Ricochet Bounces Back, Cautiously

SimHacker writes: "An article in salon.com reports that the Ricochet wireless network will be bouncing back from the dead! Aerie Networks, who purchased Metricom's Ricochet network for $8.25 million, is going to offer the service in markets where it was popular, like Southern California and the Bay Area. They're also planning to lower the price of the modem from $300 to $100, and lower the monthly flat rate fee from $80 to $50. Ricochet is hardly the perfect wireless network, but it's much faster and more reliable than CDPD, so I'm really looking forward to signing back up."

Comments

Richochet is cool

[Lumpy]

Roichochet makes CPCD look like a joke. It's based on packet radio technology and the infrastructire design is closer to a military setup than anything you can do with cellular. You can communicate modem to modem directly WITHOUT paying for service, and if you pay then you can connect MODEM to MODEM through the Richochet network.

It's really cool.. My forst wireless network was a pair of their 19.2 modems... with mods to the base station I set up I could get about 3000-5000 feet range.

What killed ricochet the first time

[t0qer]

I came into the company at a time when the .com boom was just starting to happen. I was young and didn't quite understand business politics yet, so I thought it was right to point out when things are outright ripping off the customer or in the very least preventing the company from dominating the market. My "self rightousness" cost me my job.

You see my freinds, the ricochet development cycle really ended after the first modem was introduced. Sure it got smaller and faster, or so you think. The ricochet was allways capable of 128kbps speed. There was a s register that could change the modem speed to that maximum rate, but unless you were transferring from ricochet to ricochet at a distance of 100 feet or less, you would never see 128kbps from their network. This is because the poletops were set at 9600baud.

Now to understand how you can get 28.8 from poletops set at 9600 you have to understand how the ricochet network works. Basically you are surrounded by these poletops, all shooting out bits at 9600 baud, they are multiplexed together by your modem and combined to get the desired bandwidth. Thus 9600 from 3 poletops would give you 28.8. Internally people who knew about this and thought it was wrong were fired over the years. There was a lot of them trust me.

Whenever a new modem standard like 33.6 or 56k came out, metricom would release a new "Modem software upgrade" that "contained new code!" that would magically turn your 33.6 ricochet into a 28.8 one. All it did was change the default setting of that S Register, maybe some new stuff was added, but thats about it. Nothing really magical or fancy, they fired all the real engineers that created the modem in the first place long ago. All that was left was a skeletal crew that could never really improve the internal electronics design.

When they were "Upgrading the Ricochet Network!" this was nothing more than more smoke up the ass of ricochet users. The poletops speed was simply set from 9600 to anything higher. Just a stupid S register that was allways there.

I think Ricochet's real downfall wasn't the technology, when it was introduced allmost 5 years ago, it was capable of delivering 128kbps service. So the failure can only be found in the strategy used by the marketdroids. $20@month for 128kbps wireless internet service vs $20@month for a standard 28.8 isp would have sold a lot more modems than the $40@mo ricochet $20@mo standard ISP model that they took.

They did do an amazing job creating the network, just a shame that they never put that same effort into people that acually understood the internet market. People have allways gone with the cheaper ISP simply because they want to save money. Anyways I hope no heads roll from my comment.

Oh in case you're wondering what the magic s-register was, its ats304=115200. The reason they made it so slow in the beginning is back then most motherboards were using a 8250 UART, which was limited to 14.4 speeds.

Re:What killed ricochet the first time

[Anonymous Coward]

Partially true, I worked for Metricom in the engineering department for two years before they went kaputz. The S-register you refer to 304 was actually the data terminal speed. This register controlled the speed at which the radio talked to the computer it was connected to. Setting this speed at a maximum, as most of you old school BBS freaks know is always recommended. setting the S 304 register to 0 would let it autodetect the data terminal speed up to 57.6kbps. I personally set it to 115200 to get the maximum possible throughput even though it really didn't matter with the Gen 1 radios. This is not the same as the speed at which the radios talk to each other however. The poletops for the Gen 1 network (33.6 or 28.8 or whatever you want to call it) were entirely different animals than the 128k radios. The 128k poletops could talk to the Gen 1 radios, but you could not get Gen 1 poletops to talk to 128k radios at 128k speed. This is also why it is recommended to use the USB cable when using the 128k radios. The maximum a serial connection would let you download at is about 10k because of the 115200 limitation. Most people don't have high speed serial ports, don't know they have em, or whatnot, so even though the 128k radios work with a high speed serial (about 460kpbs IIRC) it was a big hastle. I can remember trying to get test setups in the lab to work with high speed serial but there were too many cabling attenuation/interference problems to get it to work that fast all the time. Serial cables are simply not designed to handle that kind of signaling speed. USB was better suited for the fast transfer rates and is why you should use it whenever you can. Trust me, they had to roll out completely new poletops in the entire Bay Area when they upgraded systems. I remember taking a "Woody" radio home and using it on the 28.8k network until they turned on the new poletops. I am afraid you are misunderstood about the 304 register and hopefully this cleared things up a little. If anyone out there can add more to this, I'd be glad; it's been a while since I worked at Metricom.

Re:What killed ricochet the first time

[n6mod]

Because the serial ports on the poletops were always set to 9600 baud. Changing 304 on a poletop would do exactly zip to throughput through that poletop.

Your understanding of the entire system is bascially flawed. 304 is the DCE speed ONLY. Nothing to do with the radio. The original "28.8" Ricochet had an over-the-air rate of 100kbps. The original system also used the same protocol and frequencies between the portable and poletop as it did between poletops (and therefore poletop to WAP).

If you RDest'ed into a poletop, there was another S-register (one of the 800's, it's been years) that contained the actual radio speed: 100000. Change that and you just bricked the poletop. ;)

Saying that MCOM was ripping off the customer based on this bogus analysis of the system might have had a bit more to do with your longevity there. ;)

As to the "Autobahn" system, development certainly did continue after the first Phase 1 shipped. Autobahn (the 128k system) was a massive re-engineering, using higher speed (256k? Faster? Anyone know?) 900MHz links to the portables, and 2.3GHz (WCS) and 2.4GHz (Part 15) links from WAP to poletop and poletop-to-poletop. Going to WCS helped bound the latency, which was always an issue with the original network.

Yes, you could get 100kbps from the old radios in Starmode. (Check out /dev/st0, boys and girls!)Yes, Starmode was initially supported in the Autobahn system, but sadly went the way of peer-to-peer connections under the brain-dead Dreisbach administration. Perhaps Aaronson will remember the glory days of the early network and lift that restriction.

And, to get back to the original subject line, what killed Ricochet the first time was overpriced modems and service (which Aaronson seems to have corrected), horrible marketing, and serious overextension, trying to build out too many cities too fast. There were other brain-dead decisions along the way, going all the way back to announcing Autobahn in '97 and not delivering it until '00, killing Starmode and P2P, &c.

Oh, and "next to the storage area" described the location of tech support for most of the time I was there, even though that was a couple of different locations. Tech Support was next to the doors from the small parking lot when I was there.

-Z
Ricochet Tech Support 2/95-5/97