[First of all, I’ve been very busy as of late, hence why the steady stream of blog posts came to a halt. Today, I’m trying to crank out a few for you all to enjoy (or otherwise ignore) – so definitely relax and take in a little more tech :)]
In one of my earlier tech-flashback posts, I had posted pictures of my Banksia external modems, and their internals. That got me thinking – why don’t I show you guys some of my other modems as well? So here’s an onslaught of external modems from my collection. (I have a few more somewhere, they just aren’t easy to get to, so they’re not included in this post.)
So lets start off with a legend – USRobotics Sportster Flash. This modem was legendary – USRobotics eventually became 3Com, but these modems were released with a software based modem running on a Texas Instruments DSP allowing a lot of upgradability. Some guys who bought this modem in the 28.8k days managed to firmware update it to 56k, V.90, and some to V.92 contingent on having enough Flash (one of the limiting factors). This was a modem which was much better for auto-mode detection, voice support and connection reliability being a very “premium” model.
I actually wrote down the enabled firmware options on this modem – one of them is x2. X2 was a propietary 3com/USR Extension that allowed for 56k connections back when K56Flex was battling it out with x2. Eventually a V.90 standard was created by the ITU which pretty much made x2 redundant – in fact, there were very few x2 dial-in servers deployed to the point I don’t remember seeing any in Australia – and only a few in the US. I’ve heard an x2 connection thanks to the internet, and it does sound different to V.90/K56Flex, and some claim it to be superior for reliability.
The underside of the modem – it reveals the legend for the LED display and the Austel approvals. there is a cutout for the inbuilt microphone.
The rear shows a DB-25 Serial port, a speaker jack, a line and phone jack and a power jack, as usual for most modems. Older modems tended to use DB-25, and more modern modems tend to use DB-9 instead since they’re more compact.
The internals of the modem clearly show the onboard speaker, the volume adjustment buttons, RAM, DSP, line hybrid, Intel Flash, and Microphone. Lots of capacitors too.
The bottom side of the PCB clearly shows the USRobotics branding and the copyright of 1997.
Another related modem with a very similar design is the USRobotics Message Modem. This one looks visually identical to the Sportster Flash – but it has slightly different firmware options.
One thing to note is that this one supports V.92, the last modem standard from the ITU. It doesn’t support x2, which isn’t used anymore, but otherwise it looks almost visually identical. The branding is now 3com as well, showing that this is a late model modem. But this modem has a special trick up its sleeve – it’s called a Message modem because it is practically an answering machine. It’s capable of answering fax, data and voice calls and recording messages onto its onboard flash for later download, even when the computer is switched off! Unfortunately, as I don’t have the software for this one, I can’t do the download, but the modem still “adaptive answers” the modem. Unfortunately, I couldn’t get this modem open because it’s physically clipped tightly shut.
The next modem is the Swann Speed Demon – this one was donated to me by a high school friend, Ji Bin, and I’ve still kept it well. It is based on the Ambient chipset, which was bought out by Intel – and this chipset is V.92 capable by firmware upgrade. The modem itself is somewhat large physically, and has a unique curvy translucent exterior.
This modem was also available in an earlier revision that had a red translucent case. This is probably a later version PCB, that has no phone jack, just a line jack, leaving an empty hole in the back of the casing. The AMIC EEPROM containing the firmware is nicely socketed in case of “bricking” by improper firmware update. The hybrid transformer is in the bottom left hand corner, with some of the line devices unpopulated (probably for other line impedances around the world).
Next is the Netcomm Roadster II USB, another modem donated to me by a high-school friend, Lesley. This is a Conexant/Rockwell chipset based modem with an Intel USB interface chip, and is bus-powered. It too, has a blue translucent case and is relatively compact. It has inbuilt speaker and mic for speakerphone operation with external jacks for use with a headset.
The same kind of modem is available in a serial variant – this one unlovingly given to me without its power adaptor, hence the written specs in the corner:
Being a more modern modem, it sports a DB-9 connector instead of the DB-25 found in older modems, but there’s nothing special to be seen on the outside.
From the internals, we can see it’s an older Rockwell chipset which is pretty much functionally identical to the USB version. As with later devices, it’s quite common to see the flash memory which contains the firmware soldered to the PCB to save on the cost and reliability implications of using a socket.
Another unwanted modem – this I managed to get off the street, is an AcerModem 56 Surf. This one is unexpectedly not based on the Rockwell chipset, instead it uses a Motorola one. Again, modern modem, DB-9 connector.
As part of the cost-reduction of modems, you can see that a full speaker has been forgone in this modem, instead replaced by a tiny tinny buzzer (quite a common thing to do). Also, the full voice capabilities are only available through the rear connectors, by providing a headset and or external microphone.
So, lets look at another modem – the Aztech EM6800, which was another TI DSP based modem, similar to the USR Modems, but instead the firmware as a TI provided one. This one had very good voice capabilities (provided external headset is used) and performed moderately well – this one had been previously disassembled and battered.
Now back to some old 33.6k modems. Quite expected is that they are Rockwell based, and just so happens to use the 28.8k V.34 based chipset, as Rockwell did qualify these parts to work at 33.6k rates as well. Brand-wise, the first one is a KTX, the second is a Spirit, but as far as brands go, they aren’t anything special.
Interestingly, the line interface is on a plug-in daughterboard, allowing it to be replaced should it be damaged from a storm lightning surge. I guess this shows just how expensive the modems might have been and the need to repair them was a design consideration. Or maybe it was only because of the varying telephone design considerations of meeting different line standards around the world.
The Spirit modem:
Through examining my collection of modems, there’s several interesting points:
- The Rockwell/Conexant chipset is the dominant chipset. In most modems today, you will find this ubiquitous chipset – and that’s a good thing because the Rockwell chipset has a decent AT command set (very similar to the original Hayes, with some extensions), and has decent Fax mode 1 and 2 support. Some modems will have Voice and SVD, and MNP (Microcom Networking Protocol) capabilities enabled. SVD was very little used, it stood for Simultaneous Voice and Data, which allowed you to talk to the person at the other end (voice digitized and compressed by the modems and data rate stolen from the data channel) at the same time as the data transfer (which would slow down a bit due to bandwidth lost). It was an interesting curiosity, but I didn’t ever see practical use of it.
- Price reduction antics – it’s clear that as modems got along, the price pressures resulted in several changes, including the removal of Mics, the changing of speakers to buzzers, the shrinking of PCBs and the soldering of virtually everything to the PCB including the flash ROM. Some even reduced the number of parts, sockets and LEDs as another way.
- Connector Changes – DB-25 connector was well and truly out of fashion by the mid-late 90’s, and so the move to DB-9 meant smaller modems, smaller connectors and possibly cheaper cables. The extra signals on the DB-25 connector were ill defined and rarely used, and I’d hazard to say most modems wouldn’t provide much other than TXD, RXD, CTS, RTS, DTR, RI, CD, GND.
- Other chipsets did exist, and did have some benefit – and some were even DSP based which allowed flexibility as the modem modulation was defined through software rather than in an ASIC. Voice capabilities varied from modem to modem, and some were better than others – I distinctly didn’t like the voice capability of the Rockwell chipset which defaulted to 7200hz sample rate, while others did 8000hz ADPCM.
- Serial Modems were great as one could use them “driverless”. All one needed was the AT commands which were relatively standard to make a connection, although the diagnostic report data could be different in format (resulting in Windows reporting connect rates of 115200 or 57600bps port-rate). USB modems weren’t as lucky. They were also hardware, so the processing was all done in-modem, and you basically shoved it commands and data and it would “take care” of it for you. The Internal modems also came in a Winmodem flavour, which was like having a telephone-sound card and having the modulation/demodulation done in software. It was fine as long as the hardware resources were available – but if they weren’t, reliability would take a steep dive.