Really universal soldering controller

A place to document your own projects.

Re: Really universal soldering controller

Postby sparkybg » Thu May 14, 2015 6:55 pm

mrjoda wrote:For example AD7730. Yes, is it overkill like hell but you need almost none external parts.


AD7730 is no good for what I am doing, because:

- it is too slow - I have around 200 microseconds to stop the power, wait the TC voltage to come to it's real value, connect the amplifier to this voltage, wait for the amplifier to set it's putput to what I want to read, take the measurement from the adc, disconnect the amplifier from the TC, run the PID, and evenually turn the power back on. The TC on some irons is in series with the heater, remember?

- the input must be more flexible - I don't have much wires available in the connector, and it must be able to ground both inputs while the power is on. It must be able to invert the inputs, to connect them in single ended or differential mode, and negative input common mode voltage - some iron tips are connected in a way giving negative voltage (JBC C245 for example) and differentioal input is used to read them, and in the same time other irons (JBC Microtweezers) are connected in a way that the same two wires to the same connector are used as 2 separate single ended inputs.

- I need more gain steps in order to use full input of the ADC, no matter what TC is on the input - some tips TC gives 40mV at 500 degrees Celsius, some gives just 4-5mV.

- I need more offset steps. With 10bit offset steps and 10bit ADC, you have direct equality between ADC reading and offset setting - if you set the offset to a value of 100, the input falls with exactly the same amount. I am using offset correction not because the amplifiers got offset, but because PTC measurement needs offset correction to exclude the zero temperature resistance in order to use the fill ADC range for temeratures of 0-512 degrees Celsius.

And yes, I agree there are lots of parts there, but after all it is not so complicated, nor too expensive. It just needs more time to assemble.

I must admit that I didn't consider ready made analog frontend chips for this project. I will have this in mind in the future. I will even take a look at AD's site if something can be applied to a project like this.
sparkybg
Hero Member
Hero Member
 
Posts: 669
Joined: Tue Feb 26, 2013 3:30 pm
Location: Sofia, Bulgaria

Re: Really universal soldering controller

Postby mrjoda » Fri May 15, 2015 4:40 am

I collect lot of possible solution how to build controller. Universal controller is no my way so i searching possible solution for T12/JBC. One of them is use these lovely ADC. I am realy suprised because they are so cheap (a few types :) ) !

Better as Analog maybe will be some No Latency Delta Sigma ADCs from Linear ( LTC2410 11€ maybe ? or fast LTC2440 14€ ). Advantage is autozeroing ADC and realy little noise.


I dont want you dishonor your work,god no, its nice and i like it. I write you only another possible way. Maybe someone read this and try it :)


I dont have fear from lot of parts :) But this is not beginner project maybe will be good give warning to front of your topick. You spare yourselft from lot of "stupid" question" from beginer. I publish some project and trust me, this question will come :)


You are using PIC32. This is fu... powerfull MCU. How much power (in %) you need for one cycle (switch,measure, computation, control, display) ? Because this is like cortex M4, you may play video on this beast.
mrjoda
Newbie
Newbie
 
Posts: 39
Joined: Sun Mar 01, 2015 11:08 am

Re: Really universal soldering controller

Postby sparkybg » Fri May 15, 2015 5:15 am

mrjoda wrote:I collect lot of possible solution how to build controller. Universal controller is no my way so i searching possible solution for T12/JBC. One of them is use these lovely ADC. I am realy suprised because they are so cheap (a few types :) ) !

If you are making it for more than one tip, it is already an universal one. :)

I would like to see the T12 project using AD7730 you mentioned. Is it published somewhere?

mrjoda wrote:Better as Analog maybe will be some No Latency Delta Sigma ADCs from Linear ( LTC2410 11€ maybe ? or fast LTC2440 14€ ). Advantage is autozeroing ADC and realy little noise.

I dont'see a reason for using these. Every MCU has ADC inside it these days. My older project can handle both T12 and JBC C245 with single external autozeroing opamp that costs 3USD. It can be made to handle C210 with some modifications.

LTC24xx are also tooooo slow for my approach. Conversion time is 150 milliseconds and more. And you don't need these 24 bits of resolution anyway. The noise at these gains is several magnitudes higher than the LSB of that ADC.

mrjoda wrote:I dont want you dishonor your work,god no, its nice and i like it. I write you only another possible way. Maybe someone read this and try it :)

Yes, there are many possible ways for a single task in electronics. Honor/dishonor is out of question here.

mrjoda wrote:You are using PIC32. This is fu... powerfull MCU. How much power (in %) you need for one cycle (switch,measure, computation, control, display) ? Because this is like cortex M4, you may play video on this beast.

As I said, it uses floating point calculations, waveshaping, RMS voltage/gurrent calculations and so on. This means intensive integer 32bit multiplications, shifts, and so on. For example the millivolts to temperature calculation is done using polynomial with 10 members. It does this calculation using 32bit mantissa floating point numbers and completes it in around 20 microseconds. The whole waveshaping, temperature calculation, PID and so on is completed in around 50-60 microseconds. RMS current, voltage and power calculations are done in arouns 100 microseconds. All this is done between the half periods of the mains voltage, where the voltage is less than around 3 volts. You need computational power for this. This is the bottleneck of the project, and you need a good fast MCU for this. And after all, if it costs under 10 bucks, why should I bother using slower MCU? Computational power is cheap these days. From my point of view - it is better to buy a faster MCU and write the firmware in C/C++, than using slower MCU and do a lot of optimizations and write critical parts in assembler for example.
sparkybg
Hero Member
Hero Member
 
Posts: 669
Joined: Tue Feb 26, 2013 3:30 pm
Location: Sofia, Bulgaria

Re: Really universal soldering controller

Postby mrjoda » Fri May 15, 2015 10:06 am

i couldnt find any word about FPU in DS of PIC32MX564F128H... This MCU doesnt have FPU accelerator ?????
mrjoda
Newbie
Newbie
 
Posts: 39
Joined: Sun Mar 01, 2015 11:08 am

Re: Really universal soldering controller

Postby sparkybg » Fri May 15, 2015 10:26 am

PIC32 does not have FPU, but has 32x32=> 64 2 cycle multiplication, division and barrel shift, so it handles floating point pretty well. Also, I've done some optimisations for polynomial calculations with 32bit mantissa and exponent, and calculation time is more than 3 times lower, compared to same calculation when using built in libraries for single precision floating point, and 256 times more precise. You can see the code in pid.c file.
sparkybg
Hero Member
Hero Member
 
Posts: 669
Joined: Tue Feb 26, 2013 3:30 pm
Location: Sofia, Bulgaria

Re: Really universal soldering controller

Postby Sjaak » Fri May 15, 2015 10:31 am

What is the cost of this soldering controller? It looks quite complete but also costly :)
User avatar
Sjaak
Fellow
Fellow
 
Posts: 3293
Joined: Sun Jan 03, 2010 2:45 pm
Location: Hiero

Re: Really universal soldering controller

Postby mrjoda » Fri May 15, 2015 10:44 am

sparkybg wrote:PIC32 does not have FPU, but has 32x32=> 64 2 cycle multiplication, division and barrel shift, so it handles floating point pretty well. Also, I've done some optimisations for polynomial calculations with 32bit mantissa and exponent, and calculation time is more than 3 times lower, compared to same calculation when using built in libraries for single precision floating point, and 256 times more precise. You can see the code in pid.c file.



I am little suprised because PIC32 looks like adult MCU. I suspect FPU accelerator is basic for better MCU.I dont like PIC at all.

Do you work only with PIC ? Because STM32F4xx with FPU accelerator is a few mile better and faster and you dont need these optimalisation and cost same :) I suspect you using our SQRT... In FPU is it basic operation and take 14 clock cycles (80-100MHz) atc.

i have these question because i check your codes and are... interesting and complex.I have lot of questions because i am curious and one question produce more questions :D
mrjoda
Newbie
Newbie
 
Posts: 39
Joined: Sun Mar 01, 2015 11:08 am

Re: Really universal soldering controller

Postby sparkybg » Fri May 15, 2015 4:09 pm

Two weeks ago there was not any floating point calculation in the firmware of this project, and I didn't had any intention to use floating point in this project whatsoever. Even when I tried to use floating point without my optimizations, I was quite disappointed with the performance of built in floating point library, but after all, when I managed to make the same calculations more than 3 times faster and with considerably higher accuracy, it became an acceptable solution. :)

Microchip doesn't have any MCU with FPU in their current portfolio. When I need FPU, I am using Renesas RX630 family. There are many things that can be made better in the PICs, but there are also several benefits they offer:
- Single cheap programmer/debugger for all Microchip MCUs, from the smallest 8bit PIC10 to the largest 32bit PIC32MZ - no other manufacturer is even close to this.
- No complex programming/debugging settings, no different speeds, voltages and so on - you just plug the programmer, select the davice you are working with, hit a single button to program the device and you are ready to go.
- Only 3 wire programming interface, and all pins can be reused after programming
- Free development IDE and compilers from the same manufacturer
- Easy peripheral initialization and use, both with and without microchip libraries
- Decent and easy to work with documentation.

Both ST and Renesas have a long way to go before making a chip this easy to implement, despite clear functional advantages of some of their chips.

But anyway, STM32 FPU is single precision one. When you calculate 10 member polynomial with values from 10 bit ADC with added offset and other corrections you can reach the single precision floating point exponent limit quite easily, and some care must be taken to address this, which consumes additional CPU/FPU cycles. Yes, it will still be faster, maybe 3 to 5 times faster, but definitely not 10 to 15 times faster as stated in the STM32's floating point manual.

I am making a division, 20 multiplications and 10 additions with 32 bit mantissa and 32bit exponent in around 20-25 microseconds on 80Mhz MCU. If we take out the division, this gives around 2 microseconds for one multiply and one multiply and accumulate, which on 80Mhz CPU gives around 160 cycles. And this includes all - loading, calculation, storage, loop management and so on. It is not so bad for a MCU without FPU.
sparkybg
Hero Member
Hero Member
 
Posts: 669
Joined: Tue Feb 26, 2013 3:30 pm
Location: Sofia, Bulgaria

Re: Really universal soldering controller

Postby sparkybg » Fri May 15, 2015 4:20 pm

Sjaak wrote:What is the cost of this soldering controller? It looks quite complete but also costly :)


Around 70-90 EUR per assembled set of PCB-s. It is still pretty lower then original controller from JBC for example.

All parts are available from Farnell - you can check part prices yourself if you want to.
sparkybg
Hero Member
Hero Member
 
Posts: 669
Joined: Tue Feb 26, 2013 3:30 pm
Location: Sofia, Bulgaria

Re: Really universal soldering controller

Postby ferdinandk » Sat May 16, 2015 3:54 am

This is just amazing. Thank you for making it public.

I've been looking over your schematics, but I didn't quite get everything sorted out, maybe you can help me:

  • Q5, ZD1, ZD2 and the surrounding circuitry look like a linear reg, but I can't find it's output node.
  • does Q6 reduce the ripple on the -0.6V rail, and if so, do you have measurements for that?
  • with all your linear regs you a diode from the emitter to the base, what is it for?
  • I think some electrolytics have the wrong polarity (C22, C8, C11, C4).
ferdinandk
Newbie
Newbie
 
Posts: 30
Joined: Fri Apr 27, 2012 5:21 am
Location: Germany

Re: Really universal soldering controller

Postby sparkybg » Sat May 16, 2015 5:48 am

ferdinandk wrote:Q5, ZD1, ZD2 and the surrounding circuitry look like a linear reg, but I can't find it's output node.

This ensures the rectified voltage never gets under 5V. You need a flat voltage on Vin while measuring the TC, otherwise the rectified voltage penetrates the upper MOSFET drain to source capcitance and gives erratic measurement.
ferdinandk wrote:does Q6 reduce the ripple on the -0.6V rail, and if so, do you have measurements for that?

Yes, Q6 reduces ripple and adds another step of fine regulation to around 0.6V. Both the ripple and the regulation becomes good enough for the opamps using them. I've made a measurements while prototyping, but I don't have the values right now.

ferdinandk wrote:with all your linear regs you a diode from the emitter to the base, what is it for?

Protection from back voltages when turning the power off. It is a common practice. The reverse emitter-base voltage of BJT is usually around only 5 volts, and base-emitter junction can be damaged easily when the output voltage becomes higher than the input on power down.

ferdinandk wrote:I think some electrolytics have the wrong polarity (C22, C8, C11, C4).

Oh, this is because I somehow reversed the footprint of these parts. I will correct this. Thanks.
The polarity is OK on the PCB - it is only reversed on the schematics.

P.S. The polarity should be OK now. I've uploaded the correct schematics. I don't know how this happened. :)
sparkybg
Hero Member
Hero Member
 
Posts: 669
Joined: Tue Feb 26, 2013 3:30 pm
Location: Sofia, Bulgaria

Re: Really universal soldering controller

Postby ferdinandk » Sat May 16, 2015 6:12 am

sparkybg wrote:
ferdinandk wrote:Q5, ZD1, ZD2 and the surrounding circuitry look like a linear reg, but I can't find it's output node.

This ensures the rectified voltage never gets under 5V. You need a flat voltage on Vin while measuring the TC, otherwise the rectified voltage penetrates the upper MOSFET drain to source capcitance and gives erratic measurement.

I'll have to work out how the drain-source capacitance would offset the TC measurement, but I now see what you're doing there.

It's great to see so many discrete circuits, makes the design much more approachable IMHO. I haven't looked at the firmware yet, but if it's as advanced as your hardware sign me up for a pre-order :) Maybe you could order some PCBs and sell them for a profit. What case you were planning to use?
ferdinandk
Newbie
Newbie
 
Posts: 30
Joined: Fri Apr 27, 2012 5:21 am
Location: Germany

Re: Really universal soldering controller

Postby sparkybg » Sat May 16, 2015 7:45 am

You can simulate it pretty easily. Spice simulation works very well simulating this.
Last edited by sparkybg on Sat May 16, 2015 5:02 pm, edited 1 time in total.
sparkybg
Hero Member
Hero Member
 
Posts: 669
Joined: Tue Feb 26, 2013 3:30 pm
Location: Sofia, Bulgaria

Re: Really universal soldering controller

Postby ppsati » Sat May 16, 2015 8:53 am

Hi, Sparkybg.

The OLED display uses the SSD1306 driver?
I´m thinking to use the 1.3 ".

Have you already decided if you gonna sell Unpopulated PCBs.
Say something.
ppsati
Newbie
Newbie
 
Posts: 21
Joined: Mon May 11, 2015 5:54 pm
Location: Portugal

Re: Really universal soldering controller

Postby Erythros » Sat May 16, 2015 9:50 am

sparkybg wrote:BOM will be there soom. As I said, I will update the first post from time to time with new things.


Thanks.

sparkybg wrote:Unpopulated PCBs - maybe. I don't know yet.


Well OK. I understand that it is probably more work than it is worth and nowadays everybody can have PCBs made cheap in China. I have just send some quotation questions to fab houses. (Just saying that people [better to say me] are interested anyway.)

BTW, the stand sensor PCB is designed for which specific holder?

What was your design idea to include SA36-11SRWA?
User avatar
Erythros
Newbie
Newbie
 
Posts: 17
Joined: Sun Oct 05, 2014 5:38 am

PreviousNext

Return to Project logs