AxxSolder - JBC soldering controller

[eliocor]

Nice project, but I will be more interested if it will support even the T115 handle (I work only with SMD components): I own a T3B solder station, but as you can see:
https://www.eevblog.com/forum/reviews/aixun-t3b-strange-behavior-(must-invert-liveneutral)/msg5135604/#msg5135604
I'm not so enthusiast of it*!
If needed I can sponsor AxxAxx with a T115 handle + tips...


*) just to be fair, the station works flawlessly except for the strange behavior

[c0d3z3r0]

T115 is 12V, while T245/T210 use 24V. It should be possible to adapt the code (heating curve, detection (probably same connection as T210)) and use it with 12V input.

[c0d3z3r0]

thought I don't think it's confusing or too much to just ask users to avoid supplies whose negatives are shorted to mains earth, as this is the case with majority of power supplies.

Majority, u sure? Maybe true for cheapo supplies for consumer-grade laptops. Higher-quality Laptop supplies (USB-C and regular), I don't think this is true. Dell, Lenovo, Apple (with cable attached, not the short euro plug), ..., they're all earthed on the secondary.

EDIT: Oh wait, did you mean that ("this is the case")?

Oh well, and I see many Github issues with bricked devices incoming because of wrong choice of the supply

[eliocor]

It would be wonderful: a very small T115 controller powered by an USB-C (TD) power supply*!
Just something which can be carried everywhere for any electronic emergency

*) unluckily, if I remember correctly, their maximum reach is 100W/20V (not 24V)

[c0d3z3r0]

Yeah, 20V is max. but you can still run T245 with that, just with limited power. That's probably still way better than any regular soldering iron

[drksy]

@drsky Regarding regulations, the fact that gloating supplies are availavle doesn't mean that they are allowed in any use case. (Not all use cases allow choosing any one of SELV, PELV, FELV). However, in this specific one (solder station) both SELV, PELV seem to be ok.

Btw. Having a SELV SMPS and then grounding the secondary of the DC/DC converter still makes the whole construct PELV, even though the grounding is done bc of functional, not safety reasons.

I just had another random idea, that needs further evaluation, though: instead of all that isolation complexity from my post, why not have a second smaller uC, that only does measurements and heating control in the 24V circuit, communicating over isolated I2C with the main uC.

Right, in some applications like a PC power supply, PELV would be required, but not in this one. Let's say somebody is building this soldering iron controller for the first time. Which is more hassle, picking a power supply whose negative is not connected to earth? Or buying/soldering/programming a whole second microcontroller, and paying for and dealing with isolated i2c, and then have to manage two separate code projects which have to be uploaded separately, through what, two separate ports? I'm sorry, but this is just a mess! 

But what if you don't care about isolation and the error current that's passing from heater positive through the thermocouple, through earth back to negative. AxxSolder as it is right now turns off the heater, waits a little bit for the thermocouple to stabilize, then takes a measurement. Would this break if we are dealing with a PELV and a earth-grounded tip? Maybe all that's needed for ESD safety is just an added connection from the outer shell and the board ground to earth, and leave everything else as it is. What am I missing?

[c0d3z3r0]

Well, having a PELV supply will not only lead to a small error current to earth but to a short to ground (ok, nearly, it's a few ohms)

[Warhawk]

I have, the domestic shipping cost is between 8 to 10 euro, I paid total 6 euro include shipping VAT everything for 5 pcbs from JLCpcb, If is worth it then I can send.

Well, you have quite expensive shipping there. Here in Munich an envelope with the PCB is 90 cents or co 

Let me study the project more in detail. I want to check how mature this is.

[drksy]

Well, having a PELV supply will not only lead to a small error current to earth but to a short to ground (ok, nearly, it's a few ohms)

Okay I see problem now. In the AxxSolder, heater current starts from red and returns to blue, so with PELV, blue will be permanently tied to earth. But if green is also connected to earth, when applying to power to red, the supply will basically be shorted through the termocouple. Bad idea! I need to get more sleep.

The combination of PELV and grounded tip always introduces a problematic current path. If you apply 24V to blue and return to red, red will be earth and thermocouple will be permanently shorted. And if you apply 24V to red and return to blue, the supply will be right across the thermocouple because the outer shell and supply negative are both earth. How can this even be resolved with isolation? The earth connection in the heater supply secondary conflicts with the added requirement that the outer shell should be earth.

The only way this can work is to do a similar thing Aixun does: start the heater current from blue and return to the outer shell (green) through the thermocouple. The supply negative would be permanently earthed, and so is the outer shell.

[c0d3z3r0]

Well, then you create the problems the Aixun has for the Axxsolder. Thus, either your picture with SELV or my extension with various isolation measures helps. In your picture no current can flow to earth/GND of the first stage secondary side because the SELV secondary has no potential to mains earth.

[drksy]

@c0d3z3r0 Imagine you have a PELV supply and want to build a C245 controller that has a grounded tip. What I'm saying is that's not even possible to do, except the way Aixun does it - which of course as you pointed out comes with all the limitations and problems of Aixun. It's not ideal. This is why I think using an PELV heater supply in the first place is just a bad idea.

[c0d3z3r0]

Sorry, I misread your PELV as SELV. Hm, I need to think about the PELV construct again

EDIT: @drsky You are right. This edited picture helps understanding the problem :-) Btw. you had +24V/GND swapped. (I didn't fix the TC input here)



So, one can't have a PELV supply and grounded tip. Earthing is only possible by using a) battery power b) SELV supply c) toroidal transformer. If someone wants full flexibility, earthing is not possible.

Ok, next idea:
- instead of the TPS54302 converter use a isolated DC/DC converter module (will work regardless of supply type) OR: dual pcb design, so either option can be populated based on user choice
- control heater through optocoupler or isolated mosfet driver - no idea how one would combine optocoupler + mosfet driver... Maybe replace LTC4440 with HCPL-3120?
- provide a earth connector on the board + add earthing circuit like JBC does (can be left unpopulated if earthing not wanted by user)
- add a big note to document a) the options and b) the incompatibility of tip earthing with PELV / grounded supplies

[Pooya]

Yaaaay it's alive, for yours order stencil that was pain in the ass, it was so difficult to solder 

[Pooya]

with box

[drksy]

@Pooya wow! Nice! Really cool red enclosure you made there. How well does it work? Have you tested it?

Btw. you had +24V/GND swapped. (I didn't fix the TC input here)

Yeah I goofed that up. In the original JBC it's supposed to start from red and return to blue.

So, one can't have a PELV supply and grounded tip. Earthing is only possible by using a) battery power b) SELV supply c) toroidal transformer. If someone wants full flexibility, earthing is not possible.

This is one of those times you can "come agutsa" as Dave likes to say  Imagine if you try to probe the heater waveform with your earth referenced oscilloscope, clip the alligator to negative, probe the positive, turn it on, and BANG! 

Really gotta be careful with this earth connection to not short something.

Btw. you had +24V/GND swapped. (I didn't fix the TC input here)

Yeah I goofed that up. In the original JBC it's supposed to start from red and return to blue.

Ok, next idea:
- instead of the TPS54302 converter use a isolated DC/DC converter module (will work regardless of supply type) OR: dual pcb design, so either option can be populated based on user choice
- control heater through optocoupler or isolated mosfet driver - no idea how one would combine optocoupler + mosfet driver... Maybe replace LTC4440 with HCPL-3120?
- provide a earth connector on the board + add earthing circuit like JBC does (can be left unpopulated if earthing not wanted by user)
- add a big note to document a) the options and b) the incompatibility of tip earthing with PELV / grounded supplies

Maybe even another screw terminal for the 5V if someone wants just use a wall wart or something. It could also be a push-pull transformer driver like the LT3439 or the SN6507. Both of these could work with surface mount transformer which isn't going to be too much bigger than a typical high power inductor. But a power module is the simplest and most compact way to go imo.

As for isolated mosfet drivers, you could basically keep the LTC4440 + NMOS the way it is and just isolate the input to the gate driver with a jellybean PC817 + pulldown resistor.

[Pooya]

I don't know I did something wrong or it should be like that! or it has to be fixed from the firmware, when the handle is not connected,should say that (No Tool) or something like that, even the temperature should not be shown. but as you can see in the photo it shows even the handle type while nothing connected to the board.

[c0d3z3r0]

This is one of those times you can "come agutsa" as Dave likes to say  Imagine if you try to probe the heater waveform with your earth referenced oscilloscope, clip the alligator to negative, probe the positive, turn it on, and BANG! 

Not sure what exactly you are referring to. What will break? The scope? Isn't that only a problem when trying to measure the high side? I mean, my scope is still fine after measuring stuff on the Grounded Aixun. 

Maybe even another screw terminal for the 5V if someone wants just use a wall wart or something.

+1

It could also be a push-pull transformer driver like the LT3439 or the SN6507.

Yeah, there are various possibilities. The cheapest module I could find on a quick check was ~8€, the SN6507 is ~3€ without jelly bean parts, the LKT3439 alone costs ~10€
Using the module should be roughly equal costs, but saves soldering effort and maybe a bit space on the pcb.

As for isolated mosfet drivers, you could basically keep the LTC4440 + NMOS the way it is and just isolate the input to the gate driver with a jellybean PC817 + pulldown resistor.

I don't think that works because the MOSFET and gate driver need to have common GND and the MOSFET needs GND on the 24V rail as well. Since that currently used gate driver can't be driven with 24V VCC, you have the common ground problem again. Am I missing something?


Ohhh, I just realized that *maybe* there is another problem. I might be wrong, but to me it looks like the tc (and current) measurement as it is done currently, won't work in the SELV szenario because it breaks isolation as well... Isn't the OpAmp comparing the two voltages +/-, that are both referenced to GND? So we have + -> GND (RED -> BLUE), - -> GND (GREEN -> BLUE), out -> GND with common GND again...

[c0d3z3r0]

Just a word of caution about using the BAT54s as ESD protection, specifically on the handle. The leakage current on Schottky diodes is huge, at 25°C and 3V3 it's nominally 50nA and can be up to 10x worse, as it gets hotter it gets worse again. This leakage current will flow through your 470k resistor and can create large voltage offsets affecting the temperature measurement accuracy.

Why would the diodes get hot at all? They are not inside the handle but on the control board.

[drksy]

@c0d3z3r0 I meant in the situation where you use SELV in the picture you made with grounded tip but without the earth connection on heater negative, and then you try to measure the heater waveform with an earth-referenced scope, which then turns that SELV into PELV via the alligator clip!

I don't think that works because the MOSFET and gate driver need to have common GND and the MOSFET needs GND on the 24V rail as well. Since that currently used gate driver can't be driven with 24V VCC, you have the common ground problem again. Am I missing something?

You're right, it might need a 7805 just for the VCC pin of the gate driver. I imagine the gate drive IC, the mosfet, and optocoupler transistor would all live in the 24V domain and use the heater ground for their return paths. The controller PCB pretends they don't even exist, all the main PCB sees is an LED.

it looks like the tc (and current) measurement as it is done currently, won't work in the SELV szenario because it breaks isolation as well... Isn't the OpAmp comparing the two voltages +/-, that are both referenced to GND? So we have + -> GND (RED -> BLUE), - -> GND (GREEN -> BLUE), out -> GND with common GND again...

I'm not understanding this bit. Maybe you can draw it?

Why would the diodes get hot at all? They are not inside the handle but on the control board.

I suppose if some component on the pcb pisses away heat, that would heat up the whole board, and because these are smd diodes, they all start heating up. I do agree that the BAT54S's should be replaced with a low-leakage ESD-specific diode.

I want tot also point out the in the reverse engineering of CD-2BC station, the heater AC winding negative and board ground are both labeled GND. I still don't fully understand what's the best way to ground this station with a DC supply. Let's wait to see what solution Mr Axx comes up with.

[drksy]

I don't know I did something wrong or it should be like that! or it has to be fixed from the firmware, when the handle is not connected,should say that (No Tool) or something like that, even the temperature should not be shown. but as you can see in the photo it shows even the handle type while nothing connected to the board.

Have you connected the inputs for Handle_sense and Stand_sense? But yeah, it really shouldn't output garbage when nothing is connected. I assume there are plenty of improvements to make to the software.

[Pooya]

Yes needs some improvements. I didn't test yet, waiting for 6 pin handle socket. Later I will write the issues here.

[c0d3z3r0]

@c0d3z3r0 I meant in the situation where you use SELV in the picture you made with grounded tip but without the earth connection on heater negative, and then you try to measure the heater waveform with an earth-referenced scope, which then turns that SELV into PELV via the alligator clip!

I don't think that works because the MOSFET and gate driver need to have common GND and the MOSFET needs GND on the 24V rail as well. Since that currently used gate driver can't be driven with 24V VCC, you have the common ground problem again. Am I missing something?

You're right, it might need a 7805 just for the VCC pin of the gate driver. I imagine the gate drive IC, the mosfet, and optocoupler transistor would all live in the 24V domain and use the heater ground for their return paths.

Yes, that's what I meant. Not sure if a 7805 is the best option, though. (heat, but might be not relevant due to little current)

it looks like the tc (and current) measurement as it is done currently, won't work in the SELV szenario because it breaks isolation as well... Isn't the OpAmp comparing the two voltages +/-, that are both referenced to GND? So we have + -> GND (RED -> BLUE), - -> GND (GREEN -> BLUE), out -> GND with common GND again...

I'm not understanding this bit. Maybe you can draw it?

Well, voltages is all about references, right? The main question is: does the OpAmp measure a) voltages at +/- against GND or b) voltage at + with reference to -  and then compare them?
If a) then this won't work in an isolated GND szenario because there is no connection between OpAmps GND and the TC GND.
I believe it is a) and that is why isolated OpAmps exist...

EDIT: it seems to be possible to amplify an isolated signal as well. See page 3: https://www.analog.com/media/en/analog-dialogue/volume-44/number-4/articles/measuring-temp-using-thermocouples.pdf

Why would the diodes get hot at all? They are not inside the handle but on the control board.

I suppose if some component on the pcb pisses away heat, that would heat up the whole board, and because these are smd diodes, they all start heating up. I do agree that the BAT54S's should be replaced with a low-leakage ESD-specific diode.

Yeah, I would do that, too. However, JBC also used BAT54s 

Another thing I noticed in the CB-2BC schematics: JBC measures leakage current to EARTH o.O Maybe because it influences TC measurement as well...

[Warhawk]

As a guy from the semi industry, I often see engineers using schottky diodes such as BAT54 for protection. This is a bad design practice. They really leak current that varies with temperature. This leakage also differs from manufacturer to manufacturer. I've seen real pre-production designs failing specs because of it (analog front-ends and opamps).

 just my 2 cents

[sorin]

As a guy from the semi industry, I often see engineers using schottky diodes such as BAT54 for protection. This is a bad design practice. They really leak current that varies with temperature. This leakage also differs from manufacturer to manufacturer. I've seen real pre-production designs failing specs because of it (analog front-ends and opamps).

What do you suggest as a replacement?

[c0d3z3r0]

I took another look at suitable isolated DC/DC modules. There are even cheaper variants than I found initially (~8€).
Assuming ~300mA output will be sufficient (probably even lower is fine), you can get modules starting at ~2.50€ (Mouser [1], didn't check DigiKey) e.g. IEB0124S3V3, IEB0124S05, PDSE1-S24-S3-S, PDSE1-S24-S5-S, CRE1S2405SC (SMD option!), ...

Most datasheets recommend a EMC/EMI circuit (Y cap, input/output filter caps, inductor). Maybe one might just waive it in a private project, but it's probably better to do it "right", since it's just a few components more and still less components than using SN6507.

[1] https://mou.sr/3QL9cZI