Arduino control of large processing Furnace

[master53yoda]

Repost

I have used the Arduino for many projects, CNC Control of a long-arm quilting machine, many lighting controls for shutting off the lights etc in the shop, Temp control for radiant heaters that were manual control originally, Control system for use in converting a central AC system to a heat-pump.

The most complex control was for my Processing furnace for aluminum, It is running on a Arduino MEGA2560, with t-couple sensors in the pool and flue, it controls 2 waste oil burners with variable firing rate,750kbtu to 20kbtu with propane start and stabilization, the waste oil is preheated to 160F for winter operation. This furnace will melt 150-200 lb per hour, total process time including melting and ingot pouring is 2.5 hours.
My processing business is getting to the point I'm going to need to sell it, I'm 72 and it is taking all my energy to run it so I don't get to do much fun machining etc. I sell between 2.5 and 3 tons per year of ingots. I am in the process of rebuilding the control system to include 2' X 3 inch screen and keypad for adjusting set-points etc and viewing operational conditions as well as the ability to manual override if necessary, the old system required the laptop to monitor or change any settings.

I'll try to answer any questions I have other pics and an unedited video.

 

[Lloyd-ss]

Very nice. And complex. I am hoping you have some block diagrams for the whole control system and then some more detailed schematics? It will be interesting to see how you integrated the controls with the feedback sensors. Projects like that can really snow ball and keeping track of the configuration is difficult. Congrats on a successful project!

 

[Lloyd-ss]

@ajoeiam He's got the aluminum furnace thread started. Maybe he will reveal his source of the scrap engine heads.
What alloy do you state the ingots "might be" ? Sweepalloy? Just joking. I imagine the alloys would at least be consistent across an automotive manufacturer, or maybe across several manufacturers? Or maybe your customers aren't particular?

 

[master53yoda]

@ajoeiam He's got the aluminum furnace thread started. Maybe he will reveal his source of the scrap engine heads.
What alloy do you state the ingots "might be" ? Sweepalloy? Just joking. I imagine the alloys would at least be consistent across an automotive manufacturer, or maybe across several manufacturers? Or maybe your customers aren't particular?

I have been running this foundry for over ten years, my total processing has been about 28 tons of finished ingots.
The ingots test out to very close to A356 with, the only thing out of spec is the amount of FE. I have pretty well cured that by using only clean cast (no Iron) it runs about 33% higher in cost but doesn't have valves bolts etc.in the scrap. I use up to 25% clean pistons when may scrap has less than 75% automotive cast.
There is a commercial foundry in Spokane that takes all the wheels, they have certified A356 at about a dollar per lb more than I charge, They need to retain certified samples of each batch of metal, the DOT requires certified A356 be used in structural suspension castings and wheels..
I test about every y 1000 lbs to verify that I'm not getting out of spec to far.
I buy from a local recycling company that is tied into the largest automotive scrapyard in the area. I buy by the boxed pallet weighting between 750 and 1000 lbs.
My customers range from University engineering programs, art galleries and small business casting companies. to the individual home Hobby caster. about half of my sales are return customers and are multi box orders in the 90 to 160 lb range. My largest single order was 1100 lbs to an art gallery my current production rate is 300 lbs per week.
Prior to this fall I Inventoried the scrap metal and poured to order, I am currently trying to go to an inventory of finished ingots, I am still being hit by multiple box orders often enough that i haven't bee able to stay in front of the incoming orders. January and February have typically been slow months, I am hoping by pouring twice per week I can get 8 or 10 boxes in front of the orders.

 

[master53yoda]

Very nice. And complex. I am hoping you have some block diagrams for the whole control system and then some more detailed schematics? It will be interesting to see how you integrated the controls with the feedback sensors. Projects like that can really snow ball and keeping track of the configuration is difficult. Congrats on a successful project!

This may sound a little abnormal but the diagrams etc follow the code for me. I write the code and then workout the components. I came from the commercial HVAC industry and have an extensive mechanical, electrical and combustion background. I have programmed PLC and when the Arduino came out I started the programming in C++ conversion. The the new furnace controller ended up using a mega2560 with a 3.5" screen that was designed to plug into the mega560. as I progressed i found a conflict between the screen and using a SDA SLC chained t-couple boards either would work alone but wouldn't together. I ended up adding a UNO for the t-couples and using SoftwareSerial for RX-TX communication between the boards.
The other problem that I ran into was that the cheap 16 keypads were not stable enough to provide consistent key press reading into the voltage divider for one wire keypad. I went to digikey and found a high end keypad ($45) that is stable within 1 ohm between repeated key-presses.

I am in the process of wiring the system up for testing both the code and the wiring.

Below is the diagram of the old sytem

 

[GreenTwin]

I use to cut up my own aluminum auto rims, and it was a royal pain to do that, and very messy and difficult.
I ran across masteryoda selling his aluminum ingots years ago, and tried a few.
Good material, no gas, and very consistent quality.
I have been using masteryoda's aluminum ingots ever since; good material.

Masteryoda is also an expert on foundry burner tuning, and taught me much of what I know about that.
.

 

[timo_gross]

This may sound a little abnormal but the diagrams etc follow the code for me. I write the code and then workout the components. I came from the commercial HVAC industry and have an extensive mechanical, electrical and combustion background. I have programmed PLC and when the Arduino came out I started the programming in C++ conversion. The the new furnace controller ended up using a mega2560 with a 3.5" screen that was designed to plug into the mega560. as I progressed i found a conflict between the screen and using a SDA SLC chained t-couple boards either would work alone but wouldn't together. I ended up adding a UNO for the t-couples and using SoftwareSerial for RX-TX communication between the boards.
The other problem that I ran into was that the cheap 16 keypads were not stable enough to provide consistent key press reading into the voltage divider for one wire keypad. I went to digikey and found a high end keypad ($45) that is stable within 1 ohm between repeated key-presses.

I am in the process of wiring the system up for testing both the code and the wiring.

Below is the diagram of the old sytem
View attachment 163292

Interesting, something to do for the MEGA . I am thinking of making some automatic contraption for some time, but too many other things are on the list.
I sometimes have the feeling that aPLC is the better choice for such projects, because they come with relay outputs and proper sensor inputs built in.
Would you use Arduino again, or rather go with a PLC when doing it once over?
For a project at work, I paid 80USD for a small PLC with 8 Inputs and 4 outputs. (programming cable plus Power supply antother 30US) and the very simple basic programming software came with it.

basic as in not many features and functions not as in "BASIC" vs "C++". The programming is a ladder type thing.

There is also now Arduino PLCs not as cheap and easy to get, but maybe also interesting.
https://store.arduino.cc/collections/turnkey-solutions

Greetings Timo

 

[master53yoda]

actually the PLC are far more expensive and limited then a mega or an ESP the mega has 52 digital outputs or inputs as well as PWM for variable controls, and multiple analog inputs they direct connect to servos or as well as drives for stepper motors, It can feed inputs to variable drives for large motors. IDE Language has most of the capabilities of C++. The mega2560 is typically less than 20 and the ESP2 is about the same. Manny of the commercial applications have the ESP32 as the base processor.

 

[master53yoda]

I use functions to separate control sequences which really helps keep things simple, I write and test the function then go on to the next function. for example the keypad programing is a function that can be called from multiple places. The menu sections are also separate functions. I use ignition and gas valve operation as a function that is used in a couple of places such restarting the oil burners during the holding pattern.
Art B

 

[timo_gross]

actually the PLC are far more expensive and limited then a mega or an ESP the mega has 52 digital outputs or inputs as well as PWM for variable controls, and multiple analog inputs they direct connect to servos or as well as drives for stepper motors, It can feed inputs to variable drives for large motors. IDE Language has most of the capabilities of C++. The mega2560 is typically less than 20 and the ESP2 is about the same. Manny of the commercial applications have the ESP32 as the base processor.

The PLCs are definitely more expensive, but adding the safety layer to make the inputs electrically "foolproof" to avoid damages and allow for various input voltages is quickly adding up. The outputs also must be "beefed up" that adds some extra too.
With the cheapest relay boards I had some mixed experiences, burned out transistors, "sticky" relays, not very nice screw terminals.
The ESP32 can only tolerate 3.3 V and has very limited output power, so for most sensors and other things additional parts need to be attached. (e.g. relay boards or level shifters)


I found out recently that some of those relay boards work just fine with the 3.3V from the ESP32 When they have the option to power the Relay from a separate power supply. To power the LEDs (Indicator and Optocoupler) the power and voltage is sometimes good enough.

 

[Lloyd-ss]

@master53yoda Quite a rich history with the aluminum ingot processing and selling. The A356 is a robust material suitable for many mechanical applications. And I have to say that it is one of the most unique & involved full process home industries that i have run across. Finding a buyer worthy of taking over the operation might be difficult.

 

[master53yoda]

@master53yoda Quite a rich history with the aluminum ingot processing and selling. The A356 is a robust material suitable for many mechanical applications. And I have to say that it is one of the most unique & involved full process home industries that i have run across. Finding a buyer worthy of taking over the operation might be difficult.

It has been a difficult decision to try to sell it, but it is reasonably profitable for the right person. I think I probably sell more ingots on Ebay then any one else. Ebay also doesn't allow you to sell the store etc. I do have the contact in formation for my whole customer base. But it would need to be the right person.

Art B

 

[Toymaker]

This may sound a little abnormal but the diagrams etc follow the code for me. I write the code and then workout the components. I came from the commercial HVAC industry and have an extensive mechanical, electrical and combustion background. I have programmed PLC and when the Arduino came out I started the programming in C++ conversion. The the new furnace controller ended up using a mega2560 with a 3.5" screen that was designed to plug into the mega560. as I progressed i found a conflict between the screen and using a SDA SLC chained t-couple boards either would work alone but wouldn't together. I ended up adding a UNO for the t-couples and using SoftwareSerial for RX-TX communication between the boards.
The other problem that I ran into was that the cheap 16 keypads were not stable enough to provide consistent key press reading into the voltage divider for one wire keypad. I went to digikey and found a high end keypad ($45) that is stable within 1 ohm between repeated key-presses.

I am in the process of wiring the system up for testing both the code and the wiring.

Below is the diagram of the old sytem
View attachment 163292

Congratulations on a very successful & complex project,...I've built a somewhat similar forced-air burner using diesel fuel for a monotube boiler and I too use the mega2560 as the controller, so I know first hand some of the challenges you've had to deal with.

I don't want to hijack your thread but I'ld love to have your input, as this is my first Arduino project, so I have no base knowledge to build on.
I too use a small display (4") (shown here: Display Video) connected via SPI, and the biggest problem I have is very slow display update speed. My display is so slow that I only update values that have changed and I don't re-write the labels/parameters.
How many parameters are you displaying? Color screen or B&W? Does your display seem to take a long time to write characters?

 

[master53yoda]

"HiLetgo 3.5" TFT LCD Display ILI9486/ILI9488 480x320 36 Pins for Arduino Mega2560 " This display is very fast as it is parallel rather than serial, this allows a complete screen rewrite updating all variables this rewrite doesn't even have a noticeable flicker. it also comes in a touch pad style. The only problem I found with this display was the need to sandwich a external terminal board between it and the Mega. to provide access to the other terminals on the Mega. it also attaches on the one end and i had to add a mount on the opposite end to keep the sandwich from separating, on future projects i will probably use a ribbon to connect the screen rather than directly plugging it into the Mega. I have found that by super gluing the wire terminals as a connector it solves the problem of the ribbon loosing integrity. I2C has also been a bit of a problem with dedicated terminals such as " Sparkfun QWIIC family of sensors, the MCP9600 t-couple interface had to be tied to a UNO USING rx tx serial com between boards.​

The other Screens i have used have been serial and program timing slows down response time. ​

Art B