• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Answers 2 answers
  • Subscribers 63 subscribers
  • Views 250 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LP87564-Q1: Combining outputs using droop (current sense) resistors

Julian Schroeder
Prodigy 40 points
Part Number: LP87564-Q1
Other Parts Discussed in Thread: LP87562-Q1

Since the LP87562 is not available yet, I'm combining three outputs of the LP87564 using 10mOhm resistors (after the feedback point). The current balancing works well and the output noise is very low. Are there any problems with this approach that I should be aware of?

  • 0
    TI__Intellectual 950 points
    Hi Julian,

    Thank you for looking TI DCDC PMIC. We can't guarantee the device functions by combining outputs together when device is configured to different phase setup. I assume that you are using generic orderable part number LP875640RNFxQ1? If yes, you are able to change several settings via I2C, but phase configuration is not one of them. If you need LP87562-Q1 (3+1) I recommend you to contact your local TI sales office to inquiry if there is plan to release catalog version of 3+1 phase device or if the business case is big enough we will made one for you.

    thanks,

    Tuomo
  • 0 in reply to Tuomo Alatalo
    Prodigy 40 points

    Hi Tuomo,

    the four outputs of the LP875640RNF are firing at 90 degrees of each other and are thus perfectly aligned for multiphase operation. I confirmed this on a four channel oscilloscope.

    As stated below, using sufficient resistance to combine outputs (about 10mOhm), after the feedback point, achieves very good current balancing and ripple.

    The question I have is: Is there any potential problems with this setup?

    Thanks,

    Julian.

  • 0 in reply to Julian Schroeder
    TI__Genius 11020 points

    Hi Julian,

    Since the device is configured as 4-phase and automatic phase adding/shedding is used, it is possible that converter tries to use the 4th when load increases high enough even though this 4th phase is not connected. It can happen in transient situation with smaller currents as well. This can cause false faults and unexpected behavior when there is no current going through the 4th phase.

    Anyway the device is not designed to work in this kind of configuration and no validation/verification has been done when using it like you have described. So we cannot make sure it works in all cases and we cannot recommend using in this configuration.

    Is it not possible to connect the 4th phase as well in the application? Btw there is no need for having resistor between the phases, they can be connected together directly in multiphase configuration. Device uses FB_B0 as positive feedback and FB_B1 as GND feedback in 4 phase configuration. Current balancing is done measuring the current flow through internal high side FET.

    We have spins with different phase configurations available for LP875x family, so I would propose to contact local sales team and check what we have available for your application. Custom OTP spin is also possibility if the business case is strong enough.

    Thanks.

    Regards,

    Tomi Koskela

  • 0 in reply to Tomi Koskela
    Prodigy 40 points

    Hi Tomi,

    The LP87564 has four independent output channels. Thus it does not offer current balancing to use outputs in parallel. I verified this in an experiment, one output tends to take the whole load. In order to force current balancing, via the feedback voltage, I added 10mOhm resistors after the feedback points of each individual channel. The three channels, that I intended to use in parallel, were programmed to the same voltage and current limit. I did not get any errors/faults from the LP87564 and the current balancing worked fine. I varied to load from 50%-100% in a step fashion to model our real load. I understand that I could use a LP87562, the problem is that this part is not in stock at distributors or at TI. Hence I'm using this workaround. From the literature available in general, and from TI specifically, it would seem that the way I'm paralleling DCDC rails is OK, since the LP87564 outputs are switching 90 degrees out of phase and the LP87564 is using a current mode inner control loop. The LP87564 does not shed or add phases, since it is intended to be used for four independent outputs. I understand you have not tested my configuration, the questions is: What issues could I run into?

    Thanks,

    Julian.

      

  • 0 in reply to Julian Schroeder
    TI__Genius 11020 points
    Hi Julian,

    Yes you are correct, I was thinking about the 4ph single output LP87561, not the LP87564 you mentioned. Sorry about confusion.

    Adding balancing resistors in this case would be required, as you have used. This makes sure the slight differences between output voltages would be balanced out and not all current is sourced by the buck with highest output voltage. Balancing resistors of course reduce the efficiency somewhat and the load regulation won't be that great. 10mohm with 4A load causes 40mV drop. Only concern I have is that the bucks not only source, but also sink current. So if there is a lot of difference between the target voltage the bucks will 'fight' each other and there will be a lot of power burned on those balancing resistors and the total efficiency can be poor.

    Thanks.

    Regards,
    Tomi Koskela
  • 0 in reply to Tomi Koskela
    Prodigy 40 points

    Hi Tomi,

    Given that the four rails of the LP87564 are using the same reference (pg.17 block diagram), the difference between them should be very small.

    I'll take a closer look at the overall efficiency.

    Thanks,

    Julian.