JBC-mini - a low cost T245 handle controller

[poorchava]

I want a JBC soldering station. I've wanted one since I had a chance to use it at work. Unfortunately in EU they are expensive as hell, and even more so when you live in eastern europe - the most basic station is around 350€. But a T245 handle and a C245 cartrisge cost only around 100€. So I figured, that if I can build a controller for under another 100€, it will make sense.

I have analyzed all the info that I could find (teardown video, some closeup photos, a cartridge crosssection done by some chinese guy and this thread: http://dangerousprototypes.com/forum/viewtopic.php?f=56&t=5264 over at dengerousprototypes forums. I came up with block schematic like this:


This is where I am right now:



This is a very solid piece of ziptie engineering, but it's just temporary.

Power supply:

I's based on a huge 300W 12V halogene transformer, which I've bought for a few euros a few years ago. It will power the heater. This one is just temporary, as the original JBC uses a 150VA one with 23.5V secondary. 12V is being used instead of 24V, because I already had the transformer and because I could find parts that would work at 24V in my stash. Logic part is powered from a separate small 12V transformer (no idea where it came from). In final design I'll use either custom wound one, add my own low power winding by hand to an off-the-shelf transformer or use 2 separate ones.


AC switch (bottom of the stack):
It uses 2 back to back connected N-MOSFETs same as the original JBC. I didn't have enough info (photos etc) to clone the drive circuit, so I came up with my own. It uses a charge pump and two optocouplers - one for turning the FETs on and the other for turning them off. I'm pretty confident, that original station uses something along those lines too (2 optocouplers can be seem on the teardown video). Entire prototype is made out of junk: mosfets came from some PC motherboards, optocouplers from some blown SMPS. The rest is chickenfood. I have tested it with some power resistors, and it generally works as intended.


Logic power and zero crossing detector (2nd from the bottom):
Pretty simple 30V linear preregulator (zener + power transistor) to protect the circuit from surges and higher transformer output voltage when lightly loaded (an idea borrowed from user sparkybg over at DP forums), followed by an MC34063 (from junk pile) which is calculated for 6V@400mA output.
This board also contains a zero crossing detector which uses a full bridge and an optocoupler to generate positive spikes at zero crossing. I may change this to omit the optocoupler and sense the zero crossing from the transformer that powers the control logic.


Thermocouple amplifier and voltage reference (top board):
This one is not finished yet, as I don't have the opamp (MCP6V02) and cold junction sensor (MCP9701A). It has it's own 3.3V supply and a reference voltage source (TL431 - from junk). I expect to finish that until the end of this week.


CPU:
CPU is gonna be a Cortex-M0 (STM32F030). This board contains an STM32F051, but it's not a problem, since F03x and F05x are pin to pin compatible.


Interface:
Right now I'm working on the interface, which will comprise of 3 7-segment displays for temperature readout, and a rotary encoder for temperature setting.

The only relatively non-jellybean components are the MCU and the thermocouple opamp.

I'll post some updates as I go.

[bdivi]

Why not just replicate the project from Dangerous Prototypes ?

I have one working and the parts cost is around EUR20. Hardware and software are open so you can play with as much as you want.

[SeanB]

If you look at that secondary you will find it is made from 2 parallel wires, so you can rewire it to give you a 24VAC power rail and have a centre tap as well to power the logic with a simple half wave rectifier.

[Dago]

Why not just replicate the project from Dangerous Prototypes ?

I have one working and the parts cost is around EUR20. Hardware and software are open so you can play with as much as you want.

It is not for a JBC handle. They have an integrated thermocouple where you need to stop heating every once in a while (usually on mains zero crossings) to read the temperature. This enables the thermocouple to sit very close to the tip.

That is the reason why JBC is so good

[amyk]

Looks similar to the Hakko T19 scheme, same thermocouple+heater style. Maybe this controller would work too: https://www.eevblog.com/forum/reviews/interesting-936-to-t12-(fx951-integrated-tip)-conversion/

[poorchava]

sparkybg from DP forums is working on what he calls "UniSolder" which is capable of using Hakko T17/T19, JBC C245 and even JBC tweezers. I like his ideas and I have borrowed a few of those (like the thermocouple amplifier selection for example) but it's too complicated. It also feeds the heater with rectified but not filtered voltage, which means large drop on the diode bridge - I believe my design is much more efficient. I also aimed at using as generic parts as I could - MC34063 for the switcher, TL431 for reference voltage, LM1117 as LDO and so on.

@amyk:
yeah, it's similar, but the internal connections of the tip are different and thermocouple too. I don't remember specifics, but I believe the difference was, that the heater was actually grounded (i.e. connected to outer sleeve of the tip) and so was the thermocouple, which makes things easier. You can see on the picture in first powt, that JBC kind of turned that around. If the inner sleeve was on the outside (and hence - system ground), this would have been the same as Hakko.

As for the winding - it's not bifiliar, but rather trifiliar (3 wires in parallel), but they are wound as single output. If the insulating enamel on the magnet wire is not damaged, it may work. The I'll check that tomorrow.

[SeanB]

Thought as much, it is hard to wind a toroid with thick wire, so they often have a few in parallell. You likely can then use one as the logic supply and the other 2 in series for the heater supply. You will easily have enamelled wire insulation that will withstand 50V between the windings.

[poorchava]

I generally had in mind, that this transformer has seen A LOT of abuse over past few years, including some dead shorted loads, overheatings etc, so the insulation may have been damaged.

[SeanB]

Any insulation degradation would be a lot more likely to result in a shorted turn. Very noticeable as a burnt out transformer.

[Rufus]

The most difficult part of a driver for JBC cartridges is the thermocouple amp.

In the configuration you have the amp has around +/- 1.5v on its input while you are driving the heater. If you don't turn off the MOSFET switch exactly at zero current you get an inductive spike from the leakage inductance of the transformer some of which the amp will get to see. The amp needs to recover from these overloads before you can measure temperature and you have to hope the gross overloads don't upset the internal auto-zeroing.

For the best temperature control you need to measure on every mains zero crossing and quickly decide if the heater should be driven for the next half cycle or not.

The cold junction for the thermocouple is at the hand piece cartridge contacts. The hand piece warms significantly. A guess at hand piece temperature from a thermal model in software would do more than cold junction compensation on the control PCB. Taking a general ambient temperature into account in the model would help a bit.

[poorchava]

The most difficult part of a driver for JBC cartridges is the thermocouple amp.

In the configuration you have the amp has around +/- 1.5v on its input while you are driving the heater. If you don't turn off the MOSFET switch exactly at zero current you get an inductive spike from the leakage inductance of the transformer some of which the amp will get to see. The amp needs to recover from these overloads before you can measure temperature and you have to hope the gross overloads don't upset the internal auto-zeroing.

There is a clamping circuit on the input consisting of series resistor and schottky diodes to ground to shunt anything >0.3V or <-0.3. At those consitions the output will obviously saturate, but I think that the opamp will manage that. We'll see when I get the parts.

For the best temperature control you need to measure on every mains zero crossing and quickly decide if the heater should be driven for the next half cycle or not.

I plan on synchronizing entire logic to mains crossings. uC will run from internal oscillator, but measurement and everything else will be done on mains crossings.

The cold junction for the thermocouple is at the hand piece cartridge contacts. The hand piece warms significantly. A guess at hand piece temperature from a thermal model in software would do more than cold junction compensation on the control PCB. Taking a general ambient temperature into account in the model would help a bit.

The sensor on the board is indeed more for ambient temperature than anything else. I think it's possible to estimate temperature of the handpiece junction based on ambient temperature, time since the tool was picked up from the stand and tip temperature.

It think it may also be the case, that the contacts inside the handpiece are the same metals as used in the cartridge, in which case there would not be any cold junction inside the handpiece, and the real cold junction would be at the connector.

[poorchava]

Interface board is ready.
Display is a common anode type. Drive circuit is a standard one: BSS84's for switching anodes, 74HC595 for driving segments. Current limiting resistors are 160R (entire thing runs at 3.3V). There is also a rotary quadrature encoder with a pushbutton and 5 LEDs for debugging. Now I will start wiring up the uC board. Despite STM32F0 being able to progrtam it's own flash I plan to have an external eeprom for calibration data and storing temperature.

[amyk]

yeah, it's similar, but the internal connections of the tip are different and thermocouple too. I don't remember specifics, but I believe the difference was, that the heater was actually grounded (i.e. connected to outer sleeve of the tip) and so was the thermocouple, which makes things easier.

Looking at the diagram again, it seems you have heater, thermocouple, and common connections to the tip, which means the controller design won't be much different from the 936 clones that use a thermocouple too. At first I thought it was more like Hakko's integrated tip which has only 2 terminals and the thermocouple and heater are wired in series.

I think a triac would be better for control than a MOSFET, since they turn off automatically at the zero crossing.

[poorchava]

Another update: the amplifier arrived and I have put it all together. The circuit generally works, but my choice of BAT54 as input clamping diode seems to be poor. The one which is connected with anote to siangl line and cathode to ground is leaking quite a lot and it loads the thermocouple. I will have to find another diode or use much larger series resistor. Other than that the amplifier works so far. I will post pictures later.

Btw, I've made an error on the diagram in first post, inner sleeve and central rod ('core') are the other way around, so the thermocouple voltage must be measured between outer sleeve and the central rod.

Choice of mosfets instead of triac mau have been for two reasons: One - triacs have higher voltage drop than 2 mosfets in series do. Second - you cannot control triac turnoff time. You'd have to turn on somewhere in the middle of the cycle, and then triac would conduct until voltage falls below threshold. With mosfets you have complete control over turn-on and turn-off events.

[C]

Just some Ideas

What if you added a balanced RFI/EMI filter. The ac switch and transformer winding on one side. The amp  and the leads to the tip on the other side.
Should remove some of the ringing and transient noise from the amp.

Next I see an AC signal that is offset from ground by a DC signal. The average of an AC signal is 0 so by averaging samples you would get the DC signal.

If you added some Active cancellation to the amp you could keep the signal in the input range of the amp. 

How much does the heater resistance change with temp? If you had some feedback the heater temp could help make fast temp changes better.

C

[poorchava]

I think that for maximum rms currents over 6A the filter would be seriously large,  but I'll look into that.

AC signal is offset from system ground only by the thermocouple voltage.  The voltage drop across wires is around 25mV or so at 2A through the heater. As for the heater resistance temperature coefficient,  I'll check and see.

[0xdeadbeef]

Probably not what you want to hear, but but just for info: you can get a new JBC station with T245 handpiece plus two cartridges for less than 250€...
Ok, it's analog but still tempting.

eBay (including VAT and shipping in Germany)
http://www.ebay.de/itm/JBC-Lotstation-BT-2BWA-analog-mit-Lotkolben-T245-A-und-Zubehor-/231246976280
Webshop (including VAT, excluding shipping)
http://www.chiemtronic.de/produkte/jbc-loetstationen-loettechnik/analoge-jbc-loetstation-bt-2bwa/

[poorchava]

A small development:


The ac switch and zero crossing detection is fully working. I also wrote software to control the display and the encoder (which turned out to be really crappy and required additional filtering and an external schmitt trigger). I still have some glitches at minimal and maximal duty cycles, but those are only software quirks.

Now I have the analog part left.

[poorchava]

Yesterday I found some time to work on the project. Right now the setup looks like this:




I have put entire circuit together more or less. Some conclusions:
-zero crossing should be sensed from the same winding which powers the tip. Sensing from another transformer (the small one that powers logic) introduced phase shift between the zero signal and the actual zero.
-the opamp behaves well when driven into saturation and desaturates reasonably quickly. I may add some load th the opamp output to speed up coming out of positive saturation
-replacing BAT54 with 1N4148 as opamp input clamping diodes helped.
-I  must have miscalculated necessary gain, because the opamp hits the rail at about 300 degrees
-powered from 12V transformer, this is already powerful as hell and heats up fast
-analog part is quite stable even without software avaraging , i'm getting about +/- digit readout fluctuation

Interesting part:
JBC tip CAN be used as a weak substitute for a light bulb. By some software bug I have managed to get the tip glowing hot. And it still works wihtout proble and is still getting wetted by solder perfectly. This stuff is really impressive... I added a block of power resistors in series though, to limit the power.

[poorchava]

It works! Right now only with very crude proportional regulator,  but it keeps the temperature within 2 degrees from setpoint.

I will now look for an enclosure and design the board for ordering :-)

Would anyone be interested in buying a pcb,  a kit or a fully built station? I won't be asking alot, just to return the material and shipping costs.

[bitshape]

Would anyone be interested in buying a pcb,  a kit or a fully built station? I won't be asking alot, just to return the material and shipping costs.

If you keep it very low, there could be someone here (not from Europe) who wants to buy a kit.
Because in Europe you can get the analog  JBC BT-2BWA with Handle & 2 chisel Cartridges for 249,-  Euro's. Without the Handle and the 2 Cartridges you pay plus/minus 150 Euro's. The station is new and fully factory warranted.
I'm personally more interested in a analog BT-2BWA or digital CD station hacked to use with the 2 JBC TweezerTools. 

Probably not what you want to hear, but but just for info: you can get a new JBC station with T245 handpiece plus two cartridges for less than 250€...
Ok, it's analog but still tempting.

eBay (including VAT and shipping in Germany)
http://www.ebay.de/itm/JBC-Lotstation-BT-2BWA-analog-mit-Lotkolben-T245-A-und-Zubehor-/231246976280
Webshop (including VAT, excluding shipping)
http://www.chiemtronic.de/produkte/jbc-loetstationen-loettechnik/analoge-jbc-loetstation-bt-2bwa/

Also available here: http://weidinger.eu/shop/loettechnik/jbc/jbc_loet-_und_entloetgeraete/jbc_loetstationen/wl26830

[dannyf]

It seems very complicated. I have done solder stations with A1321 handles (4 wire, with thermcouple as a temperature sensor). Power comes from a laptop power supply (20v), and NE5532 as the thermocouple amplifier (with some avrs having 20x / 40x adc, you don't need that amplifier at all). No display however, other than an RGB led to indicate status.

Temperature kept within a 1c or so.

I may build one with 4-digit led display just for fun - the firmware is already written.

[poorchava]

It is not complicated. It's only a prototype, so it looks the way it looks. Esentially it's 2 MOSFETs, 3 optocouplers, an opamp a microcontroller, some shift registers, basic power supply and a handful of passives.

[poorchava]

I'm personally more interested in a analog BT-2BWA or digital CD station hacked to use with the 2 JBC TweezerTools. 

Dunno if you're joking or not, but hot tweezers are from hardware standpoint two separate tips

[bitshape]

I'm personally more interested in a analog BT-2BWA or digital CD station hacked to use with the 2 JBC TweezerTools. 

Dunno if you're joking or not, but hot tweezers are from hardware standpoint two separate tips

No, not joking.  Read more carefully what i said.
 And maybe you can make your controller compatible with the JBC Tweezer-tools, or must i buy and use 2 separate controllers from you for one tweezer-tool (with 2 tips)?