8

Pin header jumpers (like these) can be used to make optional connections between parts.

I would expect at certain frequencies that the these kind of parts become primarily reactive (the imaginary part of the impedance becomes greater than the real part). At what frequencies does this typically happen?

jskroch
  • 127
  • 4
Andrew Spott
  • 1,400
  • 8
  • 21
  • Any square wave (digital signal) would be a concern. – Harvard Aug 14 '16 at 21:21
  • 2
    Depends on many factors, including length of any unterminated stub. I don't think your criteria is accurate at all. The DC resistance is basically zero. So even at 100 kHz, the imaginary part might be larger than real part. Assuming the layout is otherwise good, it will probably work up to 100 MHz. But it depends on many things. Certainly 10 MHz should be no problem. – user57037 Aug 14 '16 at 22:02
  • There is a certain amount of pF capacitance to worry about with any header itself, let alone the jumper (which may be added, or not.) For example, the original PCI bus could only handle four connectors max (plus two ICs) on the bus, due to connector capacitive loading plus the loading of a board plugged into it. That was at 33MHz, too. I think the allowance was 100pF total, with 10pF per connector and another 10pF per board. 4 connectors used 80pF, leaving 10pF each for the remaining two ICs (south bridge and regular bridge chip.) What frequencies are you talking about here?? – jonk Aug 15 '16 at 05:15
  • @jonk I'm wondering what frequencies I need to be worried about this stuff. At the moment, I'm mostly dealing with low (sub 10MHz) stuff, but I don't know when this stuff might matter, so I was asking. – Andrew Spott Aug 15 '16 at 05:33
  • 1
    Stay under 10MHz, I'd say. That's a mental line I'd draw in the sand, without getting into the nitty-gritty details. (And even then?) If you are serious about contemplating pushing above 10MHz, you really need to search for exact loading details of the connectors. Or, better yet, start setting up your bench to test them and examine the results with a good oscilloscope on hand and appropriate testing setup and procedures. – jonk Aug 15 '16 at 05:36
  • At high freq, well into GHz, you start to see that wires are high impedance and gaps between unconnected bits of conductor become low impedance. If you are interested, you can look at the EEVblog, #199, here: https://youtu.be/J9_JxRiuS5A and take a look at what happens just after minute 15:30 on the video -- the front end for a Tektronix high speed scope. – jonk Aug 15 '16 at 05:49

1 Answers1

6

From a microwave engineers point of view, things really start to become a problem when the physical dimensions of the component are comparable to the wavelength of the signal you are considering.

The length of the pins onto which this jumper mounts are 0.230" (5.84mm) and the separation is 0.100" (2.54mm) so the signal path length is approximately 0.560" (14.22mm)

This dimension represents 1/10 of a wavelength at about 2.5GHz. To reasonably represent a square wave, I would include upto the fifth harmonic - so I would place an upper limit of 500MHz on this part. This is a heavily simplified first off starting point based on rough dimensions and rules of thumb and obviously you should verify the performance by either a full 3D EM simulation if you have the facility, or by direct measurement of a test piece using a network analyser.

N.G. near
  • 435
  • 2
  • 10