LMK5B33216EVM: TICS Pro settings not applied on LMK5B33216EVM

Hi Rob,

Can you please provide the .tcs file you are using?

I'm not following the purpose of IN1 when IN0 is already present. With ZDM enabled, it is not necessary to feedback externally.

Can you confirm the image below is the configuration you want? It would help if you provided a  block diagram so I can understand your system needs better.

Regards,

Jennifer

Dear Jennifer,

Thanks for your reply!
Actually, I have not prepared a tcs for my use case yet since, as I've mentioned in my post, what I measure on the outputs seems to be totally unrelated with even basic changes wrt the default configuration (e.g., the voltages and frequencies on the outputs are totally wrong) --- so far I've just been playing around, trying to make sense of these values I observe.
From the datasheet and the literature I thought that the GCD between the input and output frequency should be equal to the input frequency, and from other posts on the forum it looked to me that having a 1-PPS input signal led to extremely long convergence times. This is why I thought about looping back the frequency-adjusted 10 MHz signal for ZDM purposes.

Please see in the picture here what I had in mind. The overall idea is that an external PC extracts timing information from a radio signal and performs adjustments on-the-fly via I2C on the generated signals (e.g., increase the generated 10 MHz frequency by 0.001 Hz and phase-shift it by 2 ns), and those adjustments happen at a sustained rate (ideally once every 1 or 2 seconds) --- which would be incompatible with the 30-50 minutes convergence time that are mentioned in the posts discussing about 1-PPS inputs. The 1-PPS generated by the LMK5B33216 should be aligned with the 10 MHz (meaning it should have its rising edges every 1e7-th rising edge of the 10 MHz signal), with a duty cycle of minimum 25%.

Can this be achieved with the LMK5B33216?
Thanks again and have a nice day!

Regards,
Rob

Hi Rob,

1. I'm unable to click on the image link due to my company's firewall. Can you please upload a screenshot here?
2. "The overall idea is that an external PC extracts timing information from a radio signal and performs adjustments on-the-fly via I2C on the generated signals (e.g., increase the generated 10 MHz frequency by 0.001 Hz and phase-shift it by 2 ns), and those adjustments happen at a sustained rate (ideally once every 1 or 2 seconds"
   1. This is possible by using the DPLL DCO (digitally controlled oscillator) which allows you to make frequency and phase adjustments of <1ppb accuracy.
3. "which would be incompatible with the 30-50 minutes convergence time that are mentioned in the posts discussing about 1-PPS inputs."
   1. We have recently improved our register settings to allow for faster lock time.
   2. If the initial XO error is ±25ppb, it could take as low as 38s for the DPLL to lock. With 1-PPS +ZDM configured, it could take 8min for the DPLL to lock.
4. "The 1-PPS generated by the LMK5B33216 should be aligned with the 10 MHz (meaning it should have its rising edges every 1e7-th rising edge of the 10 MHz signal), with a duty cycle of minimum 25%."
   1. The LMK5B33216 can only output 50% duty cycle.
5. My apologies, I re-read your original testing procedure again. Let's focus on getting the default resolved first before working on another config with ZDM, that way we are on the same page with the setup.
6. With that, please repeat your previous EVM setup. In TICSPRO, please navigate to the Status page, issue a SWRST, click "Read Status", then send a screenshot of what you see.
   1. 

Regards,

Jennifer

Hello Jennifer,

1. Here's a diagram of the structure I had in mind:

2. Yes, the (extremely) high adjustability of the IC using the DCO is what drove us to buy it :)
3. From our previous PoC (where no external clock generator was used), the error between the reference and the desired target is initially in the 1-3 ppm range. The initial iterations of our algorithm are (relatively) high-frequency (~ 1Hz), then the system rapidly converges. Our PoC had poor adjustability and thus we had to keep a ~1 Hz frequency all along, but our goal with the LMK5B33216 is to progressively lower the interval between re-synchronizations to once every some tens of seconds --- which, if I understood your reply correctly, it is still faster than what could be achieved with 1-PPS+ZDM.
4. A 50% duty cycle would be fine by us (we have a lower bound, but no upper one).
5+6. With pleasure, thanks! This is what I've done:
  - started TICS Pro, restored the Default Configuration
  - in the Start Page, set the following values:
    - IN0 Freq 10 MHz, Interface type: CMOS
    - IN1 Freq 0 (unused)
    - OUT1 Output Format: CMOS 2.65-V. P/N = On
    - disabled all outputs other than OUT0 and OUT1
    - clicked calculate + copy + assign + apply
  - started the signal generator, with a 10 MHz square wave (2 Vpp, +1 V offset) sent into IN0_P
  - connected the oscilloscope, 50 ohm on inputs, yellow trace is the probe connected to OUT0_P, the green trace is the probe connected to OUT1_P, and the pink one is the signal from the signal generator (that also goes into IN0_P)
  - connected the board, clicked again on apply -> the frequencies of the board's outputs change -- e.g., OUT0 was 5 MHz before -- so we should have correctly applied our configuration.

Please find below (1) a screenshot of the current status in TICS Pro (after several minutes of execution), and (2) a capture on the oscilloscope.
The amplitudes are wrong (in particular for OUT1 --- no amount of weirdness on the cables should be able to make a 2V65 signal become 1/5th of itself)

I am of course doing something horribly wrong (sorry, not my area of expertise... :-( ), could you please help me in figuring out what exactly is my issue / point me to the appropriate documentation (I've read twice the whole LMK datasheet, and now I can understand most of the options in TICS Pro, but it is evident that I am still missing something crucial...).

Thanks again!
Regards,
Rob

Hi Rob,

Thanks for sending the detailed flow. I will review and work to get back to you later this week.

In the meantime, we also have the EVM User's Guide, not sure if you've seen it already. There's a section in the Appendix that contains TICSPRO guide .

Regards,

Jennifer

Hello Jennifer,

Yes, sorry, I've forgot to mention it. I've already gone over the user manual.
Should this help you in figuring out my issue, please find below the actual configuration I've used for the tests above.

Thanks again!
Regards,
Rob

first_attempt.tcs

Dear Jennifer,

Very sorry to put some pressure, but I am still stuck :(
Have you managed to find the issue in what I have been doing?

Thanks a lot for your help and understanding!

Best,
Rob

Hi Rob,

That's good to hear. Please make sure to run through the Start Page step by step and hit "Run Script" at the end to ensure the DPLL can lock to the entered input frequency.

Regards,

Jennifer