Post Go back to editing

Explanation needed for power bypassing and power quality when driving a ADuCM355 from an ADM3260

I am building an application that powers the ADuCM355 via isolated power from a ADM3260. I'm looking for a power supply topology and circuit that offers the most reliable and accurate operation for the ADuCM355.

The ADM3260 datasheet is clear that 0.1u and 10u bypass caps are needed "close" to the VISO pin of the 3260.  Also, the ADuCM355 data sheet recommends a 100nf cap at each power pin and a 4.7uf cap at the source.  However after those recommendations, I'm seeing lots of conflicting examples.  In the data sheet for the ADuCM355 on p. 27 there is a diagram of attached passive components and it shows power going through an "IS RES" (what's that?) at the source and then having a collection of beads and doubled 100nf caps and 1R6 resistors in a topology that violates the recommendation on the previous page of using single 100nf caps, plus makes no sense since some DVVD paths have the resistor and some don't without explanation.  The CN0428 schematic shows a different power topology for the ADuCM355 than the data sheet, is that because CN0428 requires less analog accuracy?   All in all, quite confusing.

So, aside from explanations of the confusing and conflicting power bypassing/cleaning topologies I mentioned above, I would like to know if any special filtering is necessary between the ADM3260 and the ADuCM355.   The ripple and noise specs of the ADM3260 seem high for the input to an AFE (analog front end), is this why there are all the additional resistors and capacitors I see on example designs (including CN0326 which is a similar circuit and has a 10uF/100nf cap pair at each power input?)    

Parents
  • Hi E, 

    The integrated isolated dc/dc converter in the ADM3260 does require 100nF and 10uF bypass caps. That is independent of any need the ADuCM355 may have for external passives. The ADM3260's bypass caps should be placed directly next to VISO and the adjacent GNDISO pin. Best would be to have no vias in the route between bypass cap and device pin adding inductance.   

    The LTM2887-I would have less noise on the output supply. That may be another option for this application. 

    Regards,

    Jason

Reply
  • Hi E, 

    The integrated isolated dc/dc converter in the ADM3260 does require 100nF and 10uF bypass caps. That is independent of any need the ADuCM355 may have for external passives. The ADM3260's bypass caps should be placed directly next to VISO and the adjacent GNDISO pin. Best would be to have no vias in the route between bypass cap and device pin adding inductance.   

    The LTM2887-I would have less noise on the output supply. That may be another option for this application. 

    Regards,

    Jason

Children
  • Wow, that LTM2887 is expensive, and I'm doing a board with three isolated power domains!   What do you think of raising the output voltage of the 3260 by one dropout spec and using a nice little LDO after it, sort of a mini-mimickry of what's done with bigger systems to improve efficiency?  I guess the other possibility is some more post-filtration with some larger capacitors in the mix.

  • Hi E,

    That is an interesting idea you have with the LDO. I haven't tried it, but it seems reasonable. Presumably the intended supply voltage to the ADuCM355 is 3.3V. You can adjust the resistor values on the VSEL pin, set the dc/dc converter output to something high enough for an input supply to an LDO. 

    Efficiency is another story... the integrated dc/dc converter in the ADM3260 is about 30% efficient. Adding an LDO is obviously going to drop the isolated supply efficiency further.    

    Have to be careful with large caps on the outputs of isoPower devices. 

    Regards,

    Jason

  • Ugh... efficiency.  I did an admittedly incomplete search of the industry to find a better combination of efficiency, compactness, and output power quality and didn't see anything.  AD has this pretty much sewn up.  Fortunately as you pointed out, I can set the output voltage to be one dropout above 3.3v, and the regulator I'm eying has 150mv dropout, so we aren't talking about worsening the already bad efficiency much.  Before I finalize the design, I'll do one more scan on efficiency.  I need 30ma per fluid probe circuit, which totals 360ma just for the probes in my circuit, kind of an embarrassment as you point out.

    The LDO I'm using just requires 1uf on its input, so it won't add much to the load impedance on the isoPower device.  I am adding a ferrite bead to help block some of the noise from the 3260 before it hits the LDO (and adjusting the output voltage to compensate.)