Vapor Phase reflow oven

[Doctorandus_P]

It is well known that Galden fluid is thermally liable, but that was not in question.
The problem is to determine the absolute minimum volume of fluid in an oven.

There are two important factors here.
First, it's "vapor phase", i.e. It's the vapor (and thermal energy in it) that does it. not the liquid, and the liquid (and I assume vapor too) is quite heavy, so you would need to fill the volume between the bottom of the pan upto just above the PCB with saturated vapor. Air is 1.2kg/m3, so assume it's twice as heavy as air and you need 5l then you will need 12gram of this stuff.

The second important factor is that you do not want to overheat the vapor. The easiest way to achieve that is to keep the heating element itself submersed in the liquid. The liquid will keep the heating element at the boiling point of the liquid. When the heating element boils dry (too less fluid) then it will get hotter, and there is a risk the galden fluid / vapor starts decomposing (and that is a very unhealthy thing, there is fluor in there).

Normally the pan has a flat bottom, and you need to cover the whole bottom with fluid, but there is no real need to. The bottom could be bent, with some kind of float switch near the lowest point (with the heating element).

Additionally: There is not much reason to filter the fluid. (As some seem to do) Any solids in the fluid will not evaporate and simply be left on the bottom. The galden fluid could become contaminated with other liquids (Flux, or cleaning fluid residue) but those won't be easy to filter out, and probably also do not matter much. Maybe you could use distillation to get rid of them. The Vapor phase thing itself is already a heater so heat it until 10 or 20 degrees below the boiling point, and then blow away the vapors that do come out. (Also such vapors also tend to be lighter, so they have a tendency to be in a "less harmful" place anyway. The high specific density of the galden stuff has an influence on quite a lot of things...

[wraper]

here are two important factors here.
First, it's "vapor phase", i.e. It's the vapor (and thermal energy in it) that does it. not the liquid, and the liquid (and I assume vapor too) is quite heavy, so you would need to fill the volume between the bottom of the pan upto just above the PCB with saturated vapor. Air is 1.2kg/m3, so assume it's twice as heavy as air and you need 5l then you will need 12gram of this stuff.

From quick googling:

For this Galden type we can ballpark the vapor density at 260 °C (assuming 1 bar here) to be somwhere between 24 and 30 kg/m³. Galden HS260 has an average molecular weight of 1210 a.m.u, or a molar mass of 1.21 kg/m³. Applying the ideal gas law

(where p: pressure; M: molar mass; R: gas constant; T: temperature)
yields 27.3 kg/m³ at 260 °C. That's within the range I identified in that plot up there, so I guess it's ok to just apply the ideal gas law to Galden to figure out the vapor density (this is not necessarily a given). For the LS 230 grade, the result is 24.4 kg/m³ at 230 °C. For comparision, air has a density of about 0.7 kg/m³ at that temperature.

[wraper]

Additionally: There is not much reason to filter the fluid. (As some seem to do) Any solids in the fluid will not evaporate and simply be left on the bottom. The galden fluid could become contaminated with other liquids (Flux, or cleaning fluid residue) but those won't be easy to filter out, and probably also do not matter much. Maybe you could use distillation to get rid of them. The Vapor phase thing itself is already a heater so heat it until 10 or 20 degrees below the boiling point, and then blow away the vapors that do come out. (Also such vapors also tend to be lighter, so they have a tendency to be in a "less harmful" place anyway. The high specific density of the galden stuff has an influence on quite a lot of things...

IME after reflow cycle galden becomes muddy but not for long. Soon all of the contamination deposits on the metal surface. So unless you filter it ASAP after every reflow, it's completely pointless. Also amount of contamination highly depends on solder paste used. LOCTITE GC 10 causes a lot of hard to clean dark deposits on all surfaces and oven will soon look like brown shit, also it results in oven pan having strong burnt smell. On other hand Chipquick SMD291SNL and AIM M8 SAC305 cause barely any contamination and no awful smell.

[Doctorandus_P]

You did your homework better then I did, but overall, it's not the important question.

More important is what causes loss of the galden fluid, and how do you prevent it. This site: https://www.vaporflow.eu/ mentions a loss of 2 to 3 grams per cycle (Depending on oven size) And with EUR200/kg this is 40ct to 60ct per soldering cycle. It also states it needs a minimum of 290gram or 750gram to work properly (Looks like bad design to me, yuch.).

 A part of the loss is because of the liquid evaporating, which forces gas out of the enclosure, and a part of the escaping vapors will be the expensive fluid. Maybe an hermetically sealed lid can help, if it is combined with some kind of bladder that can accommodate the expansion.
Another option is to actively cool the top rim of the oven, so the vapor condenses and flows back to the bottom.

Some of the condensed fluid will also be on the PCB after soldering ended. A good way to shake it off before you take the PCB out of the oven also helps to reduce loss.

And last, there will always be some partial vapor pressure in the oven as long as it contains fluid. Minimizing the mixing of outside air with inside air when the lid is off.

[wraper]

I did not see loss anywhere close to that with IMDES mini running with my own controller that does not overcook. It can easily run for hundreds of cycles with ~50ml of Galden before I see the need to add any. The largest annoyance is that Galden condenses on the lid. But there is very little unlike with original controller. I remove it with squeegee into plastic container after each cycle so it does not drop on PCB and then pour back after finishing the job.

[newto]

Normally the pan has a flat bottom, and you need to cover the whole bottom with fluid, but there is no real need to. The bottom could be bent, with some kind of float switch near the lowest point (with the heating element).

I've wondered about that before, have a small reservoir in the middle of a sloped bottom, and then you need a much smaller volume of Galden vs for a wide area, allowing for larger (or more) boards without breaking the bank on fluid. and the heating system only needs to be around or in the bottom of the reservoir

[Styno]

You have to be very (!) careful that not all the liquid Galden is evaporated because then it's possible for the temperature to rise higher and the Galden will decompose to a very toxic gas.

[woody]

Not trying to be a wise-ass here, but IMO one should not overstate the loss of Galden in a VaporFlow like machine, which is meant for prototyping and small series. The actual loss is very low as @wraper observed. Galden is bloody expensive but even if you lose 2ml / cycle this comes to €0.45 / cycle (inc VAT). I spend WAY more on solderpaste (that I buy in 50g quantities, for larger quantities become unusable faster than I need solder paste), cleaning cloths, IPA etcetera. Galden is not going to be your bank breaker.

If the need arises to operate an Imdes or VaporFlow on a daily basis you're much better off to outsource the assembly of your boards. Also because then you are going to need a decent stencil printer, a solid PnP machine and what not. Talking about breaking the bank. Apart from this these companies are much better equipped to deal with environmental issues and part handling and acquisition than most of us are in our garages.

My take is that Imdex/VaporFlow type of solder ovens are perfect for their intended market.

[Doctorandus_P]

Not trying to be a wise-ass here...
Galden is bloody expensive but even if you lose 2ml / cycle this comes to €0.45 / cycle (inc VAT).

I think you're right. The Galden stuff is expensive, but the total cost are nor much compared with the cost of a prototype. My main gripe is not really the cost of the fluid, but more that they did not even try to conserve the fluid better oven costing EUR2000.

And for solder paste. Yes, that stuff is expensive too. Why would a price of EUR400/kg be justified if most of it is tin which does not cost much. Flux is expensive, and that I also do not understand. It's mostly quite simple chemicals. Making balls of the solder is also an easy industrial process. My main problem with solder paste is it's moderate shelf life (Except for GC10 and maybe some others). If you only do 3 prototypes per year then even a small jar takes much longer to use up then it's 2 year (official) shelf life. Has anyone tried to use separate balls and flux, and then mix a small batch of paste just before soldering a protype PCB?

And the inherent ease and reliability of this process still does make it an interesting option. But do track of your units. This Galden stuff has a quite high specific density of 1.82g/ml (for HT230)

https://www.solvay.com/sites/g/files/srpend221/files/2018-10/Galden-PFPE-Heat-Transfer-Fluids_EN-v2.2_0.pdf

And apparently it's also not only used for prototypes. The video below does look like a (smallish) production site. I like the way in which they lower the PCB's into the vapor. This way

You have to be very (!) careful that not all the liquid Galden is evaporated because then it's possible for the temperature to rise higher and the Galden will decompose to a very toxic gas.

.
Yes, indeed A decent overheating detection and automatic turnoff is mandatory. But this still makes such an oven just a horeca pan (with a glass lid), a heating element and some electronics which includes the overheating detection. If I were to build such a thing. I would add at least two temperature sensors and also turn the thing off if the difference between them becomes to big. (Imagine a damages sensor, wiring or a sensor falling off, etc).

I did not see loss anywhere close to that with IMDES mini running with my own controller that does not overcook. It can easily run for hundreds of cycles with ~50ml of Galden before I see the need to add any.

That seems quite remarkable. 

The largest annoyance is that Galden condenses on the lid. But there is very little unlike with original controller. I remove it with squeegee into plastic container after each cycle so it does not drop on PCB and then pour back after finishing the job.

You could make a double layered lid with a slanted inner glass plate (I think silicone sealent is up to the temperature) or some kind of roof above the PCB from a piece of glass. Another option is to simply tilt the whole oven so the drops on the lid slide to the side. I would also make some system in which I can squeegee it directly back into the oven without an extra plastic container. Every extra surface that gets wetted will cause needless extra liquid loss.

[woody]

Another option is to simply tilt the whole oven so the drops on the lid slide to the side. I would also make some system in which I can squeegee it directly back into the oven without an extra plastic container. Every extra surface that gets wetted will cause needless extra liquid loss.

That is actually a bad and possibly dangerous idea; the layer of Galden in these ovens is really only a few mm. By tilting it part of the bottom might no longer be covered in Galden, which might heat it beyond safe temperatures. For this reason I placed my oven using a level.

You cannot get away with filling it with extra Galden to still cover the bottom, as more of the medium slows the process time considerably. Longer heating, longer cooling. And, as we all know, it is bloody expensive 

[Doctorandus_P]

Indeed, tilting the oven is incompatible with the simplistic way in which these ovens are built. Yet another reason to not buy the standard overpriced oven. With only a few mm of fluid over the whole bottom, using a level to put the thing horizontal is indeed a sensible option. Are such levels integrated in the EUR 2000 oven? Does it have leveling feet?

If you build one from a horeca pan yourself, you could deform the sides to tilt the top while the bottom stays horizontal, but it would be quite difficult to do that while keeping both the bottom and top "flat". Something like this would be relatively easy in series production (with a big press and a mold to keep both the bottom and the top flat). And of course this can be incorporated if they used a custom bucket instead of a standard horeca pan.

[woody]

Are such levels integrated in the EUR 2000 oven? Does it have leveling feet?

My €1500 oven does not have a level and does have leveling feet.

Maybe a good idea for people trying to perfect the wheel here is making the leveling automatic; that way you can keep the oven level while soldering while, during the cooling phase when the heating element is switched off and there is no longer the danger of overheating the Galden, you can tilt the oven a bit to circumvent the drip problem. Would be an interesting project with a couple of motors, a controller and an accelerometer.

Meanwhile I'll just use the oven on my bench 

[Doctorandus_P]

Just realized, if you put a few mm of liquid in the bottom, then you can use that as a level. (Assuming the corners are rounded).

[glenenglish]

Another way I have heated prototype Vapour phase ovens is to use a induction cooktop- One of those single 'burner' induction cooktops, and a thick slab of ( the right sort of )  stainless as the bottom of the tank.
If its thick enough, many times skin depth (readily calculable) , the circuit board wont be bothered.

[Electrons]

what about a hotplate (i've got weller WHP1000 laying around) does it help in making a vapor reflow machine ?

another thing why no one has built the open source crowd funded Vapor-phase One . (by pcb arts) . I have yet to find the mechanical design of this thing .

Most build threats of Vapor phase reflow seem to reached dead end wonder why ?

[IconicPCB]

Glen,

I had similar idea early in the piece.
The footprint of the induction hob is quite small and the power would be quite concentrated in a small area. This presented a problem for me.
There is a certain maximum power density, I seem to recall 4W/cm^2 in order to not cause localized fluid decomposition,

So I otped out for four solid cast iron hobs in order to spread out heat uniformly. In fact i had a copper sheet between the hobs and the base plate ( a 5 mm thick magnetic stainless steel material).

I had since lost the copper plate in order to improve heater control loop.

I measure the vapor temperature at the height of top of the PCB being re-flown such that a reasonable soak and re-flow profile is obtained.
A major pain the neck... measure vapor  temperature but apply heat to the bottom of the vessel,

[Electrons]

@IconicPCB I might have understand something incorrectly about the process. In an ideal vapor phase machine what do want your heater to do ? output constant temperature (something like hotplate) and that intern will heat the liquid and protect against over-temperature , OR you want constant heatflux and to close the loop of the heater based on temperature of vapor or liquid ?

[IconicPCB]

It all starts with the process.
Preheat the solder joint to say 150C.
Hold it at 150C for a period of time.
Raise the temperature of solder to liquidus and hold at liquidus for a period of time.

Volume of vapor is the transport mechanism.

How to achieve these temperature steps?
Sense the temperature at the height of the solder joint and control heat input into the oven.
Here is a sample oven profile.
Blue plot is the temperature cycle ( 250C ),  redish plot is the heat demand ( PWM heavily integrated to show average value ) over time 20 min.

EDIT: may be there are people with commercial equipment who could post the process profile of their ovens for comparison?

[Electrons]

is the result you posted from oven you made ? did you get your oven to work and solder properly ?

I think If I want to design an automatic machine it is much work but simple manual controlled machine is fairly easy:
1. A stainless steel deep container . and place a heater inside . or instead by a French fries deep fryer .
2. Add temperature sensor first in liquid to make sure to control the heater and keep liquid at 230 to 240 .
3. add second temperature sensor in the middle of the pot ( this will need experimentation on the place) the role is to know the temperature of the vapor .
4. add a temperature sensor that will be screwed to the board to be soldered.
5. make a manual lifting system.
6. the process will be soaking pcb in the upper level on the chamber until it reaches 150 then lower to the vapor for reflow. the temperature sensor attached to the board will be displaced for the user and he will use this information to control the lifting system.
7. when reflow is done the pot will be cooled by some means ( huge fan for example or watercooler (like the ones CPU/GPU).
8. havenot decided if the pot led should be closed completely without any hole .

After looking at Vapor phase One design ( the have a 3D pdf that shows the design ) . the place heaters similar to deep fryer and wrap thermo couples around them to control the liquid temperature . they donot seem to have temperature sensors other than those
Also they seem to have a very small fan in the Led of the pot donot know why ? doesnot this allows vapor to escape if fan is off ??

[IconicPCB]

Yes the results are from my oven and it works quite well.I do need to do some further tweaking in order to reduce the thermal mass of the internal PCB support cage,

I think immersion heaters are bad juju since it may be possible to exceed 4W/cm2 power delivered into the fluid due to heating element small area.The fan is essential in providing for accelerated cooling of the solder joint,
Prolonged slow cooling will result in poor grain structure of the solder joint.

[Electrons]

do you use a sealed or open led ? or closed with some holes ?