Harold E Price NKK

Misery Loves Company

In my previous column, I'd lost the will to live. I'd asked for reader input, in hopes that I'd find something to cheer me up. I said: "If you disagree (or agree) with any of the above, please write to my email or US mail addresses. If you have an interesting application that is running on the packet network, let me know. Aside from some character-based white pages, I haven't seen much. We seem to be working hard, but I'm not sure we're having any fun."

I had several responses, all more depressing than mine. I've included the most depressing in the "Mailbox" section at the end of this column. I had no response from anyone with an interesting application.

Real World

In the July/August Issue of Wireless, a trade magazine for cellular and other wireless data communications, I found more cause for depression. First, in an article by Ira Brodsky, "Cellular's Multi-Pronged Data Strategy," the Amateur Radio service gets a plug. Actually, it is more plug than we deserve. Mr. Brodsky states: "First adapted to radio by amateur radio operators, packet radio differs from landline packet networks..." We weren't the first users of packet radio, of course, others get that credit. We were the first to take existing technology, cheap VHF radios, cheap modems, cheap single board computers, and produce viable systems for a few hundred dollars. This was at a much lower cost than ever before. More than 100,000 were sold, probably making us the largest user, in sheer numbers, of packet radio, at least for a while.

5949 Pudding Stone Lane Bethel Park, PA 15102 email: [email protected] (Internet) 71635,1174 (Compuserve)

It is nice to get a small plug now and then. Unfortunately, the same magazine gives an indication of how hopelessly out of touch we are now. Those of you not in the industry may not have a chance to see the ads for current wireless data systems. In "The Missing Links," by David Toll, a list of leading wireless data link products is given. The systems range from a low rate of 512 kbit/s and a range of 30 miles to 8 Mbit/s and a range of one mile. All of these systems use spread spectrum, mostly in the 902 to 928-MHz band or the 2.4-GHz band and do not require an FCC license. Many are directly Ethernet or token-ring compatible.

The only thing keeping hams from running right out and buying some for the local net is the price: $10,000 to $20,000 per pair. Remember, this is the commercial price, though. Some of these 900-MHz systems come with $700 price tags for omnidirectional antennas and $5000 for directional antennas. The electronics have a similar burden of real-world engineering and marketing overhead. We need an initiative like TAPR's TNC project in the early 1980s to repackage the chips that make up these systems and build a ham-priced box. For the next column, I'll try and find out what can be added to a 900-MHz antenna to make it cost $5000. Even with standard markups, this is an amazing price. We only need a 10-to-1 reduction to bring the price into the range of local groups, 20-to-l for the individual. We've done it before.

Interestingly enough, I caught a hint of a possible project that fits this category as I was preparing this article. See "Fear of Success" in the "Mailbox" section.

Computer Networking Conference Proceedings

For some reason, the Proceedings of the 9th ARRL Computer Networking

Conference have been very popular. In addition to carrying the definition of the Pacsat data protocols, other articles are also continually referenced. The 9th has been out of print for a while. TAPR has arranged to reprint and distribute the 9th, as well as the proceedings from other years. The agreement was described in a recent announcement by TAPR: "TAPR has arranged with the ARRL to be the agent for past Proceedings of the ARRL Computer Networking Conferences and Digital Communications Confer ence Proceedings. This will apply to proceedings that are more than two years old. For example, the ARRL will continue to distribute the 12th (1993) and 13th (1994) proceedings this year, then next year TAPR will begin to distribute the 12th (1993). In this way, TAPR and the ARRL hope that these proceedings will be fully available to digitally interested hams. Proceedings are available currently from the 1st through the 11th. The 9th CNC was out of print, but as part of the agreement we have reprinted the 9th and it is again available. TAPR would like the thank Mark Wilson, AA2Z, and Jon Bloom, KE3Z, for their help at the ARRL with this arrangement."

Contact TAPR at: 8987-309 E Tanque Verde Rd #337, Tucson, AZ 85749-9399, tel: 817-383-0000, fax: 817-566-2544, for more information.

Mailbox

Lyle Johnson, WA7GXD, was heavily involved in the TAPR TNC-1 and TNC-2, the Microsat CPU, various DSP projects, and is currently working on the Phase 3D spacecraft GPS system and CPUs. He writes in response to my August "Depression" column:

"I just returned from an EDA forum (Electronic Design Automation, the software tools used to design ICs, PCB, etc). I had an opportunity to speak with some manufacturers who spend many tens of millions of dollars per year on UNIX software R&D. I asked them when I'd see lower-end UNIX tools so I could bail out of Windows and use a 'real' operating system. I was quite surprised at their responses, which were largely along the following lines:

'Which UNIX? We spend, on average, 30% of our software development budget porting our applications amongst the various incompatible UNIX platforms and implementations. It isn't a simple matter of retargetting the compiler from SPARC to PA-RISC or i486. A binary compiled for Solaris 1.0 isn't compatible with Solaris 2.0. And next year, when Sun jumps to OpenStep (their version of NextStep), that won't be compatible either. And AIX isn't the same, and HP-UX isn't the same and...

'On top of that, the GUIs are all different. Last year it was News, then it was Motif, then it was X and OpenLook. Frankly, we're pumping resources into porting to Day ton a (the next Microsoft NT system). We'll still support our UNIX customers, but we're planning on migrating to NT because:

• Microsoft allows backwards compatibility. You can still run the application you bought for your 8088 DOS 2.0 PC back in 1984 and run it on your Pentium DOS 6.22 PC you bought in 1994. I can't do that with UNIX. I have to continuously redevelop, and my clients have to continuously upgrade. Some customers won't or can't upgrade to the latest version of "UNIX." Support is a nightmare!

• NT is written by a monolith— Microsoft—for better or worse. This means the APIs will be the same for the Power-PC as they are for the 486. I can literally just retarget the compiler and have my application run on Intel; or Alpha, or MIPS, or...

• With NT, we'll be able to effectively add nearly 50% more staff devoted to making our products better with the same budget.

• We don't care that NT hasn't taken off yet. It will, because the tools people buy from us now will be better and cheaper under NT. Microsoft stuck to Windows from. 1985 until it finally took hold with the release of Windows 3.0 in 1992. Bill Gates has the ego and money to make it happen, and it'll happen.

• We aren't interested in

28 QEX

"The point is that, among people who make a living at manufacturing and supporting products for UNIX (as opposed to products for Sun, or other specific platforms), UNIX is a nightmare.

"Linux is just another flavor of UNIX. Sure, it is available for $40 on a CD-ROM, but I can't run it on my PC at home Why? My video card is made by Diamond, and there are no Linux drivers for it. And, I can't run the applications I need under Linux (MS Word 6.0a, for example, or the interface program to my multi-mode controller, or my callbook database, or...) So, I have to reboot every time I want to do something else. Not practical, even (especially?) for home use in the ham shack.

"We use Linux at work for our commercial application— DISPATCH. We love it. Yet, when we try and buy a laptop PC for our sales staff, it turns out that we have to buy expensive high-end color machines (rather than cheap $2 k WinBook machines) because Linux doesn't support the Western Digital Rocket Chip video ICs (yet). We have to spend an extra $2.5 k per laptop to run a 'free' operating system.

"If you are buying a new computer from Zeos, or Compaq or Gateway, try asking the sales person if the video cards, and CD-ROM, and (fill in the blank) will run under Linux. You get silence. If you ask what the cards are so you can research it yourself, you usually get inaccurate information (they may ship a different video card, for example, because that was the one they could buy that day, perhaps due to chip manufacturer's long lead times). We bought a lot of PCs from Gateway for the office and Linux ran just fine. The next lot, purchased just a few weeks later, didn't. Turned out they had changed the video card. The new one ran great under Windows, but Linux couldn't talk to it. We wound up having to replace the video cards.

"Bottom line? I contend that UNIX is completely impractical for the average ham, and maybe even for a reasonable fraction of the techies. I think the question is MSDOS/Windows on Intel platforms versus everything else. It's too bad, because UNIX offers a lot of resources and operating flexibility that hams could use. But, using UNIX in the ham shack is like trying to run MOSAIC and the Word Wide Web over a 1200-bit/s packet link—it just isn't practical."

Fear of Success

Lyle is touching on one aspect of the problem I call "haves and have-nots." Just after the start of the "packet revolution" everyone was at the same starting point. Most of us had an IBM PC clone and most of us had a TNC-2 clone. The required software was a terminal emulator; wiring was PTT, mic in, speaker out and ground. Now we've exploded again into a plethora of computers, operating systems, TNCs, modems, radios and protocols. The technology "haves" know what they are doing, have read the RFCs, write their own device drivers and design their own interface hardware. The "have nots," while functionally literate in other walks of life, don't know why min(++a,b) didn't do what they expected, can't figure out how to cohost Linux, Windows, and OS/2 on a double-spaced disk, and don't know how to add CTS to their data cable.

As the gap between solution providers and users grows, we end up with potential providers who are unwilling to throw themselves into the bottomless pit of needs. I can attest to the problems that come with publishing solutions. I still get the occasional letter, based on two articles I wrote for QST almost ten years ago, asking for help. Usually from some remote island, sometimes from students who have pooled their resources to buy IRCs, I'm asked to provide an interface from computer X to TNC Y to radio Z, when I've never heard of X, Y or Z. It is painful to ignore these requests, but even more painful to try to answer each one. Glenn Elmore, N6GN, says it very well:

"My recent posting has generated some questions about the availability of information, boards or kits of the 200-400 kbit/s radios we are using with PI2 cards here in northern California. I think I may need to explain the situation.

"With regard to construction articles or other detailed documentation: I haven't yet published any. The fact is that I'm afraid to. The 2-Mbit/s, 10-GHz link published first in Ham Radio and then in the ARRL Handbook has netted around 500 telephone calls, letters and pieces of email. My phone still rings about once a week about it. I think I'm nearing DXCC with countries of requests. This even though the original article is almost five years old and was complete enough to be a construction article (as proven by those who constructed it). Needless to say, my XYL isn't too happy about it.

"The 904-MHz radios, interfaces and antennas have more parts, are more complex and require working with 10 watts at 1 GHz, which is arguably more difficult than dealing with a 'canned' low-power 10-GHz transceiver. All this is simply not easy to 'kit.' I'm afraid that if I publish the plans (which are quite complete and include circuit boards and assembly drawings) I would generate requests for assistance that make the 10-GHz link experiences pale in comparison. I'm sure I'd have to change my telephone number.

"Still, I really do want to help as much as I can. The information on col.hp.com under -/hamradio/ packet/n6gn was put there to help alleviate this problem. It includes 10-GHz info, 904 antennas, ARRL CNC papers and more. I try to add to it as I find things that might be of interest.

"I hope this doesn't sound too stuffy. I'd really like for folks to be able to copy this hardware and have fun improving amateur radio networks, but I simply don't have the resources to support what I'm sure would ensue if I published a 'how to' article. Sometimes I wish that hardware were as easy to clone and codevelop as software.

"My approach is now, instead, to work on a 'layer 3 TNC' which is to be an entirely printed board, a very inexpensive combined direct sequence, direct conversion, spread spectrum 1200-MHz RF assembly married to a 68302-based digital controller. The whole works is intended to be in the cost range of current multimode TNCs but have an antenna connector on one side and host connector (nominally a parallel port but optionally Ethernet) on the other. It is to provide layer 3 connectivity to the associated host. Clearly, it needs to be an open environment and docu mented well enough that many can work on it together.

"This requires considerable software to allow economy of design. I'm attempting to perform many previously hardware functions with s/w. Additionally there needs to be a mechanism to provide user RF path measurement and qualification. Fortunately the same hardware that can recover the -16 Mcps DS SS signal can also serve to measure path loss and multipath component. 'Just a simple matter of programming.'

"This is something I'd hope would be generally available, once there is a mechanism set up to support folks so they can reasonably expect to be successful. However, this is an enormous project and may not be finished in my lifetime.

"I wish I knew how to partition a project like this so that it could effectively be pursued in a physically distributed manner. But my experience to date is that close physical proximity of the contributors is extremely important. It also requires a breadth of skills: RF, CS/networking and organizational. All contributors need to cooperate well since they each have vital contributions to bring to the table and there are often solutions in one 'domain' which are much more expensive or obscure in one of the others. There's no question that I need help, I just don't know how to appropriate it."

I'll be talking to Glenn about getting him some development help. If anyone else has a depression buster, please email. m

Learn DSP and put your knowledge to work IMMEDIATELY!

Coming to a City Near You

Scottsdale, HZ Austin, TH San Jose Long Beach, Cfl Rtlanta Denuer Ft. Lauderdale, FL

L Domain'IWhnul agios.Inc.

mm-m^wm

Hours 9-5 EST. Ask for a brochure.

Our 2-Day Advanced Course is ready. Call for more info.

By taking this 3-Day Course you will really learn DSP. Guaranteed!

0 0

Post a comment