Fritzing ??

Is it single-sided so it doesn't form a cap with the other side?

I use gold-plated double-side copperclad, FR4, 0.062 thick. A 50 ohm transmission line is about 100 mils wide, a little narrower if it's coplanar waveguide.

You really need a ground plane on the bottom side to do fast stuff.

A small part, like an 0603, would have about the same tiny pad capacitance whether the opposite side were copper or not.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

OK, thanks, I heard somewhere 50 ohms was about trace width equal to trace- gnd plane separation.

Stupid question*; (from someone who hasn't done much fast stuff) Can you run 500 ohm traces about your pcb, (w/ terminations?) and do

50 ohms with the outside world?

George H.

*When I say that I always picture Peter Falk as Columbo :^)

Get Appcad from Avago. Or Keysight. Maybe both. Nice little program.

I don't think that a 500 ohm trace is possible in this universe. A 5 mil microstrip on 0.062 FR4 is about 150 ohms, less if there's an internal ground plane.

Whether you need to source or end terminate depends on speed and length and whether bounces matter. In general, a minority of traces care.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Single ended? No, not unless you can find an insulator with a dielectric constant less than 1. Free space is something like

377-ohms. It's hard to get above that in this universe. ;-) Negative pressure?

Yep. If the length of the wire is on the order of half a rise-time less, then termination is needed. Sometimes less, to avoid humps in the edges. A lot of parts these days drive some really fast edges, though. I often over-terminate signals to slow edges down, for EMI reasons.

The difficulty with running 500 ohm traces is getting that high an impedance in the first place.

I once managed to create a 175 ohm trace on a 1.5mm thick low dielectric constant substrate (alumina-loaded Teflon) but that meant using the narrowest trace our PCB house was game to print.

--
Bill Sloman, Sydney

We used the EDS system for boards. When I first started we drew schematics by on velum and had a layout technician convert them to netlists by hand. A few years later the Toronto Lab came up with a logic entry system, TILES (Toronto Interactive Logic Entry System), and we used 3277 terminals with 19" Tektronix storage displays for the logic entry. Layouts were done on IGES systems, IIRC.

EDS was a very powerful system with a really horrible user interface. When IBM bought a hunk of Intel, it gave them the EDS system for chip development (80286 was done in EDS). The Intel engineers were horrified at the UI, so wrote their own front end. The layout and simulation tools were way ahead of their time, though.

And the really stupid thing is that the chip companies do have excellent transistor-level models of the same chips but they think that it is in their interest not to give you those models.

Everybody knows that everyone with the resources to copy the chip from the netlist also has more than enough resources to extract the netlist from their competitors' parts.

Very very stupid.

Chris

LTC seems to be a partial exception, at least if you use LTspice. The LT1028A model even has the weird noise peak at 300 kHz!

Cheers

Phil Hobbs

It would look something like twin lead, where the ratio D/d (distance D on centers, between round wires of diameter d) is ~32.

If the traces are edge parallel (i.e., same layer), the distance will be slightly less. If there is no cutout in the PCB between and around them (think ladder line

), the distance may be larger due to the increased dielectric constant.

Such large distances, even for very fine traces (32 * 7 mils is a quarter of an inch), means having even greater distances to the nearest ground, in order that the differential pair actually behave like a differential pair.

I know you didn't ask about differential pair, but now that you have(?) a mental picture of how to make a high impedance twin lead on a PCB, suppose how it would look if one of those traces was turned into the edge of a ground plane. Now you get a single trace above ground, or, well, maybe not above but beside anyway.

If you have ground on either side of the trace (coplanar waveguide), expect to need about twice the distance, i.e., half an inch from either side of the trace.

I get 0.15mm trace width, 140um thickness, 183mm gap, 500 ohms. So maybe the guess of 1/2 inch was a little optimistic.

The formula used may not be terrifically accurate in such ranges, anyway.

Since the distance between conductors is greater than the PCB thickness, it won't matter much, which layers everything is on.

Since the distance scales are so large, transmission line effects for very high frequencies (higher TEM, TE and TM modes) will get complicated on the order of 100s of MHz. Accordingly, the useful length must be longer than its width (otherwise, how do you connect to it and stuff?).

The only wideband thing that comes to mind is differential probes, the kind with loose wire leads. These usually only go up to 10MHz or so. If the leads are roughly twisted together, we can expect their transmission line impedance to be 100-150 ohms, which is a wide break from the ~1M you might hope a probe has. (The common mode depends on where the cable is laying in space, but it's probably not too much higher, 200-300 ohms. If any of that couples unequally into the differential mode cable, you're going to have a bad time!)

Obviously, we'll never get an actual 1M at high frequencies. We can calculate at what frequency the dominant pole will occur. The impedance ratio is 10,000, so the length ratio will be 10,000. If the cable is 1m long, its electrical length is more like 10,000m, or a 1/4 wave of 120kHz. Since Z > Zo, this will be a dominant pole (capacitive).

The probes can be made to look resistive at high frequencies, if the input circuit (at the far end of the probe cables) has an R+C terminating it, where R is a bit less than Zo and C is a bit more than the Ceq of the line. Total capacitance (as seen by low frequencies and high impedances) will be a bit over double that of the line.

If the probe cables can be braced like twin lead (or perhaps, use some flex cable that's printed with a good, wide, high impedance pattern), the impedance could be increased to 300 ohms, maybe a bit more, but nothing that's really obviously useful (like a 10:1 ratio into 50 ohms, making a perfectly terminated 10x attenuator possible).

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: http://seventransistorlabs.com

Oops... open mouth late at night and insert foot. A 5

Thanks, george H.

They come in other colors, now. :)

7781

You had to bring that up, didn't you?

Cheers

Phil Hobbs

That probably makes them leakier than the white ones!

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

They are sold for Ardunio projects,, so it wouldn't matter. Those are small enough to stick on a prototype 'shield'. I have never used any of those breadboards. The only type I've come close to using were a small board with brass tacks that you soldered components to. Then I bought chassis punches and some cheap aluminum chassis, instead. :)

When I was young and poor, I used to drive finishing nails into an old piece of wood and solder parts (from old TV sets) to the nails. It wasn't picosecond or picoampere stuff.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Maybe, but it did work. I started making aluminum chassis for projects in metal shop, in the seventh grade. I just couldn't go backwards to a breadboard, and I was working with RF that needed shielding. So, I mowed lawns and worked part time in a TV shop to buy tools while I was still in school. I still have some of those tools, 50 years later. :)

that's what how it was done when I was a kid in school, print out of schema tic glued on a piece of wood and brass nails

-Lasse

Brass nails? I couldn't afford brass nails.

I had a grandfather who smoked cigars, and he gave me the used cigar boxes to build things in. Nice actually, thin wood with a paper hinged cover. And I built some things on vinyl (asbestos?) floor tiles... punch small holes, stick in the part leads, solder on the back.

I also built things on old radio chassis, and on various weird military surplus stuff. Later, when I had some income from fixing TV sets, I could occasionally buy a Bud chassis.

I like Dremeled FR4 now.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics