Hardware, power supply and physical performance
Michael Fischer and Roy Ellen CSAC, University of Kent at Canterbury
This is the second part of a three-part article on computers in anthropological fieldwork, based largely on our own direct experiences in Pakistan and Indonesia. The first part (BICA 6 , September 1987: 8-11) was devoted to obtaining permissions and certain logistical matters (such as insurance and transit), and the third part will concentrate on some simple software. Here, our main purpose is to describe some relevant hardware considerations. We cover a variety of issues under a few broad headings: selection of machine, transportation, sources of power, physical performance (including handling and storage of magnetic media), and routine care and maintenance. In doing so we expand on details published elsewhere [(.Ellen Fischer anthropological.)] and which are to some extent incorporated into our present remarks.
Over a period of five years both of us have had some modest experience of field applications in remote and difficult areas. Ellen has worked in the Moluccan islands of eastern Indonesia on the social organisation of inter-island trade [(.Ellen Nuffield.)], and has used a portable battery powered computer (Epson PX-8 with microcassette tape storage) for simple household censuses, bibliographies, texts, and experimentally for the entering of field notes and analysis of local boat movements and their cargoes. In addition to the Epson, Fischer has used non-portable mains-powered machines (Apple II, Commodore PET, Commodore 64) for his work on social demography, kinship, marriage and knowledge representation in Pakistan [(.Fischer Greentown.)]. The specific machines used by Fischer are now nearly obsolete, but they are similar to more modern desktop computers.
The most important limitations on a computer for use in the field are availability of software, and capacity for data storage. For the beginning user the software is by far the most important, since the computer is useless without programs. We will say more about software in the next installment. The following discussion relates to types of microcomputer rather than to specific models, and to aspects of the hardware that are relatively independent of software.
The capacity of a microcomputer for storage of data has a direct effect on the application of the computer to your work, regardless of software. Data storage capacity can be measured in two ways: speed of access, and amount of storage available on a single piece of medium (eg cassette tape, floppy disk, fixed disk). Speed of access refers to how long it takes to connect the data on the storage medium to the computer. This can include the time taken to find the piece of medium the data is on, and to load it by hand, and always includes the time the computer takes to locate the data once the medium is available to the computer (in the cassette or disk drive).
The second measure of data storage capacity has much more relevance to machines available today: how much data can be stored on a single piece of medium? This is important because if relevant data are spread across several pieces of medium, the amount of time required to process it by computer is expanded by the need to swap disks or tape when required. Most computer operating systems (the control program supplied with the computer) assume that all the relevant bits of data are available in a disk drive, and when this is not the case each piece of software has to keep track of what is where. If the data are split amongst several pieces of medium, it complicates the whole enterprise sometimes beyond capability. There is also the problem that the operator (you) has to keep good records of where dispersed data are. This can be overcome to some extent by having multiple disk drives, but even this step introduces organizational problems. Where possible, you should use a medium which has enough capacity to store each set of data on one piece. And certainly you should avoid having your data on (say) three disks or cassettes if you have only two drives.
Cassettes. These days speed of access time is really only an issue with tape cassette storage. In the earlier days of micro-computers (1976-1980), tape cassette storage was the norm, having replaced punched paper tape as a storage medium. Although floppy disk drives became widely available about 1978 they were very expensive, and had a small storage capacity. Cassette storage was cheap, and reasonably fast for the equipment then available. These were generally machines with between 1k (1024 characters) to 16k of internal memory. Most cassettes could read in about 100 characters per second once the tape was positioned to the data. This was not too bad, since most computers could be loaded in less than a minute from the beginning of a tape. When microcomputers began to have more memory, this became unacceptable. Although times such as 5 or 10 minutes (time to read one side of a C-30 or C-60 cassette) may not seem long in print, they are crippling in practice. The machine is simply less useful than it might be. Cassettes are generally only used with smaller machines (see Notebook and Handheld categories in next section), but are useful where low power requirements and ruggedness are an issue. Today cassettes can almost always be replaced by small battery powered floppy disk drives (Radio Shack/Tandy sell such a drive), although these are currently expensive (about \z\(sp 200). However, if the computer is to be used only on very small individual data sets of 1k (1024 characters) to 4k, and cost and electricity supply are an issue, cassettes can be satisfactory.
Floppy disks. The most common format for data storage today are floppy diskettes. Floppy diskettes are small sheets of plastic coated with a surface much like that of a magnetic tape. They come in sizes of 8", 5.25", 3.5", and 3", but the only likely survivors are 5.25" and 3.5". The storage capacity of a diskette is widely variable, depending on the way the computer records data on it. The range is from about 100k (i.e. 100x1024 characters) to about 5 meg (= megabytes, or 5,000k) using latest technology. The most common values are about 150k for Commodore 64 and Apple II; 360k for CP/M and IBM micros and compatibles with 5.25" diskettes; 800k for IBM, Apple Macintosh, and Atari ST, and Commodore Amiga 3.5" diskettes; and 1 meg for Commodore 64 and IBM AT/PS-2 and compatibles. 360k characters sounds quite a lot, but in practice is rather limiting, especially if the machine's operating system uses a considerable amount of that for its own purposes. Two 360k disk drives represents about the minimum for comfortable computing if you have a lot of data, although this is in general inferior to a single 720k drive for most purposes (making copies of diskettes being a notable exception). Obviously anthropologists have made good use of the smaller disk formats, and if this is what is available, you could do worse. The main message here is that if you expect to work with large sets of data you should get the largest disk capacity you can afford. When possible use the newer 3.5" diskettes. They are enclosed in small plastic cases, rather than paper sleeves, have a higher data storage capacity, are smaller, lighter, and the drives for them use less power. They are the format for the next five years. Most computers you can now buy have these smaller drives, or you can have them fitted.
Fixed Disks. Sometimes called hard disks, these are magnetic storage devices that have a much higher data storages capacity, both in access time and number of characters. Typical sizes today found on micros range from 20 meg (20,000k) to 80 meg. Much larger sizes are available, but very expensive. Older models may be found with 5 meg to 10 meg. Unlike floppy disk systems, the medium is generally sealed in the drive, and so you are limited to one very large recording surface for data. These are very desirable for highly data intensive work. But there are a few aspects to consider. The good news is that they will work well in a moderately harsh environment, since the data storage surface is sealed. They may be mechanically sensitive however, since the mechanism must work to very fine specifications. Although the authors have not worked with a fixed disk in the field to date, they are reported to do well. One hazard that must be mentioned is that of recovering from a `crash' of the hard disk (which happens rarely these days, but not rarely enough). Occasionally the `crash' will require reformatting the fixed disk. Some computers require a special program for this that is not automatically supplied with the computer, but resides with the retailer. Make sure you have this program, and know how to use it if you take a fixed disk into the field. Of course, any data should be copied on to floppy disks or other media for security purposes. It is in principle easy to copy the data from the floppy diskettes onto the fixed disk.
Ram Disks. Ram disks are not devices at all, but rather a computer program that `fools' the computer into using spare internal memory as if it were a disk drive. Computer memory has become very inexpensive in the past few years, and it is not uncommon for a modern microcomputer to have between 500k and 1 meg more memory than is normally needed for operation of the computer. A ram disk uses this memory for data storage, and makes it available to the computer as if it were a disk drive. This generally means that the ram disk can be used with unmodified programs that expect data on physical disks, floppy or fixed. Advantages of a ram disk for field computing are that it increases immediately-available data storage, and that the power to run the ram disk as memory is often less than that required to operate a physical disk. The basic message of this note is not to ignore ram disks as an unnecessary complication to your life; learn to use them.
Main Memory. Main memory is the memory required by the computer to work directly on the problem at hand, memory for the program and the data associated with the program, plus memory for graphics or other special purposes. Memory is not really that big an issue, simply get the largest option for the particular machine you select. Some programs will not run on machines with small main memories, but any programs that are written for a specific machine will always run in the maximum configuration. There are differences in machines in terms of raw computing power, but few applications currently exploit much of that difference.
The range of available computers is now so vast and varied that it may be useful to indicate the kinds of machines which may have some kind of field application. They are usefully categorized as follows:
Your choice of machine will depend on the kind of transportation problems envisaged, and perhaps a distinction should be drawn between transportation to and from a permanent field site, and portability in the field; between fieldworkers with access to reliable and flexible four-wheel vehicles, and the rest.
If transport is straightforward and easy then larger, more flexible, machines (desktops, transportables, luggables) are possible, such as the computer-printer system used by the Dyson-Hudsons [(.Computers 1986 Current.)]. A larger but transportable system is quite realistic if a voltage stabilizer is used. The advantages of a larger system is that it usually costs less, and provides a much better environment for working; in some cases the machine is much faster. The authors would, for example, see no problems in using a highly transportable machine such as the Apple Macintosh Plus in a field environment where portability was not required. One of the authors made quite good use of a desktop (Commodore Pet), and a couple of transportables (Apple II and Commodore 64) together with a small TV/monitor carried in suitcases.
We suspect, however, that most users will be more interested in obviously portable systems (portables, laptops, notebooks and handhelds), such as the Epson PX-8 used by Ellen in Indonesia, and the Toshiba 1100+ which has had light use in the past year in the UK and continental Europe. We have already touched on some matters relating to the portability of the Epson elsewhere [(.Ellen Fischer anthropological, Ellen Fischer permissions.)], including the finer points of selecting a carrying case.
Portable or laptop computers are usually equipped with a nominal 5-6v rechargeable battery, and the Epson PX-8 5.7v battery used by Ellen may be regarded as generally sufficient for 10-15 hours continuous use. The Toshiba 1100+ has a 4.8v battery, which lasts for about 8 hours with reasonable but careful use of the disk drives. Whatever the working circumstances such batteries will have to be re-charged periodically. When using portables or laptops under standard office conditions it is usual and advisable to keep the machine permanently plugged into the mains, unless the manufacturer advises against it. If fieldwork consists of relatively brief forays from an office with a reliable mains supply, other forms of re-charging will be unnecessary. The important thing is to check the battery-level periodically, so that you are not caught short of power in the middle of something important. Power shortage is usually indicated by a light on the machine, or by a message on the screen (especially on notebook and handheld computers); in some cases you may have to run a monitoring program.
The UKC Epson PX-8 incorporates circuitry to cut off charging after eight hours of continuous charging. This is okay when the machine is not being used, but if in use the battery will not fully charge. If you notice the battery voltage dropping below 5.2 or thereabouts, pull the plug out of the back of the machine for a second and then replace it. This triggers off another eight hour cycle. Don't do this unless the voltage is visibly dropping, since otherwise you may overcharge the battery.
Overcharging is a general problem. The most common form of rechargeable batteries are NiCads, which have a good life span so long as they are not overcharged, or drained completely. Most equipment that uses them must have some means or recommendation for use that keeps them charged properly. Some, such as the Epson, do this with a combination of timer and voltage-level seeking. The Toshiba has automatic control. Others simply have a timer. For still others the user is responsible for following written instructions. In general, you will not overcharge the batteries while charging with the machine in use; most machines use about as much power as the charger can deliver.
(a) Mains supply. Increasingly, anthropologists work in areas with some kind of mains electrical source, either supplied by a public or private utility, or from a domestic generator. Even if such a supply is sufficiently reliable for recharging a portable computer, there are a number of points worth remembering.
Firstly, electrical fittings vary from country to country and may not even be standardized within a single country. For this reason you may have to change mains plugs. Flexes carrying up to and including 220v, (<6 amp) usually have wires of 0.75mm diameter, are two-core and are fitted with a two point plug; flexes carrying 240v or more (6< amp) usually have wires of 0.75mm diameter or more, are three-core and are fitted with a three point plug. Remember to check for the correct plugs and wiring before entering an area where electrical goods are not readily available.
Secondly, mains voltage varies greatly throughout the world and the more remote an area the lower the voltage. Here are some examples:
United Kingdom 240 North America 110-130 Tokyo 220 Jakarta 200 Ambon 110 Lahore 90-350
\(dg As measured by Fischer; nominally it is 240v.
Most supply voltages are 220-240v. Some cities (e.g. Madrid) have both 110v and 220v supplies to the same building, which can be puzzling. You can use a simple step-up or step-down transformer if the supply is relatively stable: for computers voltage should not vary by more than about +-15%. Almost all modern microcomputer equipment internally step down the voltages to 5-12v. The internal regulators in the machine assume at least a 20% oversupply, and can generally handle a 35% oversupply. Fluctuations within the +-15% limit should cause no problems with operation, although you might notice some heat at the top end of the range. In a well designed machine heat should not pose a serious problem, although you should never cover any vents in the case.
If the proposed field site does not have a stable supply (this information can often be acquired from the local embassy or mission) then it is necessary to use a voltage regulator, which is a device that automatically adjusts to fluctuations in voltage, usually over a \*(+- 50% range. These are unfortunately bulky and relatively expensive (\z\(sp 80 or more), but do the job quite nicely. There are cheaper, manually adjusted regulators, but these must be watched constantly. Most regulators will also serve as a step-up or step-down transformer. If the voltage is subject to frequent outrages that last longer than ~1 second, you can either accept that you will occasionally lose your work, or or you can get an uninterruptable power supply (UPS). That can be quite heavy and expensive Stlg 100-500), but it will generally prevent any risk of loss of data (see below), and it will also replace a voltage regulator or transformer. In Lahore, Fischer satisfactorily employed a Matsunaaga Stavol SVC-350N stabiliser, which had a capacity of 350 watts, to run three machines simultaneously.
Thirdly, in choosing a suitable transformer, regulator, or UPS, you must know the power consumption of your computer and attached accessories. This is usually measured in watts, and is written in most cases on the back plate of each unit, usually near the power cord inlet. It is also given in the small booklets that accompany each unit. Unless you have a very unusual collection of equipment, either 350 watts or 500 watts should be sufficient, with most in the 350 watt range. Check that your power regulator can supply the total wattage requied by your equipment.
Fourthly, fuses protect your equipment from voltage and power fluctuations. It is absolutely necessary before taking any equipment into the field to know where all the fuses are located, and to have several spares of each kind. There are two general kinds, and they should not be substituted for each other. The usual fuse is a filament fuse, which contains a simple filament or small straight bar or material which blows when the power consumption increases beyond a set level. There is also a type called Slo-Blo, which will take about 3-5 seconds before expiring. This is common in computer power supplies, and will prevent unnecessary interruption if used. But if you replace a filament fuse with a Slo-Blo you can destroy the equipment. An additional safeguard is a power bar or power strip : an extension cord unit with four or more outlets which is usually fused. These are particularly useful if you are using a transformer or regulator, as these latter generally have only one outlet and you will probably require two or more.
(b) Heavy duty batteries. Where there is no mains supply, or where this is unreliable and solar recharging not possible, it may prove necessary to recharge from an independent higher capacity battery. In order to do this you should get a separate cable with alligator clips at one end and a plug for the computer at the other. We have found that a two metre length of 1.0mm cable is suitable for this purpose. It is important to use a heavy cable to minimize loss of power.
For limited `emergency' use it is useful to include some lantern batteries in your kit. You can easily buy 6v lantern batteries from UK electrical suppliers. They are usually 6.5 x 6.5 x 9.5 cms and weigh 500 grams (17 oz), and cost in the region of \z\(sp 2.50. It is therefore feasible to take a number of these into the field if you have any doubt about their availability. To recharge, the positive alligator clip should be attached to the positive terminal (+) of the battery, and the negative to the negative terminal (-) (that is black to black or blue to blue and red to red or brown to brown). The polarity must be correct. The other end of the cable is plugged into the computer. 6v batteries are also useful for emergency backup of a RAM disk such as the Epson PX-8 has, and which are good for three or four weeks on one charge.
For more sustained use of auxiliary batteries, 12v automobile batteries have the advantage of being more powerful \*-and you are more likely to find them in remote areas, especially in the form of lead and acid car batteries. In many Indonesian villages not connected to mains supplies it is quite common these days to find television sets driven by car batteries, and many people have been able to set up in business recharging and reconstituting such batteries entirely for this purpose. When using a 12v battery the same general instructions apply as to the 6v lantern battery. However, in addition you must add a small voltage regulator to the cable. The device required when using a mains supply is bulky and expensive (see above); but for this job it can be made from components available in an ordinary electrical supplier's (Fig. 1) or can be purchased cheaply ready-made (in the form of a car `cigarette lighter adapter' available for many radios, tape recorders and other appliances).
For Ellen's Indonesian trip we inserted this gadget in the cable connecting the computer to the 12v battery by using a terminal block. It is also possible to make up an extra complete cable.
Remember also that car batteries require attention: the acid should not leak, terminals must be kept clean and free of corrosion, and must be topped-up from time to time with distilled water until the electrolyte just covers the plates. Obviously, the climatic conditions of fieldwork will affect the amount of attention such batteries require.
(c) Solar cells. There are still many circumstances in which mains and battery-driven recharging is inappropriate. We have found that at least for certain tropical and sub-tropical locations solar recharging is extremely efficient, with the added advantage of flexibility. It is particularly useful where fieldwork entails much movement and uncertainty regarding possible sources of power. The equipment is light and uncomplicated.
Except for a short period in Ambon, the provincial capital of the Moluccas, the PX-8 batteries used by Ellen were recharged, directly, with the batteries remaining in the machine, using a solar pack connected to an eight metre length of 1.0mm cable. The pack contained three cells, was purpose-made, measured 240mm x 185mm (9.5 x 7.5 ins). It had a rubber-moulded back, and weighed some 755 grams (27 oz)1. Despite high levels of cloud cover and daily precipitation, this arrangement proved perfectly satisfactory for up to four hours of computer time per day. It took between one and three hours of direct sunlight to fully charge the batteries (3 watts at nominally 6v) from a reading as low as 5.16v, even with 50 percent cloud cover. More demanding electrical equipment would require a panel with more cells, and Ellen was agreeably surprised to discover that on Geser a panel of 33 cells was being used to charge 15v car batteries which supplied the local administration's radio-telephone link. CSAC now owns two additional SPK packs, each 5 watts at 12v, which are capable of powering larger microcomputer laptops.
To recharge, place the machine in the shade (usually in a house interior) and the pack in a spot where it is most likely to be exposed to the maximum solar energy available for the maximum duration. In general it is advisable to aim the cell at the zenith of the sun to get the best average. Ellen found the eight metre cable provided sufficient distance between machine and pack for all situations encountered. If the computer does not seem to be charging well from the panel it sometimes helps to adjust the position two or more times a day, to point at the sun directly. In reasonably open environments this should not be necessary, but nearby buildings, walls, trees or bushes may give rise to creeping shadow.
There is a small plug on the CSAC solar panels that fits the end of the long cable. If possible, this should be kept under the panel to keep it dry, although water is unlikely to be a major hazard. However, the naked plug should not come into contact with metal (for example, a corrugated zinc roof) and if necessary should be wrapped in cloth or plastic. The plug is under tension, but will come out if the cable is jerked. This serves to protect the cell. It would make sense to check that the plug is secure each day before use. The other end of the cable connects to a plastic block with another plug. The cable should be disconnected if the sun goes down or when it is generally overcast. This is not critical but is good practice. Overnight connection has unpredictable results, which could be bad, though a few hours should not make much difference. Though the solar pack and cable can be left exposed overnight (and this may be especially desirable if its daily arrangement is complicated), it is probably otherwise sensible not to leave it outside. Wind, moisture, rodents, insects and prowling humans could all pose hazards. Ellen got into the habit of assembling and de-assembling the equipment each day; the whole sequence of routines took him five minutes.
In conclusion, while mains supplies are increasingly accessible to ethnographic fieldworkers, in many situations an alternative is preferable: for flexibility, for back-up, or for working beyond the reach of piped electricity. Heavy-duty batteries are a possibility where available and if they can be transported. A cheap and efficient substitute is solar recharging. The equipment is light and simple, and where computer demands are low to medium, a small knap-sack size pack is sufficient.
(a) Visual displays. Poor night-time illumination (from pressurised paraffin lamps, torch light, low voltage mains or low wattage bulbs) may cause a certain amount of eye-strain and inconvenience when using a liquid crystal display (LCD) screen, which on the Epson PX-8 uses a display of 80 columns and eight lines (measuring 225 x 37mm). The screen is set on a hinge which can be folded down to protect the display when not in use, and a view angle control on the front enables maximum contrast to be obtained. Our experience confirms that natural sunlight is infinitely better than artificial sources in the field. Some LCD displays have built-in back-lighting, and the newer screens (Supertwist) are much better than the older versions.
(b) Storage of data. Magnetic data can be stored in the field either on microcassettes, each of which hold the equivalent of 30 typed pages of information, on floppy disks or hard disks (see above).
The use of microcassettes is rather painful in some ways. Loading and unloading data and programs is very time-consuming. The running lengths for discrete operations is often too short to permit the time to be used for some other productive purpose, yet too long to remain at the keyboard. The Epson PX-8 has a rather clever cassette unit which minimizes some of the inconvenience. More RAM (Random Access Memory) can reduce tedious cassette operations. The advantages of microcassettes over 3.5 inch disks is their relative cheapness (a consideration when large numbers are required), small size, ruggedness and resistance to heat and condensation and lower power requirements.
Fischer's machines used software on diskette, and thus avoided many of the problems Ellen experienced. Over a period of 20 months there was only one hardware fault, in the Apple's disk drive controller. This was fixed on site by a modification to the operating system software, but could have been solved by purchase of a new controller card, although there would have been some delay. Fischer was quite surprised by the lack of problems using floppy diskettes in the very dusty environment of the Punjab. His primary base was a small house with open doorways and windows, and the only precaution taken was to face the disk drive away from the wind, and to store the diskettes in dust-proof boxes (which are included with many brands of diskettes).
(c) Data loss. One fear which anthropologists have of relying on computers for primary data is that in the event of a crash all data might be lost, or data might be wiped from disk or tape through user error. This is a particularly serious problem when faced with the sudden loss of power on unstable mains supplies. You can avoid this by purchasing one of the relatively inexpensive battery backups which are now on the market. These provide from five to 30 minutes of extra power. Another, rather drastic, means of protecting data is by using a printer to get near-immediate paper output. But by far the best way of protecting programs and data-files is to back-up (that is, copy) systematically onto other cassettes or disks. This should be a perfectly adequate safeguard for short field trips where back-reference is not essential.
Moisture and dust can be removed from terminals and other components with a dry cloth, camel-hair brush or swab. Heat and moisture are potentially more hazardous than dust when working in the humid tropics, and proper service on returning to home-base is prudent. We recommend taking a small tool-kit into the field to cope with minor mechanical problems. Minimally, this might consist of spare fuses; several screwdrivers suitable for the machine itself and for adjusting mains facilities and battery connections; electrical tape; a sharp knife, and swabs and spirit to clean terminals. Napoleon Chagnon (CAAN 1988) coped with insect infestation by smearing peanut butter on sticky-tape attached to the keyboard to attract them out. We regard the inclusion of peanut butter as optional.