-10dBm RF Power Reference source-Output filter needs improving.

[Xnke]

Today I put together a little RF power reference for my home shack, and after getting it assembled I took it into the engineering lab my cousin works in, and we ran it through a few tests. He doesn't do RF work other than compliance testing, and they don't have the "proper" equipment to do the tests the way I know how to set them up, so we did what we could and got what we got.



In the schematic above, you see I'm running a 3.3v CMOS clock oscillator, using a capacitor to remove the DC offset, then clipping with a pair of antiparallel diodes. This sets the RF peak voltage to +/-0.6V. After that, the clipped signal is fed into the 25Mhz low pass filter, then a 1.75db 50ohm pad, then the SMA connector.

Output level is almost perfect, 25Mhz -10.78dBm as measured on the recently calibrated Rhode&Schwartz SA in the lab. The target was -10dBm, so I only undershot the pad a tiny bit. However, there is a spur at 75Mhz that the output filter didn't knock down enough-it's still -82dBm. It's *there*, but it's low enough not to be a problem in most cases.

Is this just a case of needing another section in the filter to kill that off, or have I done something else that is upsetting the output filter? It's not *really* a big enough issue to worry about, especially if the 75mhz spur doesn't fluctuate in power after I get this thing into the aluminum case, as I'm only really concerned with *total* power output. If it turns out that I have -10.78dBm at 25mhz and -82dBm at 75Mhz, and they're both stable, then the reference is still fine. Actually, it would be OK anyway-as long as I know the total power-which is gonna be pretty slightly more than 0.100000005W total. That's pretty close, but since I'll be calibrating a wattmeter that goes down to -70dBm, it's probably important to get all the digits I can in there.

Should I rebuild the whole thing and kill that spur at 75Mhz, or just not worry about it and call it good enough for a bench reference?

[Xnke]

Tweaking the inductors up to 540nH and adjusting the first and last capacitive elements to 120pF and the filter straightens up a bit, forming a bit of a bandpass response and dropping the 50Mhz and 75Mhz transmission by another 15dB, while doing almost nothing to the desired 25Mhz signal.

[David Hess]

Harmonic filters often include a notch for the 3rd and 2nd if produced.

I would not worry about the low level harmonic if its amplitude is on the order of the error from the temperature coefficient of the diodes.  Leveled RF sources usually use a temperature compensated detector to drive an AGC loop to avoid this source of error.

[T3sl4co1l]

And, you can adjust the notch(es) by putting a little capacitance in parallel with the inductor(s).  The lowpass profile changes slightly because of this, and is easily retuned.

Tim