Best way to avoid vendor tie-in with DC/DC Converters

Question

I frequently utilize DC/DC step-down converters in many of my projects. Mostly these converters have a low voltage input (Vin_max = 30Vdc) and low output current (Iout_max = 1-2A).

There's a plethora of such controller ICs available from various manufacturers, however, they are not generally inter-changeable.

What's the best design approach to avoid dependency on a single vendor?

• Do you know of any chips that are produced by multiple manufacturers and are pin-to-pin compatible?
• Are there common alternative approaches?

My objective is to circumvent any issues related to component obsolescence or temporary unavailability, aiming to make my designs robust against supply problems.

Answer 1

At the current and voltages levels you are seeking, rather than looking for chip level compatibility, I'd suggest the best option is to use switching modules which are pin and form compatible with the very common 78xx range of linear regulators.

Multiple vendors supply complete modules in this form-factor, from well-respected brand names such as Murata to no-name Chinese modules on eBay, Banggood and Alibaba.

Just reserve the space round the pins in your PCB design, and then solder in a module from whatever source you currently favour with zero change.

(If vertical height is an issue, you can also get models with 90deg headers.)

You can even use this approach as a drop in replacement for existing 78xx systems. Noise is obviously higher than with a linear regulator, but efficiency and current limits are generally much higher, and in many older designs the swap works fine.

Answer 2

The most flexible option that comes to mind is to use a daughterboard for your DC-DC converter. On your main board, just have a connector with pins for for input, output, ground, enable, and so on, and then make various daughterboards that can plug into that connector and supply the needed output using different methods. If one chip goes out of stock, you can just design a new daughterboard with a different IC, rather than needing to redesign your entire PCB.

This is essentially a DIY version of colintd's answer; I suggest doing it yourself because you haven't specified your requirements, and you might need features (like an enable pin, a power good output, or a synchronized switching clock, or just a specific switching frequency, for instance) that the off-the-shelf modules don't typically offer (and those that do probably won't have a standardized footprint, defeating the purpose of using interchangeable modules!).

Answer 3

My vote is on SOT-23-6 or -363 devices. I have seen surprising consistency in pinouts, with the most common being 1 = BST, 2 = GND, 3 = FB, 4 = EN, 5 = VIN, 6 = SW. Examples: MP2456, LMR12007, RT7285, etc.

Possibly the next most common is 1 = GND, 2 = SW, 3 = VIN, 4 = FB, 5 = EN, 6 = BST, e.g. AP62150, TPS56120x, etc.

A couple other pinouts are much less common and can be safely ignored for these purposes.

Obviously, ratings and characteristics vary across all of these, but some are drop-in replacements for each other, and some can be adapted with other component changes. Buck regulators up to 60V 3A or thereabouts are available in this package; impressive ratings.

The next most common packages (say, SOIC-8, SON-8 or 10, MSOP-8, etc.) may show some promise, but I just haven't looked at them personally in much detail, so I don't know how general they are. There may also be a lot of parts that have similar or identical footprints within a family, or between families from the same manufacturer, which may still be of some merit -- say to avoid low stock of one part, if not the ability to shift supplier entirely.

There's also something to be said for old and generic types, like MC34063, as terrible as they are. Dating back to the days of second-source rules (as dictated by... defense customers, I think it was?), many still make the part. It does perform quite poorly (low efficiency; low frequency --> large capacitors and inductor required), but it may well be the case that that is an acceptable compromise for some applications.

Even further, one might consider discrete circuits (whether including generic functions like op-amps, comparators, logic ICs, or wholly transistor-level), but these do come at significant cost in design work, range of ratings, performance, absent features, and large layout area / parts count / assembly cost.

Answer 4

Frame shift: the problem you have to solve is to avoid vendor lock in, or in general you want to avoid depending on a single source for your designs.

For the specific chips you mention, I suspect that pin to pin compatible devices from different manufacturers are next to impossible to find. DCDC ICs are often super specific, with a plethora of features that can vary widely even within a product line of a single manufacturer.

There are a few things that you can do to mitigate this issue, they all come with their set of drawbacks:

• make provisions on your PCB for two or even three different chips. You then assemble only whatever is readily available when you make your batch. This costs PCB area, which might or might not be a problem for you
• stock critical components on your own - or at your favorite assembly house. This means you have to pay money upfront, and you have to trust your assembly house to properly stock the ICs, which is not like putting canned beans in the pantry
• you can spin your own DCDC controller, with opamps and stuff - but this makes sense only if we are talking about very specific use cases, yours seems to be pretty standard instead

Personally I would go with option 1 if PCB area allows - a similar approach is also to spin N different very small PCBs, with the DCDC & passives - each of them with a standard (for you) pinout on castellated contacts or whatever, and then you assemble those on your main PCB.

Option 2 is sensible if we are talking of huge numbers (100/1k boards per batch, a few reels of DCDCs)

Finally, avoid component brokers at all costs if you want your design to be reliable and with high quality.

Answer 5

Thanks for sharing your ideas. However, I made a little progress. I was searching for the chips with some specific parameters (Vin range, Iout, etc.). Then I focused on the packages with minimum pin count (6pin) ideally in the same package type SOT26-3. I considered, that for such packages the manufacturers do have not too many options to make every chip different in pin layout and pin function. And really this looks as a promising way. I did not spend too much time comparing the results, but within a few minutes, I found some first candidates:

AOZ1282CI - Alpha & Omega Semiconductor Inc.

R1245N001B-TR-FE - Nisshinbo Micro Devices

LMR50410 – Texas Instrument

Of course, those are not 100% the same chips, but at least have the same package and pin layout. The first two have even the same Vfb. Hopefully, I did not overlook something 😊 If not, it seems that I could make one universal layout with different chips and passive component values. It would be probably worth spending more time searching and comparing the chips. I believe that there is even the chance to keep the same passives – at least for certain converter settings.

Answer 6

I like @colintd's answer here. But when I'm looking into a part that I'm ignorant about, I do a search on a part supplier's website (like Mouser or DigiKey) to see what they stock the most of. That gives me a good idea of what's stable and popular.