Prototyping?

There's no reason to prototype stuff like this. These kinds of parts work as advertised, and are reprogrammable anyhow. You save a lot of time and expense by designing your actual product, laying out the board, and building a first article. If you're careful, you can usually sell it, too.

We only prototype small, tricky circuits, mostly when a datasheet doesn't have all the facts we need.

Today I'm debugging the code for a new product, a 4-channel DDS waveform generator with a uP, an fpga, ethernet, and a bunch of switching and linear power supplies. So far, everything seems to work, no cuts or jumpers. I'll probably change some resistor values, to tune LED brightness and maybe tweak a supply voltage or two.

Breadboarding is actually a bad habit. It wastes time and encourages sloppy design.

John

Agreed. I breadboard only when trying to use parts in really unorthodox ways (happens a lot...) or when I need a one-off to control something and it doesn't have to be pretty. Beats the usual 1-2 week wait for fab, stuffing and all the Fedex in between.

--
Regards, Joerg

http://www.analogconsultants.com/

OK... but say for something as "simple" as a UHF bandpass filter (say a standard ham band one... 420-450MHz, assume you've decided you need a 5th order Chebyshev filter to obtain the skirts you want), do you (or John) expect you can design and layout a working PCB without either...

1) Performing simulation using one of the high-end CAD tools like Genesys, ADS, or Microwave Office, which can use very good models of the capacitors and inductors you're wanting to use as well as accounting for most parasitic effects of pads and trace width variation. 2) Fully intending to perform a fair amount of tweaking (generally of capacitor values, given the frequencies involved) once you actually build the board.

???

I certainly can't do so myself, and if someone else can I'd love to learn their tricks! The fact that I've never read of such a method in amateur publication (books, magazines, etc.) suggests to me that if probably isn't doable... and the commercial guys just go with method #1 up there.

Trying to meet a customer schedule deadline, I've seen tens of thousands of dollars spent to do a quick-turn of a many-layer PCB containing relatively simply circuitry such as RF switches, splitters, etc... but going to, say,

3GHz. And I've seen engineers find out within no more than 15 minutes of receiving a stuffed PCB that their design doesn't meet some prescribed isolation or frequency response flatness requirement. I cringe, because I could have predicted before spending that sort of dough that there was no better than a 50/50 chance that it would have worked anyway. Now, from a business perspective at some point perhaps you can't increase those odds so you might as well spend the money anyway (a 50% chance of something working for, say, $25k may be better than a 0% chance of something working for free and your missing a $60k customer payment...), but what I've taken away from such experience is that one is better off going through *more* prototyping cycles *early on* in the project so that you hopefully don't have to fight such Pyrrhic battles in the first place.

---Joel

LTSpice should be enough at 450 MHz. Or even just yank a prototype filter from Williams' book and scale it. With surface-mount parts, 450 MHz isn't really different from audio.

Why tweak a prototype kluge, then move it to the real PCB, and tweak it again? If you expect to have to tweak a few cap values to get it to work, why waste time doing it twice? Besides, complex filters are almost impossible to tweak experimentally, unless you're willing to tolerate truly rotten passband response.

I don't think it will offend anybody to point out that amateurs are amateurs.

We just finished a 30-day eval of the Nuhertz filter design software, and we ordered the LC filter version today, for $2K. It's got a few bugs, but they expect to fix them soon. It's pretty spiffy; it knows a zillion filter forms I'd never heard of (inverse Chebychev, hourglass, Legendre) and can do cool stuff like systhesize filters from finite-Q, standard-value inductors.

We're trying to design a 32 MHz 7-pole lowpass for a DDS-based arbitrary waveform generator, with demanding frequency response *and* step response, and it's really hard to do by hand.

Better yet is to not prototype at all, ever, and design everything as if you are serious about selling the first one. If you assume the first few iterations will not work, you'll always be right.

John

Jim Thompson wrote in news: snipped-for-privacy@4ax.com:

Could use that in SMD varieties.

Else I do this:

"Michael A. Terrell" wrote in news: snipped-for-privacy@earthlink.net:

I use the macro mode on my digital camera. The preview zcreen can zoom enough usually after the fact. Iff need be, I'll load into the computer to view.

433MHz is considered near-DC by some folks here :-)

Really steep resonances are frowned upon because now you rely on things such as cleanliness of surfaces, board materials and stuff, and that becomes dicey. Better increase the number of filter stages.

Nah, believe it or not, sometimes I even leave the HP11C in the drawer and pull out ye olde slide rule. It's faster in finding out how to design with catalog parts because you don't have an E96 series to pick from.

If you expect tweaking you can provide either varicaps or PIN diodes and have them under host control. I try to stay away from any hand-trimming.

There are excellent books in the ham radio community on how to built UHF and microwave gear that actually works right off the bat. Mine won't do you or most others here any good because it's in German and I have no idea whether they ever translated it: K.Weiner's "UHF Unterlage", kind of a UHF and >1GHz compendium with lots of tricks on how to build this stuff when you only have a rather modest shop. I am pretty sure that ARRL and RSGB carry similar books on microwave design techniques in English. In contrast to university courses these teach you where to actually buy the materials and how to process them.

Method #1 up there becomes less costly over time, thanks to lots of online routines such as those of U of Missouri-Rolla. Many of these are just approximations but heck, it's usually good enough. A full-blown simulation will remain costly for some time but frankly I never had the urge to do so anywhere above 50MHz. Ok, I just did one at 400MHz for a client but that's the exception.

You may have to ratchet up the "input path" a notch and move to luxury materials where parameters are more controlled. For example start with Rogers instead of FR-4. Later you can always back down but usually you want something that works, and fast.

Yes, I occasionally do prototype stuff at that frequency range when trying something unusual. Or when I want to try out that new chip and can't wait 1-2 weeks for populated boards to come back. But, thing is, what good does it do if you take a piece of nice thick silver coated wire, polish it to a spit shine with Wenol paste and get the perfect filter response? You've got to migrate that to a circuit board at some point and then you are almost back to square one anyhow.

Totally OT now: Are you guys up there also having a wood pellet shortage? I had to wait in line this morning because we finally had one (one!) flatbed truck coming in late last night. Only 28 tons, meaning after the first 28 pickup trucks it was gone. I can't believe it. Ok, at least I got my 1/2 ton.

--
Regards, Joerg

http://www.analogconsultants.com/

Ah, but only *if* you have the appropriate models. Not only of the components themselves, but also of all the parasitics. I once had a hand-tuned filter on a PCB (at roughly this frequency) and then built another one with the only significant difference being that a ground pour was removed from underneath the inductors to try to improve Q a bit... the tuning drifted off by >20MHz, which was enough to put the filter completely out of spec.

Tell you what... you crank out an LTSpice design for, say, a 10MHz wide 4th order or higher filter somewhere in there (e.g., 430-440MHz) -- there are plenty of free programs like the AADE filter design that can provide starting values -- calling out readily available (DigiKey) parts, describe/sketch how you want it laid out, and I'll be happy to cut the board, measure its response, and demonstrate that this can be a little tricky (or be shown to be wrong). I'm partial to CoilCraft air springs and AVX capacitors, but any reasonable parts are fine. We should all learn something in the process...

Because some of us have an LPKF machine in-house that can cut protoboards in a matter of minutes (ahem...), so sometimes it's worth messing around with a prototype before you do commit to a real PCB. (That being said, what I believe to be your preferred method of "ugly" construction is fine too -- plenty to recommend it.)

BTW -- regarding your comment that higher-order filters are difficult to tweak: Yes, they certainly are. But it is something of an art that some people (not me) have become quite good at -- it's still quite common that something like 8th order waveguide filters for satellite (microwave) communications are hand-tuned with turning screwing, lock-nuts, and something similar to Loctite/epoxy encapsulation when it's all done.

Most of the amateur radio guys who'd try to design their own 4th order or higher UHF filters are probably regular "professional" engineers anyway. I realize that today the word "professional" implies that you're supposed to know what you're doing, but when "amateur radio" was named the distinction was simply that "professionals" were being paid for their work whereas "amateurs" weren't. As you're probably aware, in the early years or amateur radio what people built at home was often as good, more innovative, or otherwise simply better than what you could buy commercially... whereas of course today even commercial amateur radio gear is nowhere near as sophisticated as, say, commercial cell phone equipment. To some extent I think that professional RF engineers have surpassed the amateur radio guys largely because they have access to $70k+ network analyzers, $25k+ design software suites, $45k+ spectrum analyzers, etc. -- the amateur radio guys make do with older, used equipment, but it only goes so far. (The other problem the amateur radio guys have is that there's plenty of warring factions within them -- plenty consider any and all digital radio modes evil, connections to the Internet a really bad idea, etc.)

It's a good package. Jeff Kahler knows what he's doing, and in the couple of e-mail chats I've had with him he's come off as a good guy.

I just don't think that's viable without that high-end software discussed earlier or multiple prototypes. Simple example: Design me a 4x4 RF multiplexer (4 RF outputs that can each select any one of the 4 RF inputs) that's flat to, say, +/-2.5dB from 3MHz-3GHz, P1dB compression point of, say,

+10dBm (50 ohms, referenced to input). There are commercial designs that achieve those specs, so it's clearly doable... but I'm be amazed if the folks who designed them didn't go through multiple prototypes along the way.

That's a bit cynical. :-) Personally, I'd *schedule* for multiple iterations and perhaps have cash bonuses or something for getting it right on the first cut. Carrots work better than sticks...

---Joel

expect

and

Even 2.45GHz is done with discretes these days. What surprised me last week (UHF filter design) was that nicely toleranced caps where only $0.01 and inductors $0.05. That was almost as good as spotting a favorite microbrew on sale.

Au contraire. I found that there still is a decent population among the ham radio community who can design a nice GHz range filter and actually make it work. Maybe only a few percent but IME the percentage of folks coming out of university who could do that is nearly zero.

Done a bandpass like that, a few MHz up. At some point it's like cooking a gourmet meal, tweak here, add there, stir a little but not too much. In the end you can't explain how you did it. What sure helped was that new dual-core processor here. I think you prodded me to upgrade to that a while ago (thanks). This time its fan actually did come on.

Mostly, yes. But don't start with FR-4 if at the cutting edge. You can always step down later. Heck, sometimes even all the way down to phenolic.

--
Regards, Joerg

http://www.analogconsultants.com/

Hi Joerg,

OK, let's just call it Butterworth -- how's that? That asks for far less out of the resonators (with the inductor of course being the generally lousy component -- getting Q>>100 is hard if you're after both "small" and "cheap"...).

Same offer to you as with John: Let's design a, say, 5th order Butterworth filter, 430-440MHz using something like (free) AADE filter designer and then simulate/tweak a bit using (free) LTspice to get to standard component values. I can do a layout from a sketch/description, cut a board, and measure its response... and I claim the result is most likely to have band edges that are off at least, say, 5% from the simulated result. Whaddaya say?

The idea was that you hand-tune using trimmers to get the response right, then replace them with decent accuracy fixed caps that you then crank out in production quantities.

Quick survey of a couple books on my shelf...

"Experimental Methods in RF Design" (Hayward/Campbell/Larkin) -- Doesn't bother with filters >3rd order. For frequencies above ~400MHz, they suggest going with transmission line filters (and discuss how to do so). "Practical RF Design Manual" (DeMaw, R.I.P.) -- "For operation above 200MHz it is more convenient to adopt the [transmission line filter] approach..." "ARRL Handbook (2007)" -- Goes through plenty of theory, with the usual data tables and implication that, sure, go ahead, build yourself a 9th order filter at 800MHz -- just use these component values here! Actual projects are 3rd or less for VHF filters.

...

I'm telling you, higher-order UHF filters using lumped elements are a little tricky and not just a trivial "Crank out the design from tables, simulate/tweak in LTspice, build the board, results match simulation."

We have a regular wood stove and we're still working on a couple of cords that we purchased last year, so I'm not certain about wood pellet availability. I'll ask around and see...

---Joel

I don't suppose you'd care to clue us in on which manufacturer that was?

---Joel

Removing a ground pour is considered a major design change up there.

There comes a point where a spec is too tight for catalog parts. Then you'll have to move to varicaps and stuff and auto-tune routines. Mostly the digital guys gave me some space on a DSP for that. To me those things are like huge Cadillacs. I hardly ever needed more than 1% of the space they had allocated for me.

Did you have to mention that? Now I wish Harbor Freight had a cheap knock-off of that machine.

Us guys have to do that with a Dremel and watch it carefully because often those little disks shatter and the pieces fly off at high speed :-(

In one club the guys who made the best GHz stuff were a railroad service technician and a brick layer. The highest frequency the railroad guy experienced at work was 16.3 Hertz and with the brick layer you had to watch the excess material. Once somebody asked what happened to the remaining four feet of a 2" pipe. He had already used it to build an armored driveway post so it could hold a swing gate better ...

If I had the time I'd do it. But the parts won't be too cheap and it'll be largish in real estate. The trick is to make sure everything remains strictly 50ohms. And know thy PIN diode parasitics. 3MHz is a challenge because you need decent carrier lifetime.

Or just toss it over to Jim, 3GHz should not be a problem for him once he's gotten all healed from his hip surgery. It surprises me that he settled for a hospital without news server access.

:-)

--
Regards, Joerg

http://www.analogconsultants.com/

Panasonic ECJ and ELJ series.

--
Regards, Joerg

http://www.analogconsultants.com/

As I said, you can't get 5% out of catalog parts if this gear has to be affordable by folks other than the Rockefellers. If you have to be that accurate there is no way around trimming. Preferably a uC or some other computing device has to do that, meaning you'll need a test signal mux in most situations unless you can use your regular signal.

Decent accuracy caps in production -> top Dollar.

I don't like to take that approach, too expensive. A friend of ours is a tech who works in the >10GHz field and she does this kind of trimming all day long. But even there she said they are mulling to automate that.

One method that has almost fallen from grace for some reason is active laser trimming. I loved it. You design everything low, then the stuff gets loaded into a laser and ... phsssst ... a minute later there is a part with the perfect frequency response coming out the other side of the laer box. One design ran about 30k units/year and at night there was only the guard with his German Shepherd in the huge facility. All fully automated.

Hey, John, did you write that? Or a relative?

Doug has passed away? I didn't know.

Well, I just opened my "UHF Unterlage" at an arbitrary page. Fell open on page 461 and whoopdidou, a filtered three-stage RF amp for 2.3GHz with layout and all. Followed by lots of others including a power amp. Most of this book assumes two-layer because that's all hobbyists can do at home.

The intro part of this book explains in great detail all kinds of microstrip and other structures, including how to build them with a modest set of tools.

We have a wood stove upstairs. Blew through four cords last, technically ran out but then the weather turned warmer. So now we've got five cords plus whatever we collected here and there. Almond, the good stuff (but lots of ash).

--
Regards, Joerg

http://www.analogconsultants.com/

Hello Joerg,

Yeah, but I didn't get the impression that John was planning to include any sort of model for a ground pour (or not) in his LTspice simulation.

Note that it was only the ground pour directly beneath the inductors that changed. The copper pour rule for how close the copper could get to the signal nets might have changed a little as well, but in all csaes there was a proper ground plane as well underneath everything.

I calculated the capacitance needed to shift the tanks as much as they'd move, and it wasn't a lot -- something like 2 or 3pF.

There is a kit you can buy to turn those ~$500 manual mills from Harbor Freight into CNC machines... you make most all of the gears/adapter plates/etc. yourself on the very mill you're turning to CNC! I might do that some day -- just for fun.

I was Dremeling out some mount holes on a PC motherboard PCB earlier this week (long story) and had the Dremel jump up out of the mounting hole and come down on a regular signal trace near the CPU socket, severing it. @#$@#% Some

30ga. wire and 20 minutes later I had it fixed, at least.

This is impressive... but is he designing something, building it, and it works... or is there a fair amount of tweaking involved? (After which the design can be repliaced without tweaking.)

I should have specified that it's supposed to be solid-state and thus small and low power and cheap; it would be easy to do using just a bunch of those high-end coax relays, of course, that tend to run a couple hundred bucks each. (Isolation is a lot easier with relays as well... getting, say, 60dB isolation between outputs and non-selected inputs is tricky...)

Interesting, though, that you'd choose to use PIN diodes over RF switch ICs from, e.g., Hittite (they'll sell you 4:1 muxes in one tiny pakcage, for instance --

-- although the isolation is only 30dB at 3GHz). Was that just for cost reasons, or something else?

Agreed, that would be the "easy" way!

Yeah... I trust he's doing well. Surprising how "quiet" the newsgroup is without him. :-)

---Joel

Thanks!

When choosing a hospital for replacing body parts, choose competence in replacing the body parts in question and the needed rehab over newsgroup access. Unless you're not really interested in working body parts, and then you can just stay home, connected to the newsgroups.

Just Say - I hate driving to Boston, but when a relative needed knee replacement, driving to Boston beat the heck out of the local hospitals in terms of results obtained, as best we could tell. A few irritatingly long trips beats living with lousy replacement joints for the rest of your life...

--
Cats, coffee, chocolate...vices to live by

ways

expect

and

We mostly do untuned wideband stuff, DC to as high as 4 GHz, which is fine with surfmount parts on FR4. But if you want to do narrowband filters, at some point you have to tune them, or go to saws or coaxial ceramic resonators or something, since the discrete tolerances will getcha.

the

Our problem isn't to make one filter work. It's to make an entire board work: bus or comm interface, power supplies, uP, firmware, fpgas, dacs, adcs, filters, output amps... all in maybe 90 days. We don't have the time to breadboard, or even play with eval boards, except for small, rare situations. We certainly don't have time to prototype any serious subsystem. So we lay out the real thing, fully documented and formally released, and let Production build one or two for us to test. That saves scads of time and hassle - they solder better and cheaper than engineers - and if you're careful, the first one is the real thing.

Intel has, or at least had, a similar philosophy: new chip ideas were fabbed on the production lines, not in some backroom lab.

John

That's where the investment into a dip meter is absolutely crucial. Solder a cap to that inductor, dip the resonance. Hold it over such a ground plane with masking tape (_not_ usual chewing gum, please...), dip again, then you know. Hang a counter onto it if needed. After umpteen of those you can almost guestimate it.

Surprisingly, much of corporate America doesn't have the foggiest what a dip meter is. Often people stare at it. "Some kind of Geiger counter?" Imagine what would happen if I had a hazmat suit on. Only problem is, the usual dip meters have fairly chintzy variable caps, they wear out over time and then you can't find a spare.

Mostly modified versions of projects from the UHF book I mentioned. Both guys always paid good attention to dimensional precision so it usually worked right away. You almost have to be a neat freak when it comes to GHz work. According to my wife I do not fall into the neatness category.

No, I just got burned too many times by part obsolescence. Some of these chips go away the instant Nokia or some other huge customer turns their back. Hittite is a really good company though.

I guess some of the guys in Washington, D.C., are breathing a sigh of relief right now ;-)

--
Regards, Joerg

http://www.analogconsultants.com/