New 105D. replacing Nanjg-105c @ FT? (no stars!)

It seems that you are seriously misunderstanding something or multiple things.

  • None of the drivers being discussed here would ever do what you want. If you purchased a 105c for accurate 1mA output you made a mistake. This is not the thread you are looking for.
  • What driver did you attempt to use first? 20 ohms suggests an extremely low sense voltage!
  • I question the need for an accurate 1mA supply for an LED What could you possibly be doing? It seems odd is what I’m saying! Why won’t a very rough, approximate, 1mA be good enough?
  • Is an accurate 1mA really that difficult? There are many multi-channel LED driver ICs intended for that range. Surely some of them must have tight enough regulation for your purpose?
  • Also take a look at the first link under WIP in my sig, “17mm double-sided & 20mm+ single sided 4Amp+ linear driver [A17LDX]”. I suspect that this may do what you need.

Actually Google led me here.

I bought, paid for, and planned on modifying a 105c. I received something very different in the mail but didn’t realize it because I wasn’t in a position to work on it. Instead I got a D version with some pretty different characteristics.

Since it was my backup plan, I wasn’t too concerned until this thread showed there was an E version with a different MCU.

You see budget FLASHLIGHT but I’m intending the use for another effort. Yes I can get access to ICs that have lower currents but they require a PCB and fab time.

As for the precision… I realize the number of lumens or photons emitted might seem imprudently emphasized. What I work with for prototyping it’s very important.

My apologies for the chime in for an item I did not order.

I understand that you didn’t get what you ordered. The lack of Google hits for 105d was part of why I saw fit to start this thread. That wasn’t really my point though.

What I was getting at was more along the lines of “why don’t you spill the beans and maybe we’ll help you do what you at trying to do” since it seems that your efforts so far have been pretty far off base. You also specifically said flashlights in your first post. Come to think of it, you also said 10mA, then later said 1mA? You’re still welcome to spill the beans and get help.

The E version does not have a different MCU.

Error some of that is android and it’s wonderful auto correct.

There isn’t really beans to spill. I work with photometrically sensitive films. The can’t have any sort of exposure and they can only tolerate certain wavelengths.

So somewhere in the interpolated emission charts 1 to 10 ma would give me the desired photon flux such that a human eye can navigate under night vision but a radiometric sensitive emulsion would not appreciably fog.

Kinda a fun problem I’ve licked in the past.

Can always apply nd filters to the emitter to knock down the flux. Don’t have any ideas yet until we measure how many photos actually will dog the emulsion.

That’s much more engaging as far as I’m concerned. Are the films for scientific purposes?

ND filters makes sense.

What driver did you try earlier?

I do think that the QX7136 is probably a good fit. It sounds like you want single-mode, so in that application it would be completely constant current. You don’t even need an external MOSFET for such a low current, it has an internal one you can use.

I’m and idiot, but isn’t that what red lights are for?

as an FYI did you know about:

650nm-670nm Photo Red that Cree makes?

Product Index > Optoelectronics > LED Lighting - Color > XPEPHR-L1-0000-00901

All prices are in US dollars.
Digi-Key Part Number XPEPHR-L1-0000-00901CT-ND Price Break Unit Price Extended Price
1 3.43000 3.43

Quantity Available Digi-Key Stock: 872
Can ship immediately
Manufacturer Cree Inc

Manufacturer Part Number XPEPHR-L1-0000-00901
Description LED PHOTO RED 650NM 700MA

No, you are not, but not all films are red insensitive.

That’s mostly due to the type of silver chloride used. Other find that have, say, sensitivity to ir or UV might have different wavelengths of less sensitivity.

For standard bw film though you’d be right.

Well I am an idiot. I calculated I’d need a6x reduction in luminance to hit my ultra low target. That meant a 50 ohm pot. Didn’t work.

Put a 50k pot on there… Hit the ultra low luminance.

Need to do actual measurements in the sphere tomorrow.

:bigsmile:

For those that were wondering how this ended up, my ultra-low-LED flashlight was considered ‘too bright’ with a 252kohm resistor on it for sensing current. I’ve been meaning to go back and measure the current with a fluke but the last time I tried it didn’t go low enough…. and I’m not taking a flashlight over to the NIST lab for testing.

I couldn’t accurately measure the flux on most instruments. The reflector was causing most of the problems due to the focus of the beam- which was why it was designed that way.

But at least I had fun and got a SOIP programmer :slight_smile:

Ugh! I have a need to reflash 4 drivers and convert them to off-time memory with the added capacitor, but when I checked the lights I see they have 105*D* drivers installed! Ugh! I guess my best and easiest option is to replace them with 105*C*’s? Unless someone can point me to a firmware that gives me proper off-time memory without the capacitor. What I am really looking for is to have “off-time memory” on a driver set to “no memory” (it makes mode switching much much easier!). I’m very much a novice with firmware code and flashing, so it would have to be easy for me to modify and reflash (I’m currently only using STAR & STAR Off-Time and only modifying PWM levels for modes).

-Garry

Hi Garry,
take a look at a relatively new firmware @Toykeeper has written.
It’s named “biscotti”.
Written for the 105D, offtime memory without cap by using brown out detection (BOD). And she crammed in all functionality the 13A with 1 kB can take.
You find it in Toykeepers firmware repository.
Just be sure to use the right fuses, there are reports of bad timing when flashed incorrectly.

EDIT: And take a look whether you already got the driver version where pin 5 of the attiny seems to be connected to ground. You might need to cut this for biscotti to work as intended.

Thanks HQ! Quick questions: does biscotti work with regular tailcap switch lights (as opposed to momentary switch lights)? Does the off-time memory function no matter how long the light has been off? (Though in my case it shouldn’t matter since I want “no memory” once turned off for say 2+ seconds.)

Can you give me some guidance (or point me toward some) for a simple-stupid here’s what settings to change, here’s an example flash command with appropriate fuses . . .? I’m not real savvy with flashing firmware and right now just modify my existing command-line avrdude “flash command” line of text I saved in a text file. I assume the code is easy to find where I modify PWM values for the modes I want. I assume I don’t have to touch anything else in the code (except maybe to toggle memory on or off).

I’ll check into it when I get more time.

Thanks,
Garry

Garry, biscotti and similar firmware use a trick (RAM decay on a NOINIT variable, explained by TK here) to accomplish OTC-like functionality. I can’t say an exact timing, but it’s around 1/2 a second. Anything shorter than that is considered a short tap (mode switching), longer than that is considered to be the light turned off. It works out really well for typical clicky switches. The only advantage, to my knowledge, of a proper OTC is the ability to detect “medium” presses (say, more than 1/2 sec but less than 1 sec).

Don’t get too hung up on the fuse settings, but the correct ones are necessary. Assuming you don’t need to play around with the code, just download the hex file from here and then flash it to your driver using this command:

The above command was borrowed from TK’s flash.sh script, I’ve used it with biscotti and it works just fine. It works for most of the attiny13a drivers (105C, 105D, etc)

Also, mode memory can be turned off in biscotti (or is it off by default? I forget), which is what you’re looking for if I read that correctly. So if the flashlight is off for more than 1/2 second, it will start back at the first mode. This isn’t something you’ll need to change in the code, it’s user-selectable via button taps once you’ve got the light assembled.

Hmm . . I’m a little confused. Tell me this: Say I have the light set to have 2 modes (high, then low) with no memory, always starts on high from “off”. Say the user has the light on in “high” mode for a minute (or could be 2 mins, 3 mins, whatever, but much longer than just a few seconds) and that user wants to switch to “low” mode, will 1 “quick” half-click change it to “low” mode? Secondly, say the user turns the light off from “high” but 10 seconds later turns it back on, will it come on in high again?

Thanks,
-Garry

That is correct. The amount of time that the light is on does not matter, just the amount of time it’s off.

With mode memory off, whenever you’ve got the light off for more than 1/2 second, it will return to it’s first mode (high in your example).

Just FYI… Biscotti doesn’t have a 2-mode setting that starts on high, but there is a 3-mode starting on high (mode 6: 100% - 20% - 1%). I know it was an example, just thought I’d mention it.

Great! Thanks!

Wait a minute, can I not set my own modes in the code like I can do with STAR? I’m counting on being able to customize the modes exactly as needed (balance of runtime & brightness for my co-workers use underground).

-Garry

I didn’t like 105D’s either, until I modified the BLF-A6 code to have the OTC pin next to the GND pin. 0805 cap fits nicely across.

I just caught this “Edit”. I have pics of the backside of the drivers, but not the front (mcu) side. I’d have to desolder them to remove them to check. Well 1 is a Convoy M2 with a driver retaining ring; I could check that one easy. (Btw - I didn’t see any pics posted showing the trace that needs cut or any of the other issues.)

I’m thinking I’ll just plan to swap these drivers out with good old 105C’s. It’ll give me the chance to get the drivers done ahead of time without pulling the lights out of service for an extended period of time. Plus, it avoids the users being able to access a programming mode and change settings.

I just read the biscotti “instructions” in that Convoy C8 thread and see it’s a user-programmable multi-group driver.

-Garry

I didn’t realize ToyKeeper had a version of STAR modified with the same “brown out detection” until RMM posted it here. I’ll proceed with trying this out so then I can set my own modes and not worry about another user mistakenly entering programming.

-Garry