LMX2820: LD pin does not come on if LD_VTUNE_EN = 0

Noel,

Thanks for your reply. In our application, the reference signal isn't fixed so if you are referring to quick-cal, it may not work or will it? We do calibrate the VCO to obtain the vco_sel, daciset,and capcode to use for full-assist, if this is what you meant. For now we will keep the LD_TUNE_EN =1.

Onto another topic with this chip, a very odd thing is noticed. Whether in normal autotune operation or full assist, as I sweep the frequency from 5.65G to 22 GHz and back down as 1 full cycle, I would run into random locked failures. I have used freq step sizes of 1, 10, 20, and 100 MHz, and stepping rate of 1ms, 10ms, and 100 ms. It doesn't matter what freq step is used or the rate of change, the device would go into a unlock and hung state at some frequency in the 11 to 22 GHz range. It happens randomly, sometimes on the first cycle, sometimes after a couple of hundreds of cycles. It would only fail whenever the doubler is engaged, not at the same frequency or at the same frequency region. I monitored the LD pin, and if LD goes low the loop is stopped and the signal is examined on the specAn. In all cases it's found to be truly unlocked. The signal is typically within 5 to 20 MHz from the desired locked frequency and drifts around.

Now the weird thing is that the chip total stops responding to further frequency changes. I would rewrite R0 to retrigger, or I would also reprogram all other necessary registers of the same failed frequency such as the N, R, current, etc and R0, but nothing happens. When it fails and stops, I would obtain the VCO#, capcode, and daciset values. Then reprograming it to a new frequency in the doubled region (11-22G), The VCO#, capcode, daciset remained unchanged. When I reprogrammed it to a new (not half) frequency in the (5.65-11GHz), the doubler is switched out as it should (the direct VCO signal is much larger now) but the VCO#, capcode, and daciset still remains unchanged. The chip by and large doesn't respond. Further I would reload all 123 regs to default (with changes according to your registers document), and then reprogram only to find no changes, still unlocked, and VCO#, etc. remained the same. 

The randomness of failure in frequency and occurrence make it difficult to pinpoint what the issue is. The only way to get the chip to respond again correctly is to soft reset and reprogram the failed frequency. Have you noticed this behavior in the past or being reported? We have been fighting this problem noticed by a customer since last week. Any hints and help would be greatly appreciated.

Thanks

Hi Hon Yee,

With the doubler engaged, we also noticed that VCO calibration failure with small probability. This issue can be completely bypassed if we use full assist. 

Expect for the DACISET, VCO_SEL and CAPCTRL parameters, the doubler has some additional parameters that we should manually programmed in full assist mode. If you have decided to use full assist for the whole frequency range of LMX2820, I can provide the programming information to you. 

Hi Noel,

I have been looking for information to do full assist on doubler calibration since the last message,  after we figured that doubler cal en was the problem. I would greatly appreciate information on how to do doubler cal and application. Please email me. 

This will save us a lot of headaches and possible product recall.

Thanks

Hon