[quote author="diogoc"]I took a look at your PID code but it is a little confusing. What is your source for that PID algorithm?[/quote]
The source is my brain. :)
It works this way: 1. At a given moment of time we have current temperature (TempAVG), current PWM percentage, and current rate of change of the temperature for this PWM (TSlope). 2. From what we have, the code tries to predict what would be the temperature after a certain amount of time (TempAVG + TSlope), and using PI algorithm on that predicted temperature changes the PWM in order to reach the desired temperature.
Both "TempAVG" and "TSlope" are averages of past data in order to eliminate the noise in temperature measurements.
By definition, the "P" part of the PID algorithm works on the current status, the "I" part works on the past, and the "D" part works on future. So, my algorithm is indeed a PID algorithm, just in a bit more strange form. Someone with a better mathematical skills can even do a formulas on this. I just found out that it works faster and with less overshoot this way.
Moreover, many algorithms doesn't show the real temperature on the display, and the user have a feeling that the temperature is stable and without over/undershoots. I personally don't like this. My code shows the real measured temperature (averaged over the last 8-16 measurements), and you can see all the overshoots, undershoots, reaction of the code and so on. This way everyone can see what actually happens at the moment, not what the code developer wants to happen.
[quote author="diogoc"]Another question.. I realized that the heater is on or off during the whole mains period. That is, you do not use a PWM with a fixed period and varying on and off times, right? If I use a common PWM the interval between each PID calculation should be at least greater than one PWM period?[/quote]
Yes. I am using a sort of "fixed on time, variable off time" algorithm. It is simple and works great.
You have a counter (PWMCNT), and on every mains period a PWM is added to this counter. If the counter becomes more than one, you switch the heater on and subtract one from the counter.
For example, consider PWM duty of 63% (0.63). At the beginning the counter is initialized to 0. Here is what happens: 1. Counter = counter + PWM duty. (0 + 0.63 = 0.63), the heater is switched off. 2. Counter = counter + PWM duty. (0.63 + 0.63 = 1.26), the heater is switched on, and one is subtraced from counter, so the conter is 0.26 3. 0.26 + 0.63 = 0.89 -> heater off 4. 0.89 + 0.63 = 1.52 -> heater on, counter = 0.52 5. 0.52 + 0.63 = 1.15 -> heater on, counter = 0.15 6. 0.15 + 0.63 = 0.78 -> heater off 7. 0.78 + 0.63 = 1.41 -> heater on, counter = 0.41 8. 0.41 + 0.63 = 1.04 -> heater on, counter = 0.04 9. 0.04 + 0.63 = 0.67 -> heater off 10. 0.67 + 0.63 = 1.30 -> heater on, counter = 0.30 11. 0.30 + 0.63 = 0.93 -> heater off 12. 0.93 + 0.63 = 1.56 -> heater on, counter = 0.56 13. 0.56 + 0.63 = 1.19 -> heater on, counter = 0.19 14. 0.19 + 0.63 = 0.82 -> heater off 15. 0.82 + 0.63 = 1.45 -> heater on, counter = 0.45
and so on.
Why I am doing it on mains period? Because the temperature measurements are made on exact mains voltage every time, this way you eliminate the noise from the mains voltage in the measurements. And, you don't have to filter the transformer voltage with large capacitance - the electronics becomes much more compact this way, and cheaper too. The filter capacitors for 70W iron would be pretty big if you want to use real DC. And, the transformer is not loaded with high frequency, so it is happier (colder, quieter) this way. :)
Of course, you can use classic PWM, but if you continue to use rectified and unfiltered voltage and don't do the switching on the zero cross of every mains period (note - full period, not half period), the load on the transformer would not be symmetric for positive and negative part of the period, and the transformers don't like this - they will be more noisy (some transformers can be heard even with ideal symmetric loads) and will become hotter.
I hope these answers clears the things a bit. If you have more questions, I will be glad to answer them also.
I could not agree more! :) That's why I am not doing compensation at all. I am using 22 degrees for cold junction and that's it. If it is needed, I can make this adjustable, but it will not make much difference anyway. Measuring the PCB temperature is one step further of course, but again, who cares about 10 degrees difference?
I don't really agree as you have digital thermocouple digitizer chips with cold junction compensation .. in volume they are cheap and compared to the cost of the station it's really insignificant difference in price. It's big difference for you and me as we have to pay 20$ for maxim chip or 2$ for voltage reference and transistor :) but in volumes those prices go down like rock.. (not to mention that you can do cold joint compensation with a 0.1$ ntc like I do on my driver ..)
[/quote]
Yes, but cold junction of the TC is on the TC terminals, not the PCB terminals. If you are measuring the temperature at the PCB, you are compensating the temperature of the PCB, not the cold junction one. And the PCB definitely can be hotter than TC cold terminals, especially when you put it the same box the power transformer is in.
Yes, it is not a problem to put a diode ot PTC or NTC in the handle to measure the temperature there, but the very existence of the compensation is one more thing to care for. My controller is without compensation at all. It works this way too. I am compensating for 22 degrees.
You should not care for this kind of problems when working with RTD. You have one resistance there, and nothing more. You don't care about the cables, cold and hot junction, you don't have to compensate for anything, but just measure one resistance.
I agree that the TC solution with all compensations will still be cheaper than the RTD.
For me it is not "mid to low end". Don't get me wrong - i have not tried it and I will not. It is too expensive for me anyway. I've also heard people not satisfied with Weller. It is a fact!
It is easier from controller's point of view to work with RTD. You don't have to measure nothing more but one resistance. It is definitely a better solution than TC, but it is expensive. I suspect that cartridge type tips are using TC instead of RTD because it is cheaper this way. For example the HAKKO T12 tip costs around 20-30EUR here, JBC around 25EUR. If they had an RTD in them, the price would be even higher.
[quote author="arhi"] Weller uses in cheap irons TC and in expensive irons RTD [/quote] Are you absolutely positive Weller RT and NT tips are using RTD and not TC? What about this thread then? viewtopic.php?f=19&t=3583
[quote author="diogoc"]19 uV/ºc is a D type thermocouple, but I think it is expensive so Hakko would not use such type of thermocouple. From the standard types of thermocouples the N type is the closest, but even so the error is too large.[/quote]
I don't think is an expensive type of thermocouple. Most likely, it is a standart cheap thermocouple, but something in the construction of the tip changes it's output - for example the contact between heater and it's terminals which again is more or less a thermocouple.
I have made some measurements. Is is some kind of non-standart thermocouple in there. It is between 18 and 19 uV per degree celsius. This is non-standart value and I don't know what type of thermocouple can this be.
...but I am using a chinese clone of T12 tips. I don't know how the original ones perform. I will buy one original and measure it also. But it's price is 6-7 times up, compared to chinese.
P.S.: Just measured my other soldering iron's thermocouple (HAKKO 907 or something like this). 52.7uV per degree celsius. This is more like an iron-constantan type J thermocouple.
[quote author="arhi"][quote author="cowana"] Support for virtually all 24v soldering irons [/quote]
not really. most new (all new quality ones) can't be driven with this type of driver. not a big deal 'cause all those new ones don't really have cheap pencils so there's not much point in making a cheap driver for them .. but just drop the "all 24V" as someone might get himself a $150 pencil just to find out he can't use it[/quote]
Chinese HAKKO T12 clones can be bought for under $4. ;)
[quote author="neslekkim"] Is it made for some specific enclosure?[/quote]
Yes. I've got a BlackJack Solderwerks BK3000LF station, and I made the PCBs to fit in it. I made a different back plate with hole for the handle sensor 6p6c jack. Here's how it looks:
There is nothing left from original controller in it. Only the power transformer.
The sensor is made to fit in the original stand also (shown on the picture). I only had to drill 4 more holes on it (2 for PCB mount and 2 for IR transmitter and receiver).
Wish there were an inexpensive calibration for the Hakko irons. The FX888's calibration system costs as much as the iron :-P[/quote]
1. get 63/37 solder 2. put some on the tip 3. set temperature to 182 degrees 4. calibrate it so the solder is solid at that temperature 5. set temperature to 184 degrees 6. calibrate it so the solder is melted at this temperature 7. repeat steps 3 to 6 until you get solid solder on 182 degrees and melted solder on 184 degrees.
It is a bit tricky to set it solid on 182 degrees and melted on 184 degrees, but I can calibrate it this way in around 10 minutes. Calibration is done with the tip that is used most frequently. There will be difference in 2-5 degrees between different tips.
If it is too tricky for you, you can use 180 and 186 degrees instead, but once you get used to it, it is easy.
After this, if you are using lead free solder, you must clean the tip from 63/37 solder. This is easy also - put some lead free solder on tip, chen remove it, then put some more, then remove it again, and so on. If you repeat this 3-4 times, no lead from 63/37 solder will remain on tip.
The cost of all this is almost zero.
The reason to use 63/37 non lead free solder is because it is eutectic alloy - it melts and solidifies at exactly 183 degrees. Lead-free solder is not eutectic alloy and progressively softens between two temperatures, so it cannot be used for correct calibration.
