Marco Reps JBC project

[tester43]

Hi,

Here is the video from Marco Reps doing the JBC station. I am working on hobby project for HAKKO tips version - similar but different.


Here are his schematics:
https://drive.google.com/file/d/0B6S_PcWWM1YlY1lORkhDcTAzdEU/view

I do not get one thing:
He is driving the heater using 2 N-MOSFETS source to source. That's normal.
If he does not want to heat then Mosfets gets no voltage on gate and circuit becomes open = no power.
BUT there is always potential on heater wires  (AC1 on schematics) - i mean it's present at the point where thermocouple should be read.

How does he succeeds in reading TC voltage if there is voltage from AC on TC connector?
Or do I get it all wrong...

[kmike]

Hi,

I had the same problem and used the attached design for my dual iron JBC controller, see screenshot

From my experience You can only measure the TC voltage on zero crossing, but this is not a problem when You want to switch the heater on near the zero crossing

br,
mike

*edit: typo

[tester43]

this acs711 current sensor. The trick is that you are not cutting the power off but basically you wait for zero AC and when there is zero current you measure the voltage from TC? Am I correct?

edit: If I am correct then it sounds like a very noisy environment to read ~1 volt signal. It can be anything - interference, noise of any kind still present for a moment during zero, or actual TC voltage... sounds ... complicated in terms of synchronisation

[kmike]

No, the current sensor is there to correctly calculate the power going to the soldering iron, the voltage is also measured in an other point of the schematic.
The heater has a resistance around 3 Ohms, so You have to switch off at the proper time (conduction angle), so You dont deliver more than 140W of power to the iron.
I find it cheaper to do more measurements and calculations (with a more expensive MCU) than to destroy cartridges 

I have a zero crossing detector in the circuit wich triggers an interrupt, then I do the sampling of the TC.

br,
mike

[tester43]

"I have a zero crossing detector in the circuit wich triggers an interrupt, then I do the sampling of the TC." - so.... you are never cutting the power, you just wait for "best" moment and read what's there?

[kmike]

Here is a short description what the firmware does:
- wait for zero crossing
- ZC interrupt:
  - measure half period of the input AC
  - measure TC voltage
  - calculate the maximum conduction angle for the configured power limit with the required formula
  - calculate the switch off time for the next half cycle
  - switch on heater, start a timer with the switch off time

- timer limit reached:
  - measure voltage and current going to the iron to get the heater resistance for the power limit calculation
  - switch off heater
  - wait for the next ZC

If You want 140W max power (for a T245) than You have to switch off at around 55% of the half cycle

edit: If You can wait 2-3 weeks, I will publish my whole design, I am currently cleaning up the project as in its current form it cannot be published...

br,
mike

[tester43]

"measure half period of the input AC" by this you mean TIME? You measure miliseconds? and then do the actual measurement?



btw. congrats for winning over Sweden

[kmike]

Yes, I measure the time difference between two ZC interrupts (so the circuit can be used with 50Hz or 60Hz input)
I wanted to try a new for me micro for this project and settled with a TI tm4c123, so far I am impressed with it and at 50MHz it has enough processing power to do all the calculations
Some calculations and measurements could be left out probably, this is just a fun/education project for me, not a commercial design. Since the BOM cost totals at around 45€ for the two channels (I have a T245 and a DR560 iron) I am happy with it

The firmware description from my previous post is not 100% accurate, I do a lot of calculations when the heater is off, it was meant to give You an idea what my firmware is doing

Thanks

br,
mike

[tester43]

thank you
I am going with cortex m3 (blue pill) because in arm i can have easily all the interrupts i could possibly want + processing power.
Tomorrow my p-fet will arrive, and I will be able to easily turn off all power, calmly wait for 1ms, read everything I need, wait for zero and turn the heater on again.
I am working on cheaper vendor - hakko T12. Maybe when I see this working I will think about jbc...

[kmike]

The tm4c123 is a cortex m4, has a lot of timers (32 and 64 bit) and two independent ADCs wich is nice when You want to measure voltage and current at the same time

There are two inconvenient things with JBC: heater and TC ground are common (thats why You really only can measure at ZC) and the TC has a thermal coefficient of a few µV/°C
The small coefficient means a lot of amplification wich introduces a lot of noise

br,
mike

[tester43]

in hakko TC and heater are in series...

about the noise - that's exactly what I meant - it would be really much easier if it would be possible to completely shut off all voltage. I will do this using p-channel ; Found something with low resistance.
Anyway in marco and your solution there is a constant use of mosfet body diode - it also produces a lot of heat.

[kmike]

The mosfets are not even getting warm, you should not worry about that. I don't have any heatsink attached on my prototype board wich uses the TO-220 version of the transistors and both are cold. With the totem pole driver I have turn on/off times of about 3-5µs, which keeps the transistors out of their linear region

Why do You want to measusre the temperature before the zero crossing? If You want switch on at ZC, You don't have any real advantage measuring sooner. I have tried switching on at the middle of the half cycle, but my transformer did not like my idea

Can You share your schematic of the power stage?

br,
mike

[tester43]

i am fine with measurement in zero. I do it like you: wait for zero interrupt and measure. But I found that my zero signal is not perfect (see picture). It has got a long rising time. I did not wanted to fine tune regarding this time so my idea is following recurrence:

- heat as long as previously calculated
- wait for ZC interrupt
- disable MOSFET
- give 1ms of time for all voltage/noise to settle
- read TC voltage
- do the regulator logic -> number of cycles of heating needed
- go to first step

my power circut is:
- typical N-Mosfet application: something like this

- transformer (temporary 20W for the moment)
- Graetz bridge without filtering (see normal expected picture)
- N-Mosfet is driven by uC by 4n25 optocoupler. Input voltage to control mosfet gate is generated by LM7812 so gate is controlled by 12V;

[kmike]

I had the same problem with the ZC detector, I simply implemented a delay to get to the "real" zero crossing using a spare 16bit timer, at first I calculated the delay with the oscilloscope, later just programmed an auto-cal function in the firmware for this.

If you use a Graetz bridge, You don't have any advantage of using AC, You clould also filter the input voltage and use DC PWM for the heater, also now You have the losses in the 4 diodes of the bridge, not just the two body diodes of the FETs (some joking: I have paid the full price for the FETs, so I also want to use the body diodes )

The main advantage of using AC is to avoid electromigration.

br,
mike

[tester43]

with bridge I have single mosfet not like you rich people   "uh oh i will not use bridge, I want to show off and I will put there TWO mosfets!!" 

[kmike]

but the forward voltage drop of the body diode is smaller then that of a full wave rectifier (1 vs 2 diodes conducting), so in the long run the cost saving on my electricity bill justify the use of 2 mosfets! 

In a few years I can even buy another mosfet from my savings! 

br,
mike

[tester43]

with my rectifier I will produce heat. Winter is coming... cost cutting on heating. I win 

---------------------
there is one thing i do not understand - and its in every diy project of any soldering station with on internet I could find.
And I know that it's just some basic concept that shows my poor understanding. I am playing with simple projects for over a year now, but basics still chase me.

If TC is connected to transformer and resistance of heater is 3 ohms (in jbc) how is it possible that this super small voltage from TC does not get "grounded" immediately. Especially that to protect input of omAmp I will have to use resistor and zener to clamp opamp input voltage to 1-2 volts.

[kmike]

As you can see in the attached picture, power goes to C2 and you measure the Temperature between C1 and C3. When the mosfets are turned off or at ZC there is no voltage on C2 and since C1 is always grounded (and has to be always grounded because you don't want any voltage on the tip of the iron ever), you can safely measure between C1 and C3.

With a low side driver you would have 24V on the tip of the iron when the transistor is switched off, wich could destroy the device you are soldering (also ESD problems)

The TC amplifier needs clamping diodes on its input as you have written and has to be reasonably fast to recover after the voltage on C2 is switched off. The recovery should not be a problem when using AC, since at 24V AC the transistor should only conduct about 50% of the half cycle (93° conduction angle), with DC PWM you have to check it yourself.

PS.: I spent nearly every winter in the Netherlands since the last 4 years (work), it is not even cold there

br,
mike

[tester43]

correct me please if my understanding is wrong. I took parts of your image to draw how i understand it. If "yes" then I have a next question )