LMK05028: REF0 validation always fail

Hello Peter,

Can you please provide your TICS Pro configuration file (.tcs file)?

Regards,

Kia Rahbar

Hi Kia, 

attached is tcs file.

Thanks

Peter

5430.LMK05028_1PPS_config_40MHz_out_onboard_XO_and_TCXO_V2 - agxsom.tcs

Hi Kia, 

I saw one thread said it's duty cycle issue. Is it explainable?  But i can't adjust PPS duty cycle now.

e2e.ti.com/.../lmk05028-pps-validation-timer-issue

Peter

Hello Kia, 

Yes, that's thread posted by my colleague when he was working on TI EVB. So my TCS is the one I changed the XO and TCXO input frequency based on that. 

I also checked your provided tcs file on TI EVB. The important info is, with this same PPS input, as Xiao said, the REF0 is never validated on TI EVB. One day, when I cotacted scope probe to REF0 input, I found the STAT0 LED blinking normaly.  So i  populated R73 with a 47pF capacitor, surprisely I found REF0 is always validated. Unfortunately, this capacitor pach doesn't work on our own board.  That's why i posted this new thread and asked if there is any special in REF0 detection or this is just a chip bug.

Thanks

Peter 

Hello Peter,

Thank you for the additional information. Your current device settings are optimized for a 1PPS input.

I will test your new configuration in the lab tomorrow and provide you with one that locks on my end for your new XO and TCXO input frequencies.

We had initial specified the LMK05028 to support 1PPS, but have seen many issues lately so we plan to remove the 1PPS specification in the near future.

I will try my best to support you with a configuration that locks.

Regards,

Kia Rahbar

Hello Peter,

What stage of your design are you currently at? Can you switch to the LMK5C33216 or LMK5B33216?

Regards,

Kia Rahbar

Hi Kia, 

Our design is in production status. But we just start to use this function, that's why we have been talking long here.  It's better to keep LMK05028. LMK5C33216 or LMK5B33216 can meet same requirement? I saw the cost is much higher.  Our requirement is simple, a GPSDO, which can output 4x phase-aligned clocks. If GPSDO only has one output, I think we can use a fanout buffer.

If PPS validation is OK, does it mean LMK05028 should be OK？

Thanks

Peter

Kia, 

You can verifiy 40/40MHz and 1PPS on your EVB, and then we will try different ways to get REF0 validation.

Regards

Peter

Hello Peter,

Yes, the LMK5C33216 and LMK5B33216 can meet the 1PPS input requirement. The LMK05318B and LMK5B12204 can also meet the requirement. All four of these devices also have better jitter performance than the LMK05028 and have an improved 1PPS input locking.

Since you only require x4 outputs, the LMK5B12204 would be your best solution. This device has 4 outputs and supports 1PPS input.

The LMK05028 can support 1PPS inputs, but this device is older so the input structure of the device is not optimized for 1PPS.

Apologies, I have still not been able to test your new configuration in the lab.

Can I get you an updated configuration by the end of Friday?

Regards,

Kia Rahbar

Hello Kia, 

No problem, we will try your file first. Next step, we will consider clock mezzanine design. This will be flexible and decrease risk. I checked LMK05318B/LMK5B12204 is much cheaper than LMK5C33216/LMK5B33216 . For cost reason, it's better solution. What do you think?  I saw LMK05318B and LMK5B12204 are pin to pin compatible, right? same performance other than output number? 

Thanks

Peter

Hello Peter,

Yes, the LMK05318B and LMK5B12204 have a much cheaper cost. They have similar performance to the LMK5B33216 and LMK5C33216 as well.

Yes, LMK05318B and LMK5B12204 are pin compatible and have the same performance. The only difference is the number of outputs.

Can you please provide an image of your input termination for the 1PPS input?

Regards,

Kia Rahbar

Hello Kia, 

1PPS is configured as LVCMOS input, there is no input termination. there is only one 33ohm source temination.

By the way, if we change 1PPS to 10Hz, will it make REF input easier to validate?  I just checked the pulse output of our GPS can be up to 1KHz. 

Regards

Peter

Thanks Kia, I will try.

Peter

Hi Kia, 

Just want to double check with you, for chip programming

1. File-load, select your file

2. click Write all Registers

Two steps are OK,  no any other operations. 

Thanks

Peter

Hi Kia, 

We are trying. Although product is mass production state, for robust solution, we decided to use LMK5B12204.  Both LMK05318B and LMK5B12204 are out of stock, we also can't get it from distributor. Becuase of tight schedule, could you help us coordinate 100-200pcs?

Regards

Peter

Hello Kia, 

Which interfacing is prefered?  Both are in LMK5B12204 datasheet.

Thanks

Peter

Hello Kia, 

What about XO input?  Both pictures are for XO in page 30 and page 69(reference design). 

Regards

Peter

Hello Kia, 

Thanks, will use 125/375 for simplicity.

The following is draft sch. I appreciate your careful review since change's made for mass production product.

Thanks

Peter

Hello Kia, 

This is the latest one, we will proceed to PCB layout.

1. SYNCN pull to GND default

2. I2C is 1.8V in our system

Thanks

Peter

Thank you Julian.

Peter

Hello Peter,

To clarify, are you planning on using the SYNCN feature of the LMK5B12204?

Regards,

Kia Rahbar

Hello Kia, 

It's likely, so we reserve option. Is there any special consideration about this feature? 

Thanks

Peter

Hello Kia, 

One more point need to confirm, we will use a normal crystal oscillator for XO input, such as EPSON part:

https://support.epson.biz/td/api/doc_check.php?dl=brief_SG3225CAN&lang=en

https://support.epson.biz/td/api/doc_check.php?dl=app_SG3225CAN&lang=en

With 1PPS sync, we can get a very stable, precise clock output.

Thanks

Peter

Hello Kia, 

We will use 25PPM.  

BTW, Could you help us track sample status? PCB fab is pending now.

Best regards

Peter

Hello Kia,

Thank you, we will release PCB to manufacture today. 

Best regards

Peter