Fail Of The Week: How Not To Re-Reflow

There’s no question that surface-mount technology has been a game-changer for PCB design. It means easier automated component placement and soldering, and it’s a big reason why electronics have gotten so cheap. It’s not without problems, though, particularly when you have no choice but to include through-hole components on your SMT boards.

[James Clough] ran into this problem recently, and he tried to solve it by reflowing through-hole connectors onto assembled SMT boards. The boards are part of his electronic lead screw project, an accessory for lathes that makes threading operations easier and more flexible. We covered the proof-of-concept for the project; he’s come a long way since then and is almost ready to start offering the ELS for sale. The PCBs were partially assembled by the board vendor, leaving off a couple of through-hole connectors and the power jack. [James]’ thought was to run the boards back through his reflow oven to add the connectors, so he tried a few experiments first on the non-reflow rated connectors. The Phoenix-style connectors discolored and changed dimensionally after a trip through the oven, and the plastic on the pin headers loosened its grip on the pins. The female header socket and the power jack fared better, so he tried reflowing them, but it didn’t work out too well, at least for the headers. He blames poor heat conduction due to the lack of contact between the board and the reflow oven plate, and we agree; perhaps an aluminum block milled to fit snugly between the header sockets would help.

Hats off to [James] for trying to save his future customers a few steps on assembly, but it’s pretty clear there are no good shortcuts here. And we highly recommend the electronic leadscrew playlist to anyone interested in the convergence of machine tools and electronics.

30 thoughts on “Fail Of The Week: How Not To Re-Reflow

    1. Agreed, what’s the point here ? To fix badly designed kit so the are not kits anymore but fully assembled PCB?
      Reflowing PTH needs specific parts and paste in technique.
      As mentionned relflow (and board manufacturing) is a well defined process without much room forum fiddling.

      Go backup to the drawing board and design the board to be manafactured ( cheap, fast, quality :pick two)

        1. B^)

          Elliot,
          it seems that some readers have forgotten that the purpose of “Fail of the Week” is to not just to show us how/that someone failed, but provide a venue for us to contribute ideas to rectify the problem.

          1. A plate between the headers would do the most good for heat conduction. You would also be raising the headers off the oven plate so the plastic headers would get less heat and less deformation and would probably need to tape the headers in place (kapton tape) during reflow. I’m guessing you don’t have the benefit of a nitrogen atmosphere.

            Could be a little more deliberate with the solder dispense on the pins.

            Also might consider a lower temperature solder if the application will allow it.

  1. Surface mount parts often have different, higher temperature plastics which allow them to be tolerant of the reflow soldering process. Some of the parts are likely simply incompatible with reflow soldering.

    If he’s interested in (potentially) reducing board assembly cost for both THM and SMT components, that board looks like it doesn’t require any particularly fine pitch SMDs and should be largely solderable with wave soldering, if all the SMDs were flipped to the back, appropriately masked for the wave direction, etc.

      1. Yeah, that color change on the Phoenix-style connector was subtle, but clearly enough to change something dimensionally. That said, I can’t imagine manufacturers haven’t provided SMT-compatible versions of every part they offer, so there must be something out there that does the job and can – literally – take the heat.

        1. It’s been my experience that SMT connectors tend to have the solder joints fatigue and fail after not as many uses as I would have expected. It helps a lot if I buy connectors that either have plastic pins that press into matching holes in the board or through-hole tabs for the connector shell (but then it’s still a through-hole soldering project, just less of them.) And while I know professional boards use SMT connectors all the time, I spend more of my time than I’d like resoldering USB keys, cycling computer recharging ports, and cellphone data/charging ports because their solder joints have failed exactly the same way my DIY boards have failed. I’m in favor of a wiring harness to a separate panel-mount connector whenever possible.

    1. IIRC, the manufacturing way is wave soldering with the addition of a pallet which only exposes the through hole leads for soldering.
      For small batches, a pallet and solder pot would seem to be a possible solution.

  2. Nope. Not waiting through 20 minutes of talking, just to see where it went south.

    However, I did watch far enough for this:
    Make up your mind. No-clean, or water wash flux. If you’re going to wash with water (we used a regular home dishwasher at the last place I worked that did assembly), then use a water wash flux. If you’re not, use no-clean. Simple as that.

    I’ve never trusted no-clean myself, and I believe I’ve had problems with devices that used no-clean flux. These were high-impedance audio circuits, used in a high-humidity environment, and they developed high-resistance leaks between pins, and washing with alcohol appeared to fix the problems (the amps were so far out of balance, the outputs just went to the rails).

  3. I can’t help you with the reflowing of the multipin headers, but I think it’s a LOT easier to hand solder a surface-mount power jack onto a PCB than it is to run the whole board through its reflow profile.

    1. That’s easy, use SMD headers if you want do reflow. There is no other way around. THT goes thru wave soldering or hand soldering and SMD goes to reflow. And yes, sometimes there are cases where you have to use THT for example because of physical strain on headers but in most cases THT headers are just overkill adding unnecessary THT step to manufactuing. And before someone adds that headers are from oposite side, that’s not an issue, it is solved by soldering each side with different temperature profile so first you solder at higher temperatures on one side than on another.

  4. You can buy special plastic headers that are reflow solderable. It still is tricky to not deform the plastic. Alternatively, you can use wave soldering for them, but there should be no smd on the back then. Usually this step does not cost extra unless you make a massive amount of boards.

  5. What’s wrong with pin-in-paste? You still need to select your parts carefully as some manufacturers simply replace the plastic without adding extra room for the paste (a larger cavity between the board and the plastic is required as with normal THT components).

    Seems a lot easier than trying to get a normal THT component through a reflow profile IMO.

  6. Look for PIP (pin in paste) components. Phoenix, for example, has special connectors for this. They use special types of plastics and the pins are shaped specifically to make reflow effective.

  7. As someone who routinely reflows TH components at a low-to-mid volume, high-mix contract manufacturer, I have some comments for others looking to take advantage of this process as there are some valuable benefits. However, I don’t think calling this video “Fail of the Week” is appropriate. Maybe “lessons learned from a first run experiment” is a better title with a less insulting connotation.

    1) Fun facts: The process is called several names: Intrusive Reflow, Intrusive Solder, Pin In Paste, Paste In Hole, Reflow of Through Hole (ROT), or the very old term “SCRS”, Single Center Reflow Soldering. It is not a new process by any means yet dates to 1986 when Robert Clawson of Motorola ran experiments and then released an internal document entitled “Factors Affecting Reflow of Through Hole Mount Components” in 1987. Phil Zarrow of ITM Consulting released an article in the Journal of SMT in October 1999 that mentions a couple internal Motorola documents.

    2) Read component datasheets. Reflow compatible parts will be deemed as such as well as max temperatures, plastic material, etc. Also, component mass is to be considered for double sided boards where the TH part will be on the first side ran.

    3) Know your oven’s capabilities and how to set your profile correctly. A few parameters related to intrusive reflow may need to be carefully controlled. A plain toaster oven will be a stretch if your design and the part being used do not allow for some tolerances during reflow. The one the Clough42 used is perfectly capable as seen on the right hand side of the board, but his profile was incorrect for the thermal mass imbalance that exists with his board. This can be seen at 14:20 where the left side of the board with the larger thermal mass (aka copper ground plane) did not reach proper TAL (Time Above Liquidus). Essentially, the performance of the profile he used is too reliant on the board sitting on the plate. I would bet he would be successful if he played with his profile settings while monitoring thermocouple attached to a pin on the left header connected to the ground plane.

    4) Ensure you are using the proper hole size. Some reflow compatible parts will give different hole size recommendations as compared to their wave solder counterparts. Not real common, but they are out there.

    5) Determine proper aperture size for your stencil. Sometimes, you can get lucky and just make your aperture 10% larger than your annular ring of the hole, but depending on the pin to hole ratio, you may need to overprint. Also, overprinting strategies will be different for tin/lead and lead free solders as well as no-clean or aqueous pastes. In general, if you are using a no-clean tin/lead in your toaster oven (like I do at home) then overprinting is fairly easy. I usually use an elongated oval that sort of teardrops away from the hole. The solder wicks right in to the hole during reflow. His dispensing method was very good and I bet over time, he could get his “per pin” volume to be very consistent.

    6) Do not expect to see the same solder joint from intrusive reflow as you do with handsoldering or wavesoldering. Most of my joints do not have a fillet on the bottom side, but the holes are >90% filled. This is fine, even per IPC standards, as the quality determination is based upon the solder source side and in this case, that is the top, not the bottom like with handsoldering. When I have pulled off a connector housing after intrusive reflow, I have seen fillets on the top side.

    7) Avoiding additional operations like wave soldering or selective soldering is one major benefit of reflowing TH components. There are also many hybrid components, especially user interfacing connectors like USB and HDMI. Hybrid connectors may use SMT style pins for the I/O connections, but have TH pins/posts for the grounding and shell. Intrusive reflow is ideal for these as opposed to reflowing the part in SMT for the I/O joints and then soldering the posts in a later process.

    8) Commenter Wolf mentioned Selective Soldering and linked a video of a KISS 104 machine. We have 3 KISS 102 machines performing selective soldering and yes, they would be the perfect solution for this particular board.

    9) Commenter Vojtěch Vladyka mentioned SMT headers as an alternative. While I completely disagree with their statement saying that is the only way, it is a very viable option. In many cases, the retention force of the joints on an SMT header make it reliable for plugging and unplugging. Since this board is intended to be installed once, SMT headers are a slam-dunk solution. Also, many SMT headers have options for retention posts that will lock in to the PCB or can be soldered in.

    10) The solderpaste he used has a very high flux to powder ratio which results in a high slump value. This can be seen as the solderpaste flows together in many place after dispensing. For his scenario, that was an advantage, but keep that in mind if you plan to use a stencil to apply your paste. The T3 or T4 he mentions is for Type 3 or Type 4 which refers to the ball size. Type 4 has a max diameter of 38µm, versus the 45µm max for Type 3.

    11) The only part of his video that I would be ok with HaD calling a fail would be his determination that the boards were “clean”. Determination of the cleanliness of a board subjected to aqueous flux processes cannot be made by surface visual inspection alone. There is a major risk to his boards if some flux residue is still under the body of the headers or power connectors. Aqueous flux (“water wash” per chipquik’s term) remains corrosive after reflow and must be removed. I do applaud him for using the proper temp for his cleaner; he just needs deionized water to make it fully “professional”.

    12) Commenter BrightBlueJim mentioned not trusting no-clean flux. Impedance requirements can be challenging, but proper research into the specifications of the solderpaste you are using should help with that (we build 12G/4K/8K video hardware with no-clean every month). Nowadays, no-clean flux chemistry is amazing and has many advantages over aqueous fluxes where it can be used. About 80% of the product produced on my SMT lines is with no-clean fluxes. Also, where no-clean flux needs to be removed, there are many solutions (no pun intended). Three of our wash bays are set up to run with a Zestron chemical additive to aid in removing no-clean flux beautifully. The comment about pin probe issues with no clean flux is valid, which is why I avoid no-clean solderpastes that leave thick residues.

    Wow, I apologize for the insanely long post, but hopefully someone gets some value out of it.

  8. I’m not suggesting this as a good idea on a regular basis, but you can get solder pots surprisingly inexpensively from some surplus places sometimes. I’ve seen one for less than $30. Then you dip that edge of the board into the pot as quick as you can, and it’s done. They’re dangerous and probably wildly toxic, but if you have a whole bunch of boards already done and now you have to rework them, it’s not the worst solution.

  9. It might just be, that the “THT” power connector was fine with being heated up because it is not THT but actually SMD… Also, there are SMD versions of most 2.54″ pin headers available. All of those can be placed by any half decent assembly house (I already had these types machine placed by different European and Asian assembly houses without any issue).
    As for the phoenix style connector: I recommend going with actual phoenix connectors. They have SMD, THR (through hole reflow), and press-in/SKEDD options the former two being easiest since everything can be assmebled using PNP machines.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.