LTC2974 CML_MEMORY_FAULT

below is a result of the "Procedure to Identify LTC297x with CRC Error"

Yes, follow this procedure. You may have an addressing problem.  The address in the proj file (as seen in the system tree) ought to match the MFR_BASE_ADDRESS config setting and your pin strapping as set by the ASEL pins on your board.  When programming the device on your board, ensure that VDD33 is powered up to 3.3V.

Also, it is best practice to use the Programming Utility in LTpowerplay. This utility writes the config settings directly to NVM. You will need to power cycle the board afterwards to get the config into the working RAM.

Mike

ID=0x212 tells us it's the version that we first introduced years ago. Not that this is necessarily bad. 

Are you powering the chip via VPWR? 12V? Did you check the VDD33 voltage?  Also check the VDD25 voltage.  Can you post your schematic?

Mike

Mike

I am using Linear test adapter DC1508B-B and DC1613A Interface Development Tools USB-to-PMBus Controller. Checking chips that are not soldered to the PCB yet.

It powered via VPWR +5V,

Measured: VDD33=3.28V, VDD25=2.51V

Initially I was checked it in the real PCB and behavior was the same.

Below is a sch

Regards,

Alex

Hi Alex,

The DC1508 socketed board does indeed use +5V for VPWR and lets the 3.3V and 2.5V LDOs regulate on their own. Is this the board that gives you CML errors?  All devices do the same?  And not recoverable?  Can you send the proj file you use for programming?

Is the schematic you posted the one for your system board?  It's not complete, hopefully you have followed our schematic checklist for tying pins to pullups, to GND if unused, or leave floating for some.  Page 101 of the 2974 datasheet.

Mike

Hi Mike,

I am getting the CML errors on both: DC1508 and system board we made.

Have several chips which have that CML error, other chips are ok and successfully work after soldering it to the system board. Question is how to cope with those chips which have the CML error?

After the off/on cycle all settings are lost.

The .proj file attached

Hi Alexey,

I'm trying to match up your proj file with your board hardware but see the portion of the schematic with VSENSE, ISENSE, or VOUT_EN pins. The proj file shows +12V and -12V rails and in the Setup tab all the scaling and offsets, I just want to be sure the chip is sensing voltages within its limits.  It is also very important that VDD33 voltage is not too high nor too low.

You should be able to reprogram devices in the test socket and get these to recover (no CML errors). 

Mike

Mike

there are dividers for +12v and the level shifter for -12v rail, the hardware design is proven and successfully works at the customer. There are several chips 2974 which have the internal issue.

Let's consider how to recover these chips with the test board DC1508

I tried to follow the "Procedure to Identify LTC297x with CRC Error" with the script file and tried by the Programming utility of the LTpowerPlay, all logs are above

What else I could try?

Attached is a video of the sequence I made.

Hi

Any ideas how to recover these chips?

Alex

Alexey,

From the video I see that once power is applied, the CML memory fault is present. And the part does not recover after both programming methods.  It is possible that this memory error is permanent.  How many parts do you have in total?  How many have a memory fault?

Mike 

Mike,

We have 250pce of those chips with the memory fault. Got them from HK.trade disti China.

Alex

Dear Mike,

I'm working at distributor. My customer also faces this problem. They have 20 pcs failed bought from China trader.

Please advise which conditions could make LTC2974 memory fail?

Can we confirm the genuine parts getting from ADI factories won't have this issue?

Thanks & best regards,

Dear Mike,

I'm working at distributor. My customer also faces this problem. They have 20 pcs failed bought from China trader.

Please advise which conditions could make LTC2974 memory fail?

Can we confirm the genuine parts getting from ADI factories won't have this issue?

Thanks & best regards,

Hi Tuong, sometimes it's difficult to determine the origin of the parts. You can read the MFR_SPECIAL_ID and it will look like a legit number, however comparing that ID to the date code is a better indication.  Even then, it may not tell the whole story.

Mike