0
\$\begingroup\$

I have a pre-made motor controller which I want to re-purpose for a hobby project by pulling the MCU off and attaching to a custom control circuit (with a more powerful MCU, current sensors, etc). The MCU is a LQFP-48 7mm package (0.5mm pitch pins). It seems like attaching to standard SMT footprints is a reasonably common prototyping task, but I can't find any products designed to do this. What is the name of some kind of adapter/interposer/something which is designed to solder onto the PCB in place of the MCU, and how would I go about finding one of the right size?

I've found a similar product called 14 pin DIP to 14 pin SOIC/SOJ, but that's very generic and hard to search for more of. It's particularly hard to search for because there are a lot of products for going the other way (attaching a SMT IC to larger wires). Some pictures of similar products I found for reference:

14 pin DIP to 14 pin SOIC/SOJ

Just a name and/or source for those little metal clips on the edges would be helpful too, so I can attach them to my own PCB.

something for a 4-sided package

Finding some kind of SMT 0.5mm-pitch header strip would work too (solder those to the existing PCB, and then to a PCB of my own on top), but I can't find any that small.

If anybody has other ways to achieve the same goal, I'm interested in suggestions there too. Some approaches I've already considered: Soldering wires directly to the pads is hard because they're tiny. Finding other places on the PCB to solder wires is kind of annoying because it's a lot of wires to attach, and some of the nets don't go very many other places. Getting a custom PCB with 0.5mm castellated edges made for cheap doesn't seem to be possible.

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

Rather than electrically change it, I'd give serious thought to reprogramming the existing MCU with a custom firmware that makes it a delegate for more sophisticated off-board logic, using whatever fast interface you can bring out. And check to make sure there isn't a more powerful part in a compatible package.

I'd only try to solder to the board for debug purposes using fine enameled magnet wire under the microscope, but would prefer to avoid doing that for actual usage. If you do try to make such a lashup, a key could be having some sort of termination structure right next to the pads, do that right and you could potentially even use single uninsulated strands pulled from a larger stranded wire - but again, only for debug.

Or rather than making contact at the MCU, go for the FET gates themselves or their gate drive circuitry, which is hopefully less dense. Essentially, treat the existing board as an array of FETs and related parts, temporarily, on the way to making your own.

\$\endgroup\$
  • \$\begingroup\$ I already tried reprogramming the MCU... It appears to be some kind of specialized part from the same company that designed the whole PCB, and I can't find any docs on it. There's a set of 0.1"-spaced holes with most of the JTAG pins labelled, but without TCK. It also has a bizarre pinout, which doesn't match any other MCUs I can find (power pins are in the wrong place). Treating it as an array of FETs and gate drives is basically what I'm doing. The gate drives are all 0603 and SOT-23, which aren't much better for soldering, but definitely more attractive that 0.5mm pitch. \$\endgroup\$ – Brian Silverman Jan 2 at 5:12
  • \$\begingroup\$ 0603 and SOT-23 are really not small. And they have the huge advantage of not having adjacent pins immediately next to the signals you are contancting. What's the apparent MCU part number? \$\endgroup\$ – Chris Stratton Jan 2 at 5:17
  • \$\begingroup\$ XCM-K. xcmmcu.com appears to be the manufacturer, and possibly the designer of the PCBs. Looking around online, some of their other parts appear to be re-branded/stolen NEC/Renesas MCUs, but not these particular ones that I can find. Pins I've been using to try and find any kind of matching documentation: 9+10 are ground, 7+8 are 5V power, 1 is TDI, 2 is RST, 47 is TDO, and 48 is TMS. I tried talking to it via JTAG with TCK on pins 3, 4, and 46 without any hint of a scan chain (other adjacent pins are attached to on-board circuitry that rules them out). \$\endgroup\$ – Brian Silverman Jan 2 at 7:45
1
\$\begingroup\$

You've basically covered all of the common solutions in your question. There are some other specific areas to look for a connector that can be soldered to a QFP footprint, such as emulation adapters or socket converters. Googling for these terms brings up some options for other QFP footprints, such as QFP44. I suspect that these mostly target at common memory IC footprints, so it may be difficult to find a QFP48x0.50, specifically.

Another option might be to use a QFP test clip. This requires that there be a QFP IC attached to the board, but that IC must not interfere with the signals you're interested in. You might be able to make this work by simply holding the MCU on the board in reset, which should keep all of its IO lines high impedance. If that won't work for some reason, you could replace the MCU with a dummy package that has no die and no internal connections--these are available from specialist suppliers for assembly testing purposes. If you can't get hold of a suitable dummy package, then you might be able to lobotomize the existing MCU by carefully grinding out the die and bond wires from the package, but this should probably be a last resort method.

All of that said, emulation adapters and test clips are not likely to be very reliable if your prototype will be subject to handling or other motion/vibration. The most reliable solution, assuming you can't get a castellated PCB made (0.5mm is pretty small for castellations as you've noted), is probably to solder wires from the MCU pads to a header. This won't be especially easy, but with some fine magnet wire (I usually use a 34AWG Beldsol, which has a solderable enamel insulation), a steady hand, and a good microscope it is certainly doable. You will of course need to find a suitable place on the board to mount a connector. A small PCB that has a connector of your choice broken out to 1.5mm solder pads would be a big help there. The upside is that, done properly, this method can be quite reliable, especially if you glue all of the wires in place. Such bodge wire techniques, done properly, can even be acceptable on production PCBs, although you certainly want to do what you're doing more than once!

\$\endgroup\$
0
\$\begingroup\$

How about a board like this: https://www.amazon.com/QFP32-QFP44-QFP64-QFP100-Breakout/dp/B00XSC2CZS

Usually you can find these boards with the "breakout" keyword and the name of the package.

Getting a custom PCB with 0.5mm castellated edges made for cheap doesn't seem to be possible.

Not sure why not, I do it all the time, you just need some large through-hole pads that you put on the edge, half-way through the cut line. You can make a PCB and have like 20 pcs manufactured for something like less than 20$.

(sorry for formatting, I'm on mobile).

\$\endgroup\$
  • \$\begingroup\$ Interesting idea for cutting a PCB after the fact to make castellated edges. Do you have any tips for getting the plating to stay in the half-holes during/after cutting? I'm not seeing how to use one of those breakout boards. Seems like that only helps for the opposite direction (wire to the IC, vs I want to wire to the PCB it's mounted on). \$\endgroup\$ – Brian Silverman Jan 2 at 5:18
  • \$\begingroup\$ You don't have to do anything, it's on the gerber file / your pcb design, the manufacturer will cut where you put the board edge. I had no issue as to have the plating stay in place, I think there are specialized, expensive "proper" process to do that but in your case the normal way should be good enough. \$\endgroup\$ – Damien Jan 2 at 5:47
  • \$\begingroup\$ Getting good castellations requires special handling of the routing path to avoid tearout and avoid leaving copper foil in the remaining notch of the hole. Cheap fabs may or may not be willing to do the castellations, good fabs will want to do them well (and will likely charge more). 0.5mm pitch castellations are going to be challenging, because the holes will need to be 0.2-0.3mm. At that size the castellations may be smaller than the profile routing tolerance and the holes may simply be routed off instead of castellated! \$\endgroup\$ – ajb Jan 2 at 17:41
  • \$\begingroup\$ The point is, is this for a few boards ? in that case it will be fine, if you do a million boards then yes need proper process. \$\endgroup\$ – Damien Jan 3 at 3:07
-2
\$\begingroup\$

www.proto-advantage.com has all kinds of adapters available, and will even mount parts for you for a fee. Not sure about customer supplied parts, you'd have to ask on that. They will purchase parts from Digikey and mount them on adapters, I've had small A/D, D/A, and precision voltage reference chips done like that.

\$\endgroup\$

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.