The page that once had been at this location has been moved to http://www.cs.vassar.edu/~priestdo/herbert1.html , it still has a lot of useful information and and I will need to revise it.
The chord-hack chording keyboard is documented at http://www.cs.vassar.edu/~priestdo/chordhack/
This page is a description of what I did to get up and running after the demise of Herbert I. The rig discribed here was in use until early 2000 when I switched to the design I was developing for Charmed Technology.
To see the rig I am wearing now, please have a look at http://www.cs.vassar.edu/~priestdo/latest.html
For pictures of Herbert I, II, III and my keyboards, see http://www.cs.vassar.edu/~priestdo/wearable.pics.html. This includes pictures of Herbert III with components labeled.
In August of '98, Herbert, my wearable since '94, built on a Gateway handbook, started to lose it. Finally in September, Herbert went into a sleep from which there was no waking. Near the end, Herbert regressed to a past life: he kept thinking it was a particular day in the past before he existed, Jan 1, 1980 I believe. It seemed like a nice day...
He lost feeling in his extremities as well. Brushing or tapping his keys had no effect at all most of the time, then there would be moments of clarity when we would go for short walks together...
By mid-September he just withdrew, stopped communicating at all...
Thankfully for those left behind, I did manage to get him to digitally sign his organ donor card. So armed with a few PCMCIA cards, a hard disk and a keyboard, I began to work on "H2," the new "Herbert." (I have kept the name the same for a few reasons, not least of which is that there are enough overlap of parts and function, that well, I am just more comfortable still calling it "Herbert.")
In the few weeks left to me before ISWC '98 I scrambled to put together a new machine. I got the "first draft" done and wore it from Oct '98 to March '99. With Herbert II I had the oportunity to test and incorporate several design and conceptual changes. It has evolved into Herbert III, the main differences being a case that is shaped to wrap around my left side. Here is my original description of Herbert II put up just after ISWC '98.
Herbert II is a pc104 machine based around the Cell Computing P166 CardPC (http://www.cellcomputing.com/). The CardPC is in a Parvus PC/104 Scalable CPU card (http://www.parvus.com/). The CardPC has SVGA (1 meg trident chipset) built in. It will take up to 128 meg of ram on it, I have 64. I am not using a heat sink other than the case, which is against me, so I am both providing some cooling and aware of how hot it is (or isn't) at any point in time. At no point has it been to hot to use. (Both the Cell PC board and the Power Trends ISR have overheat protection built in.)
For audio display - speech output I am using a doubletalk PC/104 board from RC Systems (http://www.rcsys.com/ or http://home.sprynet.com/sprynet/rcsys/dt.htm).
For storage I have on backorder a 4-gig drive from Toshiba. But currently I have a 1-gig old IBM; both are 2.5" 8.5mm high drives.
I just got the pcmcia adapter. The power figures below do not reflect this card yet.
For visual display I am using the M1 from Liquid Image (http://www.liquidimage.ca/) mostly in 50x17, 50x27 or 80x15 mode using SVGATextMode. I am happy to share my SVGATextMode TextConfig modifications file: http://www.cs.vassar.edu/~priestdo/TextConfig.M1 (use at your own risk).
For input I am using my modified BAT keyboard (http://www.cs.vassar.edu/~priestdo/herbert1.html).
For power I am using two Sony Li-Ion camcorder batteries and a PowerTrends PT6302 (3 amp ISR) in the same way Steve Mann does in his wearcomp (http://n1nlf-1.eecg.toronto.edu/wearhow/node11.html). The only real difference is I decided to use BNC connectors for power. I did not like the idea of having to tape my power connections in place all the time as was described with banana plugs, I have been pleased with this decision.
I usually put it all in a vest with all the components in the front. This was important to me as I need to be able to move from walking to sitting to lying down on my back throughout the day and I wanted a design that would let me do that easily.
It also all fits nicely in a small hip pack. I find I am using that on occasion when the vest would not be practical - Doctors visits and the like when I would have to be removing the vest.
I am running Slackware version 3.5 which is more a reflection of how long I have been using Linux than a particular preference for Slackware. ( I started with Slackware five years ago, and can do the install in my sleep - yes, I have...)
I "live" in emacs with emacspeak (http://www.cs.vassar.edu/mirror/emacspeak/). It is simply the best audio interface there is. Thank you T. V. Raman - you have changed my life for the better.
Boxes (still oversized), cables, vest, batteries, etc., all together weigh under 6 lbs.
So, the question folks have been asking - power consumption:
note: these numbers have changed as I have been gathering more data.
When actively working--typing and listening--it draws between 5.0 and 8.2 watt/hr from the power source, average working draw stays right around 5 watts/hr. There is a momentary 11 watt peek during boot up. These numbers are without the M1. If the M1 is on and I am in text mode, add an additional 2 watts/hr for that.
In X it draws a bit more, 9 watt/hr is hi, 7.8 is active, drops to 5 watt/hr if its idle for a minute or so. Not sure why X eats so much more power, perhaps its the SVGA chip on the cpu board? Or that in combination with the added processor load? Not sure, I will have to test it more to find out, but I do not use X much, so I am not that concerned.
Of course these numbers may all change when I get the Toshiba 4 gig hard drive in to replace the old 1 gig IBM and add the pcmcia card back into the system. But I hope they go down or stay the same...
The Sonys (2 NP-F950s) are rated to store 64.8 watts. I am using 9 to 10 hours battery time a day with at least 2 to 3 hours of M1 use as well. I purchased three batteries, one in the charger during the day, another at night. So I use each battery 2 of every 3 days. I am thinking of getting 2 lower wattage Sonys so I have a light weight "emergency pack" to get me a few extra hours if I need it. But since some of each day includes tethered time, the 10 hours of battery time is doing very nicely for now.
Things that save power: Not running X saves several watts/hour, using the doubletalk board instead of visual output does as well. Using audio for reminders and cues so that the M1 is routinely off I am sure helps. But I find I am leaving the M1 on more and more...
Things that should use less power: The BAT uses almost .3watts/hour. I hope to reduce this in the near future. I may try to run the M1 off the Power Trends ISR as I beleive it to be more efficient than the power supply in the M1, but I will wait until the M1's warranty runs out before trying that level of modification :-)
I currently do not use standby or suspend mode on the CardPC, but I do use "idle" mode. This runs the P116 in a 2/8 on 6/8 off mode when it can. I would guess that is how it runs at such low power even when typing or reading in emacs.
Well, that is what I wrote up on Herbert II.
Herbert II and Herbert III are basically the same machine, with the following changes:
So, armed with a piece of aluminum, a pair of cutters and a drill I made a case for Herbert III. Would I modify the shape if I were going to do it again? Sure. Will I change it any time soon? Probably not. If I redo the case, there are a few ways I could tighten things up, a 1/4" here, an 1/8" there...
I still find that as a result of my comfort with emacspeak, the social barriers presented by an obvious HMD like the M1, my long term use and comfort with emacspeak, the additional power drain imposed by the M1 and the ease at which I am able to do what I want without a visual display thanks to emacspeak, the M1 lives around my neck when walking around. Only being put on my head occasionally. However, when in work setting (where they know me) or when I am tethered, I use the M1 extensively and I am enjoying the redundancy of information offered by a visual display. The result of this behavior is that when out "in the real world" the most common question I get is "what happened to your hand?"
I am still playing with the "ultimate body mounting position" and regular shift between wearing the case "high" - under my left arm against my ribs, and "low" on my waist. Slight preference for "high".
The cable from Ampro that allows me to put the 8 bit card next to stack is nice but very tight. If you get one, do not plan on removing it once it is in place.
I recommend making your own case. It was not that hard. I did all the bending by hand or with a wooden mallet. But as others have said before me, plan it out first. I missed filling one corner. To late now...
I have not been able to get "suspend" modes working as the parvus board does not support them this is an annoyance rather than a real problem. I suspect if I want them I will need to build some circuitry and actually solder to the parvus board...
I hope to be adding packet radio and GPS/ARPS to Herbert in the near future (funds permitting) I am excited about the idea of having access to position information in program decision making. I would also like to work on an emacspeak mode for providing position information in an audio output environmnet.
- Greg