Re: Smith Chart Amusements

See...

> >Newsgroups: alt.binaries.schematics.electronic >Subject: Smith Chart Amusements - SmithChartDoodles.JPG >Message-ID: > >I'm trying to remember from 40+ years back, when I last used a Smith >Chart... what do the circular patterns mean? > >I suspect that they are perhaps due to a distributed R-L-C network? > >What think ye all? > > ...Jim Thompson

I forgot we used Ansoft seranade in school.They offer free student versions here.

formatting link

You have to register though.

Seems like they no longer do Searanade,now its just Designer.They are good to tinker around with.Not to difficult to learn.I know seranade came with a good tutorial on an LNA. Agilent also makes good RF CAD but you have to mortage your house for it;)

Reply to
Hammy
Loading thread data ...

Of the "big RF guys" (Agilent, AWR, etc.), Ansoft is generally considered by have the most powerful free software available.

A German fellow has a very good tutorial on it here:

formatting link

The salesguy I once talked to regarding Agilent's ADS said that pretty much no one actually buys it anymore... instead companies forego buying, say, a new car or two each year instead paying for a software lease. :-)

Microwave Office -- while certainly not cheap (still a car every couple of years...) -- is a far better value than ADS, IMO, and definitely makes more of an effort to be "open" than everyone else. Ansoft's "value proposition" is that *if you were actually going to license a whole lot of modules from AWR or ADS anyway* they usually end up being the cheapest. Agilent of course also has Genesys that they acquired from Eagleware... a rather lower-featured offering than ADS, but still powerful and a good value for the money... and apparently the idea that if you get to the point of Genesys not having enough features/power for you (i.e., you decide your circuit works so well you really would like to turn it into an IC), there's a ...supposedly... clean-cut upgrade path to ADS once you hand over that blank check to the Agilent salesman.

---Joel

Reply to
Joel Koltner

Thanks for the tutorial link Joel.

I dont know why they don't lower the price of CAD packages,they would probably sell much larger volume.

Reply to
Hammy

I suspect the market is small enough that the vendors don't feel they have enough data points to know which side of the "bell curve" they're on. Genesys started out being a competitive, inexpensive offering... but very quickly saw a lot of price inflation, disproportionately to the new features added, IMO. I also suspect that the vendors figure out the only market they're really ignoring is the hobbyist market and perhaps some small private start-ups... something like $10k for a six month software lease -- while entirely out of the realm of possibility for hobbyists -- isn't that much for any company that's looking at purchasing or leasing stuff like network analyzers... or hiring any employees. :-)

Agilent (ADS) and AWR (Microwave Office) both have academic programs where they'll grant licenses for their software to universities for nothing or almost nothing per year (e.g.,

Reply to
Joel Koltner

I was in the decision-maker seat for a 100+ employee division (high-tech company). On my watch we never bought any SW that ran $10K. Didn't need any. The exception was a huge enterprise data system but that's a different ballgame.

This turns into a problem. One is that new hires immediately want some expensive SW and then you have to talk them out of it, showing them how a much lesser or older version SW can do the trick. The main issue though is that students seem to have lost touch with reality. Yeah, they can play VHDL and sims really well. But drop a few _real_ dual-gate FETs and some parts into their hand and they instantly assume that deer-in-the-headlights look.

But teach the new kids the Smith chart and stuff first. It's always amazing to see someone's eyes light up. "Oh, you mean I can just throw in 4ft of RG192 and that's it? Even coil it up?" ... "Yup."

--
Regards, Joerg

http://www.analogconsultants.com/
 Click to see the full signature
Reply to
Joerg

Hi Joerg,

Things might have been different then, but I guarantee you that if I were running a company like, say, RFMD, Hittite, Sirenza, Skyworks, or Macom today, and had, say, 6 engineers in my charge, spending $10k/year on software could easily let me be more productive with only *5* engineers.

In other words, yeah, the software's expensive, but the productivity gains are significant enough that it's a lot cheaper than hiring more bodies.

You *could* potentially convince me that you could get the same productivity out of cheaper/free software, though -- but there's also a lot of be said for not having *too* many different "point" tools running around your office.

$10k for MRP systems is cheap these days too :-), but I just don't think MRP systems provide the same sort of value for the dollar that EDA tools do... and when companies like Epicor seem to have rather sparse documentation with their products... but happily offer to "fill in the gaps" via their consultants at $240/hour (! -- I bet even Jim doesn't charge that much!), I really have to wonder whether the business model is to provide you with a complete solution (and then leave you alone) or if they fully intend for support contracting to be a major income source throughout the life of your business (kind of like McKinsey & Company...

formatting link
... although at their "team" consulting rate of more than $10,000/day (!), suddenly $240/hr looks cheap again...)

Personally, the moment I hired the first "full time" IT guy, I'd expect him to be able to whip up decent quality "parts & vendors"-type software. Take a look at this guy:

formatting link
... he works for a real-estate firm (!) and strikes me as 10x more "useful" than many of the "big corporate" IT guys I've met, who do nothing more than deploy servers, workstations, and networks... but never actually write any software beyond a few shell scripts.

It is, but the problem lies more with the educational system than the EDA tool vendors: Universities clearly have the resources to turn out high-quality code, as evidenced by SPICE and plenty of other packages. Yet these days also no universities use a non-commercial SPICE package because the commercial tools "cleaned up" SPICE and stuck some rather nice GUIs around it. Why can't universities do that themselves? Surely making a decent GUI can't be harder than making a decent transient simulation algorithm!?

Heck, look at how few universities officially support OpenOffice in deference to Microsoft Office, even when -- perhaps barring something like a thesis, for which Word is not exactly a great choice either -- I can't imagine students needing more than it for their day to day exercises.

If the universities aren't going to create and promote high-quality EDA tools of their own, the next best thing is to have them teach on tools commonly encountered in the real world. After all, if you go to a college interview with Avago, realistically it shouldn't matter if you designed, say, some spiffy microwave LNA with Puff or ADS... but I suspect you'll at least be able to relate to the Avago engineer better if you can commiserate about how much certain aspects of ADS suck!

Yes, absolutely.

I was rather amused the first time I found that, while I had two circuits that really weren't matched to anything close to 50 ohms, interconnecting the two provided almost as much gain as if they had been matched... turned out the connecting coax cable was making a pretty good matching network for one to the other.

---Joel

Reply to
Joel Koltner

It was not that long ago and it was a high-end ultrasound company. Still is, just under different ownership. Some have tried but no other company was ever able to copy our products. So we were bought ...

A lot of stuff we wrote ourselves. Had to. IIRC MatLab was pretty much the most expensive program we used. There are some good free alternatives but it isn't that expensive for a company.

Well, the MRP cost us a whole lot more than $10k. But yeah, documentation and having to rely on consultants was something we watched like hawks.

He picked some good stuff. I was introduced to Subversion by a client recently. Worked like a champ, almost like being at their building (which was thousands of miles away). OO is kind of ok but very bloated and not nearly as fast and robust as older versions of MS-Office.

I like the 2nd part of his story best :-)

The main reason: The millisecond a prof retires and a new one comes in they often toss the whole kit and kaboodle out into the street. The new guy wants "his" projects to be done. This is how some of the most excellent work of Prof.Fettweis' group at Bochum University (wave digital filters) came to grief to a large extent. Here and there some code fragments, a loose concoction of file. I was very lucky to find enough to cobble together a crude design "suite". There is no GUI, all strictly command line, and I had to clean the dreaded Borland runtime error out of some.

That candidate may be happy that I wasn't the interviewer. I'd ask him how to squeeze a little more bandwidth out of this or that amp and then hand him my HP-11C.

That, plus the Zout of a stage if often much lower than thought. One prof at our university thought all big radio transmitters had a Zout of

50ohms and he even taught that. Seriously! I had to run out of the auditorium because a major ROFL attack crept up my belly.
--
Regards, Joerg

http://www.analogconsultants.com/
 Click to see the full signature
Reply to
Joerg

If the company had a decent number of true "design" engineers (as opposed to "copy and paste the data sheet's example circuit, possibly with some minor tweaking"), I'm not surprised... such companies are unusual, as far as I can tell, and difficult to compete with.

If you purchase enough toolkits you can readliy push Matlab over the $10k mark these days.

Matlab is a good tool, although (as you alluded to) if you're looking to avoid spending Big Bucks and either don't need the toolboxes or can readily program them yourself, something like Python, NymPy, and Matplotlib make a compelling alternative. I believe someone posted this link -->

formatting link
...discussing SAGE, which packages up a bunch of the Python libraries to make it easy to transition.

Sure seems to me that building your own MRP system is a lot easier than building a cutting-edge ultrasound machine...

You're supposed to compare it to the current version of Office. No fair comparing it to older Microsoft software that was faster, more stable, but didn't have as many bells and whistles... 99+% of which are never used! :-) (What are you using, Office '97? I owned a copy of that and it worked pretty darned well...)

I figured you would! It's written a little oddly, although I expect he did so purposely, and the writing style is effective.

The bandwidth question is fair game, although hopefully you wouldn't mind if they pulled out their HP-48GX or newer calculator and crunched numbers a bit faster than what the 11C allows. :-)

Yikes... that is a bit scary!

---Joel

Reply to
Joel Koltner

All true design engineers. Most of our stuff cannot be modeled anyway, such as the behavior of transducer materials such as PZT. Basically you are in the clean room and in the lab all the time.

Yep, it's the same as using service folks for every household chore. Live will become rather expensive that way.

And Octave and all that.

Maybe, and we had a home-grown MRP before. But it distracts and the talent isn't used in an optimum way.

Office 97 and Word 2000. Good enough :-)

That would be ok but no computer.

The sad part is that you'll always have a large number of students who believe that stuff. The teachings about splitting grounds OTOH have brought me lots of work 8-D

--
Regards, Joerg

http://www.analogconsultants.com/
 Click to see the full signature
Reply to
Joerg

The difference is that you and your wife don't get paid for doing household chores, whereas having someone in-house write a software toolkit may or may not be more expensive than just buying it.

IMO it's the IT guys who should be spending their spare moments working on such software. It's not like they were hired because they were intimately familiar with Smith charts, after all. :-)

---Joel

Reply to
Joel Koltner

Sometimes the rationale is different. For example, a couple of times I got estimates for tile jobs and almost froze when looking at those five-digit numbers. Quick calcs in my head, turns out I couldn't possibly make that much money per hour after taxes unless I was a lawyer or dentist. Then did the tile jobs myself. Neighbors had tiels cracking after pros installed them. We never did ...

They are usually pulling 10hr days already. Also, why re-invent the wheel? When my father started consulting he wrote his biz book-keeping routines in APL. Mostly because he had a lot of fun doing it. They worked well for him but it was a lot of work. I started years later and received lots of valuable advice from my father. The only thing I turned down was those APL routines. Too "nerdy" for me, plus I preferred everything in English. Bought MS-Works for around $100 and within very few hours I had the whole set ready to roll. My father's comment was "Brotgelehrter", a German expression for someone who would only put his brains to something when there is an immediate return on investment.

--
Regards, Joerg

http://www.analogconsultants.com/
 Click to see the full signature
Reply to
Joerg

Hi Joerg,

What in the world are they doing? Tech support to end users or something? Once servers and workstations are set up, if it's done right, everything is quite reliable and IT guys should require very little time each day accorded to "maintenance." The bulk of their hours should be "new projects" and "tech support" (although, by hiring the right employees, hopefully you can minimize that as well... and at all the engineering companies I've worked at, "IT tech support" was always restricted to Windows, Office, and similar applications that pretty much everyone used -- never the EDA tools, understandably enough...).

Same as with EDA software... you might be able to do it cheaper and better in-house than what the commercial guys provide. Or maybe not. (But I can hire some pretty darned good programmers for much less than the $240/hr that Epicor gets for consulting, you know?)

Interesting that in my case I think spending $10k on EDA software often makes sense but I'm tempted to roll my own MRP system whereas in your case it seems like you'd rather roll your own EDA tools and spend the $10k on the MRP system. For a small business I expect that the answer to the MRP question is to start with something like much smaller, like Parts & Vendors or MS-Works as you did: No way could you pay someone to implement all that functionality for the three-digit price tag they demand... but the question is no longer so clear-cut when we're on to five-digit price tags.

(I seem to recall John Larkin mentioning that he occasionally contemplates dropping his own home-rolled MRP "system," as much as it is, and puchasing something like P&V...)

Haha... You could tell him that right then you needed a job and not a hobby! :-)

---Joel

Reply to
Joel Koltner
[snip]
[snip]

Same here, I always do my own tile work. At the old place I probably laid about 2000 sq-ft. I even tiled the walls and CEILING of the boys' bathroom ;-)

...Jim Thompson

--
| James E.Thompson, P.E.                           |    mens     |
| Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

What they were doing? It was two guys for a company of 300+ people. In the morning the accounting guys needed new fiber strung from they old digs to their new digs. The PC for the new tech needed configuration. Numerous memory upgrades here and there. Then around 11:00am a scream, compressor #3 blew a gasket and took part of the back server room with it. Lunch had to be eaten on the go in that server room. And so on.

I'd never design my own EDA. It would be like building my own pliers and wrenches. Ok, if I was stranded on a remote pacific island with nobody else there I'd probably do that.

It is amazing how fast a business blows through $10K when attempting a full-custom solution on their own. If you have three guys on that project you'll be at that point in well under two weeks. So typically a business only does that if there really is no off-the-shelf solution.

I already did that when electronics was my hobby, only building stuff when there was no off-the-shelf solution or where the industry was IMHO too freaking stupid to do it right ;-)

So when I wanted my first ham radio transceiver I looked at the complexity and parts costs, then decided to sign up for six weeks of grueling work at a meat factory so I could buy a heavily banged up used unit.

--
Regards, Joerg

http://www.analogconsultants.com/
 Click to see the full signature
Reply to
Joerg

Good morning Joerg,

OK, with that ratio of Workers:IT guys they would be quite busy. Most company don't have a ratio anywhere near that high... nearly an order of magnitude less! (See:

formatting link
... somewhere around 25:1 would be more typical.

So... you're saying the engineers at your old company were doing do (writing Matlab scripts and all) because... adequate commercial EDA tools didn't exist?

Aren't you now agreeing with my original premise that spending $10k on EDA software can sometimes make a lot of sense? :-)

---Joel

Reply to
Joel Koltner

Needs membership :-(

They must have a plum budget then. I never saw it that high except in the old mainframe days.

Nope. The reason was that MatLab is good enough for the purpose. You don't always get much of a benefit from a luxurious field solver.

I do some pretty tricky stuff here in the office yet I never felt the urge to spend >10K on EDA. And I mean the whole EDA usage, not just one program.

Just one example: A new contract required me to be able to handle AutoCAD drawings, including editing. Needless to say that is touching a rather expensive area. So I went out on the Internet to check liquidators. Found a non-AutoDesk package that was about three releases back for ten (!) bucks. Some kind of firesale from major overstock. Bought it. Installed it, got the license key within one minute, works. It's total overkill for me but heck, I won't complain after I just got a V8 luxury sedan for the price of an old moped.

It can, for example if you design a large airliner. Then you kind of need something like SolidWorks. Although, thinking back about the guys who designed the DC-3 ... :-)

--
Regards, Joerg

http://www.analogconsultants.com/
 Click to see the full signature
Reply to
Joerg

OK, here's the table converted to text (view in a fixed font, of course):

Ratio of IT Staff to Total Employees

Organization Size 25th Prcntle 50th Prcntle 75th Prcntle Count

------------------------------------------------------------------------- All Organizations 1:11 1:27 1:52 103

By Annual Dollar Volume

-------------------------------------------------------------------------

-Less than $200 Million 1:11 1:19 1:34 25

-$200 Million to $500M 1:19 1:36 1:61 20

-$500 Million to $1B 1:11 1:31 1:53 17 $1 Billion to Nope. The reason was that MatLab is good enough for the purpose. You don't

Sure, depending on the problem, that's certainly true.

OK, but say I ask you to design something like a "reasonably" tight hairpin bandpass filter at a couple of GHz... say, Q=50. You can certainly do it without a computer... with just a calculator and some tables it wouldn't even take that long -- you probably did it back in college with a slide rule at some point. :-) However, I can pretty much guarantee that if you build that "textbook" filter, the first cut of the board you do won't meet the specs because the center frequency will have shifted a bit as will the passband width. I know you're then fully capable of tweaking the design and given another iteration or two making it work, but if you had started the day with a field solver you could have been just about guaranteed of "first pass success." If we're talking Rogers material here and looking for, say, 3-day turns on PCB fabs, if it takes two extra iterations and a week or two of time, you've pretty much eaten up the cost of a field solver right there.

I've spoken to a couple of microwave filter vendors at past IMS shows about this. They told me that the biggest gain they saw when they switched to using fancy $10k+ design packages was a reduction in the number of board turns they did from 3-5 to 1-2, and that easily paid for the cost of the software.

I think our major difference here is just that you're looking for bargain software to get the job done, whereas once I'm convinced that the software is paying for itself, finding the best bargain becomes a secondary priority. I like bargains as much as the next guy -- you certainly see me pushing alternatives to Office, Matlab, ORCAD, etc. here on a regular basis -- but I really wouldn't mind paying, say, $10k for some fancy version of SPICE if I were convinced it would likely eliminate a second $250k wafer spin. At that point the "payback" is already huge, and finding a $1k tool or a free tool to do the same thing is great, but just not a high priority anymore.

Regarding your example with AutoCAD... I'd personally probably look at something like AutoSketch, since while it's a lot more than $10 at $120, it's still a lot less than full-blown AutoCAD, and -- based on my own inexperience here -- I'd be concerned that using a 3rd party package might introduce incompatibilities (not being able to read in "official" AutoCAD drawings, generate them, etc.) that could easily kill a day of effort trying to fix, completely negating the $110 savings. Of course, I'd do some Googling as well, and if I read enough good things about the 3rd party package I might still go for it, just with the "backup" plan of switching to AutoSketch if need be. :-)

Sure, we landed on the moon with nothing more than ~1MIP computers and a whole lot of slide rules and smart people after all!

---Joel

Reply to
Joel Koltner

I guess I should find the old IT guys of ours and retroactively give them a huge bonus :-)

They were pretty darn efficient.

This _was_ in the Internet boom days :-)

It can certainly help with that bandpass. However, I have observed a disturbing trend lately. New grads are often completely lost and can't do the simplest RF thing the millisecond the PC is gone because of a power outage or something. Once a major winter storm knocked us out for almost 24 hours. I suggested that our group head up to my apartment where there was a wood stove, windows that can be opened for fresh air, and beer (that convinced them...). Armed with a huge stack of chart paper and an easel we did that. It was a rather sobering session. "Now wait, how did that leakage inductance stuff work? How do you enter a coupling factor in there?" ... "Uhm, you don't. In this case you can only do that with SPICE and it looks like it took one of them big HV powerlines out."

Now don't get me wrong. I am not perfect either and regularly "forget" even simple math stuff and formulae. That's because one day I have to work on an RF design, the next day on an EMI problem and Friday on a HV circuit. It's like with languages where I have to switch a lot and then stuff becomes less than perfect in either. But one must retain the ability to find back crucial information even if the Internet or the local electricity grid is down.

For an established company, yes. For a start-up you better find that $1k or free tool :-)

Well, it worked. I could read all their stuff and their regular AutoCad could read the drawings that I revised. T'is all I usually need.

Exactly.

--
Regards, Joerg

http://www.analogconsultants.com/
 Click to see the full signature
Reply to
Joerg

Hi Joerg,

Sure sounds like it!

Have you noticed how popular UPSes are these days? :-)

I think I'd hold out until you'd offered pretzels too... :-)

Even with the Internet and computers, developing your own efficient filing system is critical, as I've come to realize lately. Creating your own notes is (at least for me) a huge benefit as well, since of course they're written in "your own" language and filtered down to just the bits you found as being tricky, forgettable, etc.

Yes, agreed... although you might be surprised at the prevalence of pirated software in small companies...

---Joel

Reply to
Joel Koltner

Even the one before was great. Yes, one as in the lone IT guy. When the company was > the simplest RF thing the millisecond the PC is gone because of a power

I had a few pizzas in the freezer.

Yes. And never, ever change that system. Saturday I remembered that I had found a "better" place for the shrink tubing. Looked in the MS-Works database file for materials, not updated to new location. Drat! Which meant I could not find it at all anymore. So a snippet of an old bicycle innertube had to do and now one of the receivers smells like rubber when you take the top panel off.

Not at my clients though. Lots of start-ups there.

--
Regards, Joerg

http://www.analogconsultants.com/
 Click to see the full signature
Reply to
Joerg

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.