muRata BPM BEI UHE or UEI isolated DC/DC (single and dual) series

zgtc

Member
2016-11-03 1:50 pm
Barcelona
Hi,

I'm thinking of building a Dual+Aux PSU with DC in, and isolated DC/DC (plus regulation) units. Power consumption not too big, around ±[email protected] for opamp based preamp boards and [email protected] for Rpi+DAC. The idea of isolation is to be able to use a single DC in and avoid sharing "different" grounds between the boards.

Has anyone tried the muRata BPM BEI UHE or UEI series? I've used the muRata site search/filters and came up with the following potential matches.

For ±[email protected]
  1. BPM15-120-Q48P-C
  2. BEI15-120-Q12P-C

For [email protected]
  1. SPM15-050-Q12P-C
  2. UEI15-050-Q12PM-C
  3. UEI30-050-Q12P-C
  4. UHE-5/5000-Q48-C

Also, just in case, anyone has implemented/designed a similar design?

Thank you
 

OlegSh

Member
Paid Member
2010-12-01 11:00 am
Germany
I have no experience with murata converters but I've implemented similar solution using TDK lambda series DC-DC converters and TPS7A4901/3001 low noise LDO regs. Works perfect! I guess murata converters should be good too. If I remember right XRK971 (diyaudio member) made similar PSU too.

Regards,
Oleg
 

zgtc

Member
2016-11-03 1:50 pm
Barcelona
Cool !

Is there any specific characteristic to look for in the datasheets? I will obviously regulate after the DC-DC, use before/after caps as defined in the DS, etc. i.e. I'm looking for ripple around or below 100mVp-p, is that good/enough/crap? Anything else to look for?

Will keep researching, thank you!!!
Thank you again
 

OlegSh

Member
Paid Member
2010-12-01 11:00 am
Germany
The ripple amplitude given in the datasheet alone is not enough to understand how good or bad the converter will perform. The main switching frequency ripple is usually easy to handle but you have to figure out its frequency first and see if the regs which you plan using have enough PSRR at this frequency. But if there are higher frequency noise spikes which can be in tens to hundreds of MHz range, they tend to propagate to the output unaltered due to almost zero PSRR of most regs past 1~10 MHz range. So you'll have to use passive filters on each rail to filter them out. You'll also have to pay attention to the common mode noise and use common mode noise filters to get rid of it.
 

zgtc

Member
2016-11-03 1:50 pm
Barcelona
The ripple amplitude given in the datasheet alone is not enough to understand how good or bad the converter will perform. The main switching frequency ripple is usually easy to handle but you have to figure out its frequency first and see if the regs which you plan using have enough PSRR at this frequency. But if there are higher frequency noise spikes which can be in tens to hundreds of MHz range, they tend to propagate to the output unaltered due to almost zero PSRR of most regs past 1~10 MHz range. So you'll have to use passive filters on each rail to filter them out. You'll also have to pay attention to the common mode noise and use common mode noise filters to get rid of it.
wow, I need to digest all the info though.

I'd suggest higher current regulators, e.g. LT3045 and LT3094 pair.
Perfect

Thank you so much again
 

zgtc

Member
2016-11-03 1:50 pm
Barcelona
Hi Sergi,

Have you advanced with your PSU design?

Regards,
Oleg

Hey Oleg!

Thank you for the interest (and all the help these days). I've had busy days, and will take the forthcoming Xmas to finish the design, but I needed to digest and understand all the info you gave me (and all the books I've read, linked in the other thread, see post 82 there).

I even wrote Steve Roberts from Recom, he's been sooooo nice answering my doubts as well. Everyone is so nice, guys, thank you! Basically I was freaking out in your thread about the caps possibly breaking the isolation. Answer: they don't.

No. What is important is that the dielectric strength of the capacitor can withstand permanent mains voltage across it – so the capacitor must be class Y. For some standards (household, medical), two Y-capacitors in series must be used across an isolation boundary. This is in case one capacitor fails short circuit, then the second still provides isolation. For other applications (industry, IT) only one Y-capacitor is required (the failure rate is close to zero).

Also asked him about 3 winding chokes:

Würth Elektronik WE-SL3 series (e.g. DigiKey 732-14096-2-ND), TDK ACM2520 (e.g. DigiKey 445-4842-2-ND) or Murata NGF series (e.g. Digikey 490-18604-2-ND) are all freely available. The need for an output filter is dependent on the application – you may get away just with an input filter and output capacitors.

In summary:

I will use input CM choke, and class Y caps shorting input/output. DC-DC type will be 1"x1" isolated converter/s for sure. Most likely the Recom RPA series but depending on the following design/cost decision:

Still undecided about the output regulation. I have 2 options, either:

- Using only one DC-DC (bipolar ±3.3V) and generate all the voltages from there. I should look for a more power converter though. One regulator per voltage, and MOSFET switches to power on/off optional circuits

- Using one board per expected final voltage. Use of the DC-DC included control pin (via opto-isolator) to power on/off optional boards, then onboard post regulation.

I have to count dollars, board space, etc.

Decisions! :D
 

OlegSh

Member
Paid Member
2010-12-01 11:00 am
Germany
Sounds like a big undertaking! I would suggest a dedicated separate PSU for every load. This will allow fully isolating all parts and avoiding cross talk through the shared power lines. Then you can use ON/OFF pin of the DC-DC converter to control the power.

In any case it won't be cheap :)