deskthority

I just picked up a nifty little USB power meter that is pretty darn accurate. I'll randomly update this topic with various keyboards' power draw through Soarer's on a Teensy 2.0 (~33 mA idle) + pull up resistors + a reset line. I'll note native USB or internally converted keyboards.

If all LEDs aren't working or none are present, I won't post a second number. For keyboards with backlighting, only no LED and worst case numbers will be posted.

AT&T 501007439 101 key BS clone: 125 mA | all LEDs 154 mA
Dauphin Datacomp DCK-84B82: 98 mA | all LEDs 120 mA
IBM 1391401: 149 mA | all LEDs = 197 mA
IBM Model F AT: 266 mA | all LEDs = 319 mA
HQK RB-2001: 72 mA | all LEDs 94 mA
JIT Technology GWT3P9KM1021S: 103 mA | all LEDs = 126 mA
??? GWT3P9KM1021S: 106 mA | all LEDs 127 mA
KSI 1293-RS: 78 mA |Caps & Num LEDs = 95 mA
LASER BNX84H80-2269-1: power LED only = 115 mA | all LEDs = 134 mA
Lexmark 1398601 Model M: 151 mA
Lexmark 1378300 M15: 148 mA | all LEDs 204 mA
Lexmark 1397961 SSK: 158 mA
MagicForce USB 68: 112 mA | worst case all LEDs = 148 mA
OEMMAX MCK-101: power LED only = 87 mA
Plum USB Keyboard: 25 mA | Caps LED = 28 mA
Siig USB Minitouch Pro Micro: 110 mA | all LEDs = 125 mA
Tandy 3000: 141 mA | all LEDs = 187 mA
Tesoro USB Excalibur SE Spectrum: 65 mA | worst case all LEDs = 500 mA
Wang MEC-101s (KB-6251): 50 mA | all LEDs 75 mA
XMIT USB TKL beta: 132 mA | worst case all LEDs = 310 mA

I will no longer accept backlighting hate that involves allegations of high current draw. Certain keyboards like the MagicForce 68 appear to use roughly the same current as a full size Model M that doesn't have any locks active with their backlights going full blast. That's pretty hilarious to me.

If you think I screwed anything up, just tell me. I'll measure whatever you think is off a second time.

Will you still accept backlighting hate on the basis that I don't like it?

Yes. That is a valid, if subjective, opinion.

Holy shitballs, dat Model F one thirsty motherfucker Oo .

Another case for the existence of a keyboard database, and now also for some kind of simple reporting engine (even if the reports are statically defined by an administrator). Supposedly you could do this in a MediaWiki table, if you can get the columns to sort properly (numerically with units, instead of as strings).

Even so, I'd want to ORDER BY is_backlit, current_draw_with_LEDs, so that I could see the range of values separately for backlit and non-backlit — it's too confusing having the LEDs of non-backlit keyboards interfering with the current draw range for backlit keyboards. It would also be interesting to see the ranges of values for different switch technologies.

It's funny how wildly similar switch technologies vary in power consumption. I really think it's a controller and circuitry regulated difference (with a few exceptions).

I'm curious to see how a Topre board compares. Also I'm curious to see how phosporglow's Colossus controller for the Model M compares (just an Atmel MCU running TMK), and to see how a generic crappy rubber dome USB Dell keyboard fares as well.

The xwhatsit (for a Model F or Beamspring) would be an interesting data point too.

Halvar posted in keyboards-f2/power-consumption-of-keybo ... ml#p193634

Halvar wrote: OK, I sacrificed a USB cable and measured what I had here with a real multimeter (currents) and the charger doctor (voltage).
...

Code: Select all

                                   Switch             Converter # LEDs lit  Power/mW
Soarer converter w/o keyboard                                   none             124
IBM Model M SSK                    Buckling Spring    Soarer    none             532
IBM Model M full size              Buckling Spring    Soarer    none             542
                                                                all              655
IBM Model M-122                    Buckling Spring    Soarer    none             679
IBM Model M2                       Buckling Spring    Soarer    none             563
                                                                all              708
IBM Model F-122                    Buckling Spring    Soarer    none             793
IBM 3278A2 xwhatsit rev3           Beam Spring        (USB)     none             106
Realforce 91UBY                    Topre              (USB)     none             150
Cherry G80-3000 USB                MX                 (USB)     none              20
                                                                all              161
Cherry G80-3000 PS/2 winkeyless    MX                 Soarer    none             449
                                                                all              519
Compaq/Cherry G80-11801            MX                 Soarer    none             131
                                                                all              262
Keybo/Cherry G81-3000              MY                 Soarer    none             130
                                                                all              266
Noppoo choc mini                   MX                 (USB)     none               9
Tipro TMC                          MX                 Soarer    none             371
SGI Granite                        Alps               Soarer    none             257
                                                                all              377
Monterey K108                      Monterey           Soarer    none             465
                                                                all              541

Topre is quite good in this table. Look at the huge difference for Cherry G80-3000 with or without indicator lights on - the difference is almost as much as total draw of the Realforce.

But I do wonder about how the measuring devices work and what they measure exactly. Those cheap USB-meters from China are obviously made for measuring continuous draw over time.
Current draw of a keyboard is not necessarily continuous but varies probably very much over a small period, and is dependent also on how you type.

Indeed power consumption of LEDs and integrated circuits have improved quite a lot through the years.
Lights draw less if you would use brighter LEDs driven with PWM.
In microcontrollers it is also about how you program them, if you would use timers and sleep mode or run in an idle loop at full throttle.

Should I redo all of my measurements to account for mW instead of mA?

@XMIT -- I'll find a random Dell for comparison at some point. I'll also bring the gadget down to TX next time I visit.

Power in W = current in A * voltage in V, and voltage should be constant over USB at 5V, so that should not be necessary unless your host has a weird USB controller.

USB port voltage varies based on load, even in excellent implementations. My USB 3 ports, for example, put out 5.1ish volts and refuse to drop. My USB 2 ports put out 4.9ish volts and drop to 4.8 under load.

*Shakes thread* Wake up!

I recently got myself a USB power meter, a Ruideng UM25C because reviews told it had good accuracy and is not too expensive. Unlike a proper multimeter setup it does also run on the USB port though.

I added "power draw" to the "dkeyboard" infobox template in the Wiki, and measurements for some USB keyboards in this thread and the previous and a couple keyboards I've had on hand.

The real reason I got the meter was for exploring ways of writing firmware for low power consumption. Like matt3o before (in the older thread), I of course also come to the conclusion that the best method would be to abandon ATmega32U4 altogether and go for some 3.3V ARM Cortex-M chip if low power consumption is desired ...

Techniques I've tried (on ATmega32U4...):
- Switch off unused units of the MCU properly. I haven't found that any other AVR keyboard firmware actually does this. Saved 1.7 mA.
- Use a timer to scan the matrix only every millisecond, putting the microcontroller in a light SLEEP MODE in-between scans: Using sleep mode saved 2.5 mA (10%) compared to busy-waiting.

The keyboard I tested on was my daily driver (...) for many years: a Phantom, which has a Teensy 2.0 where a column pin is shared with the Teensy's built-in LED, which then shines a whole lot... By turning off that pin between scans, I saved another 1.8 mA.
I also tried turning off / flipping other sensing/strobing pins in the matrix, but found no measurable effect on power consumption.