Welcome!to the Modems section
The page is intended to provide background info for those interested in modems & how they have evolved, with particular attention given to explaining some of the jargon.
If you wish for practical info on flashing a modem try Peter Curtis's site. If you are innately curious & inquisitive, read on... The parts up to 'Why is it called a 'Flash'?' are written by myself. The parts down from there are kindly provided by Graham Wood.
The terms 'Hard Modems', 'Soft Modems' & 'Flashing the Modem' are thrown around with gay abandon and, often, little understanding. I'm as guilty of this as anyone. I started using modems in the '80's & thought I knew exactly where the term "Flashing" came from. I was wrong. So, if you've ever wondered what all these terms mean, & others such as EPROM, EEPROM, & even EAPROM, explanations follow.
It's Good to Talk
At one time "hard" was the only type of modem that there was.
It began with the telephone. People liked the idea, and gradually a network of wires spread across most continents. Then computers appeared, and they got became popular too, and there were lots of them all over the place. Someone had the bright idea of connecting them together, and this could be easily done if each computer was fairly close to other computers, and thus 'Network' started to become a popular word. Inevitably the idea of connecting to far-distant computers whilst at home or on the move would crop up, and the link between computers and the telephone began.
Computers are digital - they speak in bits of information. People know of 'Yes' & 'No', but computers only know of 'Yes' or nothing. This makes computers think in black and white whereas people are also aware of the shades of grey that exist between. This makes computers much more stupid than people, which is why people like them so much. This is OK for computers, which don't know any different, but is a problem when you want to connect them to the telephone system, which was designed with people in mind.
The telephone system is analogue - it has soft and loud, and all shades in between; it also has bass and treble, and all shades in between. It also has pops & crackles & shooshes & burps & sometimes even the next-door neighbours arguing on it, especially if it is a mobile, and all of this makes it very difficult for computers to use it to talk to other computers. Still, someone finally figured out a way, & he did it by using the bits to MODulate a tone and then DEModulating the noise back into bits, and thus the MODDEM - sorry, the modem - was born.
Early modems were very slow and very expensive - just like early telephones & early computers - but they caught on and soon everyone wanted one, and they got smaller & faster & cheaper. Everything has it's prophet of doom, however, and in the case of modems this was a Mr Shannon [see Other Sources of Information] who, in 1948, had shown that the American telephone network had a maximum analogue modem communication speed of 34,822 bit/s.
In the meantime the telephone network was beginning to change. Perhaps it had caught a virus from all those modems using it, for it had caught the bug of the bits and was becoming digital also. It takes time - and a very large amount of money - to swop all the parts of a nationwide telephone network from electro-mechanical to electronic but by the nineties it was pretty much complete for most of the world's major industrial nations. This offered a tantalising prospect for modem communications - to avoid the Shannon limit by allowing computers to talk to each other in their own language... well, almost [the phone line from most homes still carries tones in an analogue fashion].
Hard Shoulders & Soft Verges - Coping with Change
In the years up to 1997 modem communication speeds had kept increasing and increasing until the state-of-the-art was the V.34 protocol of 33,600 bit/s - a good improvement on the 300 bit/s modem I was using in the '80's, but not a patch on the new technologies of ISDN and proprietary technologies of K56Flex™ and X2™ that were available in '97. Communications at this time were, in fact, in turmoil. The first soft modems had already appeared [more in a moment] and new models & standards were in such a state of flux that the modem chipset manufacturers needed to respond to give their buying public a sense of security. The modem flash ROM's were part of this response.
Modems have always had Read-Only Memory [ROM] to store the instructions that they need on what to do in which circumstances. Imagine the situation of dialling a number & it is engaged. Our ROM [Really Old Memory] gives us a whole range of different things to do - curse, make a cup of tea etc. - but Modem ROM is much more limited & usually says 'try again' or 'Hang up'. 'Flash' ROM is memory that can be easily re-programmed to give extra or different responses to meet new situations. It was the coming of V.90 that was, in particular, the new situation that modem manufacturers wanted to meet.
V.90 is written of in more detail elsewhere [detailed info & simple guide]. Introduced in 1998, & finally ratified in November of the same year, V.90 is a new method of modem communication that part-utilises the digital structure of the telephone network to bypass the Shannon limit and offer a maximum of 56,000 bit/s communications from Internet Service Provider [ISP] to consumer modem [the downstream rate] and 33,600 bit/s from modem to ISP [the upstream rate]. It is worth mentioning that 56,000 bit/s is never achieved in practice. The network has it's own V.90 limit of about 54,000 bit/s in ideal conditions, and speeds achieved in individual circumstances are very variable. Mine is almost always 49,333 bit/s.
Flash ROM has been very successful in meeting these changing circumstances, if a little puzzling for many users of modern modems. It is far quicker - and cheaper - to re-write the program for the modem ROM and distribute that program as a flash upgrade than it is to design, manufacture and distribute a new modem. The problem for these users is, I think, that they have never had to flash their microwave's ROM, or upgrade the control program on the Hoover, and this brave new world of computing and telecommunications is proving somewhat daunting. Graham Wood provides an excellent explanation of the origins of the term 'Flash' [& many others - below] so this section will be concluded by a few words on Soft modems.
If a modem has a flash ROM it is a 'hard' modem. This term has only come about because of the appearance in the '90's of the 'soft' modem, starting with DSI's SoftModem™, the "Connection 96+," in March of 1993. Soft modems are yet another attempt by the modem manufacturers to meet the challenge of changing circumstances, but they have the added bonus of being extremely cheap to produce and far lighter & less bulky than equivalent hard modems. Ordinary modems contain various components, some of which are specialised to the business of being a modem, and some of which are designed to interface with a computer. These components are implemented in a 'hard' modem as silicon chips. Graham gives a more detailed explanation below, but a silicon chip is basically a computer program implemented in silicon. The manufacturers of soft modems decided to throw away some of the silicon & implement the programs on the computer instead. The advantage of a program in silicon is that it executes extremely quickly - much more quickly than on most consumer computers - and when you realise that, as an example, V.90 modems are required to parse the incoming telephone signal at 8,000 times a second, the scale of the issue begins to become clear. Nevertheless, as cheap, fast computers started to become a reality so did soft modems.
Soft modems contain as little silicon as possible and therefore cannot be 'flashed' in the way that ordinary modems can. They can, however, be very easily upgraded simply by replacing the programs that control them with new ones. Somehow this seems more intuitive & simpler to understand for new users of computers than the hard modem procedure of flashing and then, perhaps, also installing new control programs. It also has to be said that the dire warnings surrounding the flashing process put many people off. Tales of modem woe indicate that this is not an idle concern, whereas a soft modem cannot be damaged by an upgrade. It simply won't work, until the original programs are restored, whereas a wrongly-flashed hard modem will roll over onto it's back with it's tongue hanging out & stay that way forever more [although the better flash upgrades will check first to see if the modem is suitable, & stop if it is not].
Computer mainboards are currently available with both video & sound chips integrated onto the board and it seems likely that very soon soft modems will join this trend. They are also very useful in mobile computers where their low power consumption, light weight & low cost are positive advantages. As a support professional and an owner of two different soft modems I have to say, however, that they are at present a menace to reliable communications and would recommend anyone to avoid them for a year or two, if at all possible.
What follows now was provoked by the question "Why is it called a 'Flash'?" & is provided by and is copyright Graham Wood, a long-in-the-tooth computer/electronics professional with an interest in Amateur Radio, Photography & Astronomy who has hands-on experience of these matters:
Why is it called a 'Flash'? - A Technical View
The term comes from the original 'fuselink' ROMs where all the bits in the ROM were manufactured in one state and to 'program' it the ROM programmer box cycled through the addresses and imposed a high voltage/current on those bits that needed to be in the other state. This high voltage would 'flash' i.e. burn-out the fuse for that bit of the ROM and it (the bit) would then assume the other logical state. Once a bit had been changed it was impossible to change it back.
Hence 'flashing' the ROM and the programmer box becoming known as the 'burner'. Also the phrase 'burning the ROM' is synonymous with 'putting the program into the chip'.
Chips with Windows
Those chips with windows on them are EPROMs meaning Erasable Programmable Read Only Memory. The window is made of quartz or fused silica and is used because of its extended ultra-violet [UV] transmission properties. The UV light is used to erase the contents of the EPROM and return every bit to one logical state. In physics, UV light has a much higher energy than visible light and so can do the job in a few minutes whereas visible light at the same intensity would take weeks or years or maybe not at all. The duration of the exposure is controlled by a timer on the eraser box and is usually around ten to twenty minutes. The eraser box is sometimes known as a 'cooker' since the EPROM is placed on a tray which is slid into the box then you set the timer and wait.
The UV light that erases the data does this by temporarily ionising the capacitor and this causes the charge to leak away. The programmer box does a 'blank check' on an erased chip to make sure that all the bits have been set back and none of them are 'stuck'. Hence the term 'Erasable'.
When a new program is placed into the EPROM, the EPROM programmer box cycles through the addresses and imposes a high voltage on those bits that need to be changed to the other state. This high voltage sets a small electric charge into a tiny capacitor in that bit of the EPROM so that its logical value is changed. This capacitor is part of the structure of the chip and holds its charge for a very long time.
The duration of the high voltage is quite critical and different types of EPROMs have different requirements of both voltage and duration. Typically, a duration of 10millisecs and 12.5 volts are used these days. The early programming boxes used to hit the bit with fifty pulses of 1 ms duration and 5 to 10 ms of gap between them. These would take ages to do even a small EPROM.
One I programmed in June 1981 with a control program for a job is still going strong and has never been switched off [24/7 365 day operation] apart from a couple of power cuts.
Now the ROMs in modems and motherboards that are described as 'flash' types are really EEPROMs. Electrically Erasable Programmable Read Only Memory. These work on the same 'charged capacitor' principle, but use logic gates and special circuitry around the capacitor to discharge it in order to erase the data whilst avoiding the possibility of leaking the charge away during normal use. So these types don't need a window to provide erasing capability.
Another window-less type is the One Time Programmable EPROM which is a conventional UV type EPROM but intended for small production runs where the chip is only programmed once and if a change of code is needed later, it is taken out and binned. The plastic case that these come in greatly reduces the cost of the device to make them competitive for small firms to use.
Another variant of the EEPROM is the EAPROM. Electrically Alterable.... but I think that this is just a branding thing and the above principle still holds. Both these latter types have only become practicable in the useful larger sizes during the last few years and this is due to the improvements in manufacturing techniques which reduce transistors to ever smaller sizes.
To finish off the series there is the mask programmed ROM. This is not an erasable type but has all the ones and zeros that the code comprises actually designed into the structure of the chip. This is done at design time and costs a lot to do, so it is only for large production runs that can absorb the one to ten thousand pounds it costs to set up the masks that are used to manufacture the chip. The reason for the high cost is that these masks cannot be used to manufacture any other chip.
Other Sources of Information
Flashing a modem [practical advice]: Peter Curtis's site
Shannon's Limit : 56K bits/s Data Transmission across the PSTN - Page 6
PC-Tel Soft Modems [Tech info, 1997]: Host Signal Processing
Motorola Soft Modems [Tech info]: Software Modems
V.90 [Tech info]: 3Com V.90 Technology
White Papers [this site]: Downloads & Links