2
\$\begingroup\$

I am building an appliance that should work in very wide range input voltage: In some scenarios input is 5VAC, in other scenarios it can be 250VAC.

It's not a problem to convert AC to DC, then convert it using converters to required 3.3V for the appliance. The appliance consumes current from 10mA to 150mA on 3.3V.

Requirements:

  • Wide input voltages range.
  • Small size as possible.
  • Less components as possible (footprint, size, price).
  • Less heat dissipation (efficiency in other words).
  • Low current, very low voltage, low output power.

Till now I used 2 separated power supplies for low-voltage and high-voltage scenarios (switching between them mechanically):

schematic

simulate this circuit – Schematic created using CircuitLab

There is not a risk of HV input on LV circuit: the input controlled by dual way relay. This approach has following disadvantages (for my point of view):

  • Mechanical switching is not so nice.
  • Waste of space in tiny appliance.
  • Maintaining BOM with similar by function parts.

I looked for an alternative solution to merge 2 PSUs in one. Ready PSUs are huge in size: more than inch (25mm) in one of dimensions.

In this thread it was suggested so far "the best" solution based on wide range regulator LR8. Minimal input of 13.2V is not a problem, I can ignore lower input voltages...

Disadvantages for this approach are:

  • No replacement parts for LR8 in case of short stock.
  • Supply of 1 regulator is not enough. Connecting LDOs in parallel is tricky and not trivial for me. The avoiding overload one of them and lazy load on other is a big deal.

In this thread the idea of LED driver was suggested. In common, this is good idea except following:

  • The most of LED drivers are working in range 80V and up. (At least, I didn't find lower input voltage LED driver).
  • LED driver is build to maintain constant (limited) current. In my appliance, when current not required (10mA consumption) the driver will raise the voltage to reach the preset current. This will burn the appliance itself.

Additional idea to get wide input voltage PSU can be implemented (in science fiction :) by connecting sequentially 2 regulators LTC3639 of Linear. Each regulator can deal with 5-150V range. So 2 in a sequence can deal 10-300V which maybe enough. But I have no clue how to connect them... I think this is not possible. True?

So, I ran out of ideas to solve the problem of single wide input voltage range buck converter/circuit to get 3.3V output at 10-150mA. Please, help.

What can be better: develop new research direction or improve LR8 LDO regulator in parallel?

\$\endgroup\$
1
\$\begingroup\$

Is it possible to break your load into two parts: an accurate low-current supply that is used for the microcontroller and any analog stuff; and a less accurate higher-current, higher voltage supply that powers things like relays and such?

Many of my products are built this way - my 5V rail needs to supply only 5 to 10 mA and the unregulated supply at whatever voltage I'm working at powers the multitude of relays that many of my products use. I do PWM on the relay coils as necessary to keep the average voltage at the rated value as the input voltage changes.

I have a simple but effective AC Phase Control power supply regulator that works well from about 16Vac in through 230Vac in. But it's good for only AC input (not DC) and it can supply only about 75mA. That doesn't meet your requirements.

\$\endgroup\$
  • \$\begingroup\$ Dwayne, yes, it is an option. I thought about such split of voltages. However, I didn't imagine how to implement this. My limitation was the high consuming part (100mA) expects more-or-less predictable voltage too, not scaling in wide range. I'd be glad to see your schematics, hoping it will be OK. \$\endgroup\$ – Ivy Growing Feb 7 '15 at 13:36
1
\$\begingroup\$

You need a very wide input volt range .I have had to do this a few times now so I will outline what I did to nail it .Remember that ready made stuff is unobtainium at these specs so its time to do some real design.Now imagine a buck converter that is made from discretes ,this would work albeit at a rather low duty cycle when the input DC volts are high .Of course it would work normally when the input DC volts are low like say 10V .At the high volts the efficiency would be poor because the switching losses would be much higher.This is because switching losses are based on peak voltages and peak currents .I used a different switching regime that gave greatly reduced switching losses and no diode recovery stress.If I had used a IC I would have been stuck with fixed frequency peak current mode which would have burnt up for sure.On my scheme the converter frequency changed with output loading and input voltage .Before lashing this I managed to show that the expected frequency changes were less than proportional and hence not too bad .At normal input output ratios the efficiency of my birds nest using junkbox parts was 90.1 % measured with basic equipment .I kept the frequency at a normal 60KHz so my test was comparing apples with apples .This figure is nothing to write home about and a orthodox PWM design with optimised parts should beat this .What I did find was that as I kept raising the input volts right up to the ratings of my junkbox fet the efficiency stayed constant .So if you want wide range its more important to have some switching loss reduction scheme.

\$\endgroup\$
  • \$\begingroup\$ Makes sense. Any reference design or example of such switching loss reduction scheme? Thank you. \$\endgroup\$ – Ivy Growing Apr 8 '18 at 7:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.