Thanks. I have to fix my parallel plate capacitance calculator.
What kind of circuit wants 5 high speed nodes? The GHz stuff is on 50 ohm conductors terminated in the device. The other nodes are DC bias and lf stuff.
I was referring to JL's habit of cutting the copperclad with a Dremel. No way to figure out where the ground currents are going.
In inflation-adjusted dollars, a 4-channel 500 MHz scope is cheap these days, and has all sorts of goodies that the ba's never had, like ethernet for example. So we can overkill on bandwidth. A ringy 500 MHz scope looks fine measuring a 2 ns input step.
As I've noted, some digital scopes display way less jitter if you trigger off a channel, and not through the external trigger input.
--
John Larkin Highland Technology, Inc
picosecond timing precision measurement
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
It looks inductive, but how inductive depends on the slot length and operat ing frequency. It does matter in the frequency domain--it's a great way to make an unintentional oscillator or to couple two traces that cross the sam e slot, since it's like a CM choke.
My TDS 784A has Ethernet courtesy of a Prologix dongle. It can actually control several instruments at once, though I have never done that. I just bought a second one for 5 bills in nice shape.
Yeah, horses for courses, I agree, especially in an automated test stand where you can put the fudge factors in once and for all.
There has been a lot of talk about microstrips that cross ground gaps on the signal integrity list recently.
- snip -
------------------------------------------------------------------ To unsubscribe from si-list: snipped-for-privacy@freelists.org with 'unsubscribe' in the Subject field
or to administer your membership from a web page, go to:
formatting link
For help: snipped-for-privacy@freelists.org with 'help' in the Subject field
List forum is accessible at:
formatting link
List archives are viewable at:
formatting link
Old (prior to June 6, 2001) list archives are viewable at:
Well, if you're building radios maybe. But even then, the LO has to come fr om somewhere, and randomly sprinkling nanohenries and picofarads makes that a lot harder.
My 100 MHz nanoamp front end (within 6 dB of the shot noise of 1 nA in 100 MHz--60 electrons RMS noise, total) uses an ATF38143 pHEMT cascoded with a BFP640 SiGe:C bjt. That couldn't be done Manhattan-style, but it worked OK on a tiny blob board. That has at least 5 high speed nodes.
The back side of my breadboards is a solid ground plane, so I can do
50 ohm microstrips on top. I also usually have a lot of ground on top, to make coplanar waveguide lines and to ground parts and bypass caps.
I usually have a few 2-56 nuts-and-bolts that tie the top and bottom grounds together in a few places; they are where power supplies and scope/DVM grounds connect too. I also sometimes drill vias and solder in wires to connect top and bottom grounds, but that's a nuisance so I tend to not.
People pay way too much attention to "return currents", thanks to HoJo's silly book.
--
John Larkin Highland Technology, Inc
picosecond timing precision measurement
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
Nah, they weren't that bad. My 1998 Tek catalogue lists the SD-24 at $6950. For my purposes it's more important that they're about $150 now. :)
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510
hobbs at electrooptical dot net
http://electrooptical.net
You've said that several times. It doesn't matter, *provided* there is some other plane covering the slot, or if the slot is short enough w.r.t. the important wavelength components of the signal, or if you don't care about signal integrity.
A fast signal on a track running over a real ground plane slot ends up looking terrible.
Well, I made some boards and tested them. Some were 2-side, some had additional planes.
A slot is shunted by the native FR4 all around, and may additionally be shunted by other plane layers if present. The cut does make a slotline resonator, but if its impedance is low compared to the trace impedance, or if the frequency is high compared to the signal edge, there's not much excitation of the slot. I saw the FR4 fiberglass weave more than I saw any effect from a ground plane cut, in TDR. These are time-domain signals, not sine waves, but a resonant effect should show up in TDR.
Plane layers are lossy on conventional boards, which probably kills the slotline Q. The bottom line for me is that I experimented, and don't worry about it any more. The closest that I get to a slotline is either a row of via clearances that cut a plane, or a trace crossing different power pours on a power layer.
--
John Larkin Highland Technology, Inc
picosecond timing precision measurement
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
100MHz is not high speed. Each transistor has some lead capacitance. 1.5pF at 100MHz is 1/(2*pi*1e8*1.5e-12) = 1,061 Ohms. You will have at least that much.
I don't know what you mean by a blob board. Some nodes in the cascode will be critical, others are not so important. You build the circuit appropriately.
The same thing happens with high impedance nodes at low frequency. Air wiring is recommended for the LMC660 and similar parts.
You don't go blindly in and apply the same technique across the board. Use
50 Ohm where needed, air wiring when appropriate, and Manhattan where it helps. The point is to avoid chopping up the ground plane. You need the ground.
Hmm, As someone with much less experience in fast stuff. I'll say that the return current idea, (for fast pulses I see 'shots' of charge moving around.) is one of the few things I know and use by/from H. Johnson.* I fixed a fast pulse board by a co-worker, with a bit of copper tape, and some solder mask scraping. (He had these "different grounds", star point ideas...)
I do mostly two layer pcb's, signal first, cut's in the ground plane due to power are no problem, so that's last. 'Control stuff' I think about, but as you say mostly doesn't matter.
Perhaps I wasn't explicit enough. If you were building the LO yourself, you'd have rather more than 5 high speed nodes, even in a radio.
Nope. I've posted about it before. A nanoamp, remember. Sixty electrons. Have a look in the archives if you're interested.
It has very small (~0.4 mm) copper squares over a ground plane. You make wires by blobbing solder, and it's naturally good for very small SMTs (SC-74 etc).
Yah. But there are more than 5 critical ones.
Completely different problem.
"Win awards." Did you just win the Rory award for the most gratuitous use of the word "Belgium" in a serious screenplay? (*)
Cheers
Phil Hobbs
(*) From "Life, the Universe, and Everything" by Douglas Adams.
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510
hobbs at electrooptical dot net
http://electrooptical.net
The VCO output signal in the LO is the only high speed node. Anything that needs it goes on 50 Ohms.
That has nothing to do with high speed. That is air wiring or something similar. You still have node capacitance and transistor input and output capacitances.
Copper squares over a ground plane sounds like Manhattan.
So use the appropriate method.
Same problem. Wiring components on a ground plane.
Star grounding is almost always silly, even sillier in fast stuff. Ditto splitting analog and digital grounds.
Fast things are more difficult on 2-layer boards. If you have a nice layer-2 ground plane and four or more layers, all sorts of worries (like return currents) go away.
I started highlighting the absurdities in his book, but got tired about halfway through. His writing style is reall bad... how are his lectures?
--
John Larkin Highland Technology, Inc
picosecond timing precision measurement
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
Communication is not occurring. Good night, Irene.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510
hobbs at electrooptical dot net
http://electrooptical.net
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.