Compatibility lessons from the USB

Sun, 01/31/1999 - 7:00pm
Walter S. Ciciora, PH.D

I've repeatedly been asked two questions regarding the Decoder Interface.

  1. Why do we need to have a "plug fest"?
  2. Why do we need to specify a communications protocol?

A "plug fest" is a gathering of manufacturers who have built prototypes of both sides of the Decoder Interface: TVs, VCRs, other devices with "cable-ready" TV tuners in them on the one side; and descramblers, digital television decoders, electronic program guide boxes, Web appliances and other "add-on" units on the other side.

A communications protocol is a very well-specified set of rules by which these devices communicate. Without it, there is chaos. The Decoder Interface protocol is described in the EIA IS-105.2 standard, while the electrical and mechanical aspects of the Decoder Interface standard are described in EIA IS-105.1.

I recently had an expensive and aggravating experience which drives home the answers to these questions and which should be a lesson as we move forward in a complex age of interconnecting devices which are supposed to work together.

I have owned a Toshiba Portege computer (model 320CT) since April 1998. For the most part, it has been a satisfactory performer with just a few modest problems. Because it was purchased early in the year, it came at a premium price: $3,600 plus another $500 for the "port replicator," which includes the CD-ROM drive. (The port replicator was two months late in delivery). In late November, I bought a color scanner. I decided that I wanted a top-quality scanner which would yield a serious productivity improvement. So I set aside concerns about price and paid more than $500 for a Hewlett-Packard ScanJet 6250C. The box held wonderful promises:

"HP Intelligent Scanning Technology. The Power To Be Productive."

And displayed in big type: "USB Universal Serial Bus," with a fancy logo meant to say "connectivity". The box reads: "Universal Serial Bus Standard … provides conflict-free connection, plus fast setup with no card to install. Check your PC for the USB-capable symbol …".

Well, my Toshiba Portege has that same USB-capable symbol over an appropriate-looking connector. So I carefully followed the installation instructions. I loaded the software from the CD-ROM, connected up the USB cable and clicked on "HP ScanJet Software, HP PrecisionScan Pro." But what I got was that Microsoft crashing sound, a red circle with an "X" in it, and the message: "Sorry, scanner could not be initialized (scanner not found)." This was followed by an "OK" button. Well, it's not "OK"—it's very frustrating.

So I got a chance to try out my troubleshooting skills. At about the same time that I bought the scanner, I also purchased a new Compaq Presario 5610 — 350 MHz Pentium II, 8 Gig hard drive, ZIP drive, DVD and CD-ROM drive, 56K modem, loaded software — for $1,300! It's both the cheapest computer and the most powerful computer I have ever purchased. It has two USB ports (and two IEEE 1394 ports). On this computer, the ScanJet hooks up and runs perfectly with no problems at all. Here, the promise, "provides conflict-free connection, plus fast setup" rang true.

So, simple logic says that it must be a problem with the Portege. Nosing around in the "Systems" part of the "Settings" menu, I saw that the port really wasn't turned on. It's curious to sell a product with a feature that is clearly visible on the outside, but which is not enabled.

I bought the extended warranty and decided to put it to work. Good news: it's an "800" number—I won't have to pay for my "music on hold." And of course, there is the mind-numbing menu, followed by "all of our customer service representatives are helping customers right now" (or something very similar—my mind refuses to register the details), some music on hold, followed by several repetitions of "your call is important to us." Not said is: "but not so important that we would have adequate staffing so you wouldn't have to wait."

The first young lady is in a hurry. She is telling me to click on things faster than a 233 MHz Pentium can pull them up. She tells me to go to the Web site and download a new driver. I download the driver, follow the instructions and come up with error messages that give me alternatives that require a detailed knowledge of the software. Do I want to change, delete or ignore? How should I know? I choose ignore. After complying with several requests to re-boot the computer, it appears the new software is installed. It still doesn't work! Same message — the scanner sitting just 15 inches from me cannot be found by the Portege.

I made another call to Toshiba. This time, I get Robert, who tells me that the HP Scanner is expecting to find an Intel chip, but Toshiba uses an NEC chip for the USB. This is amazing. I tell him that the ScanJet worked just fine with the Compaq, so I must conclude there is something wrong with my Portege. He tries to correct me, saying that there is an Intel chip in the Compaq. He tells me that the Toshiba is fully USB-compliant. He won't get trapped into saying that the ScanJet is not USB-compliant. So I ask him if this means that they are both compliant, just not compatibly compliant. He is getting impatient because I am running up his dwell time. (He is supposed to dispose of calls faster than this.) He suggests that I call HP. He can't tell me what good that will do. He doesn't think they will have a driver to download. But I've reached the end of the road with Robert.

So then I called HP. No "800" number; I'm paying for the call. The HP menu is much longer, but there is a relatively short wait for a person. The agent at HP confirms that the difficulty is an Intel vs. an NEC chip for the universal serial bus standard implementation. He tells me I should insist that Toshiba replace the NEC chip with an Intel chip—not a software update, but a chip update! He tells me that I need an 82371AB chip set. He also mentions P2X4. He says that the connection between my Portege and the ScanJet is "like a dead phone line." He says it will be that way until Toshiba changes the chip. Well, I'm not holding my breath.

I have to search hard for a benefit in all of this. What I come up with is a strong lesson in the importance of thorough testing of a standard before marketing products. Before the Decoder Interface can go forward, it needs a comprehensive plug fest, testing as many different implementations as possible. It must have a well-defined communications protocol.

This same lesson applies to OpenCable and to DOCSIS as well. Can you imagine a cable CSR telling a customer: "I'm sorry, but your brand XX television uses a brand YY chip, which doesn't work with our security module. Yes, it's compliant with the OpenCable standard, but it's like a dead phone line. You need to get the television's manufacturer to come to your house and put in a ZZ chip instead. Yes sir, they are both compliant with the standard—they're just incompatible with each other."

And there is another lesson in my scanner experience. I have a four-year-old Gateway 2000 computer with a SCSI (Small Computer System Interface) card. According to the HP box, I should be able to use the ScanJet with the Gateway because the Gateway is SCSI-compatible. (It has a SCSI ZIP drive that works like a champ.) But there is another problem. The ZIP drive is connected with a SCSI cable that has DB-25 male connectors on both ends. The SCSI adapter card came with a cable having a 50-pin, wide connector on one end. But the HP ScanJet needs a "high density, mini 50-pin connector."

The ScanJet didn't come with the required cable. I've been to five stores, and I can't find the correct cable. The SCSI cables I saw had the wrong connectors and prices ranging from $35 to $64! So I don't yet know if the ScanJet will work with the SCSI connection to my Gateway.

Another lesson for the Decoder Interface: cables. I tried to get a mechanical specification on the mating surfaces of the consumer electronics product and the plug-in Decoder Interface module, so that when the module evolves to a highly integrated small package, cables can be avoided. This was seen as too much of a restriction on the creative design options for the back of the TV! The cost estimates on the Decoder Interface cable will rival the cost of a descrambler module. Imagine the frustration when an installer asks if the TV or VCR came with a cable. When he gets a blank stare, he will offer to sell a cable for $20 or $30. I think you see my point.

The lesson to be learned is that "compatibility" and "compliance" are very difficult issues. The computer industry has laid an egg with its Universal Serial Bus standard. We must be very careful to avoid the same disasters with OpenCable, DOCSIS, navigation devices at retail, and (less likely) the Decoder Interface.


Share This Story

You may login with either your assigned username or your e-mail address.
The password field is case sensitive.