What is the best USB-to-RS232 adapter out there?

The ones with an FTDI chip inside.

Meindert

Meindert:

Thanks for the tip. Found FTDI stuff at Parallax,Inc. in the States

thanks!

Chris

I've tried 3-4 different USB ones and they all worked fine. I think they all use either the PL2303 or the FTDI chipset. Both chipsets are recognized by Linux out-of-the-box and are purely plug and play.

If you're using Windows, you'll have to go download drivers from somewhere and reboot 9 times.

No problem if you're running Linux.

Names are meaningless. You can buy two identically branded and labelled devices and end up with two different chipsets.

Just a warning I haven't seen anyone else post yet: What kinds of apps are you taking about?

If you're talking about simple serial communications, such as to a terminal or another computer, you're probably OK. USB has stabilized pretty. Everything I've tried in the last couple years, at least, even on Windoze boxes, has worked flawlessly.

But if you've got something more esoteric, with tight timing requirements, such as a chip programmer or a dongle, something that isn't "pure", it may be that no USB device will work with it. For example, no USB serial or parallel port (that I'm aware of, at least) will allow you to toggle a control line faster than 500 Hz (the time base for USB frames is 1 ms).

I only mention this because embedded programmers are more likely to have such applications than the general population.

Regards, -=Dave

Just to add a bit more to that: if these are *very* legacy, e.g. DOS with direct UART access, you're out of luck.

Under Windoze, or any other OS with a hardware abstraction layer, it should be fine.

Steve

Have a read of:

you will see that this uses the FTDI chipset, is manufactured by Easysync, an FTDI company, and works where many others fail.

Read the comments from customers on that page.

Don...

I have had problems with the Prolific chip. It's a driver issue, not a chip issue, but Prolific provides both, so the difference is moot. (The driver has problems with certain function calls...)

I don't see these issues with FTDI based units. And I've heard good things about the TI and SiLabs controllers.

As someone pointed out you can find different chips in the same product from the same maker. IOGear changed chips a while back, and you can only tell them apart by the revision code.

Expect to pay a bit more for one where you actually know what chip is in it.

Hope that helps.

Do they still work. My IO Gear converters work fine.

If you are using a standard driver. Why not use more end points and a non standard protocol USB JTAG Emulators certainly toggle the JTAG faster than 500 Hz

You can get CDC (Communication Device Class) drivers for the AT91SAM7 series so you do not have to bother with special chips like FTDI to implement the serial port.

I can't stand the suspense any longer. What is a "cuel laptop"? I can't make any sense of any possible typo or misspelling.

He means 'kewl'.

Best regards, Spehro Pefhany

"cuel" := "cool"

It took me a long time to figure this out. I guess you and I are not cuel anymore...

Well, I thought he missed the "r" out, since I think it's particularly cruel that these laptops have no serial ports :p

A USB serial cable is only $10 if you keep your eyes open. I, for one, am quite happy to pay the $10 in return for the cheap laptop prices that mass-market volume has brought. The mass market doesn't need serial ports, so mass-market laptops aren't going to have them.

Apparently you missed the smiley...

Yes, because they don't bit-bang JTAG with the PC software. They send data and commands down to a microncontroller in the dongle, which then does the bit-banging locally, without having to work through the USB latency.

Some even have a CPLD/FPGA in there to handle the serial protocol in hardware (freescale HCS12 BDM protocol for example is very hard to do in software as it is locked to the target clock frequency, rather than timed by the programmer)

USB is a good interface for embedded programmers - but you have to put the detailed target interaction on the other end of the USB cable, and only sends higher level things that can tolerate some latency over the USB.

It's an inconvenient, but fairly general trend, that the faster PC's become, the greater latency they suffer in interacting with the outside world. Absolutely in terms of processor clock cycles, and often even in terms of time - a good old ISA pentium machine with a real parallel port is a better platform for CPU-based realtime tinkering than the latest multi-GHZ no-legacy box. On a modern box, you're back to having to use IO processors to handle anything that can't tolerate substantial turnaround latency.

everybody else covered it.

i would add that i have had the occasi> Gang:

What bugs me most with modern laptops is that they often lack a parallel port. For most uses, a USB-based serial port will work perfectly well (and it's easy to add several), but for simple, cheap and easy-to-use programming adaptors nothing beats the parallel port.

... snip ...

There are three essentials that are missing from far too many machines. If the public simply insisted on them they would be present. All are cheap.

1. Real serial port(s) 2. Real parallel port(s) 3. ECC memory.

The first two are for usability, the last for reliability.