resetting a filter

Of course you would. And I'd like someone to offer me a job.

If you could have solved the problem with a few cheap 0603 parts, you wouldn't have started the thread in the first place.

Not me!

It's a discussion group. We need electronic things to discuss.

The RLC+diode thing works pretty well. That's what's on the board now.

I don't think I had that lecture. But unless you propose to drive it and receive it with balanced levels, it's grounded on both ends, asymmetric, so capacitive coupling will be real.

Only the magnetic part.

So the delay would be determined in assembly and not adjustable?

Why couldn't that be done with a resistor between 2 inverters?

For an assymetrical delay, maybe an OC gate with a resistor in series with its output, and a regular gate and resistor, with both inputs driven together and both resistors driving another inverter. Is Tiny Logic still available?

What is wrong with a manufactured delay line or a slowish buffer?

?-)

Yup. If you take a balanced line and drive it single-ended, the fields are equally divided between the odd mode (normal for transmission lines) and the even mode, where the currents in the two wires have the same sign and magnitude. IOW, that case is halfway between a transmission line (good) and an antenna (bad).

Cheers

Phil Hobbs

ote:

ns.

, but

ble

cost $10

akes more

le, it won't

lume requirement

ing to sell you

leadtimes.

qually

n with

ner and

he

of

on to send Ethernet data

ed text -

I've got a spool of bifilar wound phosphur bronze. I've forgotten the gauge, but 100" ~ 50 ohms.

We could sell some spool of wire where you could, short, then open circuit the end, and look at the reflection. I'm thinking a spool of coax will do better than bifilar.

George H.

A little brain fart there. Not that I have much of a brain, but I should have realized that a tiny logic OC gate is just a transistor.

ore

t.

e,

ime

n,

the

Cool, can you post that?

George H.

(or a link for someone else.)

ed text -

rote:

taff

r 42

ore

t.

e,

at

elds?

t -

So coax is better? (in this regard)

George H.

That would be AWG 36, assuming it's 5% Phosphor bronze.

Best regards, Spehro Pefhany

Note that the version with the diode has essentially the same delay on both pulses (red trace) but the simpler one (blue) shifts some on the second pulse.

Actually, the downstream logic gate will be a comparator, so I can tune the delay with a DAC. This is sort of a poor boy's linear ramp.

It's complex. The signal is unbalanced on the ends but maybe sorta balanced near the middle. Using transmission lines to make delays is appealing at first, but not so much after you've done it a few times.

Coax generally has a much better ratio of delay/risetime than pcb traces or twisted pair. Better shielding, too!

Delay is linear on tx line length, but risetime goes as the square of length. That hurts. And skin effect makes the output waveform really ugly, with a fast component and a slow, annoying drool.

The first transatlantic telegraph surprised people in this respect.

Unless it's a 'U' part, aren't they three inverters in series, with the last "just a transistor"? If it were "just a transistor" why have a VCC connection? ;-)

That depends on what kind of 'coil former' is considered. Ones I've seen looked like a footlong soda straw. The twisted pair was spiraled onto it, with enough space between windings to minimize coupling, and a return-winding 'layer' on the sodastraw was possible. Or maybe it was two coaxial sodastraws...

That could be a problem. The board is 2" x 2", and it's brickwalled with parts.

fields?

Lots better. Ideally the fields are all identically zero outside the shield, rather than just falling off rapidly.

Cheers

Phil Hobbs