Cascaded isolated power supply - rules of thumb for total power selection

Thread Starter

MaximinBlanc

Joined Mar 2, 2020
5
Hello everyone,
As a hardware designer in power electronics, sometimes I must deal with multiples supply to create 3.3V and 5V for electronics but also isolated +15V/0V or +15V/-9V to drive MOSFETs in HV or LV... Basically this supply can be LDO or isolated chip (like in SIP or DIP package). All of these can take à 5V input and generate the proper supply value in output.

I can be interested in creating a global 5V supply (from the main 230V line) and them put as power supply as I need in cascade with this main 5V supply. It can be cost effective to have one main supply with lots of functionality (or safety/iso requirement) and then several cheaper supply instead of tens of expensive supply which all takes the 230V as input.

I attached an example of what I mean.

I can calculate the power for each DCDC with the load I have.

My concern is about the choice of the total power of the MainACDC converter.
Most of time the consumption of electronics is very low but the standard for isolated DCDC supply is 1W minimum (some 0.5W or less existes but it quite rare and without lost of brand or package). So let say my 3 DCDC supply are 1W rated but the consumption of each never exceed 0.3W in worst case
Can I choose a MainACDC of 1W rated (1W > 3x0.3W) or I have to select a 3W MainACDC because of 3x1W rated DCDC supply ?

I believe the answer is not trivial because of load regulation/stability... If the main supply is not able to give enough charge maybe the total system never start up ?

I would be very interrested by your point of view on this and the rules of thumb if it exists.

Regards,
M. BLANC
 

Attachments

dendad

Joined Feb 20, 2016
3,906
That will not be efficient. Going to 5V first, then back up again.
I'd use, maybe a 24V (or an old laptop 19V) supply and then use separate regs for each supply.
The 24V (or 19V) supply wattage will need to be a bit greater than the combined wattage of all the others.
For example.
5V @ 1A = 5W
3.3V @ 1A = 3.3W
9V @ 2A = 18W
Total, 36.3W
Use at least a 40W main supply, if all the regs are switch mode.
So, 19V @ 2.5A will be ok.
Old laptop 19V supplies can be got for $0 often, and are ideal for this.

BUT!!! are you aiming to use linear regs?
If so, add all the currents up and that is what the main supply will need to be rated at.

1 + 1 + 2 = 4A
So, 19V @ 4A minimum. (76W).
You see, switch mode regs are way more efficient, but can be noisier.
Work out what you need.
 

Thread Starter

MaximinBlanc

Joined Mar 2, 2020
5
The 5Vdd is a good choice for main electronics in my design because most of the chosen electronics works in 5V (electronics team has made some choice I cannot change). The 1W supply is the good order of magnitude (few MOSFET running 50kHz + few ACPL and AOP...) so the supply should be like SIP7 RECOM series of equivalent.

I am bulding a new industrial product so I wont use an old laptop supply (thanks for the idea anyway).

If I have well understood, the main supply must be rated for the sum of all cascaded supply rate even if they are not fully charged. But I'd like to know why (start up problem, regulation...) maybe some issues can be handle by other solution (soft start up, start up sequence : electronics first and then MOSFET...). My concern is about reducing size. In the final product I will probably build my own supply (with 4-5 windings in a same core transformer creating 3.3V, 5V, 15V...) but yet I prefer limit the issues and use commercial isolated DCDC (or LDO) supply.

About size reducing, I don't want to use 3W supply if I use only 1W or 2W in normal operation.
 

dendad

Joined Feb 20, 2016
3,906
I've designed industrial control equipment for the past 30 years.
One of the things I've found, Over Engineer it!
So, if you need a 1W supply, use a 2W one!
Quite a lot of the equipment we installed is still working 20+ years later.

(The main problem we have with our old gear is electro caps drying out after years on continuous operation. Replace them, and it usually works again.)
Industrial control needs to be very reliable as lots of money can be lost with a failure.
Oh, most industrial control stuff runs on 24VDC.
 

Thread Starter

MaximinBlanc

Joined Mar 2, 2020
5
Sorry if I persevere in my question but the reason why cascaded supply must fullfill some requirement (like main converter rated power should more than XXX...) is still unclear.

When I design a 100W power supply, I garanti it will work up to 100W whitout regarding the load (in some limit of course, the control is effective up to a certain step load). I dont know above 100W but I don't say to a customer "well it's 100W but if you could limit to 50W it woul be appreciated". The guy will say "OK so your converter is 50W, I go somewhere else".
For sure in constraint environnement (medical, aerospacial, automotive...) it is not cost/volume/weight (whatever) effective to always oversize. Maybe the lifetime is reduce from 30years to 10years, I really agree with you and I would rather always build extra-longlife system but this is also the reality and longlife system is not always a claim from customer.

So if someone would like to complete the topic with more specific reasons than "over engineering" ; with all du respect to dendad which I thanks for his experience share, or maybe I didn't catch the entire substance of what he meant. I will appreciate a lot

Thanks

Maximin
 
Top