LMH1219: Cable Length Problem in 4K

Hi Patrick,

I would recommend emulating the schematic from the typical application of LMH1219.  You could try replacing several of your components on IN0 accordingly.

Also you mentioned that you set trace width at 10 mil.  Can you confirm that the layout / stack up was designed to create 75 Ohm single ended traces on IN0+/- ?

Thanks,

Drew

Hi Drew

Thanks for your reply. Yes I calculated my 75OHM impedance with Si9000 and Allegro Impedance Workflow and it was right. Do you think it happened just cause my PCB or Schematic is wrong? Are you sure it is not about the registers?

I want to connect LMH1219 to driver(LMH1228).Here is my block diagram and it is very simple. Also you can see my impedance based on Allegro impedance workflow.

Hi Patrick,

Your block diagram from LMH1219 -> LMH1228 makes sense.

Regarding LMH1219, I'm not aware of any registers that will address this.  Currently there is double termination since the device has termination.  At the least, try removing 75 Ohm to GND and see how this impacts performance.  You can try replacing other components with zero ohm resistors.

A couple other thoughts, you could share your layout with us and we could review.  Also, have you tried taking any TDR or return loss measurements?  These might help us better understand the issue.

Thanks,

Drew

Hi Drew

Based on your advices, I changed my schematic to this and now it is compatible with datasheet but there is no differences.

I sent you my pdf schematic and Allegro PCB. About TDR actually I have no tools to measure it. 

LMH1219-Problem.rar

Thank you so much

Hi Patrick,

Thank you for sharing your schematic and layout.  Please allow us a couple days to review it.

Did you try reworking the components in your current design to see if this helps?  If you see improvements by doing this, this would indicate that we understand the root cause of your issues.  It might be worth trying before having another board fabricated.

Thanks,

Drew

Dear Drew 

Thank you for reviewing our design and take it your time. Yes we made another PCB based on your help and it improved but it was less than 5 meters and now we can get just less than 45 meters while we expect to get more than 65 meters cause datasheet says LMH1219 can give 75 meters length in 4K with Belden 1694A

Thanks

Hi Patrick,

Thanks for the update.  We will review and provide an update later this week.

Thanks,

Drew

Dear Drew 

Do you have any update about my problem?

Hi Patrick,

Apologies for the delay.  Looking over your schematic, it appears that 2.5V-EQ-DR is connected to 1.8V via FB10 and FB12 as well as FB7 and FB8.  I'm a bit confused by this, am I misreading the schematic?

Thanks,

Drew

Dear Drew 

As you can see the value of FB12 and FB14 have DNP at the end it means we didn't assemble these 2 part. So these voltage you mentioned are not connected. I have signal on my equalizer but the length is not right. 

Thanks

Hi Patrick,

Thanks for clarifying, I misinterpreted which components were DNP.  I don't see any issues with your schematic, but would like to check with a colleague on the layout.  I see you have an option to get SDI signal from either a connector or PCB finger connection depending on capacitor placement, but I am concerned about the impact this layout may have on the high speed signal.

Thanks,

Drew

Dear Drew

I cut the routes on my PCB to get just from BNC connector and so there is no return path in the way of my SDI signal. In this situation I got 45 meters with Belden 1694A but as datasheet says this part will give 75 meters. right?

Hi Patrick,
Yes, you should be able to get 75 meters with this device.  Is 45 meters with your previous layout, or with the one you shared?

Thanks,
Drew

Dear Drew

In the best case I can get 45 meters. I want to get more than 60 meters but I don't know what should I do.

Hi Patrick,

I understand, let me discuss with my team.  You should be able to get what is stated in the datasheet.  How are you testing the maximum reach?

Thanks,
Drew

Dear Drew

I give 4K-SDI data from Phabrix tools to the BNC connector on the input track near the equalizer and also I got the SDI signal from BNC connector on output track of the driver and it is close to driver too and give it to the phabrix tools and compare signal's specification with the resource. In this situation the maximum Input cable length is 45m while datasheet says it should be 75m. I need to mention that I have no length on the driver side.

Thanks,

Patrick

Hi Patrick,

Thanks for clarifying.  In general, I don't see any outstanding issue with your implementation.  Have you tried taking any return loss measurements of the input?

What BNc connector are you using?

Thanks,

Drew

I didn't measure return loss yet cause I don't have tools to test it but our BNC is BNC7T-J-P-GN-ST-TH1 and it is 12G.

Hi Patrick,

Thanks for clarifying.  I took another look over your layout, but I still don't see any issue that stands out.  I know SDI can be sensitive to return loss, so this is why I was curious about this.

When you say best case is 45m, what is the failure mode for 60m?  Is there an increase of errors on the Phabrix tool?  Have you checked if the LMH1219 has CDR lock?

Thanks,
Drew

Dear Drew

Yes. We face with CRC, Jitter and rise/fall time error when we give more than 45m to equalizer.  About CDR lock I am not sure. I will check it today and reply you soon. 

Thanks.

Hi Patrick,

I would like to confirm that I understand your observations correctly.  It sounds like you are seeing increased jitter and rise/fall time errors on the LMH1228 output when the cable length to the LMH1219 is increased beyond 45m.  Is this correct?

Thanks,
Drew

Dear Drew

Exactly. This is my problem and I don't know is for my equalizer or driver?

Hi Patrick,

This seems very strange.  The equalizer and drive both have CDR circuits that retime the data and filter out jitter.  We would not expect the input cable length on the equalizer to significantly impact the driver output jitter.  We would expect CRC/bit errors if the equalizer cannot equalize the signal, but not increased jitter/rise/fall time.

How are you measuring jitter and rise/fall time.  My understanding is that this should be measured with a 1m cable from driver to Phabrix equipment using the color bar pattern.  What is your data test pattern?

Thanks,
Drew