LMH6518: DC maximum input range, datasheet error (URGENT PLEASE READ!)

Hi Adrian, 

To address your questions, the DC differential input voltage should be limited to 0.4V. The original specification of 1V maximum refers only to transient signals, which will not be present for continuous duration. A long-time DC input exceeding 0.4V can cause a systemic increase in the offset voltage. 

The datasheet is still scheduled to receive an update to the specification. I will check on the status of this activity and try and get the changes prioritized. 

Please let me know if you have any further questions. 

Best regards, 

Jacob 

Hi Jacob, we have been manufacturing products using 1000s of these for a number of years. It is deeply disappointing that you have not released a PCN to advise customers on this issue once it was discover over 2 years ago! 

1) Can you give me some info on the root cause of the issue, i.e. is it the 0-20dB attenuator that gets damaged? 

2) Can we continue to use parts that have be subject to over voltage if we accept the offset error? or will they be out of spec in other regards or perhaps less reliable?

3) we have noticed that if you put a fixed over range of 600mV the offset drift in that direction over a number of days, then if you reverse that with -600mV, it appears to go back in the other direction. So there look to be an element of self correction? 

Hello Adrian,

We appreciate your questions and acknowledge your push for results.  I am already working on this issue from the various angles discussed; I intend to discuss with design about the internal behavior of the LMH6518 which contributes to the behavior we are investigating.

I will address your questions about root cause, continued part use, and potential 'self-correction' as I resolve those pursuits.

Additionally I am keeping a eye on progress for the datasheet update.

Please provide any updates as you encounter them, or additional questions as you may think of them.

I look forward to providing you, at minimum, a partial update by end of this week.  I have received the related email from our field and I am also aligning with them on this topic.

Best,

Alec

Hello Adrian,

I am actively working on the datasheet edits.  There were additional edits and some formatting/migration settings to resolve.  The process is generally a bit extended past what you may expect; would it be alright to provide you a draft copy of edits once complete?  The actual datasheet revision takes time to go through TI processes.

Regarding your questions, I am looking forward to meeting with one of our designers this week.

Best,

Alec

Hi Alec, 

Yes I would be happy with a draft for now thanks. My technical sales engineer is Jorgen Larsson whom I believe has been in touch since he got back from vacation. He can forward the draft datasheet or you can email direct if he will pass on my details.

Thanks

Hello Adrian,

I have been communicating and working with Jorgen; I appreciate your understanding on typical timeline.  

I will update you on any additional developments as I discover them, either directly here or via Jorgen.

Regarding your set of three questions earlier, I do think I can share a bit of information now:

1. Our designers have spent time looking at internal structures; there is currently not a verified root cause, though we have theories.  I cannot elaborate further at this time.

2. It does appear the unexpected behavior is limited to the observed voltage offset.  In this short term while we are investigating, it is up to your expertise and discretion to advise continued use/operation.  As I will address in the next answer, there are 'recovery' options which can 'alleviate' offset.

3. Much like the answer to your first question, the apparent 'recovery' when the differential voltage is swapped in polarity has been examined.  This 'self correction' as you mentioned is not yet fully understood to be dependable or repeatable, but it does appear to dictate the polarity of the differential voltage has effect on how the voltage offset is generated.

If you have a specific figure of merit which is blocking your ability to support customers, please ask me.  I would like to do what I can to help in this interim time while there is not yet a datasheet revision.

Best,

Alec

Alec, Jorgen has hopefully communicated that my tests at ~0.4V have show significant offset drift over 7 days on all 32 units I have tested. The test was done at Max gain, however its has affect all ranges. The error occur at the output of the chip regardless on the gain setting though it does change when you change the input amp from 10dB to 30dB, it doesnt change significantly when changing the step laddar settings. 

This a fundamental problem with the part. Have you considered electromigration for this part? Its a common problem that happens when you subject RF attenuators to DC for long periods, hence many are AC coupled. It causes conductivity changes in conductors.

I can confirm that left or 7 days without a DC voltage applied, the LMH6518 is very stable <2mV of output offset drift seen at a fixed ambient temperature. 

Hello Adrian,

We did receive your work & description of behaviors from Jorgen.  I do appreciate you following-up on the thread; I did not want the presence of two communication channels to be confusing, but I think the discussion here is not repetitive.

Your point of electromigration is valued; thank you.  The HSAMPs team is working on understanding of how to effectively replicate the behavior in our lab & discern root cause.

I am away from the lab on work travel this upcoming week; I will make sure to inform my colleagues to watch this thread (as well as for any emails from Jorgen).

Thank you again Adrian.

Best,

Alec