We were setting our system to work with a relative low LO and generate a simple CW tone. For this purpose we've set the LO to 90 MHz and feed a 20 MHz tone to obtain a 70 MHz output. To our surprise, the output signal seems to have a huge amount of spectral harmonics (due to the mixer between the LO and the input signal) and the signal at the oscilloscope looks terribly bad. Changing the input frequency for the tone didn't make the situation any better.
We wanted to verify this behaviour with the reference design and we got the same outcome (system being controlled by IIO tool). Even with higher LO frequencies (we tried several frequencies up to 1GHz), we see that the harmonics being generated become a significant part of the signal.
I wanted to know if this is a known issue and how to solve it? Please note that we are only looking at the transmitter side.
Are you measuring the harmonics of LO or tone?
Can you please share the screenshot of your harmonics measurement?
I've run the following test:
ADRV9009 reference design for Intel Arria10 + IIO for system control
ADRV9009 dev board
Profile: 122.88 MSps, 100MHz BW TX (default profile)
LO freq: 90 MHz
Signal at Tx1: Single CW @ 20MHz, -12dB, 90 phase shift
Tx Attenuation: 10dB
With these settings these is what I get:
1) Spectrum analyzer (3G Span):
As you can see the signal looks terrible. The frequency spikes are at frequencies:
250 MHz, 470 MHz, 610 MHz, 830 MHz, 970 MHz... which are intermodulation frequencies:
250 = 3*90-20 (-10dB)
470 = 5*90+20 (-15dB)
610 = 7*90-20 (-18dB)
830= 9*90+20 (-20 dB)
970= 11*90-20 (-23 dB)
Are these normal levels? Is there something I can do about it?
Ok thanks for confirming the behavior
Just one more question. Does the receiver use the same LO (i.e. square wave)?
Same is true for the RX. Without external filtering, the RX will observe the signals present at the odd integer multiples of the LO attenuated by 1/n.
Would an external LO (pure sinewave) solve this issue or is there any internal element within the ADRV9009 that would turn it to a square wave as well?
Even with external LO (pure sinewave) you will see this as this internally converted to square wave.