Jeroen Belleman wrote in news:r7bjkh$1ohi$ snipped-for-privacy@gioia.aioe.org:
There are exactly ZERO teflon sheathed wires in a nuclear reactor.
It turns teflon to a powder. I wonder how delrin fares.
Search Amazon for "edge-launch SMA connector." Ones that look just like your link cost a dollar or less with free Prime shipping.
That's incredible.
The fat center pin adds capacitance that matters for really fast stuff. Just the connector pins in free space, no PCB, are below 100 ohms. We did an ATLC em simulation to optimize PCB stackup. If you cut away all internal PCB planes near the center pin, and put ground on the bottom only, it's pretty good.
The more serious microwave SMAs have a tiny flat center pin. We are about to test some layouts.
-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
Yes, I find the Amazon prices incredible too, often more 3 times the AliExpress price. I just try to guess what stock I will need in the future, so I can wait the extra month for delivery to the Netherlands. I'm not pressed for time anymore. And, since I only use them for 868 MHz the quality is good enough (contact resistance and robustness is more important than SWR at that 'low' frequency). And they did not have bent pins like the one on the Amazon picture :-).
For very critical analysis you might add the center pin solder fillet.
On that PCB, I do expect reflections by impedance discontinuities at the intermediate locations, like when testing from J11 to J10 at the J27,28,29 pads. Isn't that more critical than the difference between round and flat pin? Why not add GND traces with the same width as the pads, to prevent these discontinuities?
Regards, Arie de Muijnck
I meant incredibly low. Under $1 for an gold plated SMA edge-launch, with shipping, is effectively free.
I just try to guess what stock I will need in the
Those boards (we have six) will be sheared up into a bunch of txline experiments of various length. J27-28-29 might make minor discontinuities in TDR, but that's the worst part of the layout, with cheap connectors. I don't think the intermediates will have much effect on overall TDT wave shape. We could always hack them too.
Why not add GND traces with the same width as the
That would make this a coplanar waveguide experiment. We might try that some day. The immediate problem is to help us decide when FR4 is good enough for microstrip circuits.
On the more serious microwave connectors, the center pins are tiny, which is more important than their shape.
The serious microwave ones, like J7, have more distance between the ground pins. A coplanar waveguide with that spacing is indistinguishable from a plain microstrip.
-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
I see. Interested to see the TDR results.
I assume the GND planes under the wide and narrow tracks are on different depths to get the same impedance? That would give a nice indication on FR4 and track width performance. Less FR4 depth = less loss, but narrower copper track = more loss? It might compensate.
Regards, Arie de Muijnck
Our real question is how far we can run a 50 ohm trace on FR4 and get a decent x picosecond rise time, and also which connectors work better to deliver that edge.
We'll do some experiments and write it up. I'll post it here. It may take a while, given the current chaos.
If we don't like the numbers, we might try some microwave laminates. Right now, we want to drive a Mach-Zender modulator with a roughly 50 ps wide pulse, and the critical PCB trace is longer than we'd prefer, a bit over half an inch long.
Yes. The two on the right have the general plane, 25 mils below the surface. That transitions to full-board, 62 mils, ground for the two on the left.
That would give a nice
I think that less depth is more loss. Might multiply.
Our experience suggests that the ones on the left will be better. Wider traces have less skin loss. Cheap FR4 boards are dominated by skin loss; the current density is mostly on the bottom of the layer 1 foil, and the bottom has the drecky black oxide treatment for adhesion.
-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
o
...
Philmore? Their stuff used to be worse than Calrad's early Japanese crapnec tors.
They have been around for a very long time. One early product was Galena cr ystals an cat whiskers.
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.