| ??? 10/03/04 03:55 Read: times |
#78613 - RE: Hating USB Responding to: ???'s previous message |
How many times have you had to futz with things to get RS-232 to work? Very, very few, actually. And normal users even less frequently. It's only those that have a lot of weird devices (often home-brewed) and a lot of unlabeled cables that get into that problem. Hey, I have some and can think of other serial devices which don't have Linux (or Mac OS) support. Let's see -- programming dongles from Xilinx. A nifty I2C EEPROM programmer from Microchip. The dbx DriveRack 260 DSP. A Rane DSP. BSS Omnidrives. A Video Instruments test-pattern generator box. Etc. And USB just makes it even more unlikely that they'll be supported since it requires both a Linux version of the application and a Linux version of the USB driver. You can choose from any of a dozen or more 8051 micros that have a built-in USB Serial Interface Engine. You don't have to do the most difficult stuff; the micros take care of it for you. Most of the chip vendors provide extensive application notes, example firmware, example host software, even a host driver. Cypress, with their EZ-USB series, is the big gorilla here (although their parts are typically the most expensive). Silicon Labs has a couple of USB parts. So does Philips. And TI. I wrote the keyboard demo code for the TUSB2136. I can assure you it's significantly more complex than MOV SBUF,A. I haven't used USB-ready devices from other manufacturers but there's just so simple they can make it. Well, look at it this way: us engineer types are NOT the majority of users. Quick: name a device that the Average Computer User would hook up using a serial port. Printer? Nope -- USB. Modem? Nope -- USB (or a PC Card). Mouse? Ditto. Yeah, some people still use USB to sync their Palms to their computers, but I think the newer PDAs all use USB. My Palm m130's USB cradle worked for all of about half an hour before it died with a known USB problem (i.e. a USB implementation problem that was Palm's fault) for which there was no solution other than to get a serial cradle. Which I did. Spent an extra $20 for the serial cradle the next day and have had no problems with it since (and it works under Linux!). And what do these stupid icons mean?" Of course, making things easier for the end user makes the developers work harder. Like I said, I'm not against USB. I'm just against USB replacing DB9 RS232 which costs all of a few pennies to include on a motherboard. Heck, my laptop even has a cut-out on the plastic case for where a DB9 should go. I'm sure the motherboard has 9 holes where a DB9 could go, too. Once this thing is out of warranty I'll probably open it up myself and slap a DB9 on there. Rather than complain about USB -- embrace it. Learn how to design USB devices and how to write the firmware. Thars money in them thar hills! I wrote the TUSB2136 keyboard example code for TI a few years ago. I had a lot of fun learning all about USB and writing a USB device would still be fun today. But the code to write a simple HID keyboard was several C files and I wouldn't even want to think about writing it in assembly. Even when I was developing the USB keyboard firmware I was using the serial port to get debug messages out, confirm where the code was going, etc. To this day I get email from people that are having problems getting USB working on TI USB devices. And several 'C' files is significantly more bulky than MOV SBUF,A. I don't want everything to be DB9 RS232. I just wish the dang computer companies wouldn't try to save 40 or 50 cents and not include even a single DB9 that was clearly already part of the design. Regards, Craig Steiner |
