Infrared IO "explorer"

ian · June 6, 2024, 11:34am

Oh, wow, nice, thank you. I haven’t checked in on them for years.

Last night I spent a few hours sifting through the available IR demodulators, trying to make sense of what is most useful. This is a better perspective: include a range. The questions I guess is to include a set on the PCB, or, use a header on the PCB and include a bag of demodulators?

Also great to know about the TSMP58000 IR detector. I specced the QSE159, which is similar, but I absolutely cannot source it in china. TSMP58000 is marked as obsolete, but knowing there is something common is a good start to finding an alternative.

Ordering a bunch of demodulators now with the intention to test. I’ve been leaning towards Vishay, but Everlight (also Taipei) has an impressive range too and I’ve had good luck with them in the past.

ian · June 6, 2024, 7:10pm

From AnalysIR

Model	Frequency	Description	Gain	Typical Application	Note
TSSP4038	38kHz	1 x IR receiver	Fixed Gain	Barrier	IRM-3638T?
TSMP58000	20 → 60kHz	1 x IR receiver	–	Learner	Obsolete, QSE159?
TSOP34436	36kHz	1 x IR receiver	AGC 4	RC5/6
TSOP34438	38kHz	1 x IR receiver	AGC 4	NEC
TSOP4840	40kHz	1 x IR receiver	AGC 2	SONY
TSOP2256	56kHz	1 x IR receiver	AGC 2	RCA	End of Life

Evidently Vishay did a huge part culling in 2020.

henrygab · June 6, 2024, 8:57pm

TSMP98000 (or 95000/96000/97000) appear to be the replacement for TSMP58000.

All the Vishay parts listed above show as having >1k in-stock at digikey, even though they all had variations listed in that PDF. Perhaps only certain packaging / variations of each model had stopped production?

I just want to ensure you don’t exclude the part line from consideration because certain variations are no longer made.

Here’s a good page for easily finding the Vishay IR Receivers:

henrygab · June 6, 2024, 10:38pm

For reference, here’s a mapping from remote protocol to frequency(ies) that the protocol uses, based on Vishay’s site:

Protocol	Carrier Frequencies
MCIR	`36`
Mitsubishi	`38`
NEC	`38`
Panasonic	`36`
r-map	`38`, `56`
r-step	`56`, `38`
RC-5	`36`
RC-6	`36`
RCMM	`36`, `38`
RECS-80 Code	`38`
Sharp	`38`
Sony 12 bit	`40`
Thomson RCA	`56`
XMP-1	`38`
XMP-2	`38`

ian · June 7, 2024, 11:21am

	Model	Frequency	Typical Application	Note	Everlight	Opti-fake
Vishay	TSSP4038 (3元)	38kHz	Barrier	Widely Available	IRM-3638T (0.78元)
Vishay	TSMP58000	20 → 60kHz	Learner	Obsolete		OS1200MW-A1 (3元)
Vishay	TSMP98000 (7元)	20 → 60kHz	Learner	New TSMP58000
Vishay	TSOP34436 (8元)	36kHz	RC5/6
Vishay	TSOP34438 (8元)	38kHz	NEC		IRM-3638J7(1.6元)	OS-883YM-MS (1.2元)
Vishay	TSOP4840 (5元)	40kHz	SONY
Vishay	TSOP2256 (5元)	56kHz	RCA	EOL

Table of findings, prices in RMB (元).

Thank you @henrygab for spotting the updated part number for the TSMP58000 (TSMP98000), and for suggesting to contact Chris at AnalysIR.

Recommendations

Chris is an expert in this area, and he was kind enough to make some recommendations:

IR LEDs:

TSAL6100 (10 degrees)- we use this the most or pair with the next if 2 LEDs are installed.
TSAL6200 (17degrees)
TSAL6400 (25 degrees)

All similar with just different emission angles.

Remotes use circa 300mA peak - we use jumper selectable resistors to allow varying the current up to 300-400ma.

IR Receivers Modulated output:

20-60 kHz - TSMP58000 (EOL but use the direct replacement that has a smaller bandwidth)

IR Receivers demodulated output:

38kHz (& 36kHz, 40kHz) - TSSP4038 - we find this works better for wider frequency range from a single receiver
38kHz (& 36kHz, 40kHz) - TSOP34428 (rejects some out-of-bounds signals)

The 2 options above will have you covered - but your pick…for a testing device(like BPV5) you don’t really need an AGC as the environment can be controlled…but for commercial deployment of end user devices you would need to match up the AGC & protocol very closely.

Thoughts

I was considering 3 or 4 LEDs, but now I’m backing off that. One narrow (TSAL6100) and one wider (TSAL6200,TSAL6400) LED in series on a single driver is board space and design friendly.

TSMP58000/TSMP98000 passes along the actual modulated raw signal if the modulation frequency is between 20kHz and 60kHz. This is useful for measuring the modulation frequency for cloning/copying (learning) a signal. The USB IR Toy used the QSE159 for this, but it doesn’t reject any frequencies so is a lot noisier.

The center spaced 38kHz demodulators will also handle 36kHz and 40kHz in low noise environments. This worked a treat on the USB IR Toy for years with no major incompatibilities. With Chris’ confirmation, I don’t think dedicated 36kHz and 40kHz demodulators are needed.

TSSP4038 is a ‘barrier’ type, meaning it activates when hit with a 38kHz modulated IR light for any length of time. Useful for odd protocols, but needs a lot of software processing.
TSOP34428 adds a tiny bit of logic that rejects signals that are too long or too short (rejecting buzz from florescent lamps, etc), so it is more likely to trigger on a valid signal.

56kHz is where I’m tempted to depart from Chris’ recommendation. While 38kHz is still 85-90% responsive to 36kHz and 40kHz signals (as shown in the chart), it really does nothing for 56kHz signals. Now, in my experience 56kHz remote signals are rare, but so are micro SIM cards and we included that on the SIM card adapter. If feasible, I’d like to include a 56kHz demodulator.

Sourcing

Vishay

We’re in this weird no man’s land for the Vishay parts. Just about everything except the 38kHz parts are part of a custom run. Vishay distributors have very little of anything (“None of that’s in China”), so we turn to the part ‘scalpers’ who dig up little bits of end runs from other manufacturers’ projects.

If we were doing 10K TVs, a Vishay distributor would work with Vishay to import MOQ, or make a part to spec that isn’t normally stocked.
If we were doing 1, 10 or even a one-off of 100, we could use the limited supply available through the scalpers and not worry about long term availability
We’re shooting for batches of 100 on a regular basis though, so we’re too small for a custom run but need more long term stability than the scalpers can supply.
TSSP4038 is the only part that is consistently available on the market in reliable quantities.

Everlight

In this situation I usually hunt for the “next best local option” that is either manufactured locally, or is widely available in China. In this case that is Everlight. Here’s the deal with Everlight:

Tons of distributors! This stuff is available everywhere cheap, where as Vishay is just scraps.
However! The distributors have no idea what any of it is. They can’t recommend equivalent part numbers, and they don’t have datasheets.
Everlight has a nice website, but the product pages don’t actually have any product listings! They list distributors world wide, but without contact info! there’s no QQ or WeChat link to talk to them. They defo don’t want to deal with us.
I can find some old (2004-2020) catalogs, but the catalog just lists part numbers without specs…
I’m left making my best guess based on parts available at szlcsc and the dodgy dated datasheets they have available for some parts.

My best guess is that IRM-3638T is equivalent of TSSP4038. IRM-3638J7 seems like a light-touch rejection demodulator like the TSOP34438. I bought samples of both and will compare them to the Vishay parts with a scope. The important part is these are widely available.

Opti-fake

Now, let’s talk about Opti-fake. We were looking for a sample of the TSMP98000 (new/updated TSMP58000). Opti-fake is a Vishay distributor and lists the part on 1688, so we make an inquiry. The response is basically “psst psst, add me on WeChat”, we’re now following a guy down a back alley.

The rep sends us datasheets for their “own version” TSMP98000 (OS1200MW-A1) and TSOP34438 ( OS-883YM-MS). We got samples of both, and will compare to Vishay parts with a scope.

Scalpers

Finally, that 56kHz demodulator. The scalpers can get TSOP2256. I don’t know how steady the supply will be, but I’m inclined to include it and hope we build a relationship over a few batches that gives us better access.

Next steps

I realize this is a wall of text, but it is also my notes for the eventual hardware design documentation

We tried to get 5 samples of each sensor. They will arrive to the Shenzhen office early next week, and make it to me a week after that. If there are enough to go around, I’m happy to mail some out to anyone who is interested in playing along.

ian · June 8, 2024, 2:42pm

Just a simple little board to test the various IR detectors in an actual PCB with closely placed components.

HEDoRAH · June 11, 2024, 6:02am

I don’t even know what i’d use it for, but old palm pilots and other devices used IrDA for UART communication. I was looking into it, because i was wondering if flipper zero could talk to some of these vintage devices, but it’s IR hardware is not fast enough. irda can go up to 115200 baud (in the MHz).

there was even a protocol for data transfer OBEX: https://www.irda.org/standards/pubs/OBEX13.pdf

I know a lot of PDAs, printers, old cameras etc had IrDA for serial comms right?

It would be right up buspirate’s alley to bridge this gap of allowing serial communication over what was a common protocol (albeit not anymore). i have a few old toys/devices in my ebay wishlist like the Sega IR 7000 communicator and Cybiko PDA. would be cool to toy around with these old devices that may have IrDA

For that matter we need an RS-232 plank dont we?

i’ve been meaning to buy this for f0: Flipper Zero - Flip'n Cereal RS232 by Rabbit-Labs™ | Rabbit-Labs
you just need an rs232 to RJ45 adapter

EDIT:
this board is pretty interesting for f0 as well:

It outputs from 3.5mm audio jack, and you use an adapter to DB9, RJ12, or RJ45

ian · June 11, 2024, 9:11am

I’d really like to get into IRDA. It’s pretty antique now I guess, but it still seems fun for con badges and stuff. I’ve looked into it a few times, and there are still IrDA RX/TX modules available. I have no idea where to get a test target - I guess an old palm pilot on ebay?

RS232 is a good idea. I’m working on that. Dual port you think? for loopback, sniffing, and man in the middle?

HEDoRAH · June 11, 2024, 12:42pm

its a pretty deep rabbit hole. apparently they even had IrDA “routers” that could provide network access up to 4Mbps. there were different protocols for these later implementations. from a post on reddit about the IrDA groups protocols:

SIR – 115200 baud seemed like the most common protocol of the 90s/00s.
MIR – 1Mbit is just HDLC bit-stuffing, and just changes the protocol to 1/4th wave (250ns active for “on” waves), rather than 3/16th wave on FIR. Seems like its still fundamentally a UART, just running at 1MBaud for the 1Mbit signal. Whether the diode is fast enough for this is another question though. Its probably too fast for ATTiny, but the other microcontrollers might be fast enough to interpret this?
FIR – 4Mbit, now stuffing more than one bit per timeslot. This is starting to look a touch math intensive.
VFIR – 16Mbit. Looks like some kind of weird encoding scheme (Run-length limited - Wikipedia)
UFIR – 96Mbit
GigaIR – 1Gigbit. I’m sure they just stole the specs from some fiber-optic cable.

I think we would mainly be interested in SIR. i didnt know about any of this other stuff until now.
here is what an IR access point looked like:

–

tamagotchis of that time could get items through IrDA:
https://www.reddit.com/r/tamagotchi/comments/xkts5r/guide_everything_you_might_want_to_know_and_more/

i believe its the Tamagotchi ID and tamagotchi P (which some are collectible so probably not the best cheap target for now)

–

I think this might be a good bet. The Palmpilot Tungsten E2. it definitely has IrDA but is modern enough to also have bluetooth 1.1 and some other more modern featurs, and this guy has hundreds of new-old stock for 40$

The predecessor may be slightly cheaper, the Tungsten E:

and before that it was the Palm m500 series:

all of which were in the IrDA era

–

PS: and yes, i didnt think about passive sniffing for rs232. thatd be sick

ian · June 11, 2024, 4:28pm

I’m going to look for knock off tamagotchis

HEDoRAH · June 11, 2024, 5:36pm

Dude, this is getting far off topic but there is a whole community of people who mod and hack tamagotchis. im in a discord with some of them and they use your old buspirate v3/v4 frequently

there are so many different versions throughout the years. I remember seeing a CCC talk about the tamagotchi GO that used these little hat add-ons for items. but they dumped roms and achieved arbitrary code execution to add their own items and stuff:

–

I know a lot of progress has been made on reversing the Pix, and 4u series is very popular. it goes really really deep and some are very collectible

. Heres some stuff on the IrDA of the Ps
https://www.reddit.com/r/tamagotchi/comments/xkts5r/guide_everything_you_might_want_to_know_and_more/
and someone wrote this tool to explore the P1: GitHub - jcrona/tamatool: A cross-platform Tamagotchi P1 explorer <–includes realtime ram editing?

the current generation (uni) has an ESP32:

heres an interesting article on how they are using AWS to manage millions of them:

other interesting things:
tama hive: http://tamahive.spritesserver.nl/
emulator actually showing the assembly instructions in real time: TamaTown test harness

–

The takeaway for me is we should aim to make BusPirate5 a tool that attracts this community, but im not sure ill be breaking into their projects anytime soon. there are people that are already putting a ton of energy into this.
i would like to at least get one of the old ones that may be a fun target to mess around.
Maybe the P1 with the IrDA?

EDIT:
Also, the m1x/On/meets generation had actual evolution through breeding. IE genes/DNA. This sounds really interesting to me, but not sure what hardware/peripherals that gen used.

here is a site showing a ton of hardware disassembly of the different versions:

ian · June 11, 2024, 5:53pm

There are a ton of clones and knock offs from 9rmb. But, I can’t tell if any use the irda. Will look a bit further.

henrygab · June 12, 2024, 2:42am

Another toy … from 1998 … was the “Furby”. Creepy things that initially talked a nonsense language, and over time switched to English. But… they could “learn” (switch faster) if they were near other Furbies … thanks to IR communications.

ian · June 13, 2024, 9:19am

Found a really helpful app note at AnalysIR from Sharp.

The spreadsheet uses datasheet values at each current to calculate how many IR LEDs we can use. The values all shift depending on the current, this is using the extremes. We should be good to go with 2 IR LEDs running at 100/200/300mA.

ian · June 14, 2024, 10:43am

This will probably be the final layout.

I received a “complete” set of IR demodulators from AnalysIR yesterday. The Shenzhen office shipped the “alternative” non-vishay parts yesterday. This should all arrive for testing/debugging in 8 days or so.

HEDoRAH · June 22, 2024, 1:38am

Didn’t know some of those e-Ink shelf displays use IR for programming:
http://furrtek.free.fr/index.php?a=esl
i cant seem to find a place to buy singles. and most i find use RFID hmm

ian · June 24, 2024, 10:40am

That’s a great read!

The change to French half way through was a surprise

HEDoRAH · June 24, 2024, 12:27pm

bahahahah keep you on your toes. do you speak french?

ian · June 25, 2024, 9:39am

I can get around in Mandarin and Germanic languages, but I’ve never studied romance languages. Google translate helped me out though

ian · July 5, 2024, 11:56am

20240705_133351
A box box arrived a few days ago with 5 different planks. The IR plank is one of them. I was super excited to finalize the IR LED transmit current setting resistors and soldered it this morning.

Too many things on the stove means meals get burnt - the transistor footprint is wrong. Sent out an updated board and will try again. If there’s any time I’ll try to dead bug the transistor in the right footprint.

There is a board to test the various manufacturer’s IR sensors. Digging into the datasheets though, I’m not really sure I want to mess with the cheap stuff. This requests a 0.1uF PLUS 10-47uF. That makes me nervous they’re compensating for something.

This one is also >47uF. That’s a lot. Vishay is 0.1uF-4.7uF

Ok, talked to the factory. They say “most customers don’t use the filter capacitor anymore to save cost, 103 (0.1uF) or 104 (0.01uF) is probably fine”. Which tells us little. So, I’ll try both with a 4.7uF cap and try them under a scope.