Is a USB to GPIB dongle/convertor a difficult project ?

Compared to USB or Firewire, it's simple.

But the hardware sure isn't.

Do you use the Prologix? If so, does it work well to get screen prints and stuff?

It's $150 a pop (both direct and via Sparkfun).

The Prologix adapters are unique, in that they use standard serial-port communication from the computer (USB-to-serial) and add a serial-to-GPIB (IEEE-488) interface with a microprocessor. This means you write standard serial-port programs to communicate with your GPIB device, without using GPIB drivers. For example, you can use raw PowerBasic or whatever. The Prologix adapters are also inexpensive, the new 4.2 model comes in a nice box for $150.

It should never have been born! What a hunk of junk.

Yes. Didn't make it work any better though. What a mess.

products

Unlike GPIB, USB and Firewire work. It took USB a while, though.

OTOH GPIB cables look like they could be quite useful to tow a truck :-)

And yes, GPIB is IMHO ill-conceived. My worst experience: Unscrewed a connection, didn't pay attention for a split second, cable ricocheted across the lab table and my coffee cup went sailing. Kleenex time. What a mess!

You can actually do a simple gpib interface with just a parallel port but then you have to deal with drivers.

Is there any link where somebody did that? Drivers would otherwise be the big roadblock. Me and the world of software are miles apart.

Cost as much as the truck.

I had one setup that the cables had to be starred off one device to get it to work. There was a 12" GPIB stack of connectors hanging off one device. In another phase of the moon they'd have to be daisy-chained.

Did that this morning. Dumped half a 20oz cup of coffee all over my desk and side. Fortunately, my laptop was the opposite direction. Yes, a mess.

I read the spec for an hour or so before I figured out that "message" was ieee-speak for a logic level on a wire. There's one state diagram that looks like a pot full of pasta and peas.

John

We did one test system that used a DOS pc, with an ebay RS-232 to GPIB converter talking to a vintage HP 5370 time-interval counter, all programmed in PowerBasic. Somehow it works.

Incidentally, the Console Compiler (Windows) version of PowerBasic is very slick. I coded about 1400 lines of inventory-control database stuff this weekend, and it works great... compiles to a 50K EXE and runs blinding fast. Now when you look at the inventory record for a part, you can hit a key and launch an Explorer window that includes datasheets, pics, notes, browser links, whatever, about that part.

John

I bought one of the earlier (bare-board) versions through SparkFun, and stuffed it into the case from an old Ethernet MAU interface. I've used it with a Tektronix TDS-series 'scope and a Linux system.

It does work. I haven't tried to use the GPIB-target print-screen feature, except to try it once and see that the scope did spit out some data through the interface. I used it more extensively to do waveform data capture, using a simple C program to send commands and retrieve the data (something akin to the classic "chat" program). The equipment lashup was a bit idiosyncratic - an old HP RF signal generator (with AM or FM modulation turned on), into a Systron-Donner spectrum analyzer, whose X and Y sweep outputs fed two inputs on a TDS420A 'scope. I triggered the scope on the rising X input, did a one-shot capture of the Y input, and (when I got one I liked) used the Prologix to transfer the curve data to my laptop system, where it was decimated and fed into OpenOffice.org to create a graph I could add to a presentation.

The speed is adequate for transfers of this sort. I don't think I'd try to use this type of GPIB interface for controlling measurements or devices in situations where there are tight timing requirements. And, the basic I/O metaphor used by this sort of interface is a lot higher-level than the GPIB-bus-bit-banging API used by the NI interface cards and driver libraries... I'm not sure it would be easy to interface it to LabView.

I still have some Commodore dual 5 1/4" GPIB floppy drives, and a stack of 4023 GPIB printers.

I think they mentioned LabView interfacing in the doc section. Supposedly while you can use a program written for an NI card directly there are ways to make that transfer work.

I'd really only need it to spit out the screen contents. Verbatim, not as a CSV file or such. IOW straight to the LAN server, hop into the office chair, send it on to the client, "Here, found the problem and figure 2 is how it looks after the fix".

My solution was to buy a Durabook (mil-type laptop). It's supposedly spill-proof.

If Jan Axelson were 25 years younger I suppose she'd have written, "GPIB Complete" by now. :-)

I used her "USB Complete" as my main reference the one time I did a USB project where I did have to take apart and assemble USB packets in software myself. (These days I just use, e.g., FTDI ICs, but I was purposely looking for the "experience" on that project... and the FTDI-like approach fails if you have high-bandwidth or isochronous needs...)

WHOOOOHHHH, I want a copy!

Nation Instruments has a trade in program so you can buy their obscenely overpriced gear at merely overpriced levels. It used to be a really complex matrix, but now it appears that any NI carcass gets you the discount. Of course, it is still $100 more than the prologix device.

I've bought the NI 232 to gpib boxes at local flea markets, They go for peanuts. $3 However, there is some nuance with Keithly gear.

All that said, the prologix board looks good. It has been my experience that the FTTI chips work on everything (including X64), and I probably wouldn't use labview.