resetting a filter

ote:

They do, but they don't happen to be competing in your little niche market, so you don't notice.

Production get shirty about that sort of solution - not unreasonably.

I certainly don't ignore the parallel capacitance of my inductors - I do rely on the regular LTSpice library model of my ferrite beads, but when there isn't a built-in library model I put something together.

And I wasn't "whining about you ignoring parasitics" - the parallel capacitance of an inductor determines the frequency range over which it looks like an inductor. It's not a parasitic, but an absolutely fundamental feature of the device, which you can only rarely get away with setting to zero.

It's never good "for an accurate simulation" of any circuit. You may have got away with it on this particular circuit, but all you are saying its that you didn't need a particularly accurate simulation.

It didn't make very much difference, so it can't be said to be any worse than your mindless over-simplification.

-- Bill Sloman, Nijmegen

$10

ore

won't

equirement

sell you

imes.

Enamel isn't a great dielectric. My experience went the other way. I needed a transmission line transformer, and wound something small with Filotex 55R 1,2mm OD coaxial cable. It worked, but it wasn't all that cheap. As a reality check, I tried a twisted pair made with 38 swg enamelled wire - mainly because the coil winding shop stocked both 38 swg with both red and green enamel (one was self-fluxing and one wasn't (or something like that).

It worked just as well - because I had room for more twisted pair it fell over at 150MHz rather than 500MHz but that was entirely due to the extra length of the twisted pair winding, which put the low frequency end down at about 50kHz, rather lower than fewer turns of the Filotex could offer.

send Ethernet data

When I was knowledgeable about Ethernet, back in 1980, it used proper coax, and a single Ethernet might span a university campus. It's changed a bit since then, and nobody has been paying me to keep up.

Bifilar windings are a nightmare in production? Your production staff can't be up to much.

Enough to know that you are improvising.

-- Bill Sloman, Nijmegen

Got any open-drain parts? ...Jim Thompson

Interesting. I'd no idea at all, either way. But a quick search did turn this up:

It says, "In 1988, the log, with the moth still taped by the entry, was in the Naval Surface Warfare Center Computer Museum at Dahlgren, Virginia."

Perhaps the book was transferred in the intervening period.

Jon

Yes, now I find this:

Jon

R is usually put across the diode, then you point the diode in the direction you want the delay.

Tim

I don't do that. I only simulate little snippets that I can't predict the behavior of, like this little RLC thing. I never simulate whole products. It takes too much time, isn't necessary, and the device models are usually bad enough that the sim doesn't prove much.

We don't prototype, either. It's too complex and wastes time. We only brassboard little snippets that we can't predict from data sheets or past experience. We lay out real boards, formally release the docs to production, and expect to be able to sell rev A.

That is production documentation. Entirely different.

When's the last time you said something pleasant?

won't

requirement

you

Ethernet data

I'm not going to ask them to measure and twist and strip and solder 42 ga wire. It's under 3 mils in diameter, practically invisible. And what do we do with a couple of feet of that, on a small PC board?

You've been winding one transformer for months, or years, and you're still not done.

I'm designing electronics that works and sells, and you're not.

Usually? That causes the diode current to taper off to zero, which extends the discharge tail. Keeping the diode current flowing keeps the dynamic resistance low, which makes the discharge tau low.

=3D30

I gave you the link to the Smithsonian site which shows they have possessio= n of it, assigned it an Object ID: 1994.0191.01, and have it on display the= re.

More fool you.

Hierachical circuit design lets you simulate snippets of a complete (and consistent) circuit design. Having the sim run right may not prove much, but it does prove that that part of the net-list hasn't been corrupted by the wires getting hooked up wrong (which is the kind of human error that happens often enough any automatic proof-reading mechanism is very helpful).

That strikes me as sensible, particularly with fast circuits where the actual printed circuit layout is a crucial part of the design. Again, simulating a snippet with estimated track lengths, track inductances and track capacitances provides a sanity check on the transfer of the design into the documentation.

Everybody wants to sell the first example of whatever it is they are producing. If your design process proceeds by such small increments that you can do a "new" design in a fortnight you really ought to be able to manage it most of the time.

It shouldn't be. Keeping them separate looses you a useful form of "proof-reading" check.

r.

It must have been last week. Some nitwit claimed that I thought that I was smarter than everybody else and I reminded them that there a few posters here whom I do treat with considerable respect - Joerg comes to mind, and Spehro Pefhany, amongst others.

More recently John Fields nailed krw for being the contemptible waste- of-space that he's always been and I posted a brief congratulation.

-- Bill Sloman, Nijmegen

ote:

ut

st $10

s more

it won't

e requirement

to sell you

adtimes.

lly

ith

and

to send Ethernet data

You wind it onto a small transformer coil former - I mentioned RM core coil formers, and EP cores coil formers tend to be even more compact. Here's the data sheet for the EP7 parts, which includes an surface- mounting coil former.

People have been coping with 42 ga wire for years - it's not rocket science. And the wire is perfectly visible under a binocular microscope (which you've got to have for fine SMD work anyway) although I can't remember ever having trouble seeing it (though I am short-sighted and tended to take off my spectacles when I was working on SMD boards).

I got the transformer and two different inductors wound a couple of weeks ago. Finding someone who still has a small coil winding machine did take about a week, but it all went quickly once I found a local coil-winding firm. I still haven't got around to putting the double screen on the transformer and over-winding it with the last - single- layer - bifilar winding, but I will.

Not at the moment. but I've quite done enough of it to know that you aren't quite as "insanely good" at it as you like to think.

-- Bill Sloman, Nijmegen

$10

more

won't

requirement

sell you

leadtimes.

send Ethernet data

That's insane. You're talking about a 42 gage twisted-pair delay line, and you propose to wind it up, unshielded, onto a bobbin?

Try it and let us know how it works.

$10

more

won't

requirement

sell you

leadtimes.

send Ethernet data

He'll get right to it, after he finishes his oscillator.

Do it thyself bifilar winding machine:

I usually use two different insulation colors. There are plenty of other plans and coil winding abominations found on the Internet.

Bifilar wire and trifler are commonly found in RF torid transformers, double balanced mixers, power dividers/combiners, etc. The first time I saw the insides of the cans, I was amazed that it could even be done by humans. At the time, the tiny torid were made by MCL in Brooklyn, New Yuck.

My guess(tm) is that the delay line would be simply wound flat onto a long rod. It won't be very small, but it also won't leak signal between multiple layers.

Some bifilar bonded wire suppliers:

However, if you don't like bifilar, maybe something like this:

but that brings us back to where we started. You can buy passive delay lines in a small package. Sigh.

Everything considered, I'd rather pick-and-place a few cheap 0603 parts.

Even 48 awg is only recommended for 1.4MHz to 2.8MHz

Skin loss doesn't stop transmission lines from working as delay lines - it just makes them a bit lossier

Wind the twisted pair onto a surface mount EP core coil former. I've already posted a URL to the EP7 part. The EP13 part - CSHS-EP13-1S-10P-T

- may be a bit big for you, but see

You don't seem to know much about what's possible in a serious production environment. Your clowns may not be able to wind 43 swg twisted pair onto a EP former, but you should be able to find a sub-contractor who could do it for you, and then your laughable excuse for a production facility should be able to reflow the surface mount former onto your printed circuit board.

Why not? Did you miss the undergraduate lecture where they told you that twisted pair transmission lines don't generate significant external fields?

Each twist does have an external field, but they alternate in sign and cancel pretty effectively.

I haven't got the test gear. My transmission line transformer worked fine using exactly that construction.

It won't leak much - each twist leaks in the opposite direction, so it cancels pretty effectively

It's as fine as I could reliably wind on the lab coil winding machine in the George Kent development department in Luton, England in 1975.

The coil winders down on the shop floor could get down to 46 swg.

Our tension control was a pulley on a long and flexible arm - not exactly complicated.