DIY JBC Nano

[Ysjoelfir]

How do you get along with a single supply for the thermocouple amplifier? I am working on a JBC soldering station project myself (which also works pretty nice ) but used a dual supply because I just couldn't get the damn thing to work - the offset was huge in relation to the actual thermocouple voltage.

[Xyphro]

Hi Kai (nice name :-)),

I selected the OPAMP carefully and the characterization measurements of the temperature to voltage curve did not show issues.

My testpoints for this curve were:
- room temperature = 0V
- When I put the running iron into water I get 100 degrees (in the video it was just a wet sponge)
- The point, where it begins melting solder (around 190°C)

Cold junction compensation I left out for the reasons given in my inital post + an inaccuracy of maybe even 20°C would really not matter here.

Best regards,

Kai

[SiliconWizard]

Hi,

I'm also considering building a controller for JBC T210 irons (C210 tips). I have one question that seems to pop up regularly without much definite answer, so if you guys have any idea...

The original JBC controller seem to drive the heater from 24V AC (or 23.5V). Driving it from 24V DC with PWM seems much simpler overall. Is there a definite reason why we should drive the heater with AC (such as premature corrosion??) rather than DC? What do you guys think?

Then, in case driving it with a balanced voltage is really to be preferred, does it seem possible to still use a DC power supply and use a H-bridge?

Thanks!

[ZomBiE80]

https://create.arduino.cc/projecthub/sfrwmaker/soldering-iron-controller-for-hakko-907-v-2-fc75d7

[Ysjoelfir]

Heating with DC lets the tip degrade faster (in theory, I don't have numbers how much) because of electromigration. So heating with AC is prefered. As I wrote, I don't know how much faster this really is. I have built a prototype of a benchtop version JBC controller for T245/T210 which works with AC (actually with 30V AC because I didn't have an appropriate transformer... Damn that thing heats fast!) and a small, portable Version (just 25x25x50mm + external PSU) which runs on an 24V DC Notebook powerbrick. Both work very well. I would expect the idea of using an H-bridge could be an apropriate way to compensate for electromigration. But is it really necessary, meaning, is the result worth the not insignificantly greater effort?

[SiliconWizard]

Thanks!

I don't have a clue either whether the degradation is significant with DC, and suspect that other factors like tip oxidization especially when using mainly lead-free solder would probably be predominant.

As you just said, the main point in using DC would be to have a simple design that can be powered with external power bricks and/or batteries, which I find handy. On top of that, since the heater and the thermocouple share a common, as I reckon, that should be a bit easier to get reliable measurements when using PWM, because you can control exactly when it's off.

The H-bridge approach would not necessarily cost a lot more. There are integrated H-bridge controllers that embed the controller, the power transistors and even current limit + current measurement, which would allow some added protection for the tips.

The AC-approach probably costs significantly less overall (and additionally has probably a bit better efficiency), so that makes sense for high-volume production. But for prototyping or a for a low-volume production, that would just be a matter of a few bucks more.

Since you have tried both the DC and AC approach, have you noticed any difference in heating times and recovery?

[Ysjoelfir]

Regarding the costs of the H-Bridge: I thought especially about the temperature sensing, not the heating itself. If I am not overlooking something, that could be potentially difficult.
AC is not necessarily cheaper than DC - with DC you can just use a FET, directly controlled by the µIC. If you want to switch AC with FETs you need an isolated voltage source and two FETs. Using an SCR is possible, but I experienced quite a big loss and quite a lot of heat so I decided to use FETs instead.


I did indeed notice a difference in AC and DC. A note first: I have also an original JBC AD2200 which is the AC reference I refer to. As already mentioned, my DIY version uses a 30V transformer which lets the tip heat up even faster than the original station and recovery is... well, instant? But I am pretty sure the lifetime of the tips will suffer quite extensively.

Comparing the AD2200 with the DC DIY station I noticed that the selfmade station seems to be a bit (but not significantly) faster both in initial heatup time and recovery/regulation. I suspect that this is because I can measure the temperature much faster and more often with DC since the AD2200 switches only at mains zero crossing, so in the worst case it has to "wait" around 10 milliseconds whereas with DC I can decide that I want to measure the temperature right about "now".

[SiliconWizard]

I will think about it in more details, but I don't see at the moment what would make the thermocouple measurement trickier when using a H-bridge. The measurement would take place when the H-bridge is completely off anyway (which is a possible state with most integrated H-bridge controllers). One of the thermocouple connection is shared with one of the heater connnection. It would then either see +Vcc, GND or Hi-Z depending on the state of the bridge. I don't know if shorting the thermocouple (when the Heater + is in the GND state) could degrade it? I would guess not. Anyway, if you can think of specific concerns for the temp sensing that I may be overlooking right now, let me know.

I've found nice H-bridge ICs at Infineon for less than 3€ (by 1). They contain fault detection which is handy and would cost quite a few additional components if done just with discrete parts.

The main concern I'm having is with EMI. Direct PWM with relatively high currents, balanced or not, can cause bad EMI issues. And a soldering tip can make a nice antenna. That may be one of the reasons why it's often not used, or just on amateur gear mostly. The nature of the heater elements may naturally reduce EMI but I'm not sure about that. This would at least be a concern to be adressed when considering doing this on commercial products IMO.

As for cost, what I meant is that if you take the overal BOM into account, the DC only approach may cost more if done properly (on commercial products). The AC approach basically just needs a transformer, an SCR and ideally an optocoupler (and then some low-power rectification and regulation for the logic part). The SCR approach seems to be quite common in a lot of commercial soldering stations.
OTOH, a decent, approved DC power adapter in the order of 100W+ may itself cost more than this in high production volumes. Just a thought.

[SiliconWizard]

Just a quick update. I've worked on this a little bit. And yes we were overlooking something obvious, and it's not with the temperature sensing but with the heating part!

Since the thermocouple and heater are in series in JBC tips, and the tip shell is itself connected to this chain, you can't ground the tip shell while imposing any voltage other than 0V on one of the connections of the heater, which defeats the possibility of using a h-bridge scheme to drive it.

So basically, if we want to drive the heater with an average DC voltage of 0, we have to keep the driver isolated from the rest of the circuit, so that it's floating relative to the system ground. This is what is achieved at a relatively low cost by using a separate secondary of the main transformer and an optocoupler. Doing this with a single supply of 24V DC would require to use an isolated DC/DC converter and isolate a few digital signals. Now that's beginning to add complexity and significant cost.

From what I gathered looking at the UniSolder (a quite popular soldering controller) schematics, it actually seems to use either DC or rectified AC, so the heater doesn't seem to be driven with DC-balanced voltage, but instead, either DC or rectified AC. So it doesn't solve the potential electromigration issue, unless I've missed something.

As for the thermocouple conditioning, I'm using an AD8223 for signal amplification, with a reference voltage of 1.2V to keep it in its operating range for input signals down to ground and even a bit lower. Works well with a single supply.

[C]

Think you are missing some things.

For JVC you have three contacts on the plug in tip.
Between tip and controller you have wire which has resistance.
So at a distance you will have an AC signal on temperature reading.
So a whole bunch of resistances that change with current.

If you use the AC to your advantage, you could measure temp at any time.
You could do this by having the equal of a second connected tip, A reference iron.
If you think about this, it the tip wire is at ground at the controller, then a properly adjusted pot will create a matching AC like the heater power.
The pot is adjusted such that the temp value does not change based on the AC heater current.
Your thermocouple amp does not see a large AC input any more because the pot adjustment removes it.

Note that the pot must be adjusted by control circuit as the resistances will change.

[dymbo]

Think you are missing some things.

For JVC you have three contacts on the plug in tip.

Biasing on wrong assumption leads to big mistakes.
First please search for real photos of C105 series tips discussed here. Those only have two terminals like Hakko T-series

[C]

Think you are missing some things.

For JVC you have three contacts on the plug in tip.

Biasing on wrong assumption leads to big mistakes.
First please search for real photos of C105 series tips discussed here. Those only have two terminals like Hakko T-series

With just a little thinking you might see that with idea you can keep the thermocouple amp happy and ready to read value. No time delays coming out of saturation with two connections.

[dymbo]

Think you are missing some things.

For JVC you have three contacts on the plug in tip.

Biasing on wrong assumption leads to big mistakes.
First please search for real photos of C105 series tips discussed here. Those only have two terminals like Hakko T-series

With just a little thinking you might see that with idea you can keep the thermocouple amp happy and ready to read value.

Please expand. Mean, I got the Idea of filtering out the AC so the AMP to hear thermocouple only voltage. Just didn't get "the whole bunch of resistances", pot adjustment and other little details. E.g., thermocouple T isn't read by resistance, but by the voltage... Also, the AC is about to introduce much of a noise that's unlikely to get easily filtered.

[C]

If you think on the total problem.
It takes time for heat in the heater to get to tip.
The greater the differential between heater & tip the faster the change.
To know heat energy you need to know watts.
When combined with temp sensor you have more facts for control.

You can then compute the problem and see the response from the temp sensor.
The feedback can now compute how big the mass the tip is
touching and respond faster. 
So with the facts a processor can do better control.
So when you pick up the iron you get a response curve that will change when tip is not in air.

Above is what you want.
Thermocouple amp is amplifying a small voltage & a big voltage puts output in the rail talking time to recover.

You have a tip with two connections.
From tip to control box you have wires with resistance.

To get watts you need a current sense resistor. Only place to connect is non-tip ground.
The power end of current sense resistor is the spot to connect the virtual iron which is nothing more then a resistor divider that gets fancy.
With larger value resistor in parallel to current sense you get a low current version that changes same as the connection to the handle and back to control.
Proper adjustment of tap gets voltage difference between two inputs to Thermocouple amp in range.
Fine adjustment removes wire resistance.


 

[SiliconWizard]

The discussion is kind of drifting towards thermocouple measurement, which is not the issue I was talking about.
There may be schemes for thermocouple measurement which can reduce or avoid having to shutdown the heater and wait for a delay (when dealing with PWM) or wait for zero-crossing (when dealing with AC drive). But that was not my concern here. This waiting time doesn't bother me actually, since I'm controlling the whole cycle with PWM. The only drawback it has is that we can't have 100% duty cycle. I don't mind. I'm aiming at C210 and maybe C245 tips, and I'll get more than enough max power even with the reduced max duty cycle.

The issue is with the tip grounding.

If you ground the tip for ESD protection, as is done in original JBC stations, I can't figure out how we could drive the heater with a H-bridge, unless it's isolated from the rest of the circuit. It's that simple.
Of course if the tip is not grounded, this is no problem - but I wouldn't feel comfortable with this. I want ESD protection.

[C]

SiliconWizard

For the three connection tips this is easy a floating power supply.
For both you build circuit that keeps Thermocouple amp happy then you have no problems.

[SiliconWizard]

Well - when using a DC power supply, typically a "power brick", the solution would be not to connect the outter shell of the tip to the internal ground but just to an external earth connection through a 1 Meg resistor for instance. To avoid issues, the DC power supply would have to have no internal path to earth. A lot of high-power bricks actually have their ground (-) connected to earth and have a mains plug with earth. So that would be a problem. Completely isolated power bricks do exist but they are pretty expensive.

So really, the H-bridge approach seemed interesting but it turns out to be a lot of trouble, especially if we want to use external DC power bricks.

Can anyone who worked with the UniSolder design confirm what I understood of it, that it actually can't drive heaters with a balanced voltage and thus can't avoid electromigration (if that is ever a real issue)?

[C]

First think about PWM.
Great at creating noise.

For a three connection tip, could be some good reasons not to use 50/60 cycle AC, but you could still use AC with a power brick. Higher frequency gets you smaller lighter transformer.
Could be some vantage for two connection tip still.

The important thing is to keep the Thermocouple amp from being effected by power state. Then A two connection tip has a limit of getting good temp reading after some time with out power applied. The less the effect the quicker you can get a reading which leads to part of better control.
The important fact is that things change based on temp & current.
If you collect good facts then micro could adj the circuit as needed to keep Thermocouple amp happy.

To get fast control your micro needs to learn how the tip responds, Each tip will be different.
For example to get from room temp to soldering temp could be X seconds of power then temp slowly gets to tip at time Y with little overshoot. To gain some facts, you might break this up into more power shots, but need to remember that a temp reading is not final when doing this.
With out that knowledge, I think you will find it will take longer using normal control loop.
Most diy designs I have seen gain no knowledge of how much mass the tip is in contact with and just try to control with temp.
With out mass in control loop you have bigger swings and/or slower response.

So stop thinking how to power and think about all the information you could collect and keeping Thermocouple amp happy. This gives best control loop to control power.
But better control costs in getting better input facts.
Some temp controls add a heat sense to current sense resistor.
A lot of electric range burners do this for control.

So a great iron has many facts/sensors in it's control loop.

Start with a lower voltage to check your circuit and verify proper operation. Use your micro for more then a dumb temp control.

Think you will find that you will have a better iron if you can use a higher wattage power source.
JVC gets a lot of advantage from that high wattage source of power but control becomes more critical.

ESD protection works with 1 meg due to isolation.
With grounded power, you need to try to keep tip at ground.
With high power and wire to handle you will still have a change at the tip.

[dymbo]

Well, I didn't get your idea first.
Actually, I'm about to build such a state machine to watch for thermal system response and advance the heating power based on the times it gets to heat up & cool down. As for the AC, I thought about building either the AC PWM, or full PWM SMPS with filtered AC sine output, like those used for Vector [motor] Control or Field Oriented Control with same state machines watching system's thermal dynamics.
I agree there's much room to predict such a system states. Also, there's the heater resistance T drift which is possible to get measured and serve as a second thermal sensor (PTC).
But first that sounded to me like an idea to filter out the AC to read thermocouple voltage remainder.
But the very first is that I must wait for about 1-2 months before I get the micro tweezers and the very tips! 

[C]

With temp change a thermocouple is a varying DC voltage.

Thinking on two connection tip.
That has to effect the circuit. You should see a DC offset in the AC. The problem is that it's a small signal and needs to be made large to get a reading.
So if you have one path that has a thermocouple & a second path that does not have a thermocouple and you power both paths with the same signal ( the power ) you should be able to see a difference.
You have wires in path that change but should effect both half's or AC the same.
You have the heater resistance that should change based on heater temperature but again should effect both half's of AC the same.
If you put a current sense resistor in both paths and measure the difference you should see the voltage of the thermocouple voltage. You would be building a Wheatstone bridge.
https://en.wikipedia.org/wiki/Wheatstone_bridge
If Rx = cable to handle & tip. Then with proper adjustment of R2
you should get a AC balance with  thermocouple effect showing.

Sensing the heat of current sense resistor should give an idea of heater temp.

By taking a lot of samples, you should get a lot of information.
Rate of thermocouple helps build profile of tip & gives hint of mass tip is in contact with.

Would not be that hard to build some test plots or graphs to see what you can get from data.
Just need a few differential input ADC's that sample at same time to remove time errors.

Starting to wonder if they could be using some audio ADC's & DAC in a circuit.

[dymbo]

You would be building a Wheatstone bridge.
https://en.wikipedia.org/wiki/Wheatstone_bridge
If Rx = cable to handle & tip. Then with proper adjustment of R2
you should get a AC balance with  thermocouple effect showing.

That's what I was thinking even before your post, but as a secondary option to get more effecient control loop.
Indeed we must measure the power applied to the heater to diffirentiate the response.

Sensing the heat of current sense resistor should give an idea of heater temp.

I believe, it's kinda overshooting approach. First, we loose energy on such a resistor. Indeed it's a low value chained with the handle circuit to get any reasonable to measure T change on it.
On the other hand, the heater resistance has a drift over T range which can be measured as long as we measure the power. Just need some "calibration" stage on the tip change. And no power loss, just a high value R in parallel to the handle circuit.

Rate of thermocouple helps build profile of tip & gives hint of mass tip is in contact with.
Would not be that hard to build some test plots or graphs to see what you can get from data.
Just need a few differential input ADC's that sample at same time to remove time errors.

Yep, it's about just a simple maths. The state watchers should be calculating correction factors to the main PWM cycle.

Starting to wonder if they could be using some audio ADC's & DAC in a circuit.

The best way to find out is to tear down the real JBC nano station. Unfortunately, we don't have one to reverse-engeneer.

BUT! We could be building even better, should they haven't realized neither of the ideas. Honestly, I wonder if they did. More likely they would have built it around fancy algorythms.
Have you ever seen the Hakko T-series oscillograms? Not even a hint on a precise PWM control, it just skips N half-waves from a transformer!
To me it looks ridiculous nowadays, moreover considering the price 

[C]

Sensing the heat of current sense resistor should give an idea of heater temp.

I believe, it's kinda overshooting approach. First, we loose energy on such a resistor. Indeed it's a low value chained with the handle circuit to get any reasonable to measure T change on it.
On the other hand, the heater resistance has a drift over T range which can be measured as long as we measure the power. Just need some "calibration" stage on the tip change. And no power loss, just a high value R in parallel to the handle circuit.

one path for Wheatstone bridge is the current resistor and handle/tip. This resistor produces the most heat and will be apx of heater temp.

The TS100 soldering iron could supply a lot of the firmware.

The ridiculous is the prices even for the handle.

A great iron is all in the huge power and fast control compared to lesser irons. Making the most of all the data/information that can be collected and using it in control.

In addition to Wheatstone bridge, the AC is suppling a dither signal. Might gain even more data with a low power dither.
having three states for power. This might help with time from heater to tip.

Then it might help to have control loop make the tip oscillate about set temp some. Could help sense tip touching mass, but would be a very slow.
 

[SiliconWizard]

Just found this oldish thread: https://www.eevblog.com/forum/projects/electromigration-in-soldering-iron-heaters/

There is still no definite answer, but as I suspected, AC driving may be mostly used for cost reasons. Electromigration in soldering tips heaters probably has little chance of getting significant during the tip's lifetime.

So for now my controller won't drive the heaters with a 0V DC offset. Doing this with PWM DC control would be too expensive as it would need isolation as I explained above. I will still use the H-bridge to be able to use both T210 and T245 irons with the same connector (connections to the heater are reversed as explained here: http://dangerousprototypes.com/forum/viewtopic.php?f=56&t=7218&p=61175#p61175 )

[hexpope]

Hi all, I am also in the process of designing a controller for the JBC line of products.  I think I will be using DC for the initial tests.  I noticed a lot of you guys like to use OPAMPS given the that the TC of the iron is in the uV range on the JBC.  Has anyone tried using the TI ADS1115 at all in their tests and leaving out the OPMAP from their design with good results? Looking at the specifications its a fairly decent ADC

[SiliconWizard]

I noticed a lot of you guys like to use OPAMPS given the that the TC of the iron is in the uV range on the JBC.  Has anyone tried using the TI ADS1115 at all in their tests and leaving out the OPMAP from their design with good results? Looking at the specifications its a fairly decent ADC

I'm using an ADS1115 in my design, but with an AD8223 as a front-end, with a gain of 25.

An ADS1115 alone will give you a couple problems. When used single-ended with very low voltage input signals, the SNR will probably not be that great. And yes, you will have to use it single ended, because the TC in JBC cartridges is in series and one of its ends will have a low-impedance path to ground.

Also, if you plan on using C210 cartridges, the TC has a very low temperature coefficient of a little less than 10 µV/°C. Given than the smallest full-scale range of the ADS1115 is +/-256 mV, and that you get half of that when used single-ended, you'll get a theoretical LSB of ~7.8 µV. Thus in a real-world environment, I would expect an effective resolution of several °C, which exceeds what I would like to deal with. Then again, for a simple controller, you may find this appropriate. Feel free to experiment and report back. There are ADS1115 breakout boards, you could prototype this in a couple of hours and see what you get.

I am currently in the process of finalizing my prototype and will post my results once it's done, probably in a new thread.