Yet another controller for JBC T210/T245

[SiliconWizard]

This is a long overdue thread. This project was completed over a year ago, but I just didn't get to creating a thread about it until now... (now that most of us spend a lot more time inside...)
I had talked about it in a couple previous threads about JBC controllers, so here it is. Sorry for the pictures, I think they don't really do justice to the OLED display I used (which looks great in person).

Main features:

• DC controlled (12V to 24V)
• Can handle both C210 and C245 tips (with their corresponding handles), detected automatically
• OLED display (256x64, 16 grayscale levels - looks nice as it can display reasonably antialiased bitmaps/fonts)
• Small unit using external DC adapters
• Auto standby (goes to 175°C after 5 min, and shuts down after 15 min, settable)
• Very fast time from ambient to set temperature (under 2 s for a C210 cartridge)
• Isolated UART interface

I've actually used it almost exclusively for SMD work for about a year now.

Design-wise, the heater control part is done with a H-bridge controller (TLE9201SG), which contains various protections, current limiting, and detection of open load and short-circuit, which is handy. The H-bridge serves a purpose: it allows to deal with both T210 and T245 handles with the same connector (turns out the wiring for the heater is different.)

The thermocouple sensing is designed around an AD8223 instrumentation amp and an ADS1115 16-bit ADC.

The MCU is an STM32L433RCT6.

Isolation for the UART interface around an ADuM2201.

Disclaimer: the whole controller could have been designed with cheaper parts. Cost was not a concern on this.

The temperature control algorithm is a take-back-half algorithm that I improved for faster reaction (JBC cartridges, especially the C210 ones, react extremely fast!)
You can see there for the basic idea: https://www.electronicdesign.com/technologies/analog/article/21773207/take-back-half-a-novel-integrating-temperaturecontrol-algorithm
It works very well and was a lot easier to tune than a classic PID.

I remember there was one question about driving the heaters with DC instead of AC, and how it could possibly damage the tips much faster. I haven't noticed that in practice, but I don't use it heavily either (just for prototyping, not for any kind of production work of course!), so this is not a definite answer on this point. I can just say that it works fine for my use.

[SiliconWizard]

For anyone curious about the regulation algorithm, I had worked on some simulations before implementing this in the firmware. I'll attach the respective graphs. For the heater, I used a simple model, and simulated heat dissipation (which you can guess on the graphs) to simulate soldering joints. Although just simulation, the PID version took much longer to "tune" than the TBH algorithm, and this was the best I could come up with.

Although the PID looks more than adequate, the ITBH controller just performed better, so what the heck, that's what I implemented in the firmware. I didn't implement a PID to compare both on real hardware, but it peformed close to what I simulated so to me that was a win. Due to a simple model for the simulation, it's possible that a regular PID would have performed just the same, but my ITBH algorithm was not requiring any more computing time so I didn't bother. That was an opportunity to try something a bit different.

[hugo]

So even though is a long overdue thread you're still not going to show us the schematic ... 

[digsys]

yeah <ugly crowd forming at the door> show us the schematic !  Serious, will you be making your design available? Selling it?

[SiliconWizard]

I may make this available, or open-source it. Not completely decided yet.
I've also considered just giving away the schematic and not the firmware, as an option. I'll see. I don't think creating a thread in this section about a project automatically means you're going to share the full design. If there's enough interest, I may reconsider it though. Just giving the main parts I used and the regulation algorithm as I did is already enough to get a good idea of how it's designed IMO.

Meanwhile, I can answer any specific question. And we'll see how things go.

[sorin]

Sorry but if you don't give the firmware the entire post is useless in my opinion.
The schematic for a soldering iron is very simple, anybody can design it in 10 minutes. The difficult part is the PID tuning.
Its ok if you don't give your schematic or firmware but at last you should give some useful infos, but nothing ... "

ps. Even the link that you provided don't work properly (the images are not displayed).

[TheHolyHorse]

Sorry but if you don't give the firmware the entire post is useless in my opinion.
The schematic for a soldering iron is very simple, anybody can design it in 10 minutes. The difficult part is the PID tuning.
Its ok if you don't give your schematic or firmware but at last you should give some useful infos, but nothing ... "

ps. Even the link that you provided don't work properly (the images are not displayed).

He just wanted to share what he's been working on, doesn't mean he's obligated to give you everything. Don't expect others to do the work for you.

Also the link works just fine, that's a problem on your end.

[SiliconWizard]

He just wanted to share what he's been working on, doesn't mean he's obligated to give you everything. Don't expect others to do the work for you.

Yep. This sections' description is "Show off your projects or other stuff you are working on." I think some people are confusing the EEVBlog forum with an open-source repository doubled with a free design service.

Some people also don't seem to realize that some of us may not be allowed to completely share a design depending on a number of factors. Just realize that unless we have decided to open-source a design, or put it in the public domain, sharing it without any kind of license is an issue that we don't necessarily want to deal with.

Sorry but if you don't give the firmware the entire post is useless in my opinion.
The schematic for a soldering iron is very simple, anybody can design it in 10 minutes. The difficult part is the PID tuning.

Having actually worked on that as shown, instead of just talking about it randomly, I do think the opposite. PID controllers are textbook stuff. You can find them anywhere. Tuning can be a bit lenghty but it's not hard, and obviously it entirely depends not only on the iron (/cartridges) itself, but also on all the driving electronics, so it's completely linked to the schematic as well.

The schematic is the "hard" part IMO, not that it is really hard per se, but just because we are dealing here with basically undocumented stuff. A few other people have reverse-engineered those JBC cartridges, not all with complete exactness though - so getting it right is the tricky part. Thermocouple sensing is also not trivial, especially here where you have to deal with the heater that is basically in series with the thermocouple.

Its ok if you don't give your schematic or firmware but at last you should give some useful infos, but nothing ... "

I thought you just said above that without the firmware it was useless. Looks like your are already changing your mind a couple sentences later.
I gave minimal info so far, but I think enough to get people who'd be interested to look for more info by themselves (talked about the TBH algorithm, you can find a handful of links about it online, talked about the main ICs I used...)

ps. Even the link that you provided don't work properly (the images are not displayed).

Those are all forum-attached pictures, no reason they wouldn't work. Sometimes they are a bit slow to appear when you want to see the full size though - depends on the server and your connection obviously.

Anyway, parentheses closed. If anyone has specific questions, I'll gladly answer them. I can also post parts of the design that I think would help the most without disclosing the full thing (like how the handle is wired exactly, how the thermocouple sensing is implemented, and the regulation algorithm as pseudocode.

[sorin]

Also the link works just fine, that's a problem on your end.

      You dont even have read it. Try to open  (Fig. 1) or (Fig. 2) and see what happen.

I have worked with PID controllers 3-4 Years ago, and of course they are not rocket science but if you want to implement a good Soldering iron a simple PID is not enough. You need more, for example, you can detect the thermal mass of what you are trying to solder from dT of the temperature and change the pid parameters, you can also detect from the software if the iron is not used and automatically put it on stand by.

Thermocouple sensing is also not trivial, especially here where you have to deal with the heater that is basically in series with the thermocouple.

To me don't seems so complicated, all the soldering irons use the same principle Resistance + Temperature sensor (Thermocuple, PTC or the resistance is used as temp sensor). In your case i thing that is Thermocuple so just turn off the power for breef period of time and measure the temp.

Sorry if I have offendet you, is is not my intention.

[TheHolyHorse]

Also the link works just fine, that's a problem on your end.

      You dont even have read it. Try to open  (Fig. 1) or (Fig. 2) and see what happen.

I did open the link, but you're correct I never clicked the Fig1 and 2. But "the images are not displayed" isn't very descriptive, I opened the link and images were showing.

Anyway enough of this it doesn't matter

[Anyzis]

I have a question, how do you detect which handle is connected to the controller? Do you rely on the extra pin in C210 connector or some sort of pulse probing for the heater?

[SiliconWizard]

I have a question, how do you detect which handle is connected to the controller? Do you rely on the extra pin in C210 connector or some sort of pulse probing for the heater?

Yes, it relies on pin 6 of the RPC1 connector. In T210 handles, pin 6 is connected to the tip, in T245 handles, it's left open.

Attached is the connector pin-out from my schematic.

[wizard69]

High SiliconWizard,

Thanks for this post.    I've take an interest in DIY soldering stations and the various solution seen there.   Your post is extremely valuable if for nothing else because of the link to the "take back half" algorithm.   I haven't seen this before so it is good even if a new soldering station doesn't happen.

[SiliconWizard]

the link to the "take back half" algorithm.   I haven't seen this before so it is good even if a new soldering station doesn't happen.

My "improved" version reintroduces a derivative term, whereas it's not there in the basic TBH algorithm. It allows decreasing under/overshoot significantly while still being relatively simple. You can say it's then close to a classic PID, except that the "TBH" step is a nice way of dealing with integral windup, and effectively makes tuning coefficients much easier.

[Ozan Bayram oglu]

Hi
Can you share super project, schematic and software?

[SiliconWizard]

Can you share super project, schematic and software?

Hi, as I said earlier, I was not ready to do this. But I will definitely consider it. I'll let people know if/when I do.

[pebcac]

Glad I found this post.

Can you give more details regarding the H-bridge and how it's driving the C210 cartridge? Does the heater current go through the C210's thermocouple?

Why are you using an ADS1115 instead of the STM32L4's internal ADC?

[SiliconWizard]

Glad I found this post.

Can you give more details regarding the H-bridge and how it's driving the C210 cartridge? Does the heater current go through the C210's thermocouple?

Go from the schematic extract I posted for the connector pinout. (RPC1 connector - Hirose, relatively easy to source, I highly suggest using that instead of butchering existing handles for no good reason IMO.)

* For T210:
- Heater on state: RPC1_HEATER+ to Vsupply, RPC1_HEATER- to ground
- Heater off state: RPC1_HEATER+ and RPC1_HEATER- : both high-Z

* For T245:
- Heater on state: RPC1_HEATER- to Vsupply, RPC1_HEATER+ to ground
- Heater off state: RPC1_HEATER- and RPC1_HEATER+ : both high-Z

For both handles, the thermocouple sensing is between RPC1_HEATER+ and ground.

Why are you using an ADS1115 instead of the STM32L4's internal ADC?

Thermocouples generate very low voltages. The sensing stage is as follows: input protection, then an AD8223, set with a gain of ~25, and a voltage ref of 1.2V (to avoid having to generate negative voltages.) Then the ADS1115 is set as differential input (between AD8223's output and REF), with a further programmable gain (internal). The goal was to get an approximate resolution of 0.5°C or better.

You may manage to get something workable with a simpler approach and the internal ADC of the MCU, but I didn't want to bother. I would probably have had to use a much higher gain with the front amp, which would have led to other issues.

And whatever you do, the thermocouple sensing can only be done when the heater is OFF (with this DC design).

[shangaoren]

Are you sure about the heater beeing in series with thermocouple? i have a fried tips that no longer heats but i still can have the thermocouple value between body and one of the pins

[SiliconWizard]

Are you sure about the heater beeing in series with thermocouple? i have a fried tips that no longer heats but i still can have the thermocouple value between body and one of the pins

Well, given there are exactly 3 pins, yes. I don't see how they could be wired otherwise, and I think there are a couple JBC patents that show this. If they were independant, there would be 4 pins. If they were in parallel, there would  be 2 pins. I don't see any other possible configuration?

But you can drive the heater bypassing the thermocouple.

[pebcac]

* For T210:
- Heater on state: RPC1_HEATER+ to Vsupply, RPC1_HEATER- to ground
- Heater off state: RPC1_HEATER+ and RPC1_HEATER- : both high-Z

Makes sense. For T210, HEATER+ and GND are used to both power the heater and measure the thermocouple voltage. Not at the same time, of course.
Since they're in series, the heater current is going through the TC. It's interesting that the TC can handle amps of current.

Are you limiting the T210 heater voltage (24V down to ~7V) in any way, or just relying on the current limiting feature of the H-bridge?

* For T245:
- Heater on state: RPC1_HEATER- to Vsupply, RPC1_HEATER+ to ground
- Heater off state: RPC1_HEATER- and RPC1_HEATER+ : both high-Z

For T245, the heater polarity is swapped because the cartridge wiring is different and the heater can be powered directly without going through the TC. Got it.

Thermocouples generate very low voltages.

Do you remember the tempco for the thermocouples?

I've seen posts that suggest 15.7uV/C for the C245 and <10uV/C for the C210.

Are you sure about the heater beeing in series with thermocouple? i have a fried tips that no longer heats but i still can have the thermocouple value between body and one of the pins

For C210 the heater current seems to flow through the thermocouple and for C245 the heater can be powered separately.
The C210 schematic is from Xyphro's Nanosolder.
C245 pinout:


The cartridge you fried, was it a C210 or a C245?

[shangaoren]

Are you sure about the heater beeing in series with thermocouple? i have a fried tips that no longer heats but i still can have the thermocouple value between body and one of the pins

Well, given there are exactly 3 pins, yes. I don't see how they could be wired otherwise, and I think there are a couple JBC patents that show this. If they were independant, there would be 4 pins. If they were in parallel, there would  be 2 pins. I don't see any other possible configuration?

But you can drive the heater bypassing the thermocouple.

Sorry i should have been more precise, i meant // with common point on tip body, we're saying the same thing

The cartridge you fried, was it a C210 or a C245?

I fried a C245, i found the same wiring as the one you linked (thermocouple from top pin to body and heater from middle pin to body)
Is there a reason to power heater throught thermocouple ? 

[pebcac]

Is there a reason to power heater throught thermocouple ?

For the C245, no.
For C210, the internal wiring is a bit different, and the heater and the TC are in series without a tap in the middle, so there is no way to power the heater without passing current through the TC.

[shangaoren]

This is a patented technology : using the thermocouple as an additionnal heater 

For C210, the internal wiring is a bit different, and the heater and the TC are in series without a tap in the middle, so there is no way to power the heater without passing current through the TC.

[SiliconWizard]

Is there a reason to power heater throught thermocouple ?

For the C245, no.
For C210, the internal wiring is a bit different, and the heater and the TC are in series without a tap in the middle, so there is no way to power the heater without passing current through the TC.

Note that this isn't a problem - the thermocouple is an integral part of the cartridge's construction, and not a separate, discrete part inside of it. It can handle the current just fine.

The base patent: https://patents.google.com/patent/EP1086772A2/en