Hello!Recently I've bought a LTC6430-20 demo board DC2076A-B to estimate amp together with my ADC's front-end. And when I measured signal by ADC I've seen that 2nd and 3rd harmonics is pretty high, much higher than datasheet's values (for 300 MHz with same conditions I've measured about -64dBc for 2nd and 3rd vs -75dBc and -77dBc in datasheet, ADC's SFDR about -71dBc).To be shure that amp works well I've measured IM3 level as stated in board's manual. It says that if I apply 500MHz and 501MHz signals with 2dBm per tone than IM3 levels "should be approximately –98dBc". But in spectrum analyzer I've seen much higher level:
Generator's signal was filted by bandpass filter with more than 90dB supresson, so input signal is pretty clear, I've seen it on analyzer. Also i've try to change lab power supply to external board with linear regulator, but got same IM3 level, not a dB better.Board's config is SE to SE, as it had come to me.What's the possible reasons of such huge inconsistence? With that performance I will not reach nesessary harmonic's levels
We measure the OIP3 of every LTC6430-20 in production test. So its unlikely that it’s the part having an issue with OIP3.
Your spectrum analyzer plot shows your tones at 2dBm which means you're applying the LTC6430-20 output directly into the Spectrum Analyzer.
A common problem is the Spectrum Analyzer. Every top notch Rohde & Schwarz / Keysight spectrum analyzer will have an input IP3 of 25dBm. The output OIP3 of the LTC6430-20 is nearly 50dBm. So the Spectrum Analyzer (SA) input gets hammered an you only see its receiver non-linearity. To help first try adjusting the input ATTEN on the SA and see if the IM3 products go down. If the settings on the SA ATTEN are not fine enough step wise, you can add external ATTEN. We typically use 16dB of input ATTEN and 20-30dB of ATTEN internally on a Rohde & Schwarz that is above 23dBM IIP3.
Also move the tone spacing further apart to say 5MHz (instead of 1MHz currently) as the phase noise of the sources can get you when the tones are too close. As you know, good 2nd order suppression >50dB is required. Tying the 10MHz REF out of the GEN's is something also that we always do to help.
We also use some attenuation (e.g. 4dBm) at the input of our combiner when doing this test to help with providing some isolation / matching. What combiner are you using for the test? I guess you could also do an evaluation of your setup (without the LTC6430-20) by looking at the IP3 at your combiner output to make sure it is not a limiting factor.
Let me know if these steps help or if you have questions?