Reflow Oven Controller - Attiny 461-based

[wraper]

Diode is a wrong type of temperature sensor first of all. And there is no guarantee it won't die after a few dozens of reflow cycles.

[wraper]

I normally work with Eagle's net classes to ensure spacing between the high voltage parts. This works also fine with copper planes, see attached files. This also has the big advantage that the DRC warns you if you route traces too close to each other.

As a general guideline I'm using https://www.mikrocontroller.net/articles/Leiterbahnabst%C3%A4nde

3mm between L and N
6mm between L/N and Low-Voltage
6mm between PE and everything (not mentioned in the article, but should be save).

I don't see any good reason to connect PE with GND, I'd rather leave it floating. By referencing GND you remove all your galvanic isolation without any reason.

What is the purpose of PE if you leave it unconnected? You are left without any protection it should provide. What isolation? for what reason GND should be floating against oven enclosure. And if you don't connect oven enclosure to PE too, you are an idiot. PE shouldn't be connected only if this is double insulated device which usually is not the case for high power devices.

[moemoe]

What is the purpose of PE if you leave it unconnected? You are left without any protection it should provide. What isolation? for what reason GND should be floating against oven enclosure. And if you don't connect oven enclosure to PE too, you are an idiot.

I never talked of leaving the oven enclosure unconnected.

But it's still legitimate to leave your GND floating. In fact, I don't know any power supply or lab supply that has GND referenced to anything, so all devices connected to them are also floating. Are all these designers idiots?

[fcb]

What is the purpose of PE if you leave it unconnected? You are left without any protection it should provide. What isolation? for what reason GND should be floating against oven enclosure. And if you don't connect oven enclosure to PE too, you are an idiot.

I never talked of leaving the oven enclosure unconnected.

But it's still legitimate to leave your GND floating. In fact, I don't know any power supply or lab supply that has GND referenced to anything, so all devices connected to them are also floating. Are all these designers idiots?

Huh?  All my bench supplies, and all those I have come across in my many years have the chassis connected to earth!  Please give the model number of the ones you have found that don't have an earthed chassis so I can be sure to avoid any labs with them in...

[moemoe]

I never talked of leaving the oven enclosure unconnected.

But it's still legitimate to leave your GND floating. In fact, I don't know any power supply or lab supply that has GND referenced to anything, so all devices connected to them are also floating. Are all these designers idiots?

Huh?  All my bench supplies, and all those I have come across in my many years have the chassis connected to earth!  Please give the model number of the ones you have found that don't have an earthed chassis so I can be sure to avoid any labs with them in...

It seems you're confusing GND/Enclosure/PE.

To clarify:
GND is the low voltage side's GND/0V
Enclosure is the metal outside, which should be always connected to PE, of course
PE is the green/yellow wire, or whatever color it has in your country, connected to "real" earth outside of the house/N in your distribution

[wraper]

I never talked of leaving the oven enclosure unconnected.

But it's still legitimate to leave your GND floating. In fact, I don't know any power supply or lab supply that has GND referenced to anything, so all devices connected to them are also floating. Are all these designers idiots?

In the lab supply negative terminal isn't connected to the ground because you can use the output in various ways. For example, connect multiple PSUs in series. There usually is earth terminal too with a metal plate included which you can attach between negative terminal and earth.
Laptop power supplies usually have negative contact connected to the earth. In the the PCs GND is connected to the PE too. Almost every oscilloscope excluding portable ones. If negative terminal isn't connected to the enclosure (PE) in the PSU itself, doesn't mean it shouldn't be grounded. You may want to make one grounding point somewhere else to avoid ground loops.

[moemoe]

Laptop power supplies usually have negative contact connected to the earth. In the the PCs GND is connected to the PE too. Almost every oscilloscope excluding portable ones. If negative terminal isn't connected to the enclosure (PE) in the PSU itself, doesn't mean it shouldn't be grounded. You may want to make one grounding point somewhere else to avoid ground loops.

I just measured all my Thinkpad Power supplies - none of them hast GND and PE connected.

On my battery charger (BC700), Bat- is not connected to PE (or, to be more precisely, the external power supply doesn't conenct them, as it has a Euro plug).

On my printer (Brother MFC), the Network/USB shields are not connected to PE.

[wraper]

I just measured all my Thinkpad Power supplies - none of them hast GND and PE connected.

On my battery charger (BC700), Bat- is not connected to PE (or, to be more precisely, the external power supply doesn't conenct them, as it has a Euro plug).

They are double insulated devices.

as it has a Euro plug

Europlugs are only designed for low-power (less than 2.5 A) Class II (double-insulated) devices that operate at normal room temperature and do not require a protective-earth connection.

[wraper]

On my printer (Brother MFC), the Network/USB shields are not connected to PE.

They likely are not connected to GND directly but through the capacitor and likely bleeding resistor too.

[eneuro]

Diode is a wrong type of temperature sensor first of all. And there is no guarantee it won't die after a few dozens of reflow cycles.

Anyway, f someone wants sense temperature at low cost why not? 

Accurate Temperature Sensing with an External P-N
Junction
http://cds.linear.com/docs/en/application-note/an137f.pdf

Diode-Based Temperature Measurement
http://www.ti.com/lit/an/sboa019/sboa019.pdf

However, I'm interested if there is some kind of diode testing in this project MPU software, to ensure we hav enot faluty diode, eg. at room temperature 25*C we expect on 1N4148 some voltage, so if this diode is damaged I;d like to see the massage on LCD or "SOS" diode health status blink to let us know tha we ahve take another 1N4148 and replace inside reflow oven 
So, there is no problem that a few cents 1N4148 fails, but some kind of management needed

Another thing is, how temperature sensing is implemented there? Using formulas mentioned in linked application notes or by experimental data?

I never tried this method of higher temperatures sensing, but I like this idea 
There is -2mV/C forward voltage drop mentioned in this fan controler based on 1N4148 Vf sensing:
Using 1N4148 As A Temperature Sensor
http://electronics.stackexchange.com/questions/80131/using-1n4148-as-a-temperature-sensor

I'm interested in using those cheapy 1N4148 as temperature sensors in my custom reflow owen based on hot iron ceramic heater, so I'm thinking to close such 1N4148 diode inside reflow oven and starting from room temperature ~23*C  made measurements of Vf and temperature up to lets say 100*C, than plot this data to see relationship between temperature as this Vf changes and make: linear or maybe better parabolic approximation to xtrapolate above 100*C up to 300*C.
How do you think, such approximation of real data and extrapolation to higher temperatures makes sense or we have to do all this "hard" math and use delta Vf measurements (using eg. two constant current sources) to be able to estimate sensed temperature as mentioned in linked above application notes? 


Of course I will try sense temperature inside kitchen oven using this calibrated 1N4148 temperature sensor and we'll se if I will get something which was preset at this oven temperature scale 

[wraper]

Anyway, f someone wants sense temperature at low cost why not?

Because one fried board/batch of boards very likely will cost more than a decent temperature sensor (+ coresponding circuit) like PT100 or thermocouple . It is just not wise idea (and waaay over spec) to use semiconductors at so high temperature.

[eneuro]

It is just not wise idea (and waaay over spec) to use semiconductors at so high temperature.

Semiconductors are used at such temperatures since you reflow them in ovens, maybe not so many times, so as I said probably we need to manage this and lets say find amount of reflows that will degrade its performace to the point that they will be useless
I'm thinking about something like arbitration logic in my solution-use three those cheapy 1N4148 diodes inside reflow oven, than connect to three MPU ADC pins and choose only two closest Vf voltages and skip third reading, so when one of those diodes fails during reflow it shouldn't affect this process.
When one of the diodes gives temperature readings far away from other two than fault diode blinks and output which diode is cause of this fault, so we know which one we need to replace...
I do not know if it will work, but I'll test it next week to see what happends later in kitchen oven

Update: If we're talking about specs we can find in absolute maximum rating for 1N4148
http://www.nxp.com/documents/data_sheet/1N4148_1N4448.pdf

Tstg storage temperature -65*C .. +200*C
Tj junction temperature 200*Cmax

So, yes 250*C is above its specs, but I'd like to see when (after how many made cakes in kitchen oven) this 1N4148 fails 

[free_electron]

It is just not wise idea (and waaay over spec) to use semiconductors at so high temperature.

Semiconductors are used at such temperatures since you reflow them in ovens,

and they undergo even higher temperature when they are being produced , BUT : THEY ARE NOT POWERED UP ! No electrons are flowing through them. !!!

you will create electron migration and other strange effects if you do this.

[wraper]

It is just not wise idea (and waaay over spec) to use semiconductors at so high temperature.

Semiconductors are used at such temperatures since you reflow them in ovens

And if you look in the datasheet for the soldering info, there will be something like 3-5 times for a few seconds at peak temperature.

Tstg storage temperature -65*C .. +200*C
Tj junction temperature 200*Cmax

Should be some special flavor of 1N4148 from NXP as in other datasheets Tj is 165 - 175^OC

[eneuro]

This diode will sit inside the oven right ?
What will you use as solder to connect to its wires ?

I want use this small custom PCB to hold diode in reflow oven


No need to solder anything


I've in my toolbox something like this for PCB and heatsinks assembly

[eneuro]

Tstg storage temperature -65*C .. +200*C
Tj junction temperature 200*Cmax

Should be some special flavor of 1N4148 from NXP as in other datasheets Tj is 165 - 175^OC

I'm more concerned for the moment about this maximum permissible continuous forward current as a function of ambient temperature   


Unfortunatelly, I can't see schematics of this reflow oven controller since there is no PDF version included (I use different PCB design tools), so I have no idea how much current is provided througth this diode 

Anyway, looking into another schematics attached below, I looks like while 10V zener was uses and there is at least 10k resistor from diode to gound, less than 1mA will flow, which seams reasonable since from NXP graph above for temperature ~175*C they mention ~25mA maximum permisible continuous forward current, so close to 200*C, lets say 199*C we get ~1mA from proportion.

However, maybe we could quite safelly use short pules of 1mA current at 200*C (and beyond) only when we want sense temperature, so hopefully maybe we could be still at quite safe  current levels, since probably pulse current allowed will be higher than continous 

[free_electron]

This diode will sit inside the oven right ?
What will you use as solder to connect to its wires ?

I want use this small custom PCB to hold diode in reflow oven


No need to solder anything


I've in my toolbox something like this for PCB and heatsinks assembly

not a good idea... poprivets are aluminum... not guaranteed a good electrical contact. plus you are probably creating four thermocouple junctions  as well as you have dissimilar metals... (cable -> board - board - diode cathode , diode  anode ->board - board -> cable.

i would try this out first before you develop this thing... my gut feeling say 'trouble'....  even if it works that diode die due to electromigration effects.

you are running the thing waaaay too hot. and powered up

[eneuro]

(cable -> board - board - diode cathode , diode  anode ->board - board -> cable.

I didn't show you how diode terminals will be connected with copper wires, so this what you wrote is only your guess, not mine 

So, I can tell you and reveal more details, that aluminium rivets shown on those pictures of course will not conduct temperature sensing current, but... will work as a metal glue, which will press 1N4148 ~0.6mm in diameter copper terminals with ring copper connectors together, so there will be copper to copper connection 

BTW: I was pretty sure, that someone will notice this you pointed out and what can I say, I've shown those aluminium rivets ... in the hope someone will try to find issue there 
I'm very happy, you noticed this and this maybe very usefull information-it could be a trap for young electronics players 

[free_electron]

and what will you use as wire ? it better be fiberglass insulated or it will melt ..

i am still convinced this will not work right and the sensor will die prematurely.

[eneuro]

We'll see what happends
I've made quick prototype PCBs, to be able install  those temperature "sensors" inside my small reflow oven and another one in kitchen oven:


I think, I'll connect such three sensors to ATTiny85 ADC's and make I2C interface on two other pins to be in final solution have I2C temperature sensor ready to connect to another MPU, so basicly this temperature sensor MPU will do nothing more than monitor those 1N4148 diodes, compute estimated temperature, do arbitration logic  and maybe store in MPU EEPROM something like temperature accumulated over time in 0.1 second steps, to be able estimate how much average heat those diodes experienced in their lifespan while used during reflows processes until one of them failed and we left with two good diodes, but need report fault condition, so I2C interface will have faulty bit for each of those diodes, estimeated temperature of each of them and at another adress final temperature after arbitration logic rules applied-this is information reflow oven main MPU will look for, but other PC i2C optoisolated master will log some performance data on disk to futher analysis at the same to see in real time on PC what is going on inside reflow oven 

I hope, that not all diodes will fail at the same time, so I will be able catch the moment when one of them fails 

[SaabFAN]

I'm using the 1N4148 as the temperature-sensing diode. It is connected to VCC via a 1K Resistor and the ADC. It will be connected with crimp-contacts and a 300°C rated cable. Or the pop-rivet method. Kinda like that idea
Idk how long the diode will last, but I'm not planning on soldering dozens of boards each day. And it is the easiest way I could find to measure temperatures that high without paying a fortune for the sensor.

I haven't really started writing the firmware for this thing yet, as I was able to manually solder a few boards with the oven (turning the heaters on and off with the switches on the oven, while monitoring the temp with a Multimeter and its thermocouple). Adding a sanity-check for the diode is a good idea btw. I will certainly do that. Not sure how exactly, though
To get a reasonable accurate temperature-reading, I'm implementing a calibration-scheme with fixed measurement-points at 25 or 50°C intervals. Values between these intervals are interpolated.
Completely frying any board should be prevented by the thermo-switch inside the oven, which turns it off at a temperature of about 250°C. And also by watching the whole thing when in operation.

About the interface: I'm planning to use these widly available 2-line 16 Character LCDs combined with a i2c-Port Expander,.
Input is provided by buttons or switches, whatever I can find in the basement, which are also connected to another i2c-Port Expander. This allows up to 8 buttons, which are read in regular intervals.

Attached is the schematic.

[fcb]

AFAIK no commercial or even hobby level reflow oven uses semiconductor junction temperature sensing - it seems really daft as thermocouples are cheap, come with long leads and are well characterized.

The diode is not designed for cycling at the temperatures your going to need - at least not for long.

[eneuro]

It will be connected with crimp-contacts and a 300°C rated cable. Or the pop-rivet method.

In some application pop-rivet method might be good, but in the case of my prototype reflow oven based on flat heater metal plate I realized that it will be easier assembly this temperature sensor like this

Of course instead of normal wires used there for temperature calibration only and to make quick test, in final design high temperature resistant wires should do the trick 

I've installed four such sensor in my reflow oven concept (based on flat iron heater element, but unsure if I will be able get decent temperatures, so in final design probably high temperature resistant concrete used in fireplace will be used instead of wood, but reflow control software will be the same)


Today realized, that I can disable ATTiny85 Reset pin and use it as  4th ADC, so I've four 1N4148 based temp sensors inside this oven (this temp sense MPU will have bootloader capable after restart to reflash using only one pin, so no need to use huge bloody few pins ISP connector and Reset pin no longer needed- so I have 4 ADCs available and two pins for software I2C slave implementation)

NOTE: Those orange and black pipes... are nitrogen inlet and outlet to replace oven atmosphere during reflow with nitrogen of course
I will need a few more ATTiny85 in SO8M package to controll everything, but buying more those MPUs t other projects too, they will be cheap, so it will be fun connect them together to one I2C network.

Highest priority now has my ATTiny85 custom 1 wire bootloader, since I've a few projects where I'd like to use it, so bootloader software development time right now
Can't wait to reflash this tiny monster with versatile, password protected 1 wire bootloader to be able disable Reset pin and implement i2C interface for those four sensors.

[timofonic]

Any news?

[SaabFAN]

I haven't found the time to install the controller into the oven. Partially, because manually controlling the temperature worked pretty well for several boards so far - switching the heaters on and off while monitoring the temperature with the Multimeter.