Model M - stock LED specs, anyone?

[rootwyrm]

Well, this is pretty dang embarrassing.

Just got in a fresh M for mod/repair, and I have somehow lost my detailed notes on the electrical. Some ages ago I tracked down the Num/Caps/Scoll LED specifications for M, M5 and M13, wrote it down, and somewhere along the line, lost said piece of paper. Obviously, "duh, <=5V <112mA!", but I had every single detail including angles and wavelength. I don't suppose any of you out there would happen to have these specs, or the manufacturer part number for the 3mm rounds and 5x2 rectangles? All I can recall is that they were somewhere around 10-15mA and I think ~3V forward. I'd check what I ordered previously, but it has literally been years.

Also somebody swiped my nut driver and this Craftsman one doesn't fit - anyone got a recommendation for a replacement?

TIA!

[wcass]

i don't have the specs, but i do want to watch this thread. i have been planning to change my led's to red (to match the keys) but have been putting it off as my solder skills suck.

[False_Dmitry_II]

Yeah, that is the one remaining thing on alot of my boards that is clearly standard and original. On some of them I'd almost want to just stick a piece of tape on the inside and have no light... hmm that would be pretty easy to do...

[DeltaGunner]

That IBM with the red keys and red LED's would look veryy slick. Im quite jealous actually.

[Oqsy]

That board looks like it was left in a muddy field for a year or two. Nice LED swap though

[wcass]

i'm planning on doing an in-key caps lock LED on an M but i need some help with the soldering - any volunteers?

i have extra LEDs, resistors, and light tubes in exchange for the help.

[Oqsy]

I'm confused about how you plan to put it in the key. BS mechanisms are very picky about interference in their range of motion. Unless yr doing the LED off in a corner of the keycap, I foresee problems.

[wcass]

the LED would be under the plate. a 3mm wide light pipe takes light up throu a rivet hole under the top left of the caps lock key. it might be visible nicely from there, but it not, i'll drill a hole in the keycap and back-fill with clear epoxy or run another light pipe through the key leaving an open air gap between the two light pipes. i'll do all three and show pictures of the results so that others don't have to.

total cost of parts is < $5

but again, i need help - my hands shake like a junkie in detox.

[Oqsy]

I'll trust that you have worked it out, I just need to see it to understand. I'm a diagrams and exploded drawings kind of guy.

[rootwyrm]

LEDs are cheap.  Just dump in some Radio Shack ones and fuggedaboutit.

Replacement LEDs, while cheap, STILL have to match original specs. Otherwise, you increase the draw on the PS/2 port, and enter the "blah blah incompatible etcetera" crap. And you still have to match voltage regardless. Leading me to wonder if you're just here to troll or what.

the LED would be under the plate. a 3mm wide light pipe takes light up throu a rivet hole under the top left of the caps lock key. it might be visible nicely from there, but it not, i'll drill a hole in the keycap and back-fill with clear epoxy or run another light pipe through the key leaving an open air gap between the two light pipes. i'll do all three and show pictures of the results so that others don't have to.

Keep us posted there.. that'll definitely be interesting to see. Also it will definitely work that way with the light pipe. Good catch there. I can help with selecting an LED that'll make sure it's visible. There's some wizardry involved, but it's nothing all that bad - all about viewing angles and wavelength and millicandela.

[Soarer]

You could easily change the resistors as well, to match whatever LEDs you find. The problem would not be that newer ones (of the same colour) draw more current, but that they might be a LOT brighter. LEDs of different colours do tend to have different forward voltages, and so draw different current, but even so you'd be unlikely to damage anything by not changing the resistors.

[The Solutor]

Replacement LEDs, while cheap, STILL have to match original specs. Otherwise, you increase the draw on the PS/2 port, and enter the "blah blah incompatible etcetera" crap. And you still have to match voltage regardless. Leading me to wonder if you're just here to troll or what.

A led is a led, the FW voltage is more or less the same for any kind of led of the same color.

Going from a plain green led (not an ultrabright one) to a red led is just matter of replace it. You can have  a little increased current keeping the same resistor but nothing to be worried about.

If the end result is too bright just replace the resistor with one with an higher value.

[~Blood~]

all about viewing angles and wavelength and millicandela.

wavelength = color

However, mind that if you replace a blue/green LED with a red LED, you will probably have a very weak red if you use the same power, because the eye isn't nearly as sensitive when it comes to red light.

the maximum visibility should be around 555nm wavelength. the voltage has to fit, however parts are often built with 15% or more tolerance, so dont sweat it.

oh and please dont take thoes cheap blue LEDs, because they do not use a natural mixture of colors (RGB diode = white) they really sting in your eyes. I think they are made of a green LED and some yellow fluorescent bull****.

[The Solutor]

However, mind that if you replace a blue/green LED with a red LED, you will probably have a very weak red if you use the same power

If he replaces a blue led with a red LED has the RED led will be driven at 2X/3X the currend, so the problem could be the opposite, too bright red, or current out of spec.

because the eye isn't nearly as sensitive when it comes to red light.

The eye is way more sensitive to RED than BLUE (the alarm signals and the stop light on the traffic lights are RED for a reason), usually the blue leds appears very bright because they are damn efficient, way more efficient than a classic red led, and still more efficient than the modern super bright red leds.

the maximum visibility should be around 555nm wavelength

Correct, it's a green-yellow color.

oh and please don't take those cheap blue LEDs, because they do not use a natural mixture of colors (RGB diode = white) they really sting in your eyes.

I guess you are referring to the white leds...

Btw RGB leds has nothing of natural, they emits three spikes of color to emulate the white, like happen on TV or monitors, this obviously works and has the advantage of being adjustable, but has nothing to do with the sun light that emits a continuous spectrum.

I think they are made of a green LED and some yellow fluorescent bull****.

They are an UV led with a white phosphor or blue led with a yellow phosphor, they provide a more continuous spectrum than RGB LEDs and the quality depends on the phosphor used as happen on any fluorescent lamp.

With quality phosphors the white leds have a very very good light quality, less flexible than RGB leds but, often, more natural.

[rootwyrm]

If he replaces a blue led with a red LED has the RED led will be driven at 2X/3X the currend, so the problem could be the opposite, too bright red, or current out of spec.

Well, here's where things get really fun. There's multiple LED types and voltages on the Model M. There's 4-wire pin-lead (1389989, 1392993 and 1351523) and there's an 8 contact film cable (1393291) - these all have LED takeoffs. 1398011 takes LED power and signal off the matrix block connectors. That only covers M, not M5 or M13. Also not F. Also not the further reduced size PCBs with crimp connections. (Oh how I love those.) Obviously there's also the issue of how they're wired - it's not consistent either.

Regardless, the reason I stick to specs is twofold. One, it avoids the "oh well now it's drawing too much current" issue. Two, it's easier than spending all the time fiddling with two multimeters and holding LEDs in place to make sure I'm under 135mA and at the correct fV.

The eye is way more sensitive to RED than BLUE (the alarm signals and the stop light on the traffic lights are RED for a reason), usually the blue leds appears very bright because they are damn efficient, way more efficient than a classic red led, and still more efficient than the modern super bright red leds.

Correct, it's a green-yellow color.

Standard wavelength for the 'classic' green look it 565-569nm, not 555nm. It appears as 545-565 typically because of lenses, diffusion covers, etcetera. The "standard" green LED is a T1 (3mm) @ 2.1-5Vf with a <=80deg viewing angle. (Obviously, no such thing as standard other than the color. And the reason for green? Easy to make, accepted indicator of 'good.')
Solutor is absolutely correct - blue is the most 'invisible' color whereas red is one of the more 'observed' colors. My practice is Num Lock green, Caps red, Scroll amber. (There's a reason for the amber - I use scroll lock frequently.) Realistically speaking: a modern green or red can be equally efficient to a blue. It just comes down to the process. Blue's uniqueness is the process; Indium Gallium Nitride (InGaN.) Also because Blue leads to White, the Holy Grail. (Not to be confused with Monty Python's.)

I guess you are referring to the white leds...

Boy don't you wish it was that. (I can't say more than that sadly. Suffice to say: the really bad whites are even worse.) Regardless, I wouldn't use white LEDs in keyboards anyhow. High outputs and high temperatures, and the miniatures generally are knockoffs or more blue than white.

With quality phosphors the white leds have a very very good light quality, less flexible than RGB leds but, often, more natural.

 
It's actually somewhat less the phosphor and a lot more the process itself as well as the temperature control. White LEDs run hot, which is why you typically see them on pucks - those are IHSes. The big pads? Heatsinking. You get the idea.

[The Solutor]

Well, here's where things get really fun. There's multiple LED types and voltages on the Model M. There's 4-wire pin-lead (1389989, 1392993 and 1351523) and there's an 8 contact film cable (1393291) - these all have LED takeoffs. 1398011 takes LED power and signal off the matrix block connectors. That only covers M, not M5 or M13. Also not F. Also not the further reduced size PCBs with crimp connections. (Oh how I love those.) Obviously there's also the issue of how they're wired - it's not consistent either.

That said a green led is a green led they are made with the same semiconductor and have the same VF so they drain the same current.

Then there are some differences in the efficiency so a newer led can be driven with less current to emit the same light.

What is sure is that even if you have the exact specs, you will not find a direct replacement for a led sold 15 years ago, you will find a different led with slight variation in the wavelength, efficiency and so on.

Regardless, the reason I stick to specs is twofold. One, it avoids the "oh well now it's drawing too much current" issue. Two, it's easier than spending all the time fiddling with two multimeters and holding LEDs in place to make sure I'm under 135mA and at the correct fV.

It's easier to buy a random led, measure how much current it drains and eventually replace the resistor, btw my board has 108 leeds and current it isn't an issue, how a single led can change something, even if overdriven ?

Standard wavelength for the 'classic' green look it 565-569nm, not 555nm. It appears as 545-565 typically because of lenses, diffusion covers, etcetera. The "standard" green LED is a T1 (3mm) @ 2.1-5Vf with a <=80deg viewing angle. (Obviously, no such thing as standard other than the color. And the reason for green? Easy to make, accepted indicator of 'good.')

Main reason for green was that at the time was considered cool because initially they were rarer and most expensive than the usual reds, so something with green LEDs was considered less cheap, more professional.

Boy don't you wish it was that. (I can't say more than that sadly. Suffice to say: the really bad whites are even worse.) Regardless, I wouldn't use white LEDs in keyboards anyhow. High outputs and high temperatures, and the miniatures generally are knockoffs or more blue than white.

A withe led is fresh exactly as a blue or modern green led, only when the illumination is the main purpose high power LEDs are used (even 7 watt).

And no they aren't more blue than white, only the daylight ones are bluish. Nowadays you can find daylight LEDs (6500K, bluish), standard white (4500K, just white), warm white (3000K, like incandescent lamps) and even golden white (also incandescent like but with a different phosphor).

This is my modded Xarmor with warm white leds



I have just replaced the original blues, they are not hot, nor are draining abnormal amounts of current

[Soarer]

rootwyrm,

Wow, you do like to make things complicated!

If you are fitting a different colour LED, is it even possible/practical to find one with the same Vf?

Surely it's far easier to swap the resistor as well, if needed. And to test the LEDs before fitting them - a variety bag of resistors and a 5V supply is all you need, and the 5V supply is easy to find thanks to USB! A breadboard would make life easier, but a couple of clips or terminal blocks would do just fine.

Also, since If is related to Vf, there's no need to use two meters when testing.

[xwhatsit]

Man you are making it complicated. An LED of the same colour will have approximately the same voltage. Also, get one that is as high current rating as you like -- the keyboard controller will have current-limiting resistors for each LED. Given that the resistance of an LED is not likely to change very much, even at different current rating (it is a semiconductor after all, it's not like a lamp filament), and the forward voltage is about the same, the LED will draw approx the same current as the previous LED did. The current-limiting resistor assures that.

So just buy some LEDs!

If you want a different colour, the forward voltage might be different. So look at the existing resistor value and the typical voltage drop for a green LED (2.2V is about normal), and work out what you need to replace the resistor with in order to keep the same current.

[Soarer]

Subtract the forward voltage (V_f) from the supply voltage (V_s), then divide by the desired current (I_f) to give the required resistance (R):

R = (V_s - V_f) / I_f

Then realize that it's _way_ too bright as a lock indicator and go to plan B, which is trial and error with various resistors from 220Ω to 2.2kΩ until it glows nicely.

(BTW, It's a mistake to think of LEDs having resistance, but I suspect xwhatsit was just trying to simplify).

[Soarer]

∆∟∩∞Ω

[False_Dmitry_II]

I kinda like the no lights at all. You just have to be aware that that is the case, which I actually was not when I first plugged this thing in. Before I even wrote the post about the tape, I had spraypainted this board with no holes cut out of tape so it just went all over the indicator light region. I didn't even notice I couldn't see anything through it for awhile.

[Konrad]

I'm guessing this M is nominally rated to draw 5V ~100-125mA ... while the PS/2 port is rated up to 5V 275mA (in theory) and USB is rated up to 5V 100-500mA. Adding a few extra 10-20mA LEDs is unlikely to cause any problems; indeed, many keyboards (and keyboard mods) use numerous LEDs without issue.

For your convenience PS/2 and USB both provide 5V power, so the keyboard controller and circuitry will all operate on 5V, and almost all LEDs are designed for 5V circuits. Many LED technologies exist, but for your purpose all that really matters is the form factor (size and shape) of the LED package, the colour, and which resistor value you should put it behind. Lower-power (say 10mA or even lower) LEDs will tend to be dimmer than higher-power (20-30mA) LEDs, and varying the resistor value can further control the intensity of the LED; the resistor basically drops the voltage below 5V before it reaches the LED, simple stuff.

Here's a convenient calculator to determine which resistor you need - http://www.hobby-hour.com/electronics/ledcalc.php

Note that the LED colour is determined by the semiconductor chemistry it's built from, which is another way of saying that different LED colours have different voltage values. The calculator I linked already factors in the correct values, although you should always use the actual specifications for the actual part whenever they are known. 10mA LEDs will suffice for this application, regardless of whatever IBM's bureaucorporated index of part specifications might state. You might encounter miniaturized SMT LEDs, often advertised as "high flux" (because the ratio of radiated light energy per unit surface area is much higher), these would be much more difficult to work with yet could fit within keycaps quite easily. You might see LED specifications which advertise all sorts of optical properties, focus angle, dispersal, etc ... these basically boil down to brightness intensity (in mcd) and how the plastic package acts as a "lens" to direct the light output ... none of this stuff is critical for your application, unless you want to seriously nerd out on LED details.

Ripster's got it right - just use whatever LEDs you can find at Radio Shack or salvage (free) from some dead McHappy toy or random junkbox item, component values are not critical. Really, about the worst that can happen is you blow out the LED with too much sustained voltage (it'll get a tiny bit warm and might even make a tiny popping noise, yet it will visually appear undamaged aside from no longer working, sorta like a little plastic light bulb). If you insist on purchasing your LED then $1 is an outrageous price (in fact, you could buy a Made-In-Cheapland junk gizmo with a whole bunch of LEDs at your local dollar store), while if you paid a dime for your resistor then you got robbed.

[The Solutor]

All correct but this is not exactly what happen.

the resistor basically drops the voltage below 5V before it reaches the LED, simple stuff.

The voltage is keept, more or less, constant by the LED itself, then you will have two different constant voltages connected together, leading to a theoretically infinite current, practically a very high current, the resistor limits that current.

[Konrad]

Wow, Rip, two Italians here now, an endless source of tasteless jokes. Not quite as much fun as Canadian jokes, of course, but oh well you've gotta work with what you've got.

Hmm, you look a little older than before. Must be the glasses.

[The Solutor]

Wow, Rip, two Italians here now, an endless source of tasteless jokes.

In Canada the ohm's law is considered a tasteless joke ?

Good to know ! :lol:

[rootwyrm]

I'm guessing this M is nominally rated to draw 5V ~100-125mA ... while the PS/2 port is rated up to 5V 275mA (in theory) and USB is rated up to 5V 100-500mA. Adding a few extra 10-20mA LEDs is unlikely to cause any problems; indeed, many keyboards (and keyboard mods) use numerous LEDs without issue.

On an M5/M13 you have in theory 550mA @ 5V (PS/2 A+B, and yes, the supply is joined at the keyboard.) But as I point out often; see the "incompatible with Model M" motherboard list. It's long. Peak draw on this M is measured @ 120mA, while this motherboard is known to top out at ~138mA per PS/2 port (combined ~275mA.) USB isn't happening here either; but it's 100mA per unit load for 2.0 with a max unit load per port of 5 (500mA). Won't bore you with the details there. Translation: power budget is very tight.

For your convenience PS/2 and USB both provide 5V power, so the keyboard controller and circuitry will all operate on 5V, and almost all LEDs are designed for 5V circuits. Many LED technologies exist, but for your purpose all that really matters is the form factor (size and shape) of the LED package, the colour, and which resistor value you should put it behind. Lower-power (say 10mA or even lower) LEDs will tend to be dimmer than higher-power (20-30mA) LEDs, and varying the resistor value can further control the intensity of the LED; the resistor basically drops the voltage below 5V before it reaches the LED, simple stuff.

Unfortunately, I'm looking at a 1389986, which means the LEDs are not straight fed. Instead, they appear to go from a 7406 TTL through 2000 ohm resistors. (Presuming I'm still reading those correctly - red black red gold should be 2K +-5% yes? The resistors are definite though.) There's some obnoxiousness about this particular controller I've dealt with before, but I found the notes on modifying it, so that'll do for now.
However, candela rating is not defined by power - it's defined by process. If you want to destroy your eyes (and use a Belkin USB adapter,) you can fit 500mcd 20mA greens built on AlGalnP process. Or REALLY destroy your eyes with 672mcd (680mlm!) blues running 10mA @ 3V. For comparison, your typical 5v 10mA green is 8mcd or less. (EDUCATIONAL! Also blinding. But educational nonetheless.) Millicandela and luminous flux (mlm at current) are the measures of brightness.

Note that the LED colour is determined by the semiconductor chemistry it's built from, which is another way of saying that different LED colours have different voltage values. The calculator I linked already factors in the correct values, although you should always use the actual specifications for the actual part whenever they are known. 10mA LEDs will suffice for this application, regardless of whatever IBM's bureaucorporated index of part specifications might state.

If we want to nitpick, IBM used every single manufacturer out there. So there's actually some variance, including on current. (I have one here that's got 6mA LEDs.) It's typical IBM-ism; you've got IBM and Lexmark both building, with different supply chains. Sanity does not apply here, but it's IBM, so we already knew that. "You have read an IBM Manual Scroll. --more-- You are permanently confused." Really the core problem is the different controllers involved. Some days I wonder if it wouldn't just be easier to call Unicomp and have them build a custom - but it'd certainly be less fun. (... wait, this is fun? What is wrong with me. Sigh.)

none of this stuff is critical for your application, unless you want to seriously nerd out on LED details.

I think we've established that I nerd out on LED details sufficiently. Unfortunately, I'm also subject to strict NDAs relating to LED technology, so I exercise an abundance of "OH GOD DON'T SUE ME."

Ripster's got it right - just use whatever LEDs you can find at Radio Shack or salvage (free) from some dead McHappy toy or random junkbox item, component values are not critical. Really, about the worst that can happen is you blow out the LED with too much sustained voltage (it'll get a tiny bit warm and might even make a tiny popping noise, yet it will visually appear undamaged aside from no longer working, sorta like a little plastic light bulb). If you insist on purchasing your LED then $1 is an outrageous price (in fact, you could buy a Made-In-Cheapland junk gizmo with a whole bunch of LEDs at your local dollar store), while if you paid a dime for your resistor then you got robbed.

 
Pft. I'm a perfectionist. I'd rather pay the grand sum of $0.18-$0.53 per LED to get new known parts. (Oh NOES! Not even two whole dimes to almost three quarters!) These are keyboards that I use for literally a decade or more. The problem is that I very much do NOT want to fry this controller, which is a distinct possibility because of the way it was implemented, and I don't want to have to redo it again in a year. Replacing the controller is a huge hassle, because it's the ancient wired LED board and not the thin-film connector style. And that's time I'm not getting things done, when I'm replacing internal components.

Oh, and I still could use a nut driver recommendation. The one I borrowed to disassemble the keyboard was a nightmare, way too tight a fit, gave myself a heart attack. Someone's got to know of a good nut driver for M disassembly, right?

[Konrad]

You apparently already have the LED part specifications you need. If not, then just extract one or more of the existing LEDs to meter out of (keyboard) circuit while you push their thresholds, then pick replacement LED parts which comply with your experimental spec data. As a perfectionist (and big spender, lol) you could easily obtain LEDs matching almost any particulars you desire.

I shall stand in firm solidarity next to Ripster (eww) and champion our claim that you can just use any random Radio Shack LED parts you like without issue. Engineering is about building economically as much as about building perfection - do you want to spend a week searching for the ultimate part now so it's more likely to last 15 years, or do want to instead just spend 15 mins making it work well enough with the gamble that you'll need to do so again in perhaps 5 years?

LEDs are of no consequence, you can even tape a few extras inside the chassis for later if they concern you. I can well understand your fears about burning the microcontroller, since it's probably impossible to find an exact OEM replacement outside of cannibalizing from another keyboard. Even so, there's literally thousands of different controllers available which can fully duplicate the functionality. You might do well to document the circuit and (if not codelocked) read a copy of the firmware.

[The Solutor]

On an M5/M13 you have in theory 550mA @ 5V (PS/2 A+B, and yes, the supply is joined at the keyboard.) But as I point out often; see the "incompatible with Model M" motherboard list. It's long. Peak draw on this M is measured @ 120mA, while this motherboard is known to top out at ~138mA per PS/2 port (combined ~275mA.) USB isn't happening here either; but it's 100mA per unit load for 2.0 with a max unit load per port of 5 (500mA). Won't bore you with the details there. Translation: power budget is very tight.

Translation power budget is  way more than enough

However, candela rating is not defined by power - it's defined by process. If you want to destroy your eyes (and use a Belkin USB adapter,) you can fit 500mcd 20mA greens built on AlGalnP process.

You can just use an high efficiency led and drive it at 3 or 4mA, solving even the inexistent problem of the max current drained

Pft. I'm a perfectionist.

No offense but you look more paranoid than perfectionist.

It's a freaking LED, if you are worried about frying the controller just start with a big resistor and then lower it until you get the the right amout of light you like.

[Konrad]

I note, from numerous random internet findings, that the "incompatible with Model M list" appears to often be caused by particular motherboards, and seemingly more often, from all manner of PS/2 adapter gizmos.

Modern computing has always been more about compatibility than about standards anyhow ... you might just need to bite the bullet and move your Model M platforms into the world of USB.  Today's cheap PS/2 garbage is really more of an afterthought than anything else, when present at all.

[Soarer]

Well, with 2kΩ resistors, you could fit a short in place of the LEDs and no damage would occur. With 5V and an LED, it would limit current to about 1mA, which doesn't sound right at all, for an older 'board. 2kΩ was commonly used as a pull-up resistor in conjunction with open collector buffer outputs (like the 7406).

[The Solutor]

which doesn't sound right at all

Indeed something  like 100-200 ohm, depending on the current chosen looks like a more reasonable value.