I have noticed this many times over the years.
For no apparent reason, the EGTs go for a walkabout, by as much as 20F, with a sort of sawtooth waveform with a cycle of some tens of seconds.
It isn’t a duff sensor, because all the EGTs go up and down together. It isn’t flying through pockets of air at different temps. The engine power output does not change – the speed is rock solid. The CHTs do not change.
It could be a duff instrument, but why only on the EGT channels and not the CHT channels?
Switch to Celsius and see what you’ll get.
A/D converter sampling ?
It happens only sometimes.
Fuel flow fluctuation changing the mixture?
Check for good ground connections, and test with transponder and DME off. Transponder and DME are sources of high RF peak output which can easily be pickup by these wirings, especially as the instrument impedance is high and the signals are very weak, just some millivolts.
Also be sure that it uses engine ground only, multiple grounds, or single airframe ground or ground bus is not suiteable for this instrument! Multiple grounds may form loops which act as pickup for noise.
On a Saratoga I had to use additional shielding to prevent interference.
A very good point about the transponder or DME.
My memory is not good, but I do recall having visited this issue before, a long time ago. So I went through my flight notepads (which I have all the way back) and 2 years ago I tracked it down to the wingtip strobes!
The thermocouple wires are not shielded – as is normal in GA wiring for these instruments.
Fuel flow fluctuation changing the mixture?
If I thought it was that, I would not fly the plane at all. I would immediately get the fuel servo overhauled. Anything looking like a fuel system issue is a total no-fly for me.
Wing strobes could indeed give same type of interference, could be the case. Try to seperate wiring as far as practical, or add shielding.
The orginal wiring indeed lacks shielding (or twisting) which makes is prone to electro magnetic interference. As you can’t change this wiring for shielded wiring (special thermocouple wiring), you could only add shielding in the form of braiding. This is what I had done and solved the problem at the Saratoga. This AA59569R36T0500 MIL spec braiding is tinned copper braid. It requires to get the wiring disconnected, to be able to get the braiding over the wiring. Connect the braiding to ground for maximum efficiency.
This is one of the “problems” with modern instrumentation. A variation like this, where we are talking about a variation of less than 1%, you would NEVER pick up on a analogue instrument.
It is one of the things I dislike about the “glass cockpit” modules, with the Airspeed and Altitude readout. When handflying, it jumps from number to number, which can be distracting, where as on an analogue gauge the needle movement would be very small, and more intuitive.
RPM readout on the EDM models as well. It can easily jump plus or minus 10 rpm (the smallest increment). On the analogue RPM gauge you would never see that difference.
Sometimes modern instrumentation is overaccurate. Then it becomes a distraction.
..it’s like a CD player with a scratch on it..it get’s it right until something screws up the ADC sampling…DME, XPDR pulse…well my take..as the JPI are known to have fast probes/converters maybe the ADC converter get’s messed up by an EMI pulse… 
