SMD assembly: Hot air, plate or oven?

I’ve soldered a few boards with my hot air rework gun and that works fine. You can grab anything that wants to walk off somewhere when the solder flows.

For batch production of say 10’s of boards, whats going to work better? A hot plate or an oven?

The only issue I can see with an oven is that you cant watch the soldering process taking place, so anything that might shift is going to need kapton taping into place?

1 Like

“Walking off” shouldn’t happen in an oven. If it happens, there’s some design problem or problem with applying paste or putting the components in their places.

They can of course move if you blow the air on them. But it should be possible to heat the board from the bottom.

I have worked with a rework station which has a changeable nozzle. If I took off the narrow nozzle it gave less blow but to larger area. That was good for heating the bottom of the board. The board must of course be set so that you can work under it.

It’s also pretty quick, one board takes a couple of minutes (depends on the size and density of the design of course). You can learn to set the temperature to a sweet spot which melts the paste but doesn’t burn the board. Keep the airflow 10…20 seconds in one place until the paste melts, then move it.

I have a D2 PAK regulator that wants to move a lot. I’ll look into why, when I do he next batch of PCBs.

I think the solder pads for the two legs and tab are about 2-3mm too close together.

I’ve no issues using the hot air, but would like to do more boards at a time.

It may also be the large exposed pad area, maybe there’s too much paste there or the copper area is large so that the component can float on it. Out of curiosity, can you give the datasheet and the footprint?

I agree with the comment to use a larger nozzle. I use a 852D+ rework station with what looks like a 7mm nozzle. I use a tiny nozzle that’s “pinched” almost closed to blow parts off a board for rework. My latest board has has a high density of parts on both sides in some places, so I’m not sure how that could even be done other than carefully with a hot air gun. But doing it in volume? No idea other than that I’d have to do a simpler board. And with 250 parts on a board, I wouldn’t want to do more than a few.


There is some often sold beige SMT oven with a big drawer, and apparently its quite garbage. There are some video’s for modifications before starting to use it (I remember something about removing not heat resistant tape from the insulation) Apparently it also has problems with non -even heating.

One of the things on my list is to make a SMT oven from a small kitchen oven. I’ve already got a donor oven with a glass window, and bought some extra 300W IR heating elements from Ali. I do plan to shield these heating elements, so the PCB is not directly heated by IR radiation, which causes un-even heating, depending on color and reflectivity.

Another option is to use a hot plate in combination with hot air. In this case you just use the hot plate for pre-heating. This may or may not give you problems with evaporating flux before it can do the job it’s there for.

I’ve used all the common approaches:

  • Hot Air
  • Oven
  • Soldering Iron
  • HotPlate (actually, a pan on the stovetop which is really the poor man’s version)

My favourite by far was my “Hot Plate”. It was just so easy to heat the board up and fix any problems with tombstoned components on the fly. It’s also pretty quick.

If you’re just doing single-sided boards, I’m all for the hot plate approach, I like it the most.

1 Like

I currently work for a (very) small manufacturer, about a dozen employees in total.

For several years we have successfully used an electric skillet as our “tabletop reflow oven”. (Here’s one, of several, online pages discussing the method.) There has been a fair amount of cut-and-try to the process, but we made it work and you can’t beat the price. (Our skillet was US$5.00 in a second-hand store.)

Our boards range from about a dozen parts, up to around 100 parts. Passives are 0805 or 1206; semiconductor packages have lead pitch down to 25 mils or 0.65mm. (But it is MUCH easier to work with 50 mils and 0.95mm.) Hand-placed parts want to have a larger courtyard area than automated placement.

We typically do boards in batches of 5 to 20 units. Depending on size (obviously!) there is room in the skillet for 2 to 8 boards at a time. We actually stumbled onto the skillet method by accident, after having problems doing hand-soldering of a chip with an exposed back-side thermal pad. Even using two soldering irons at a time, we couldn’t get consistent, thorough, solder-wetting of that thermal pad. Since the skillet heats the entire board plus all the components at once, soldering the thermal pad is not a problem.

It’s important to be consistent and uniform for the process to work. Definitely apply solder paste with a stencil, and be prepared to try different stencil thicknesses and/or stencil-shrink parameters. Our solder is Kester EP-256 from CML supply. The specified shelf-life is 6 months under refrigeration; we get about 2 years when it’s kept in a desk drawer (and rejuvenated once or twice with a squirt of isopropanol).

A glass cover on the skillet is not essential, though watching things happen speeds up the initial process calibration.

Half an inch (1 cm) of fine sand in the skillet improves uniformity of heat distribution.

We pass everything under a 10X binocular microscope to find bridges and dry joints. Decent wide-field microscopes start around US$200 new; used units from top-line manufacturers (Nikon, Bausch & Lomb, et al) are sometimes as low as US$100.



Do you ever do pcbs with parts on both sides? If yes, go to the “toaster oven”. I put parts on 1 side and toast the board. After it cools, I pasted, place and toast the other side. You don’t dare touch the oven or pcb until it has cooled down. Otherwise the bottom parts get knocked off.

What method would you recommend to a newbie to find this “sweet spot” on the temperature dial ?

Can you recommend vendors for these ?

1 Like

I’ve not had good luck finding a hot plate that gets hot enough, so I use the hot air method. I bought a Dremel drill press and modified it to hold my hot air unit. I use a large nozzle and slowly lower the nozzle to the board in some sort of emulation of a soldering profile (in my imagination, but it seems to work). Rarely blow a component off the board. Working with lots of 0402 parts, and with a bit of practice. it goes quickly. I’ve built about 200+ boards that way with good success.

Does the nozzle reflow solder on more than one component at a time, or do you repeat the process for each SMD ?

About 3 years ago, my employer bought a binocular microscope from “Amscope”. It has a 1X objective lens and two pairs of wide-field oculars, 10X and 20X. The microscope head is mounted on a boom arm, cantilevered out from a weighted base.

We have used the 10X magnification almost exclusively. I recall only two or three times when I popped in the 20X oculars, to get a closer inspection of a potential problem after locating it under 10X magnification. If I could change any feature of the optical system, I’d ask for a wider field of view, or possibly greater depth of focus. Even those would be “convenience” or “wishlist” requests, not really a shortcoming of the instrument we have. One thing I WOULD like to have is a diffused lighting source, rather than the single-point illuminator.

You definitely want an extended boom-arm stand, or possibly an articulated arm. An articulated arm would let you fold the microscope up out of the way when not being used, but I wouldn’t put out a lot of money for an articulated arm until I was convinced it doesn’t have problems with vibration or jiggles.

I don’t know what the working distance (from the objective lens to the target workpiece) is, but there is adequate clearance for my fumbly fingers, small tools, meter probes, soldering iron, etc.

A photographer may be critical of our microscope’s optical performance but I have no complaints. Focus seems to be flat across the field of view, I haven’t noticed any geometric distortions, nor have I seen any color fringes or halos around sharp edges. Of course, my super-annuated, tri-focaled eyeballs may not notice such things unless they are very severe.

I believe the company’s bottom-line cost was right at US$200, after shipping charges (the thing weighs at least 10 lbs (4.5 kilos)!) and a new-customer discount. Amscope offers several hundred models, with various combinations of magnification, stand styles, illuminators, CCD camera mounts, etc. I have also noticed several vendors offering microscopes with very similar appearance to Amscope’s products. I don’t know if these are identical instruments with different marketing nameplates, or different performance grades of the same design, or significantly different designs in similar enclosures. And, of course, there are the “usual suspects” offering used/surplus/pre-owned equipment.


1 Like

I used the largest nozzle I have, and once the board is sufficiently pre-heated I bring the nozzle closer and watch as the paste begins to coalesce. Once the melting has started, I move the board and nozzle slowly toward more components. The larger nozzle usually covers multiple passives or one side of a chip at a time.

I have read that amateur solution for thermal pads are just big plated holes under it and soldering it from backside.

As I recall, that design had copper zones on both sides with a lot of stitching-vias linking them. We tried several approaches, using multiple soldering irons and heat guns, but couldn’t get reliable, consistent results. Using the skillet to solder that one IC - the rest of the board was still thru-hole parts - made all the difference. We gauged the effectiveness of the soldering job by observing how the paste solder flowed through the stitching vias to the back side.


From what you write looks that you didn’t tried to replace lot of stitching-vias with one 3mm hole to be fully soldered from backside.
But I only read that it is the solution. I have never tried it myself.
We were soldering our products ourselves till 2005. At the beginning (88-96) manually and later using semi automatic placement and simple oven (but destined to SMD).
Our oven didn’t had any inspection window so we couldn’t see soldering process. We used a thermocouple (made of very thin wires) inserted in soldering paste at element pads and registered the temperature during the process and based on it we setup our oven parameters.
In 2004 we tried to order PCB production at contract manufacturer who not needed the orders to be 10 000 pcs or more but was able to do 50 pcs for us. We found that calculating our time used for element and PCB ordering and then soldering everything ourselves using external service costs about the same or even a little less. During 2004-2006 we gradually moved all our production outside.

We are very small (without “()” around very). 7 people working, from that 2 in direct manufacturing.

Using external services has one more advantage. In 2006 we got an order to do 800 pcs of that:

If we were to do it on our own, we would not be able to meet the 3 months in which it had to be delivered.

800 units in 3 months! That’s quite a complex board to build. 8051 training computer? I remember using something like that back in the 80’s.

As a set to each education unit we also had to manufacture probably (don’t remember well) 5 external boards allowing for some exercises (some of them are visible at link I have given previously). We had designed 10 and I just don’t remember how many of them were ordered those time. So in total we had to deliver 4800 PCBs or more.
We got order in April 2006 and the original delivery date (30.06.2006) was ‘kindly’ shifted by officials to 31.09.2006. But in my opinion it is a perfect example of clerical stupidity. After all, everyone knew then that the RoHS directive came into force on July 1. The contract manufacturer with whom we have recently cooperated said he has not yet mastered the lead-free process and does not know if he will reach it by September. We also preferred to have it done in classical Pb-process as we were forced to give a 3-year warranty. So, our decision was - we have to do it before 1.07.2006.
We used other manufacturer for all those external boards. Our main job was to check and run all of these boards.
We were lucky to have just recently mastered outsourcing production to a contract manufacturer. If not, we wouldn’t have had a chance to complete this order (the biggest one we ever made).

When we designed it (1993…1995) some schools in Poland used it without PC (they couldn’t afford computers). So we embedded into it simple assemble code editor (display has only 2 lines). You selected mnemonics and their parameters from the menu. Then you were able to run your program with full speed or step by step. There were two reset levels - absolute or ‘to your program’.
If you had PC you could use written by me 8051 assembler and environment allowing to debug your program.