S-parameter test sets

The Customer wants easy. The Customer wants simple. The Customer wants plug-n-play. The Customer doesn't want to think. USB is a nightmare on the inside, but there are dedicated pepole who put up with it, because the Customer wants crap like my little USD139 camera that just plugs into my USB port, and just shows up. I suddenly have another drive, with a bunch of jpg files on it. And USB also supports my HP transparent scanner - a stunningly good deal - USD79 bucks! For a SCANNER!

So, yeah, for us engineering-types, USB is an abortion looking for a coat hanger, but for the End User, it's a dream come true. I can't help but think of that roommate I had several years ago, who had a Macintosh. It used some kind of serial daisy-chainable interface, where you'd just plug your peripherals into each other. One socket on the back of the computer itself, and one little cable that went to the keyboard, and then another little cable from the keyboard to the printer, and another little cable from the printer to the ...

Well, you get the point.

At some point or another in the loop, somebody had to do the grunt work.

Thanks, Rich

and best not to think about how the display appears on the 3577 or 3585.

Cheers Terry

In the mid '90s I looked very closely at USB, thinking to use it for a serial bus within high-power motor controllers, so all IO devices could be separate cards and plugged in at will.

One detailed read of the protocol was enough to kill that idea. What a dog! Trust computer scientists to pick the most complex method of doing anything.

Cheers Terry

not as such, no. But I can do a magnitude-only plot of A (or B, or R). so when considering excitation-and-response, it behaves the same as a speccy with tracking generator.

And it uses the same display module as the 3585. I read the service manual last night, and it sure as hell looks like there is a "digital link" in there. Not that I am unhappy with the result.

Cheers Terry

Hello Terry,

Magnitude-only on the 3577 is a successive sampling process with a very slow ADC. So it relies on perfect tracking of generator and receiver. Looking at a spectrum does reveal that "there is something at x MHz" but the amplitude will jump around even at very slow scan rates.

Regards, Joerg

its been a while since I looked (despite desinging USB products quite recently, I never write software so dont care about the protocol), but it struck me as being overly complex. I should go read my USB 2.0 book again, and try and recall what else I thought was crap. IIRC I didnt like the hierarchical structure, but this was a decade ago.

Cheers Terry

Intel is very much a fan of trying to do all the processing on their CPU -- they'd rather have people build $10 "WinModems" (signal processing is done on the CPU) than $10 "smart" modems!

I had a very illuminating experience with winmodems, on my first W95 computer. I had never even heard of a "winmodem," it simply was the one that came with the computer. In 1996, $1000.00 was a lot of money! [still is, in my book -- Pig Bladder]

Anyway, I was just using the included software for email and junk, and it was molasses-slow - I'd had better responses on the BBS with DOS!

Then, for some unknown reason, I started investigating Linux. The first thing _any_ of the Linux crowd were saying back in those days was, "Get a real modem. Linux doesn't support winmodems." I had no idea.

So, I went shopping, and found a "REAL(r)" modem or something like that - maybe "Realtek" or something, that was guaranteed to work on DOS.

I was astonished. I was flabbergasted. Even in Windoze, with the new modem, my connection seemed a hundred times faster. Not that it _was_, mind you, it just _seemed_ to be that way. During that edjamacaishunal experience, I also got turned onto Eudora, the mail program that doesn't have to be on-line to compose email, but still, the difference between a winmodem and a real modem was significant.

Of course, it's been some years since I've used a modem - I presume that the 100MBPS ethernet cards don't depend on the CPU for their signal processing, although, we all know what ass-u-me means. ;-)

Thanks! Rich

Not directly, but there's plenty of time spent digesting TCP/IP packets. In fact, from what I've heard, the TCP/IP stacks are relatively complex in that some "starightforward" implementation would be dog slow and bring the CPU to its knees, so they have "fast path" code (for when all the packet come in the correct order, there are no errors, etc.) as well as "special needs" code (dog slow, but shouldn't be called that often). Apparently this is sophisticated enough that people write journal papers about it!

There are such things as hardware level TCP/IP routers out there, but the $25 wireless routers one finds at Fry's don't fall into that category. :-) Still, it's absolutely astonishing that for $25 you get a fully functional, 54mbps

2.4GHz radio, a router, firewall, etc., all in a little box no bigger than a hardback book. I remember looking at Telebit (are they still alive?) netblazers back about a decade ago, and you got less functionality in a box that at the time sold for >$1500!

---Joel

P4's and up suck on numeric processing. That's why I switched last year to AMD Athlon 64... runs circles around a P4.

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
|  Analog/Mixed-Signal ASIC\'s and Discrete Systems  |    manus    |
|  Phoenix, Arizona            Voice:(480)460-2350  |             |
|  E-mail Address at Website     Fax:(480)460-2142  |  Brass Rat  |
|       http://www.analog-innovations.com           |    1962     |
             
I love to cook with wine.      Sometimes I even put it in the food.

Hi Rich,

you are dead right. I hooked up my digital camera to my pc for the first time the other day. I plugged it in and it played. And yep, scanners sure are cheap nowadays.

and the corollary is that when you have control over the hardware that gets plugged into the interface, things can get a lot simpler. At the time I was looking for a *cheap* hardware solution that did its own buffering and error checking, so as not to burden the poor old system processor, which was being beaten to death numerically.

Cheers Terry

Nice, thanks for the explanation - I have seen that behaviour while playing, and wondered why.

Cheers Terry

P4s suck at integer arithmetic (they forgot the barrel-shifter integer multiplier, D'oh!) and but aren't so bad at FP. AMD64 is a far better choice, in any case.

--
  Keith