Youyue 858D+ some reverse engineering + custom firmware

[bitseeker]

Welcome to the forum, maximilliangreat. Thanks for the teardown of the new version. It's cool to see how modern it has become.

As for the capacitor, it looks like it's just filtering the power rail for the fan. So, unless you notice some problem with the air flow, I wouldn't worry about it. Of course, if it bothers you, go ahead and change it. Any 105°C cap of the same capacitance and the same or greater voltage rating will work. Be sure to install it in the same orientation, too.

[Magiciaen]

I have an Atten 858D+ with Holtek uC, and came across this adaptor board for installing an Atmega 168/328. I thought it might be useful for some folks:

https://oshpark.com/profiles/cn

I rebuilt my unit with high quality components, including the handpiece, which uses an ebm-papst fan with tach output (it has greater air displacement as an added benefit). Attached is my progress thus far. Instead of installing this adaptor board though, I've decided to make an entirely new board for the whole unit with OLED display. The tach wire needs some signal conditioning circuitry in any case. I'll post all the schematics and code over here when I'm done.

[bitseeker]

A complete replacement PCB with OLED. That sounds interesting. Looking forward to seeing the results.

[Magiciaen]

A couple of things I was thinking of:

1) Fan control - it's straightforward to implement a rotary encoder to control the fan speed, but I'm not sure it's really a benefit. PWM of the fan would eventually damage it, so you'd be controlling a digital potentiometer like the AD5290, which would directly replace the pot in the existing circuit. Those kinds of digi pots have current limits to them though, about 5mA continuous. The 2N3904 in the circuit sees about 3mA from the potentiometer at the minimum setting, so that's ok. With a digital pot you can directly switch the fan to max for cooldown, but as others have already done here, you could just short out the potentiometer with a transistor. I'd like a push button 'mode' switch integrated into the rotary encoder, but you can also get potentiometers with those kind of momentary switches. What route do you think is best? One way or the other doesn't really affect the design/build time.

2) Bistable power relay - will definitely integrate one of these to cut off the heater from the mains when in standby and in any kind of fault.

3) I might replace the heating element with the one from the Quick 861DA, which is 1300W apparently. It will fit into the 858D handpiece with a bit of modification. The ebm-papst fan delivers over 200 lpm at at the maximum setting, so potentially, this thing could have the performance of the Quick, and you'd have access to all the different nozzles.

[bitseeker]

Bistable relay, as in latching in both on and off positions? I'd probably opt for a normal one so that the relay can't get stuck in the on position (e.g., if the controller dies).

[Magiciaen]

That's a very good point, thanks. Have decided to go ahead with the rotary encoder, as it can also be used for scrolling through menus. The whole circuit is going to be a lot more elegant than it was before I hope. A summary of the features:

- 18F4550 uC
- 128x64 OLED display
- Rotary encoder with pushbutton
- AD5290 digi pot
- MAX31855 thermocouple converter (replaces old op-amp circuit)
- Better power supply solution (30 VAC toroidal transformer -> 28 VDC -> 12.5V DC-DC converter -> 5V DC -> -5V charge pump)
- Fan tachometer measurement + Fan voltage from ebm-papst fan (display of 1->8 pot setting, lpm, rpm and voltage)
- 4-layer board with proper isolation slots
- Proper mains design, with non-latching relay isolation, good heatsinking
- LED indicator and cradle override switch on handle

The code will be directly based off the exceptional PD/PID control and safety features written and worked out by madworm and others here.

Hopefully, it won't end up costing the same as an 861D 

[oolloo]

That Samsung matches the pinout for power/gnd, and the fact that pin 4 is input-only also matches how it's used in the circuit. ADC is also available on the right pins.

A little more searching on the Internet about how to read/write those MCUs (the phrase "tool mode" in particular) lead to a forum about programmable remote controls, and eventually a patent and programming software and utilities (written in Visual Basic, it seems) for them... anyone want to do some more RE'ing with all this info?

I just received one 4 years after you guys off amazon, its chip is 20 pin samsung s3f94c4ezz-dk94

is this one that can use the board directly or do I have to snip this and add that and solder a trace?it says Sigma R700  smd rework station greek sigma symbol logo , came in a fancy card board box plastic foam inside for compartments, unfortunately gun wasn't wrapped some fingerprints and micro scratches, gun wire attached to case not removable, same gun with fan inbuilt.

I opened it to find the samsung and 2 other chips the samsung is in a socket, the other 2 are soldered on.

they are :
board id IC2
 8 pin
 mounted vertical parallel to 20 pin with markings:
ATMLU442
02C  M   B
4X0437D


next chip also 8pin
mounted horizontal
board ID IC 3
texas instruments
79W0G1FE3      hard to make out last 3 digits
LM358P


I've not powered it on yet looking for the checking grounding etc you-tube.,   

also the gun harness and fuse as well as power are held in with white hot glue, none of it is stuck to case all 3 points rotate,with resistance.

I had read this site before ordering, and was hoping for the Atmel, as it says its a direct replace right?

I was also hoping that these Chinese factories (as they often do) implemented the mods mentioned here, obvi because they're pirates.  certainly the description on amazon seemed to match.

how can i secure these 3 through-hole ? 
add more glue?
remelt glue?
use the station on itself?

[stj]

if your really safety concious,

use 2 relays for the heater,
wired in series - one using the normally open contacts and the other using normally closed.
and have feedback from both.

this is how gas valves are switched in professional boilers.
if the computer crashes or the buffer fails and all outputs go high OR low the output is off,
and if either fails directly, the feedback detects it and the mcu shuts it all down.

[oolloo]

Anyone have a source for the adapter boards to suit Samsung S3F94C4EZZ ic's ?  I PM'ed wguibas a while ago but haven't heard back from him, and he seems not to have been active on this forum for a few months.

Tried the attachment under this post: https://www.eevblog.com/forum/reviews/youyue-858d-some-reverse-engineering-custom-firmware/msg782035/#msg782035

What worked for you?

 I have a clone called Sigma R700 from Amazon, looks nice (no rear bezel, only front)
has the same samsung chip as yours. 

[Magiciaen]

Thanks for teaching me! Fantastic idea, I'll definitely put this in!

[steve1515]

Thanks DGM, that's an excellent summary this thread has been lacking!

The important bit you're missing from me is the name of the 'unknown name board' - as shown in the attached picture, it is marked AF858D, and dated 20161212.

In case you want to add it, the details of the adapter board I developed for this PCB are below:

• Schematic: attached
• PCB: https://oshpark.com/shared_projects/NQGwEEH0
• Features:
• TL431 reference
• Vfan and current sense support
• ISP header
• MLF-32 ATmega package

The ADC scaling factor for this board when used with a 2.5v reference was 2.4 in my case.

I just got one of these 120V rework stations from Amazon and it came with one of the AF858D boards with date code 20161212 just like yours. I want to mod it using your mod board, but I was hoping you could provide your firmware with edits that you made a long with a list of parts that you used. Would you by any chance be able to post it up here?

Thanks! 

[steve1515]

Thanks everyone for looking at the Rev. A sch and pcb and giving feedback. I corrected the LED digits (misread the F/W assignments) so PCB Rev. B now....

Hi floobydust. Could you please post your latest PCB revision schematic? I was able to find your Rev A linked, but I couldn't find B even though you mentioned it.

Thanks!

[floobydust]

I had already updated the 858D+ station schematic I drew with the LED digits fixed, so it should already be the 'Rev. B' as I was calling it. Reposting it here.
Note Line and Neutral were reversed (at the switch) in the unit as I received it, so turning it off meant hazardous live was still present at the hand wand and connector from triac leakage.

[steve1515]

Thanks. I did notice that one, but I was referring to the one for your adapter board. I see rev A here: https://www.eevblog.com/forum/reviews/youyue-858d-some-reverse-engineering-custom-firmware/msg1206460/#msg1206460

Sorry I should have been more clear. I was looking for the Rev B of the adapter PCB or what ever was the latest. Would you be able to post that?

Thanks 

[floobydust]

I was trying the Mega328PB and interrupt-driven tachometer for the fan and then I thought why not just use a Arduino Nano and then the project was all over the place.
What sections would you like to look at, I can dig for that.

[steve1515]

I was wanting to look at the BLDC fan-speed detect circuit that you ended up going with. It may be what you have in Rev A and if so, then I'm all set. I wasn't sure if you had any updates though.

[floobydust]

This tach circuit works very well. But it needs a MCU timer/counter or an interrupt becuase of the 1/F relationship to RPM.
The fan curve has RPM/speed and airflow as a complex relationship that is non-linear.

[steve1515]

This tach circuit works very well. But it needs a MCU timer/counter or an interrupt becuase of the 1/F relationship to RPM.
The fan curve has RPM/speed and airflow as a complex relationship that is non-linear.

Hmm. Ok. I was thinking that I was going to read the output of the circuit as a voltage on the ADC. I suppose it would make sense to count the pulses. But that confuses me a little... Madworm's original circuit also pulses, but is read on the ADC.

I suppose that the ADC would show higher voltage if more pulses are seen during it's sample period??

Is there any issue with just reading the voltage at the ADC vs counting the pulses?

[Magiciaen]

This will probably be of limited interest to everyone, but I've knocked up my own version of the 858D. The board fits in the existing case, and uses the same transformer, but has OLED display and rotary encoder control. The thermocouple is measured with a MAX31855. It also has an accurate tachometer circuit. I used a EBM-Papst RLF 35-8/14 N fan in my unit, that I bought 'new' off eBay. It had a tachometer wire, but apparently they aren't sold with one ordinarily. It seemed the fan was opened and a wire was tacked on to some transistor in there. The output from the fan is only 50mV or so, but my breadboarded circuit gave spot on measurements of rpm. The handpiece definitely has to be rewired to work with the circuit, since I needed to connect PE to ground, and the thermocouple is no longer connected to ground. The MAX31855 tests the thermocouple before using it, and it shouldn't be connected to ground for that testing procedure to work. Per stj's excellent suggestion, there are two complementary non-latched isolation relays. In all, the circuit should make the 858D a lot safer and more fun to use, in addition to it being a lot more accurate. It should be able to drive the heater from a Quick 861DA, so that can be retrofitted to the handpiece if one wants more power:

https://www.somersetsolders.com/quick-861da-replacement-element/p657

I've almost finished routing the PCB, and after I've built it, I guess the real fun will start with programming the software. I'll post the entire project with code, schematics, CAD files when it gets done. It's definitely possible to cheap out on parts and build it on a budget, although that isn't what I've done. My brother bought me my 858D as a birthday present, not realising the shocking state of its insides. Rather than throw it away, I decided to make this circuit, as I am a sentimental chap.

[floobydust]

One big problem with the 858D+ is AC mains noise on the thermocouple signal.
The hand-wand cable has the thermocouple conductors (mV) right next to the heating element conductors (100's V).

I don't have success with Maxim thermocouple IC's like MAX31855 and older MAX6675.
They float the thermocouple in order to do fault detection and you get false faults due to RF pickup or mains hum.
Ensure you have ferrite beads/common-mode choke and RF caps at the TC inputs in your design. A nearby cell phone causes readings to go all over the place.

[Magiciaen]

Really good tips, many thanks! Will definitely add these.

[floobydust]

This tach circuit works very well. But it needs a MCU timer/counter or an interrupt becuase of the 1/F relationship to RPM.
The fan curve has RPM/speed and airflow as a complex relationship that is non-linear.

Hmm. Ok. I was thinking that I was going to read the output of the circuit as a voltage on the ADC. I suppose it would make sense to count the pulses. But that confuses me a little... Madworm's original circuit also pulses, but is read on the ADC.

I suppose that the ADC would show higher voltage if more pulses are seen during it's sample period??

Is there any issue with just reading the voltage at the ADC vs counting the pulses?

The "data slicer" I use to pick off fan speed, gives a pulse-width and frequency that changes with the fan speed.
I didn't try turning it into an average DC signal for an A/D.

I found that fan speed is not so useful....
For safety, the temperature climbs very fast if the fan is not working. The original 858D+ firmware looks at this rate of change and will shut down if it climbs too quickly. Take a look at the derivative of temperature term.
You could take fan speed and model airflow for super precise control, but as I said earlier you still don't know what nozzle (restriction) is being used. Each nozzle could get characterized and put on a menu. But you still could be close/far away from the part being heated, so end temperature is not precise here.

[Magiciaen]

The "data slicer" I use to pick off fan speed, gives a pulse-width and frequency that changes with the fan speed.
I didn't try turning it into an average DC signal for an A/D.

I found that fan speed is not so useful....
For safety, the temperature climbs very fast if the fan is not working. The original 858D+ firmware looks at this rate of change and will shut down if it climbs too quickly. Take a look at the derivative of temperature term.
You could take fan speed and model airflow for super precise control, but as I said earlier you still don't know what nozzle (restriction) is being used. Each nozzle could get characterized and put on a menu. But you still could be close/far away from the part being heated, so end temperature is not precise here.

That's a really interesting topic. Since I'm able to measure rpm, fan voltage and temperature derivative, together you should be able to characterise nozzle restriction. The fan voltage vs rpm curve would yield different flow (lpm) along the curve at different backpressures. I think the voltage vs rpm curve itself would also change with backpressure. As the temperature derivative would modulate with flow, all together you should be able to deterimine backpressure, and hence nozzle size. You wouldn't need to enter the nozzle size manually I'd imagine.

[DGM]

That's a really interesting topic. Since I'm able to measure rpm, fan voltage and temperature derivative, together you should be able to characterise nozzle restriction. The fan voltage vs rpm curve would yield different flow (lpm) along the curve at different backpressures. I think the voltage vs rpm curve itself would also change with backpressure. As the temperature derivative would modulate with flow, all together you should be able to deterimine backpressure, and hence nozzle size. You wouldn't need to enter the nozzle size manually I'd imagine.

This is similar to a tangent I wasted a bunch of time thinking about:
https://www.eevblog.com/forum/reviews/youyue-858d-some-reverse-engineering-custom-firmware/msg1448669/#msg1448669

I had intended to simply design a thermistor anemometer for getting mass flow information.  Like I said in the post, I'm pretty sure that between power and RPM, there isn't enough information to unambiguously derive mass flow.  If pressure were measurable in-process, things might be different, but that's not really an option.  I got a spare blower to experimentally verify that claim, but I never did. 

[Magiciaen]

One big problem with the 858D+ is AC mains noise on the thermocouple signal.
The hand-wand cable has the thermocouple conductors (mV) right next to the heating element conductors (100's V).

I already replaced the cable for my wand to a screened 20AWG version, as the original cable isn't safe. It was probably 26AWG if I remember. My cable is 8.1mm in diameter, which is at the upper limit of practicality, but this sort of cable would be the ideal solution:

https://www.conrad-electronic.co.uk/ce/en/product/486018/CPC-2-2xdata-1xpower-720045000-Kabeltronik

Separate mains and data sections, with screening for the data lines. A bit too thick at 12mm. But then again, the cable on a hot air station with blower in the main unit is about that thickness.

Thanks again for your heads up about the noise issue, I've now added a filter I designed and simulated, and it seems to work virtually at least. (Two common mode chokes for low and high RF bands, ferrite beads for differential noise, cascaded 1uF X2Y capacitors with matched series resistors, 10nF X2Y capacitor, and biasing of the floating signal lines between 3.3V and ground using 1M resistors.) Some practical testing will definitely be needed to see what works.