Noise in Rx I and Q

System Development
1

Hello,
We got the Rx part working but see some noise in the I and Q signals. This is Q

Rx_noise_Q
Rx_noise_Q.png1365×456 5.36 KB
This is I
Rx_noise_I
Rx_noise_I.jpg1344×456 98.4 KB

The noise in Q seems to be stronger in I. What could the problem be?

The setup is as follows:

  1. CLKGEN set to 200MHz.
  2. ADC and RxTSP running at 50MHz
  3. LNA, TIA and PGA gains are set to 0
  4. RF Input signal is at -5dBm and 2MHz away from LO
  5. All blocks inside Rx TSP are bypassed including GFIRs
  6. HBD is set to 1 (no decimation)
  7. No calibration is done
  8. RBB filter is bypassed
  9. Please ignore the amplitude number (it has been scaled by 16)
    Looking forward to your help (like always) in fixing this problem,
    Regards,
2

Hi @EnthuMan,

It looks like one IQ data interface line is failing/stuck. Or you are missing samples at these points.

3

Hello @Zack,
We checked TSG and constant values in I and Q and didn’t notice anything wrong there. (Why does it seem that Q has a stronger spike compared to I?).
Do you strongly feel that this is a problem on the digital line (and not a setting in side the LMS)? We will proceed as per your suggestions.

4

Hi @EnthuMan,

I think it is coming from digital part.

I would not trust this kind of test. Could you set RxTSG to NCO mode and capture data. Draw the data using raw values instead of scaled.

5

Hello @Zack,
You were right, there is a problem with the data reception. The top plot is channel A and bottom is channel B (ch A has the problem). The acquired data also behaves in the same way (the spikes are in chA and not in chB).

NCO
NCO.png1572×834 10.4 KB

Please advise on the next step?

6

Hi @EnthuMan,

Not sure what do you mean. Let us call them AI, AQ, BI, BQ just to avoid confusion.

Anyway, as I wrote before, look for stuck pin of LML interface. Soldering issue probably. One option is to provide data samples from RNDGEN in LML and see with a scope if every pin is toggling. Another option is to provide data from TSG configured as DC. Set for instance AI, AQ and BI to 0x000, while BQ to 0x800. Check if DIQ[11] is toggling with a scope or with your software. If yes, then set BQ to 0x400 and check DIQ[10] and so on.

7

Hi @Zack,
AI and AQ have the problem while BI and BQ are error free. Could this rule out a soldering problem?

8

Hi @EnthuMan,

Assuming RxTSP settings are OK, then something wrong in LMS7002M.

9

Hello @Zack,
The problem has been resolved. The decimation settings were not correct (the decimation was set to 1 (2^0)).
Now we are able to receive the signals properly.
Thanks for your help.

1 Like