NONE of you idiots have paid attention to his space limitations or remarks.
Let's end it. Using a LENGTH of trace to solve his problem is NOT feasible, nor possible in his situation.
Even 16 layers of zig zags. So try a different slow ride.
I'll bet you make nice, rounded hairpin turns instead of squared off 'zig zags' too.
I could do stand up,but the audience would all have to be engineers or technically inclined, and the jokes would only work once, so I would always need new material. It would be a failed career.
e:
tot
text -
I was thinking that you could terminate the end of the trace and the tap with the 'right' impedance and get rid of most of the reflection.
Say John, (If you don't mind that I drag this thread a bit further away from your original question.) What's your take on the reflection from 90 degree bends in the trace.
Here's a wiki link, scroll down a bit for the bend 'stuff'
I never done any controlled impedance trace designs, so consider me an idiot. I can imagine that you get a tiny bit more C from the corner. But does it make that much difference? (With ~100ps edges, the edge is spread out over centimeters of trace length...) I've used microwave plumbing (in the distant past) where the bend radius was pretty close the the wavlength, and that never caused any 'significant reflections' that I can recall.
Thanks,
George H.
Stripline conductors can be as close as a couple dielectric thicknesses between conductor edges without too much crosstalk.
Getting a few ns of delay doesn't take up much area using stripline, say at the edge of a board-- maybe 0.3in^2, and parts can be mounted overtop a buried delay line.
However, the delay is probably not well controlled, it's not adjustable and may not be stable enough, so it's probably unsuited for John's application.
Hey, a microstrip delay could be adjusted by adding a dielectric on top of the strip. Like having a plastic ID label that is stuck on the zig-zag at different spots to tune it. Let's see the copiers figure that one out!
-
I occasionally put a test trace on PCB layouts, a couple of SMA connectors with a 50 ohm trace between them that tours all the layers. With a 20 GHz TDR bandwidth, I can see the connector transitions, vias, and what looks like the fiberglas weave, but I usually can't resolve 90 degree bends. In theory, they do add a little extra capacitance, and the fix is to clip the corners a little.
I don't know if that one helped, but it did work pretty well. I did use ATLC to model the edge-launch SMA connector, which was the hard part on this one.
Here's a test trace that includes angles and vias and crossing plane splits. On the TDR, the vias are visible discontinuities but the angles aren't.
Impedance matching matters more and more these days. Logic and FPGA chips and things like cheap crystal oscillators often have sub-ns edges, and LVDS type things get below 100 ps. Even a SPI clock can mess up of it's not impedance controlled and terminated.
This EM stuff is fun, where the speed of light becomes very real.
If you nedd to know more about controlled impedance traces, ask Sloman. I'm just a tinkerer.
-- John Larkin Highland Technology Inc www.highlandtechnology.com jlarkin at highlandtechnology dot com Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom timing and laser controllers Photonics and fiberoptic TTL data links VME analog, thermocouple, LVDT, synchro, tachometer Multichannel arbitrary waveform generators
and back to Q looks like a forward biased c-b current path?
Crazy talk. We design high-performance high-order filters and pulse shaping networks all the time in our shop, and we worry about inductor Q, but seldom about shunt capacitance. In my example, the inductor is feeding a 10 pF load through a 200 ohm resistor, and it has a shunt capacitance below 0.2 pF, and I'm just delaying a logic signal. Here, neither inductor Q nor shunt capacitance matter, it's obvious that neither matters, so why include them in the model?
I only model things that matter. It only takes a few seconds of consideration to exclude the things that really, really don't matter.
You are so determined to act as an authority that you quit thinking. Bad engineering practice. Go put another turn on your transformer.
-- John Larkin Highland Technology Inc www.highlandtechnology.com jlarkin at highlandtechnology dot com Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom timing and laser controllers Photonics and fiberoptic TTL data links VME analog, thermocouple, LVDT, synchro, tachometer Multichannel arbitrary waveform generators
I'm after a 3 ns delay, several hundred MHz equivalent bandwidth, so I'd need a couple of feet of that 42-ga twisted pair. Its rise at the end would be a horrible drool thing, from skin losses, and the drool tail would defeat my goal of having a delay that has minimal memory of previous events. There would be reflections, too, because the pair impedance would be a function of frequency, so there's no simple resistive termination value. I have TDRd twisted pairs (but not 42 ga!) and these problems are real. Small diameter twisted pairs are very lossy, much worse than coax. Even CAT6 is ugly in the time domain, and needs a lot of equalization to send Ethernet data reliably.
Not to mention it being a nightmare in production.
You don't know much about this stuff.
-- John Larkin Highland Technology Inc www.highlandtechnology.com jlarkin at highlandtechnology dot com Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom timing and laser controllers Photonics and fiberoptic TTL data links VME analog, thermocouple, LVDT, synchro, tachometer Multichannel arbitrary waveform generators
I've played with that. Just a plastic sticker doesn't do much. A piece of single-side copperclad, copper up, on top of the trace works, but of course does messy stuff if it's not over *all* of the trace. I suppose you could use a metal slab, with spacing above the board tweaked with setscrews.
Really, I used to do stuff like this, chunks of coax and exotic PCB traces. Surface-mount lumped components are so much better.
LLNL did an arbitrary waveform generator that used reflections off a transmission line to make waveforms. They machined a brass plate as the line, one plate for each desired waveform; they looked sort of like those Christmas-tree air fresheners. We replaced that with an electronic ARB.
-- John Larkin Highland Technology Inc www.highlandtechnology.com jlarkin at highlandtechnology dot com Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom timing and laser controllers Photonics and fiberoptic TTL data links VME analog, thermocouple, LVDT, synchro, tachometer Multichannel arbitrary waveform generators
:
figure
nto the
ottky
ctor.)
it.
saturate
current
?b-c and back to Q looks like a forward biased c-b current path?
ction
into
to
ble
here
ein
n,
ly
eto
t s lrs
agh-
I, yet.
And you've got away with it so far.
Because it's relevant and easily accessible information. If the circuit stops working a few years because purchasing has bought a cheaper inductor with a different shunt capacitance, it's handy if you can pull the original value out of the Spice model.
Trying to find the original data sheets for old parts is one of the bits of the electronic engineer's job that I've not got any nostalgia for.
That you think matter, at the time.
And even less to exclude things that might matter, if they changed.
You can't recognise thinking when you get your nose rubbed in it. You are in fact advocating not thinking about any aspect of a circuit that isn't actually giving you a problem at the moment, as tinkerers are prone to do. It's a habit that can get you into trouble from time to time.
You do seem to be an authority on that ...
Manyana.
-- Bill Sloman, Nijmegen
Cool. I bet their jitter was even better than yours, though the programming time would sure have stunk.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 845-480-2058 hobbs at electrooptical dot net http://electrooptical.net
Don't know about the jitter; we run a couple ps RMS from an async optical trigger. If their impedance was low and the reflection amplitude low, they may have been worse. That air freshener is a big antenna, too.
They did develop software that turned a waveform into an n/c machining program.
-- John Larkin Highland Technology Inc www.highlandtechnology.com jlarkin at highlandtechnology dot com Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom timing and laser controllers Photonics and fiberoptic TTL data links VME analog, thermocouple, LVDT, synchro, tachometer Multichannel arbitrary waveform generators
figure
the
inductor.)
saturate
current
and back to Q looks like a forward biased c-b current path?
yet.
Obviously. What an engineer does is a quick mental, or maybe calculated, thing to determine if trace resistance, or pad capacitance, or prop delay, or part parasitics, are within orders of magnitude of making any difference. Sometimes you just know this from instinct or experience, sometimes you have to actually stop and think about it.
It would be insane to do full nonlinear EM analysis of a slow circuit, or to include parasitics that won't change things enough to matter. Engineering involves knowing or calculating what matters and what doesn't... otherwise you'd never get anything done.
You, apparently, never get anything done.
Thousands of manyanas. I wonder if it will oscillate in your lifetime.
-- John Larkin Highland Technology Inc www.highlandtechnology.com jlarkin at highlandtechnology dot com Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom timing and laser controllers Photonics and fiberoptic TTL data links VME analog, thermocouple, LVDT, synchro, tachometer Multichannel arbitrary waveform generators
rote:
ug.
feet
rd
d not
st
,ed text -
Wow, Hey your dropbox launched paint on my computer and stuck your drawing in it.
Thanks for the nice reply.
I had a =93massive failure=94 in regards to impedance matching a few weeks ago. We=92re marketing some ultrasonic teaching equipment in the US for a German company. They do a lot of medical applications, and I was adding some more physicsie things to it. One idea was to make an anti-reflection 1/4 wavelength plate to go between the transducer and some piece of metal. (Like the AR coating in optics). The transducer put out pulses at about 2 MHz, and had an acoustic impedance of about
3, brass has an impedance of 40 (in the same units) and that of aluminum is about 10. (So close to the needed geometric mean.) I got the correct thickness of Al for 2 MHz, and slapped a piece on each end of a brass cylinder. It didn=92t do squat to the amplitude of the pulses. Though it did change the phase around a bit. The next day I finally realized that you can=92t make a 1/4 wave plate for a single pulse. (Duh!) Using a CW source it works great.George H.
.highlandtechnology.com=A0 jlarkin at highlandtechnology dot com
figure
the
schottky
inductor.)
saturate
current
and back to Q looks like a forward biased c-b current
For some reason, TD&H don't do it.
Because you've now iterated through five
Official edge launch connectors are a convenience. But you can butt-solder a flanged SMA to the edge of a PCB and get just as good matching to microstrip.
You chose to ignore the parasitics. So did I. You whine when I do it.
Zero is a good value for accurate simulation of this circuit. 1 pF isn't.
-- John Larkin Highland Technology, Inc jlarkin at highlandtechnology dot com http://www.highlandtechnology.com Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom laser drivers and controllers Photonics and fiberoptic TTL data links VME thermocouple, LVDT, synchro acquisition and simulation
figure
the
inductor.)
saturate
current
and back to Q looks like a forward biased c-b current path?
yet.
I don't get into trouble over simple stuff like RLC step response. I don't think we've ever had an LC filter break. I sometimes get into trouble because a transistor or a complex IC is inadequately specified, or its Spice model is bad, things like that. But even then, not very often.
You don't get in trouble because you don't do anything.
Absolutely. I've seen a lot of bad designs, a few of my own. The biggest design problem isn't things that don't work; it's the things that don't sell.
-- John Larkin Highland Technology, Inc jlarkin at highlandtechnology dot com http://www.highlandtechnology.com Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom laser drivers and controllers Photonics and fiberoptic TTL data links VME thermocouple, LVDT, synchro acquisition and simulation
-586x169.jpg
I agree that the note in the log book implies the term was in use before, but in video and live presentations Grace did claim that the word usage originated there.
The line in the log book looks like it could have been paranthetic and squeezed in later to support their endless jokes after that.
My favorite story from her lectures was the part where during the war, she was asked to calculate the cost in lives and resources to take each island in the chain to Japan. She did that, but also calculated the cost of NOT taking every island along the way, the risks of hopping OVER islands along the chain. The result saved a considerable number of US lives.
Grace herself claimed it started among their group. Her reputation is one of exceptional honesty.
The origin of the computer bug
Although Hopper is often credited with having found the first bug, she was always careful to mention when she recalled the story that she was not present when the bug was actually found. When she was involved with testing the Mark II Aiken Relay Calculator at Harvard in 1947 =96 not 1945 as mentioned in many accounts of this story =96 the machine malfunctioned. Upon investigation the operators found a moth stuck at Relay #70, panel F. Using tweezers, the moth was removed and taped to the computer log with the explanatory text =93First actual case of bug being found.=94 The story was put out that the computer had been debugged and the term =93debugging=94 had been added to the computer world=92s vocabulary. Hopper often said =93From then on, when anything went wrong with a computer, we said it had bugs in it." The original logbook page, dated September 9 and with the moth still taped to it, is located in the Naval History Center at the U.S. Naval Surface Warfare Center in Dahlgren, Virginia.
She's full of manure. T.A. Edison was using the term in published material = as early as 1889.
=20
The wizened and delusional old senile sycophant fool is again full of manur= e:
"After World War I, the Versailles Treaty gave Japan a mandate over former = German colonies in the western Pacific; specifically, the Mariana, Marshall= , and Caroline Islands. If these islands were fortified, Japan could, in pr= inciple, deny the US access to its interests in the western Pacific. Theref= ore, in 1921, Lieutenant Colonel Earl Hancock Ellis of the US Marine Corps = drafted "Plan 712, Advanced Base Operations in Micronesia," a plan for war = against Japan which updated War Plan Orange by incorporating modern militar= y technology (submarines, aircraft, etc.) and which again included an islan= d-hopping strategy.[5] Shortly afterwards, a British-American reporter on n= aval affairs, Hector Charles Bywater, publicized the prospect of a Japanese=
-American war in his books Seapower in the Pacific (1923) and The Great Pac= ific War (1925), which detailed an island-hopping strategy. The books were = read not only by Americans but by senior officers of the Japanese Imperial = Navy,[6] who used "island-hopping" in their successful southeast Asia offen= sives in 1941 and 1942.[7]"
The log book is in the Smithsonian and not some worn-out waste dump militar= y museum.
Get a clue, most history as recalled by military related people/activities = is 100% manure.
So use the RCD delay. You still haven't mentioned what's wrong with it. Just not novel enough?
Tim
-- Deep Friar: a very philosophical monk. Website: http://webpages.charter.net/dawill/tmoranwms
back to Q looks like a forward biased c-b current path?
You know, I should read what I type more often, I meant to say spatial charge.. but what ever..
Now if I could only figure out how to kill fruit flies coming from some unknown place in my home? The climate has been bad here lately and these little buggers are just loving my home. It has been suggested to use a HV fly swatter near the source, well that would be fine, if I knew where the source was...
Jamie
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.