# Design of a low cost/low power Isolated AC/DC SMPS

I am looking to replace a classic 50Hz transformer/rectifier circuit with an SMPS. The input is to be 230VAC (EU) compliant and the output should be 5V/0.2A minimum while 5V/0.5A is a nice to have.

The cost of the current circuit is less than 2€ (50Hz transformer (1.05€), rectifier, regulator).

I've started by looking in to the lowest cost SMPS controller to find that the total BOM is quickly exceeding the origincal circuit's cost (100kHz transformer 1.1€, controller 0.2€, rectifier, optocoupler).

So the bigger transformer costs less than the smaller 100kHz transformer.

Should I be looking for an SMPS controller with higher switching frequency (does that exists) to get a smaller (cheaper) transformer or is it currently hopeless to design an SMPS that is cheaper that the classic 50Hz transformer centered design?

What is the best approach. Example: find the transformer first and then the controller?

Where's the state of the art of Isolated AC/DC SMPS at today ?

Edits:

• looking at pricings for volumes of 2k to 10k/annum.
• The current power solution parts cost $2.5 (50Hz 220V/1.5VA transformer, bridge, caps and LDO). • Must I conclude that the best AC/DC SMPS solutions are using around 100kHz, 15x15x15mm transformers, and that smaller (/cheaper) solutions using 1MHz for instance, a smaller transformer do not exist/are not available? My conclusion after consulting several sources: - For a custom design the cost of an isolated SMPS is higher than simply rectifying and regulating the transformer output. Thanks to all contributers for your feedbacks! • "BOM is quickly exceeding the origincal circuit's cost" - If your customers won't pay a little more for a superior device that will save them money in the long run then you have a marketing problem, not an engineering problem. And this is a shopping question. – Bruce Abbott Jun 18 at 6:36 • An AC/DC SMPS is not considered superior as it is providing the same functionnality. In case there are effective needs of space savings, efficiency, etc, then we can start talking about superior design. The marketing problem for this product is that it is "entry level" so cost is important. @BruceAbbott I've been checking around and one feedback is that SMPS is interesting when there is a lot of power consumption, but not interesting for low power consumption. – le_top Jun 18 at 10:27 • "problem for this product is that it is "entry level"" - Like I said, it's a marketing problem. I hope you don't have any competition, because "many OEMs are now starting to demand “greener” power supplies as a way to differentiate their end-products" digikey.com/en/articles/techzone/2015/aug/… – Bruce Abbott Jun 18 at 20:38 • @BruceAbbott The competition for the entry level product has a classic transformer/rectifier solution. I did not check the power balance, I am not sure that SMPS is (much) better here in terms of total power consumption. We're working to win a few$ on the product, not add a few $. Anyway, this is an occasion to check on the state of the art of SMPS . – le_top Jun 18 at 23:24 ## 2 Answers You won't be able to make this low qty/yr cheaper than buying direct. The cost of failure exceeds the savings unless you have done this before. But choosing a transformer 1st tells me otherwise. You always start with detailed specs. You still have to select many sources and perform DVT to their specs and your requirements with accelerated stress tests. This is probably what you omitted last time. • You cannot expect$0.1/W prices in a 2.5W supply.

• You ought to budget 1 month of work doing Design Verification Tests (DVT) with at least 10 units from each supplier.

• This should include a teardown reverse-engineering analysis. If you know how to design, then do this. This is essential for high volume users who rely on high quality from the cheapest sources.

Best quality comes from Japanese parent companies like TDK, OKI, etc.

This looks like the best CUI model distributed by DK. $4.43 @ 540 = qty. They do not make them but are custom made for them under their US brand. • I found a similar product as the CUI from RECOM: RAC01-05SGA . Pricing is a bit higher, but still more than the current solution. If you can, it would be interesting to add$/W estimates that could be targeted. The BOM of the product that is being optimized is currently $12 including the PCB. The parts required to get 5V regulated by LDO cost$2.5 . – le_top Jun 18 at 10:42

Where's the state of the art of Isolated AC/DC SMPS at today ?

For 5V and less than an amp? Inside the nearest name-brand phone charger. You might want to price out wall-warts from CUI. For small production I'd be tempted to build a wall-wart inside my box, but I suspect that UL and CE would both look askance at such practices.

You didn't mention production volumes -- if I were you, once I got slapped for suggesting to use a wall-wart, either inside or outside the box, I'd look for semiconductor companies selling off-line power supply chips, and see if I could entice an apps engineer from one of them in the door to recommend a circuit, with part numbers and everything.

• This is exactly correct. You can get a single chip flyback converter for 5W output with integrated FET from several of the big players like TI or Power Integrations. There are reference designs available on their websites, but offline power is not a trivial task with all the regulatory and safety requirement. Best left to experts if you have the choice. – John D Jun 17 at 22:39
• @JohnD Many big players have single chip converters and some propose online design tools, but smaller 100kHz transformers do not look cheaper than 50Hz transformers so it looks like the isolated SMPS is just increasing the cost by requiring the converter chip and the optoisolator. – le_top Jun 17 at 22:48
• "Best left to experts if you have the choice." That's what was driving my suggestion to just use a wall-wart. – TimWescott Jun 17 at 22:53
• @TimWescott We'ld rather not use the cheap "wall-wart" adapters. I'll spare you the details of the experience we had where we used 5V/1A adapters for a non-critical 5V/300mA load - they could not always provide that current at start-up and the some systems ended up rebooting erratically in loops at occasions. I've examined a demonstration board from one of the big players (3W 5V/600mA) and that is what the pricing is for, I was especially surprised with the transformer cost. I was expecting better solutions than that. – le_top Jun 17 at 22:55
• I suspect that to get transformer cost down you need to buy truckloads from factories in East Asia. I did a quick check (which I should have before); the cheapest assembled wall-warts from DigiKey are a bit more than US\$3.50, which is consistent with a BOM cost less than a Euro -- but, that's from an East Asian supplier, and I'm sure they make truckloads of them. Eventually the line transformers will get more expensive, but apparently not today... – TimWescott Jun 17 at 22:59