay will massively improve FM TX ;-) the antenna is parallel to stylus and would get coupled to such l/415:14
Joerg-Neo900re stylus sensor, I also pondered mechanical switch (EEEWW!) and even capacitive "touch" sensor directly to a GPIO pin that senses statical charge movement in stylus plastic and change in capacitor value of the sensor pad to verify status inserted/removed15:16
houkimealas. In stock on Digikey but no datasheet on the ams website after all. Such a mess.15:25
houkimegoing to next candidate.15:26
DocScrutinizer51bydy, its a PITA15:28
DocScrutinizer51btdt even15:28
houkimedatasheet is present though on alldatasheet.com15:37
houkimeneed to check remaining other candidates regardless15:38
houkimethough all these sensors are so expensive... 5-6 dollars for a proximity thing. oh man...15:40
Joerg-Neo900prohibitive pricing15:45
houkimeit might be even better idea to have led and sensor separate at this point and sensor just have a programmable interrupt on too much/too little light.15:45
Joerg-Neo900rule of thumb: sourcing * 3, so almost 20$ for stylus sensor isn't a selling point15:46
Joerg-Neo900those reflexive lightbar sensors were always a PITA15:47
Joerg-Neo900prolly there are more sensor concepts where LED is separate component15:48
Joerg-Neo900I see soap dispensers with lightbar, sold for like 8 bucks15:49
houkimedue to small distance needed in neo900 led could be so weak that it can be just always on wherever phone is on.15:51
houkimeso no flashing circuitry or whatever.15:51
DocScrutinizer51duh, that would cause at  least 0.5 to 1mA15:51
DocScrutinizer51and sensing unmodulated light level is a nogo regarding noise rejection and reliability15:52
Joerg-Neo900if you really wanna go that route, you need a LED controlled by a CPU GPIO to pulse it with a duty cycle of <0.001% and detect the pulses on a A/D converted photodiode. Nasty but feasible16:04
Joerg-Neo900even I would opt against that stylus detection feature based on this concept. But in the end it's your design now16:05
Joerg-Neo900"even I" since I'm notorious for my featuritis16:08
claponthi everyone16:12
Joerg-Neo900hi16:14
houkimeSI1151-AB00-GM looks kinda nice compared to ams stuff.17:28
houkime1.8 bucks17:28
houkimeno integrated LED but LED driver instead17:29
houkimeinterrupt pin, all things programmable17:29
houkime1.8v all across (excluding LED which has separate power) possible17:29
houkimeand the LED can be basically just any LED under 5.5v17:30
houkimeboth visible and IR17:30
houkimemaximum sensitivity in blue area17:31
houkimeso it is bring_your_own_LED proximity sensor thing17:31
houkimehttps://www.silabs.com/documents/public/data-sheets/si115x-datasheet.pdf17:33
houkimeit also configurably limits LED current17:34
houkimethough maybe 1.8 bucks + LED is still a bit much.17:35
houkimeJoerg-Neo900, what do you think?17:35
houkimeshould look cool though if optical - you can probably look in the hole and see it blinking.17:36
houkimeif duty cycle and measurement pulses count permit that is)17:37
houkimeit has autonomous mode, so you just set up a timer via i2c and it measures every now and then.17:45
Joerg-Neo900sounds excellent, price is bearable. I suggest blue LED then, best efficiency, power LED from VBATT_SWITCHED18:32

Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!