QS-5100 Reflow oven, any tips, warnings etc ?

[NF6X]

I guess that if I were you I would run exactly the same profile that I did... then we can start looking for the differences.. firmware version, thermocouple, cold junction...  board version...  I guess Id like to see a plot of your data.  Im not sure what you mean by "wild temperature profile"

I still had some test data left on the logger's memory card. Here is a plot with the thermocouple taped to the top of the board, using the oven's default temperature profile:



I just plotted one channel there, but the measurements were pretty consistent at the other three locations I probed. Measured temperatures were a lot higher than the profile settings, with big overshoots at each step. Next time I work with the oven, I'll take better data, including logging the actual profile settings. If I recall correctly, the setting for that first shelf, just under 200ºC, had a 150ºC setpoint. The soak stage ended up well above reflow temperature.

I would be open to working on an open source oven controller to retrofit into that oven....   

Unless we identify a serious problem with the controller card design, I would be more inclined to want open source firmware for the original controller. We could clean up the user interface for English-speaking users, and maybe add options such as running the fan slowly during soldering to even out hot spots in the oven?

Do you think that the design of the IR tubes is correct?  i.e. is the platform sound sans temp measurement.

I don't have enough experience with these things yet to form much of an opinion on that. I certainly like the QS-5100 platform more than the T-962 platform, but even the inferior (in my opinion) T-962 platform appears to be giving many people good results after being modified and upgraded.

[engineer@iotexpert.com]

Default?  I notice that the oven kept the last profile it was on... presumably in flash...

That is considerably hot for sure... I wonder if you oven was on profile 4 or 5?

If that is true I wonder if the oven has problems with the hotter temperatures...

I guess I love to see what your oven does with the same profile that I used.

As far as open source firmware... sure... that would be better.

I guess that I am looking to forward to seeing some data where you are for sure about the profile... because if the graph you have is really profile 1 then this thing is way far away.

Alan

[NF6X]

I finally got around to making a new test run of my QS-5100 after getting some thermocouple probes that can handle the oven temperature without their jackets burning. My oven settings were:

• PREH: 150 deg. C 01:00
• HEAT: 200 deg. C 01:30
• SLDR: 235 deg. C 00:30
• KEEP: 220 deg. C
• COOL: 150 deg. C

Probe 1 (blue trace) was in contact with the board, and probe 2 (green trace) was in the air just above the board. The probes shifted a bit ruing the cycles, as the Kapton tape holding them in place got hot. It looks like the board temperatures ran around 15 deg. C hotter than the oven settings.



Here is the raw data, including pictures of my test setup:
https://github.com/NF6X/Qinsi_QS-5100_Tests/tree/master/test01

[NF6X]

I ran another test with the oven's original temperature probe disconnected, and replaced with a thermocouple taped to the board surface. As I expected, the temperature profile looks much better. See the blue trace:



I plan to come up with a less kludgy way to have the oven controller feedback probe in contact with the board being reflowed. I have an idea of how to accomplish that...

[NF6X]

After the successful experiment in which I replaced the oven's air probe with a thermocouple taped to the surface of the board being heated, I finally completed my modifications to permanently install a contact thermocouple in my oven. Unfortunately, I'm getting severe temperature overshoots, much like with the original air probe. I don't understand why yet, because my previous experiment with a contact thermocouple in the controller's feedback loop worked very nicely.

I bought some K-type thermocouple elements with ceramic bead insulation which should be able to tolerate the heat inside the oven without scorching like the fabric-insulated and fiberglass-insulated probes I've been using for logging. I also got ceramic-bodied thermocouple connectors for use inside the oven chamber, and a regular plastic bulkhead thermocouple jack to reroute the controller's thermocouple input to an external connector. I installed a ceramic jack passing through the front of the drawer, built a plug-in contact probe for installation inside the drawer, and hooked it all up to the controller with an external cable:









I set up a test board, with a second thermocouple also taped to the board for logging the board temperature with my Pax Instruments T400:



Oven settings were:

• PREH: 150 deg. C 01:00
• HEAT: 200 deg. C 01:30
• SLDR: 235 deg. C 00:30
• KEEP: 220 deg. C
• COOL: 150 deg. C

Results were poor:



I tried again with a different contact probe in which I removed the ceramic insulator bead near the junction, in case the bead was absorbing too much IR and throwing off the temperature reading. I also placed both the feedback and logging probes at the same location on the test board. Results were still poor:





I don't understand yet why I got such poor results here. I have some additional experiments in mind. Raw data is all available at:

https://github.com/NF6X/Qinsi_QS-5100_Tests

[NF6X]

I performed another experiment:

• Controller feedback via thermocouple taped to board.
• Temperature logging via another thermocouple taped to board. Logged on channel 1.
• Ceramic-insulated probe plugged into logger channel 2 to log its output.

Channel 1 agreed pretty closely with temperature displayed on the oven's LCD, and its plot looked satisfactory to me. I'll be happy if I can get my oven to reliably produce temperature profiles like this without having to tape a probe to the board being soldered.

Channel 2 showed that the ceramic-insulated probe was reading well below the other two probes. I think this is consistent with the temperature overshoots I saw when using this probe for the controller feedback.



Maybe I just don't understand how to install thermocouples yet? I am under the impression that the ceramic-bodied connectors I bought should have contacts made with the same materials as the thermocouple probe wires, so that the connector temperature shouldn't affect the readings, and the cold junction should remain at the controller board. Is my understanding incorrect? Are the readings so far off because of the hot connector in the wall of the drawer?

[invzim]

Here are my T-962 and QS-5100 teardown videos:

Nice!

Got this one recently.

The build quality as as far as I can tell very nice, I fired it up at home and no foul smells like I read about the T962.

It is NOT as accurate as I hoped, and I too experience pretty severe overshoot.  I don't have fancy measuring equipment, just a simple UNI-T UT136C with the probe that came with it.  It seems to be around 5-13 degrees off at the end of a period, and 14-24 degrees off at the start of a period. 

I'm only going to use it for prototypes, but it WOULD be nice to be able to consistently do 4 layer boards with qfp 0.5mm and BGA parts at home.

Too bad I'm not an atmel guy, as a modified firmware for the ATmega64A(?) that reads the probe a little quicker, and maybe using the fan to fix the overshoot seem very doable.

I'll do dome more experiments and will post what I come up with in terms of playing with the profile.