Measuring temperature from a T12 soldering tip

[MikeLemon]

Hello,

I'm interested in making a T12 tip based temperature station powering it right from the AS power using a Triac and max6675 type K thermocouple amplifier

I already took care of the powering part of the circuit including zero crossing sensing and activating the triac with an octocoupler but sins the Thermocouple of the actual tip is connected in series to the heating element I'm not sure if it is possible to hook it right up to the sensor inputs of the max6675.

Can the 15-20V 50Hz RMS AC connected right to the max6675 chip hurt it?
if not what is the way around?

Thanks for your time.

[KL27x]

I've started down the same rabbit hole.

What I have done is use an SSR to connect the opamp to the T12 + lead. With timer synced to the zero crossing, it opens and closes at the appropriate time to prevent high voltage from reaching opamp input. I also have clamping (low FVD schottky) diodes on the opamp input to keep the voltage +- 0.3V from ground rail.

... I can't say I got everything working just peachy. But I haven't burned out my opamp, yet. IIRC, I have also tried clearing the input from any leakage/transient by temporarily shorting it to ground with a FET before the read, and only reading it on the one direction. But maybe I need FET going both directions... dunno. Anyhow, my reads are getting messed up while the heater is on, even though the read occurs in the zero crossing when the triac is off. Something (Leakage or emi or residual voltage?) is messing everything up.

The guy behind "Black Jack Solderwerks" made a universal controller which can use T12 tips. If you google, he has schematics available for download for free. In forum post he mentions he has to protect the opamp from high voltage, but I haven't looked at his schematic. I assume he can use a regular signal FET to switch the opamp in/out of circuit, since he uses rectified unfiltered DC rather than AC.

*Quick peek at the datasheet shows absolute max V  Vss-0.3V to Vcc+0.3V on the inputs for your MAX chip. Max supply voltage of 6V. So I bet +-20ishV*1.7 will cause a problem, yeah. I imagine it will short the T12 heater supply to Vss/Vcc up until the ESD clamping diodes in the chip burn out in maybe <100mS. Then the heater will work at full voltage again... and your MAX chip will be dead.

If you get this thing working on AC, reading only at the zero crossing, be sure to post your success. I haven't seen anyone else doing it this way. All the clones run off DC. SparkyBG's version uses the aforementioned rectified but unfiltered DC. I have never seen anyone share details of their actual FX951/950. Apparently, youtubers only teardown cheap Chinese clone stations. And even then, with their racks of $10,000.00 worth of testing gear, the height of analysis boils down to looking for shoddy soldering joints to laugh at. LOL.

I have reverse engineered one of the cheapo (DC) T12 boards, and it (apparently) can read (at least approximately) the thermocouple with just an lm358. I say apparently, because after I got done drawing it out and plugging it in, it doesn't work, lol. So I never got to see if it was just some gimpy power control or if it was actually achieving thermostatic effect. Took a month to get, so.. lol.

[DBecker]

Most of the cheap T12 controllers use a LM358.  Complete with it's relatively high offset voltage and noise, and edge-case performance with inputs near V-.

It works OK, largely because the boards and tips are individually calibrated.

The typical input circuit is a 10K resistor with a diode clamp to the '+' opamp input.  Some even omit the diode clamp (!).  Most use two gain stages totaling 50x-100x, but some use only one stage (!!).

If you do use an AC waveform, control in increments of full cycles not half cycles to avoid a persistent DC bias.  And skip a full cycle when doing a measurement, even if you nominally want 100% power.

[DBecker]

I have reverse engineered one of the cheapo (DC) T12 boards, and it (apparently) can read (at least approximately) the thermocouple with just an lm358. I say apparently, because after I got done drawing it out and plugging it in, it doesn't work, lol. So I never got to see if it was just some gimpy power control or if it was actually achieving thermostatic effect. Took a month to get, so.. lol.

It probably died the first control cycle after you plugged it in.

The cheap T12 controllers I've seen use a P channel MOSFET on the positive supply, with the gate controlled by the full applied voltage.  That works (doesn't blow up immediately) with an 18V or 19V supply from a laptop power brick, but blows up immediately if you believed the spec (e.g. "30VDC").  Some claim to even support 24VAC, which could hit 50V when switched off at the wrong time.

[KL27x]

^I noticed this setup on mine in time to avoid that mistake. NPN pulling down the PFET gate with just a resistor in between. I set my PSU to 19.5V, which in addition to the voltage drop of the NPN should have been fine.

I have never seen a PFET with max VGS higher than 20V, and I looked this one up. 20V. Even if the FET were blown, it's not the problem. I'm getting no signal anywhere. Just a green LED turns on and the circuit draws like 4mA.

Most of the cheap T12 controllers use a LM358.  Complete with it's relatively high offset voltage and noise, and edge-case performance with inputs near V-.

It works OK, largely because the boards and tips are individually calibrated.

The one I have is doing something very interesting (if it actually works, which I'm assuming it does; I probably broke this one). There's no micro or timer. The lm358 is doing EVERYTHING. It's a strange circuit and I'm still not figuring it out. Perhaps when I get time to put it into proper digital format I'll post it for the forum to have a look. I think the second opamp is used as a comparator, and there are some caps for the timing. The control is just a 10k pot, and there's no display. Just a RG led.

Total active parts: 7805, npn transistor to drive a PFET, lm358 dual opamp. That's all she wrote.

[DBecker]

There are a handful of PFETs with a +-25V gate.  I replaced the blown MOSFET on that board with an AO4413 that I had around, and stuck a note on it.  After that I always checked the circuit before trying out a new controller.

That still limits the input to less than 25VDC, and then only if you are willing to run a gate at the maximum voltage.  I never run close to the limit on my own designs, since there is always an inductive overshoot.  It's silly that they designed it that way, since it would only use one more resistor to have a controlled gate voltage.

The board with *only* a LM358 is a simple thermostatic control.  A controller has the potential to be considerably better with even the cheapest microcontroller, although most firmware does little better.

[KL27x]

The board with *only* a LM358 is a simple thermostatic control.

If it were true thermostatic control, I'd be thrilled it could be done with a circuit so simple. My hunch is telling me something fishier than that is going on.

Anyway you slice it, I see 3 people and only 2 chairs. This is quite the juggling act if it can produce true thermostatic control. 

[KL27x]

Here's what I got so far on it. This is pretty much IT. The FET drive is an NPN, as I discussed. And there's a 7805 to create the 5V rail. This is what I believe to by the entirety of the important bits of the control circuit, minus the LED indicator. The green LED is on all the time, and the red LED switches on with the PFET. Simple enough. The squiggly line on the left edge is the wiper of the control knob pot, which is connected between 5V and ground. (The "20V" at the top left is incorrect. It is the positive end of the TC and/or the output of the PFET. So it's a conditional 20V, it's not tied to the 20V rail. )



If this can read a thermocouple in one stage, I don't see how that is possible. Its not wired as an amplifier, at all. There's no feedback between opamp A output and its inputs. This circuit had me scratching my head until I finally just decided to plug it in and measure some signals. And that didn't go anywhere. I probably have some major mistake in the schematic. Not sure though why it was DOA. All I did was probe it, remove the opamp, and put it back on.

[DBecker]

The first stage is a comparator. The second stage is a one-shot.

When the temperature (thermocouple output) is lower than the setpoint, the R-C timer is reset through the diode.  This trigger one pulse period of power to the heater.

This implements simple thermostatic control with a sampling period.

[KL27x]

Cool, thanks. I see it, now. The diode on the left lifts the TC output off the ground rail, so the lm358 can read it. And it also wastes power when the soldering iron is switched on. Curiously tiny diode, considering. It's probably 0402 size-ish diode.

When the one-shot fires it automatically causes the comparator to switch high/off. But D2 rectifies it, and the caps ensure that the pulse continues for the predetermined time. Then it switches off when the caps charge, and eventually opamp A will fire low, again, when the input+ falls to lower than input -, starting the cycle all over. 

So I swapped the opamp for a new one, and it's still totally dead. But at least now I know how to debug it.  I'm curious to see/scope how well this works. There's zero amplification of the TC output. I suppose when I cleaned the board with alcohol, maybe I threw the resistor ladder off whack. The entire range of the knob must be changing the voltage just a couple tenths of a volt. And the resistor values must be quite on the spot for it to work. The gimpiest part is I don't see why the positive input of opamp A would ever go higher than the FVD of the diode. This doesn't reek of any precision repeatability. Seems like there should be another resistor in there. The FVD change with current would also seem to be working against it. I'll have to triple check if this schematic is correct.

Edit. doh.. ok, so the diode just clamps the input to max 0.6ishV (what I measure on DMM for the diode). And when the pfet is off, it should fall to level of the tc output, below the FVD of the diode? So the lm358 is indeed reading TC directly, even though it isn't rail to rail opamp...  I wonder if this wouldn't work better if the lm358 was lifted off the rail by putting a diode between ground pin and ground rail. Voltage at negative input adjusted, accordingly. Well, damn, I think I got that backwards.

[KL27x]

Anyways, as far as OP's original question, you don't have to necessarily need an active solution for protecting the opamp input. A high value resistor and a clamping diode will suffice, if you take that resistor into account in your opamp circuit.

[DBecker]

Yes, the input diode is just a clamp.  An LM358 with a 10K input resistor might actually tolerate the full voltage swing, but the recovery time and wonkiness from current injection will be a problem with this circuit.  (With a microcontroller the firmware can just wait a bit for the LM358 to stabilize, but not here.)

In this circuit the LM358 is acting as a comparator, not an amplifier.  The thermocouple output will be in the range of 10-20 millivolts when hot, as will the reference voltage set by the temperature control. 

[KL27x]

K. I think I got some of that.

Tada! I got it sort of working. I lost a resistor when I swapped the lm358. And when I looked at the traces and schematic, I realized it was what I put in the schematic as the "cap" on the positive input of opamp A. I'm pretty sure it was a resistor, though. 0402 sized... gone like a ghost.

Also, somehow the NPN managed to die on me... no obvious short. I happen to have same size 3904 and that fit.

Anyhow I put in a 105K resistor for the unknown resistor/cap. I also broke the diode next to it. I had to bodge in the wrong package.

But for some reason adjusting the knob doesn't seem to affect the set temp... it just hangs around 270C. Sticking the tip in a glass of water, it draws close to 2A and the led stays mostly red. So it is definitely working. I'm sure I could fix the rest of it if it was worth the bother, but this pcb is smaller than a postage stamp and it may have earned a place in the garbage.

Thanks, DBecker. I'm going to order another TWO to finish the schematic and do some proper testing. So I got another 3-5 weeks to ponder what I learned.

[MikeLemon]

Yeah so like I said earlier I can already control the power cycle at zero crossing with an MCU to stop the voltage from interrupting the reading now I assume the way to go is to make a buffer that closes and open the output at a voltage close to 0V shut down the power cycle and make a ready.

Now I'm not that much of a thermocouple expert so the MAX7765 it already hooked up for me on the MCU because I used another one for my hotair part of the station which is finished.
Also just to be clear the thermocouple of the T12 is K type right?

[KL27x]

K type? I don't know, but I would make same guess as you.

[DBecker]

The unknown component is likely a small capacitor.  It will shunt any RF energy picked up the antenna / cord to ground.  It's not needed for all use, but people that solder tend to have transmitters (AKA 'prototypes') running nearby.

You can model it as an R-C filter with the input resistor.  You'll want the time constant to be considerably faster than the pulse period, but the filter should knock out anything remotely close to RF.  The original designer probably didn't even put that level of thought into it... "I'll leave a footprint in case we find a problem later".

I believe that T12 tips use N type thermocouples, but the exact type is only important for calibrated use.  If you support a production controller where the temperature is locked down and tips may be swapped, then it matters.

[KL27x]

^Thanks for the sanity check re cap/resistor. I didn't see the purpose of a resistor, but seeing is believing and all. (And I stuck a large value, which in my mind did the least damage, lol). But with your vote of confidence, I went back and changed it to a 0.1uF cap.

The thing was still stuck, concerning adjustment. But I went this far, and I tweaked the resistor between wiper and input to reduce the impedance just a tad. Lucky me, the only 0402 I have are 105K. Stacked over the 1.8K on the board, this appears to be perfect. 180C at lower end. 350C just about smack dab in the middle.

I have a 24 V transformer here and I'm going to bodge in a zener on the FET gate and rig it up. It's a sad looking board by now with broken traces and wrong sized parts and lots of tiny bits of wire. I won't be able to trust it, but now I got something to do with the extra T12 handles (and the clone makers favorite knife tip) which I have accumulated. I think I'll soon have an iron for the garage for welding plastic.

In case anyone is curious, the range of voltage on the T12 TC I am seeing is roughly 3mV at 180C, 7 mV at 350C. This might be the only useful thing I posted in this thread, lol.

Damned if I can figure out how resistance ladder of 1.8K on top, 50R on bottom produces roughly 0.003mV. Then with a 5V 10K pot  with 1.8K resistor in between wiper and node produces no change. Entire range remains 0.003V. But putting 105K resistor on top of the 1.8K resistor allows pot to change voltage between 0.003 to 0.012V. I wonder what the math is on that.

Behavior:
Unlike say Hakko 888, which heater turns on solid and remains 100% until it reaches set point.... then it abruptly changes to much lower duty cycle (and there is definitely thermal lag of the tip while it reaches equilibrium after this point), this circuit is 99% duty cycle at first, but it gradually slows down to a steady low duty cycle as it gets closer. And after this duty cycle reaches steady state, the tip is still lagging and begins to reach equilibrium after this point. I'm assuming no fault on the circuit. I think because the sensor is right on the heater that this is necessarily going to be the case that it will reach cutoff point well before tip is up to temp. I think it's half a dozen or one, six of the other, as far as reaches equilibrium from cold start, but T12 is able to respond faster to temp drop. Overall appearance/behavior appears to be identical to a Suhan 616 clone I have (even though it's a different pcb), and this appears to work perfectly fine in practice.  Turning the knob produces instant change in behavior, stopping heater completely when you turn it down, and going very high duty cycle the instant you turn it up. And short of PID, I don't think this needs much improvement. I guess I would say that it works.

(And I have to admit that after observing this behavior I was wrong to underestimate benefit of PID. It should in the least improve the performance by a good measure no matter the unknowns of joint mass. Using 888, seems like the sensor is some distance away from the heater so overshoot kinda balances out lack of PID. But in T12 tip, any PID should give an observable improvement.)

Sorry to OP to take over thread with my own drama.

[MikeLemon]

The unknown component is likely a small capacitor.  It will shunt any RF energy picked up the antenna / cord to ground.  It's not needed for all use, but people that solder tend to have transmitters (AKA 'prototypes') running nearby.

You can model it as an R-C filter with the input resistor.  You'll want the time constant to be considerably faster than the pulse period, but the filter should knock out anything remotely close to RF.  The original designer probably didn't even put that level of thought into it... "I'll leave a footprint in case we find a problem later".

I believe that T12 tips use N type thermocouples, but the exact type is only important for calibrated use.  If you support a production controller where the temperature is locked down and tips may be swapped, then it matters.

Well then how do you suggest I proceed with these project? I just want to read the tip's temp and control with with an arduino through a TRIAC...

[.rpv]

Hi guys, sorry for hijack a bit the post but maybe this also could help anyone out here... I'm doing a similar project with the t12 clones and when I was getting documentation about how other people are doing this project and I found the ts100 soldering station which it's open source and they provide the schematics (I got it from banggood's website), but I got a few doubts about it, I'll be very grateful if anyone can help me with a clue about why they did it on that way...

Schematic.


1. they use a dual p-mosfet (Q1A,Q1B are CJQ4953) but it's one opposed to the other (back-to-back, front-to-front?, I'm hobbyist and learning electronics  ), why it's this arrangement?, I mean, it be better to both point at the same direction, isn't?
2. the 3.9v rail it's connected through a 39k resistor to the thermocouple/heating element, I haven't any clue about this.
3. the TMP36GRTZ (U6) it's a temperature sensor to provide cold junction compensation for the thermocouple reading, but isn't connected to the opamp (SGM8551), is it some kind of software compensation?

  Thanks!.

[DBecker]

1) The MOSFET is set up as a bilateral switch.  At a glance, it appears that this circuit might support AC input.

2) The pull-up resistor is used to detect a missing or intermittent temperature sensor

3) They might be attempting cold junction compensation, but it's probably not effective.  That compensation needs to be done at the first dissimilar metal connection.  For a T12, that's in the handle.  Measuring back at the circuit board is pointless or worse. 

[.rpv]

1) The MOSFET is set up as a bilateral switch.  At a glance, it appears that this circuit might support AC input.

2) The pull-up resistor is used to detect a missing or intermittent temperature sensor

3) They might be attempting cold junction compensation, but it's probably not effective.  That compensation needs to be done at the first dissimilar metal connection.  For a T12, that's in the handle.  Measuring back at the circuit board is pointless or worse. 

Thanks!

[KL27x]

Measuring back at the circuit board is pointless or worse.

Thanks! I learned everything I know about TC from google, and I came to this conclusion, too. I find it curious that SparkyBG makes statements suggesting otherwise. He talks about cold junction like it's on the pcb in the station. And he makes comment about temp stability on a cold or hot day putting importance on ambient temp.  I feel like few degrees of ambient (to the handle) don't matter, all that matters is tip heating doesn't affect the cold junction.

It would appear to me the sole purpose of the super long T12 tip is to keep the cold junction away from the hot end. There are shorter T13 tips. And Weller also makes theirs shorter, too. I wonder how short you can do and still get this working right. If I understand correctly, at some point you could end up with thermal runaway. If the cold junction at the other end of the tip gets warmer, the TC output will go down, and the control circuit will up the power, and the TC will eventually rise higher in temp. Positive feedback, runaway train? Probably not so much, as higher temp differential will mean more heat dissipation in the metal in the tip body. But I am curious if leaving iron on for a couple hours, the temp will end up 10 degrees higher as the handle/tip warms.

Ideally, it seems to me that if you could make suitable TC with metals which can be made into good contacts, themselves, you could make the tip very short and most of the TC would be integral to the handpiece. They would plug together and become one longer TC. I bet this is completely impractical, though.

(probably best to ignore my ensuing pontification)
All told, I am pretty close to dropping T12 experiment. Using these irons for weeks now, I'm not that impressed. Improvement over conventional ceramic heater station is not significant (compared to 888 anyway.. I'm sure there are some anemic ones, like my old velleman). I find tip style/type is still 99% of the equation and T18 has everything I need, so far. Just your tips are now longer/taller, and they require more space to store and access in any convenient way. It doesn't help that ergos of the handpieces is meh. Actually, Bakon handpiece is the best shape IMO, but it won't hold the tip quite tight enough, no matter what I do. 951 handpiece is too light and skinny at the back when reaching far distance, for my tast. And I find the two piece construction is more annoying than helpful. At least on the clone, the parts fall apart when you remove the tip... so it's only a benefit if you have a foam piece for every tip, and I dont' have the space to spare to put giant tip holder on my bench. Already the size of T12 tip is annoying without the handpiece. When I need 80.00 spatula tip, I will be happy to have T12 station. That all said, I'm talking 24v clone, not true 951... maybe the clones are still lagging in the power and response, but I find with T12 clone I run practical same temp and I have to change temp and/or change tip just as often. And to me the actual time to change tip is not the main factor. Seeing and accessing all tips in small space is more beneficial to how I work (cramped chaos).

[DBecker]

There are significant advantages to the T12 style tips.  The heating element is better coupled to the tip, and the thermocouple may be optimally placed for the tip shape.  Integrating the thermocouple with the heater wire makes the whole assembly more compact, and more readily placed close the heat transfer surface.  Putting the temperature sensor close to the surface is a huge benefit for thermostatic control, and even benefits predictive control.

The long T12 tip seems to be more about keeping the handle cool for the user than cold junction compensation.

Before you give up on cheap soldering irons, try a digital controller with an aggressive predictive control coupled with a power supply close to maximum voltage. 

[KL27x]

The heating element is better coupled to the tip,

I believe this is true and that is has real practical benefit. BUT.....

What I notice is that this is most prominent/effective in the tips which are already large and fat and stubby. So if I compare a 3mm bevel, the T12 is a bit more responsive to temp change and will stomp the gas faster. I truly believe the T12 is better, here.*

When I compare to longer, skinnier tips, this is not the case. There is no perceivable/practical difference. I am pretty sure the smaller tips like pointed conicals and small diameter bevel tips and the like have very, very little or even no copper in the last third of an inch of the "stalk." The benefit of T12 design is not here, anymore. Anyhow, in practice, I see no difference. The sensor being on the heater is not the advantage. The sensor being close to the tip is the advantage, and in the smaller tips it is perhaps not nearly as much of an advantage.

PID is another matter. If PID makes it that much better, it might be something. If the 951 will remove TO220 heatsink with BR tip without changing the dial from 340C, then I'll be impressed. If I can use a BCF1 tip to solder 16 gauge wire to a plated thru hole, then ill be impressed. (If it can do these things, tell me and I might shell out for BJSW or Hakko; at least it would be in the running with Metcal) But short of that, it is not going to change much as far as what I do, and the overall system I'm liking (or highly used to, at least) is the older stuff that is simply result of 100 years of evolution from the world market leader in industrial pcb soldering. As stated, part of that is I am always using right tip because I have all of them right in front of me. (I don't have to have tall stand and constantly decicde which 3 or 4 are going to have a half handle balloon stuck on them, etc.) The system overall hinders me a bit, and I gave it a month. (I work largely under microscope and tip holder/display with easy access is not amenable with T12 tip due to the length with the rest of mess I have around). At first, I thought for sure I'm going to break the 888 at some point doing tip changes. But 8 years later, still original handpiece/heater. Perhaps if handpiece ergos were worked out, T12 gains enough ground. But the 2 piece shuffle with the hakko styl (and meh ergos), and the tip spin/fall out thing going on with the Bakon are putting stick in the gears. Also it's annoying that you can't stick the handpiece in the holder with the tip out, cuz there's nothing to make it stay in the stand. The perceptible flex (in both styles) don't do any additional favors. Ironically, my favorite thing about T12 is how the iron doesn't wiggle at all in the stand, but this is part that I made myself, and I could do same thing for the 888.

*But I already have a mean fat tip on the iron, I can turn up the temp, easily, for that rare occasion where I need to. And it is not much trouble. I still have to change to smaller tip where necessary. I have never had to go above "7" on the ole 888. If push came to shove, I think Hakko 888 will probably be capable of MORE, simply through thermal mass and user control, alone. That's my guess.