Hi If you are going to do a full boat implementation and work out all the isolation issues and packaging, the question becomes:
Will it be better bang for the buck than a ~ $200 HP 5370? Bob On Dec 17, 2010, at 9:29 AM, jimlux wrote: > Interesting discussion.. > comments interspersed > > Chris Albertson wrote: >> Jumping ahead to design. No one wants a serial RS232 interface. they >> don't even make computers with RS232 ports much any more. Those guys >> that designed equipment that forced people to load costom USB drivers >> just did not think. There is no need for that. What you do is make >> you project appear to be some "standard" USB class and then the OS >> (Linux, Windows or Mac OSX) will already have a driver. That VNA >> should have presented itself as a serial port. And then the software >> could read from a serial port. But of course there would be not >> physical RS232 device. >> If you have to select an interface I'd rather have any wireless type. >> WiFi or Bluetooth. > > wireless interface and RF test equipment is a bad combination. If you're > trying to measure small scale performance (e.g. timing at 1E-10 levels), > small amounts of RF leaking in/out causes problems. This is one of the > things that separates good test equipment from great equipment. It's hard to > get better than 100dB isolation from packaging, and if you're looking for > things at the -150dBm level, something at 0dBm is huge. > >> But if you are building a modular system you do NOT want to pick one. >> You just make a project standard to use (say) I2C, SPI, "two wire" ir >> whatever. Then the counter module is controlled by i2c and if you >> want to connect it to a computer you build the USB module but if you >> want a stand alone no-computer instrument you build the "front panel" >> that has LED numbers. >> That is the entire ont is "modular", you avoid this kinds of decisions >> and allow for easy upgrade as technology changes. > > > IR/fiber optic interfaces are very intriguing. Too bad that the plastic > fiber stuff costs more than conventional wires/connectors. > >> Other questions to resolve are "how many slices to cut the pie into". >> I would argue for "very small" single funtions bulding blocks so we >> don't have the HPSDR problem of years of time to design each one. > > Against that: every connection causes potential troubles. A better solution > for the generalized case is to put multiple functions together, but don't > necessarily connect them all. Think of the old IF strip chips. Oscillator, > amplifiers, variable gain stages, detectors, all on the same chip but the ins > and outs brought out to pins. I don't think you want to bring them out to > connectors, but, rather, provide a way to do interconnections, etc. > > > >> Selecting a physical chassis to use willl take time. I like the idea >> of using a disk enclosure because then you can buy a 1U or 8U rack or >> an old PC chassis, If you modual looks like a disk there are plenty of >> things it can fit into. > > "old pc chassis" is a very limited life item in a particular configuration. > do you mean my old IBM PC? Or an AT? or a tower case? or a midsize case? > When you say disk drive size, do you mean "5 1/4" floppy/CD-ROM/DVD" or > something else. > > Packaging is going to be critical for high performance. Look at boxes from > COMPAC, for instance. > >>> From my experience, for something like this to take off one person has >> to take ownership of the project and run with it, make the web site, >> write some golas and build "something" that works. Only then do other >> jump in and help. >> It really would be good to have a Time and Frequency Instrumentation >> Project as currently the state of the art seem to be that you simply >> buy something from a Chinese eBay reseller. This is hardly what I'd >> cal "innovation." > > > > _______________________________________________ > time-nuts mailing list -- time-nuts@febo.com > To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts > and follow the instructions there. _______________________________________________ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
