ADM2587EBRWZ fall over after short time

Hi,

I am using ADM2587EBRWZ.  Circuit diagrams attached.

I am using x2 of them as normal RS485 comms.

I am using x8 of them in receive mode only for translating between differential pair RS485 lines to x8 single control lines going to a MCU.

Altogether then x10 of them.

After building prototype and putting power on the board, it appears that the regulators supplying 5V to the devices run ok for a while and then suddenly some of them and eventually all of them starts to get hot, draws more and more current and then stop supplying correct voltage etc.

I then removed all of the regulators and powered only one of the devices from a bench power supply.  So it appears the problem is not the supply.

Not sure if the device is perhaps getting unstable for some reason and if so why...or what else could be the problem.

I asked Analog Devices back when doing the design for the circuit to check my circuits and they said it was ok...

Can someone please double check me or perhaps what else could the problem be ??

PDFPDF

Parents
  • +1
    •  Analog Employees 
    on Feb 20, 2020 2:08 PM

    Is it possible that LM2940 is going into thermal shutdown? Do you have a heatsink on these regulators?

    The voltage drop is 8.8V and an approximate value for the load current could be 0.36A (note the datasheet maximum Icc for one ADM2587E is 0.12A with 120Ohm load, but likely 50% more with 54 Ohm load = 0.18A, double for each regulator serving two ADM2587E).

    This would suggest power dissipation of >3W. Max junction temp for the regulator is 125C, and 85C is max ambient (per ADM2587E), so max temp rise of 45C. So the ThetaJa would need to be 15C/W which is below the actual ThetaJa of the regulator package options. So a heat sink is required.

    I would suggest using a more efficient voltage conversion (e.g. buck regulator from the ADI Power by Linear portfolio LTxxxx), unless simply adding heat sinks to the existing regulators is simpler for the design.

    Regards,

    Conal Watterson

Reply
  • +1
    •  Analog Employees 
    on Feb 20, 2020 2:08 PM

    Is it possible that LM2940 is going into thermal shutdown? Do you have a heatsink on these regulators?

    The voltage drop is 8.8V and an approximate value for the load current could be 0.36A (note the datasheet maximum Icc for one ADM2587E is 0.12A with 120Ohm load, but likely 50% more with 54 Ohm load = 0.18A, double for each regulator serving two ADM2587E).

    This would suggest power dissipation of >3W. Max junction temp for the regulator is 125C, and 85C is max ambient (per ADM2587E), so max temp rise of 45C. So the ThetaJa would need to be 15C/W which is below the actual ThetaJa of the regulator package options. So a heat sink is required.

    I would suggest using a more efficient voltage conversion (e.g. buck regulator from the ADI Power by Linear portfolio LTxxxx), unless simply adding heat sinks to the existing regulators is simpler for the design.

    Regards,

    Conal Watterson

Children
  • Hi, 

    Thanks for the reply.

    At room temperaturem it should be fine but yes I am aware of that problem for higher ambients...any suggestions for better regulator for this device ?

    I found the actual problem...polarized tantalum capacitor the wrong way arrond...oopa...silly silly mistake...

  • 0
    •  Analog Employees 
    on Feb 27, 2020 12:16 PM in reply to NeilElectronics

    Hi Neil,

    It looks like LT8606 would be suitable to convert 13V to 5V for two ADM2587E, using five of these per how you are using the existing regulator. Alternatively a multi-output option like LT8606 could work. Or one much larger power buck like LT8609 for all ten ADM2587E.

    If you can isolate all ports together as a bank rather than individual channel to channel isolation, another option might be to use LTM8067 isolated DC-DC for isolated 13V to 5V and power the Vdd2 of ADM2483 x10 with this, powering all ten Vdd1 with just one LT8606 or similar.

    Regards,

    Conal