LMX2820: Clock & timing forum

mike leggett

Part Number: LMX2820

We still see a 10% failure rate for instcal when the PDF is 200Mhz where we see no failures when the PDF is set to 50Mhz. As it says in your data sheet, the instcal results remain until the power is removed from the device. We see that this is true, but in the case where the instacal fails, that failure cannot be corrected without cycling the power to the device. Note that writing to the reset bit in register 00 does not reset the instacal so that is no help.

Is there an undocumented bit/register that we can write to to reset the instcal so we can avoid cycling the power in the event that instcal fail. Note that if instcal fails, most frequencies will not lock as I suspect that is sets the wrong oscillator, cap or current.

Also, while we feel we have successfully implemented the instcal from the vague instructions provided, can you provide more complete and definitive instructions for instcal, noting the differences when the doubler is used or not. Note that "wait a while" is not definitive. What is the worst case wait required for instcal to complete in seconds?

see previous help requests that were not completely answered.

thanks,

Mike

over 3 years ago

0 Noel Fung over 3 years ago

TI__Guru** 107190 points

Hi Mike,

Give us some time to look into this, we will get back in next week.

0 mike leggett over 3 years ago

Prodigy 80 points

this is "next week", right? I got an email from ajeet Pal asking questions about our workaround that are already described in my post. Please answer the questions that I have asked.

0 Noel Fung over 3 years ago in reply to mike leggett

TI__Guru** 107190 points

Hi Mike,

as far as I know, instcal runs on fosc instead of fpd, do you have fpd = fosc?

in order to get an appropriate instcal, we should allow sufficient time for the engine to run the calibration. INSTCAL_DLY is used to set up this specific time. Did you set this register up according to the equation? when you say instcal fail, did you mean, after instcal, you cannot make it lock to any desirable frequencies?

There is no specific bit to reset instcal, including the hidden bits. If you set INSTCAL_EN from 1 --> 0 --> 1, that should reset the calibrated instcal values.

if we write RESET =1, it will reset everything to the silicon default value, we need to write all registers again to setup the synthesizer.

The exact time taken for "while a while" was not well documented, I was told it is in ms range, we should be able to see the LD pin turns back HIGH after calibration is complete. Enabling the VCO doubler will increase the calibration time of instcal. Frequency switching time after instcal is not affected.

Instcal operation procedure as described in the following post is correct.

https://e2e.ti.com/support/clock-timing-group/clock-and-timing/f/clock-timing-forum/1047349/lmx2820-register-programming-order-for-applying-instacal-data?tisearch=e2e-sitesearch&keymatch=%25252520user%2525253A490382#

0 Ajeet Pal over 3 years ago in reply to Noel Fung

TI__Genius 12560 points

Hi Mike,

We already have taken it offline. Hence closing this thread now.

Feel free to open again or create new thread for any queries.

Thanks!

Regards,

Ajeet Pal