Electrical – At what frequency does jumper wire stop functioning properly

jumperpcbvlsi

I have designed an IC with some digital circuitry requiring a 25 MHz clock signal. Now I'm designing a PCB to perform testing. My question is, would normal jumper wires be sufficient to pass a 25 MHz clock signal, or would it create signal integrity issues? Should I go for an SMA connector?

Best Answer

Well the wavelength of 25MHz is 12m in free air. Let's call it 6m in the PCB (assuming Dk of ~4, calculator here). At 1/10th of that you are still at 60cm. So transmission line effects should not be a problem at that speed if they were a sine wave.

Now let's assume a square wave an use Bogatins rule of thumb here. Let's assume a 10% rise time. The period of a 25MHz signal is 40ns. A 10% rise time is 4ns. The Bandwidth of that signal is 87.5MHz, which works out to 3.4m wavelength in free space or let's approximate 1.7m in a pcb. At 1/10th wavelength you are at 17cm, so still probably safe unless the jumper wire is obnoxiously long. So again probably safe even for a square wave.

Ok so we are ok on transmission line effects. What about rise times?

Let's do another rule of thumb from Bogatin. 50Ohm transmission line has a capacitance of about 3.3pF/inch. Our wire is not impedance controlled and may be much lower, let's be aggressive and assume x2 that and 6inches of jumper wire. Well that works out to about 3.3pf/inch*6inches*2 = ~40pF. Let's say our driver has an output impedance close to 50ohms (tho usually lower). Well the 5RC time constant becomes 5*50*40pF = 10ns. So our rise time is now about 25% of our period, which may become a problem if our driver is weaker and say closer to 100ohm output impedance or the clocking requirements are stringent. We will see rounding on our signal but the rise/fall time may still allow the system to work depending on high/low voltage thresholds.

The next problem is EMI/EMC. Will we become a radiator or will we accept interference from an external source (fluorescent lights, other broadband devices? etc.). Here the answer is more complicated because it will depend on the loop area, geometry of the wire and geometry of the ground plane. This is why shipping products don't use jumper wires for these higher frequencies, especially for something as sensitive as a clock.

The other problem will be ringing of your system. What is the inline inductance in the whole system? Is it close to resonating at the 25Mhz? Again the answer is very geometry dependent.

So in conclusion probably the SI is ok just from the jumper wire but the real kicker is functionality in various real world scenarios. This should work for something like a 1 off breadboard test but is a bad idea for any serious design going forward. Definitely do a twisted pair for the clock or twist it with a ground wire if it's a single ended signal. Shield it if you can, even if that just means grounded tin foil (but be careful as the shielding will increase the capacitance and may start increasing rise/fall times).