[4]

Electromechanical Project: Induction Heater

[3]
[10] Like what you see?
Click here to donate to this forum and upgrade your account!

Norppu

Active Member
Registered
Joined
Jun 15, 2018
Messages
74
This thread contains instructions and drawings to create an induction heater.

BlockDiagram.jpg

The internet is bloated with diverse projects for induction heater but unfortunately only a few of them have any chances to become an useful tool and most of them will fry out in a matter of days or even seconds With this design I harness my very long experience on electronics and try to cope with the machining. Yes, this project involves quite some interesting machining including (but not limited to) piping fittings for the heater coil and radiator, water cooled heatsinks for rectifier, FET array and controller power electronics.

At the business end of the induction heater there is a ZVS driver. I chose a chinese 1800W driver which I will modify to be able to fully deliver the power to the output coil. At the moment the coil connections are flimsy and the supplied coil is a joke. However, one cannot buy the separate components with the price of this unit. Cheap Chinese units are available on all common auction sites.

IMG_5046.jpg

The coolant will be circulated with a pump. When choosing a pump for a project like this one should look for a device with an adequate yield. A centrifugal pump is generally adequate but those must be placed under water level as they have no suction. I chose a ROTEK 24V WPCD 11.5l/min@0.3 bar device. Although this pump is submersible it will be running outside the water reservoir.

IMG_5049.jpg

The generated heat must be dissipated out of the cabin. For this task I chose a 115 x 210 mm Mocal oil cooler with BSP fitting. Thank god those fittings are not tapered. It will still be a challenge for me to machine those fittings. Luckily those must be made of aluminium.

IMG_5051.jpgIMG_5052.jpg

The coolant flow and temperature must be monitored. For flow monitorin a flow meter is used. This is a device (YF-S402) that sends pulses when the fluid is flowing. For temperature monitoring two Philips KTY-sensors are used. One for coolant and another for cabin ambient temperature. Those are very cheap and more than accurate enough for the task. In the picture I have disassembled the flow sensor. This unit should not be placed on the pressure side of the coolant flow. Best place is right before or after the radiator.

IMG_5086.jpg

The blower (fan) will be one or more PWM controlled PC cabin blowers. For this project I chose a super-silent blower that has the PWM control and RPM monitoring facilities. Most of the blowers (4-pin PWM connector) have these features inbuilt.

Blower.jpg

All this needs a controller. I have designed one and the schematics, etching masks, gerber files for PCB manufacturers and whatnot are included in the attached eagle.zip. With these files it is possible either manufacture the PCB home or let some PCB prototype-manufacturer take care of that. The chosen components are widely available either directly or be easily replaced.

UiMockup.jpg

The controller needs a program. The microcontroller is a cheap Atmel AtMega32L which has just enough I/O for this task. I will attach the program (written with avr-gcc) once I get it debugged.

There will also be an User Manual in English, German, Swedish and Finnish languages.

In case You want to see all files involved with this project these are available at Hobby-Machinist Directory in my server.
 

Attachments

Last edited:

extropic

Active User
Registered
Joined
Nov 24, 2014
Messages
557
Welcome to Hobby-Machinist.
Very impressive second post.
I don't have the nerve (to much common sense) to open your .zip file yet but I will watch this thread and hope it turns out as well as it has started.
 

Norppu

Active Member
Registered
Joined
Jun 15, 2018
Messages
74
Welcome to Hobby-Machinist.
Very impressive second post.
I don't have the nerve (to much common sense) to open your .zip file yet but I will watch this thread and hope it turns out as well as it has started.
Thanks !

You can look the individual files in my server. The zip file is here just to conform with forum rules. In case You are interested in microcontroller programming or electronics designs then it might be worth to look at.

The zip contains the PCB design for this project. I have already ordered five of those from a Chinese PCB manufacturer. As always those will contain at least 10 bugs but one has to start somewhere.

I will also add more plans for the water cooled heatsinks, common layout in the cabin and of course the Atmel code. The unfinished code is already there in the directory to see. Link to that directory in in the post.
 

markba633csi

Platinum
Registered
Joined
Apr 30, 2015
Messages
4,147
Hello, what do the letters ZVS stand for?
edit: found it : zero voltage switching
Mark
 
Last edited:

brino

Active User
H-M Supporter - Gold Member ($25)
Joined
Jan 2, 2014
Messages
3,673
Holy Crap! @Norppu, what a great thread!
....and you just joined....

Welcome to the Hobby-Machinist!

I am watching this thread.
Thanks for sharing your expertise.

-brino
 

brino

Active User
H-M Supporter - Gold Member ($25)
Joined
Jan 2, 2014
Messages
3,673
This unit should not be placed on the pressure side of the coolant flow. Best place is right before or after the radiator.
why?
thanks
-brino
 

Norppu

Active Member
Registered
Joined
Jun 15, 2018
Messages
74
why?
thanks
-brino
Two reasons.
The plastic case of the flow sensor is not trustworthy. It will most probably generate a leak after being 5 years in use and having a pressure. I will have this apparatus placed in a manner that if (when) it leaks, the leak will go into the reservoir.

Another place for the flow meter would be between the pump and the reservoir. However, it is a bad idea to have flow restrictions in the suction side of an centrifugal pump as that may create cavitation and premature failure of the pump impeller.
 

brino

Active User
H-M Supporter - Gold Member ($25)
Joined
Jan 2, 2014
Messages
3,673
I don't have the nerve (to much common sense) to open your .zip file yet but I will watch this thread and hope it turns out as well as it has started.
Great computer hygiene!

I have download the file and then uploaded it to VirusTotal.
They test the file with 61 different virus and malware detection tools and it came out fine.
Here is the link to the results:
https://www.virustotal.com/en/file/4dc3c397b70ee2e4ee5c9b0f7ebace85989a0e87a6eb82bc7eaed0741feb8979/analysis/1529165825/

The file also tests clean with my locally installed tools.

-brino
 

Eddyde

Bronze
Registered
Joined
Oct 13, 2014
Messages
1,504
Wow, Norppu awesome thread and welcome to HM! Members like you are what makes this site great.
I bought one of those Chinese induction heaters and found it to be pretty much useless. I will put your design on my project list.
 

Norppu

Active Member
Registered
Joined
Jun 15, 2018
Messages
74
Today the post brought me the PCB for the controller. The quality of this is superb. My hands are itching to get some components into that. Unfortunately I am in an assignment quite far away from home . Luckily my vacation starts in 2 weeks and then I have a whole month time to do things like this.
IMG_5091.jpg
 

Norppu

Active Member
Registered
Joined
Jun 15, 2018
Messages
74
Drawing for the water cooled heat sink for the controller unit is now there. There is a PDF file attached and if You rather use some CAD software to look into it, the there is a dwg file in the drawings folder for this project.

Corrosion

While I was drawing this particular item I started to wonder about the corrosion on the radiator and the diverse water cooled heat sinks. I am not worried of the copper coil corroding away as it will not be made a permanent item of this apparatus. The coil will be having fittings to attach the coil making it easily interchangeable.

What made me really worried was a lot of information about aluminium getting pitted under presence of copper ions. Pitting corrosion is probably the worst as it can perforate the radiator in one year or even faster if it gets really nasty. The radiator and the heatsinks will be electrically isolated so there should not be electrolytic corrosion which would be even faster if it comes to that.

I have been searching information on how to prevent pitting on aluminium and the only advice which is present everywhere is to use glycol based and it should be free of nitrites, silicates, borates, and amines. Finding such an antifreeze is nothing but easy. Further more the additives in coolants may attack some of the plastic parts which is not good.

So the thing is that aluminium will not corrode if the coolant is pure water (rain or distilled water) and if there are no other metal ions present. Basically this scenario is like having separate aluminium parts in water without interconnecting them. No corrosion - will last forever.

But .... this device contains the coil which must be made of copper. It is also the source of most of the heat and it will release copper ions into the water for sure. These copper ions (according to my reading) will then be attracted by the aluminium parts which will get pitted.

How do I remove the copper ions from the water before they reach the radiator ?
Would a "filter" filled with aluminium swarf do the trick ? It is my thought that the swarf filter would atttract most of the ions before they reach the radiator. It is easier to replace the swarf than replace the radiator of the heat sinks.

Funnily enough, it seems like peppermit tea would be one way to remove copper ions. So, am I actually trying to build an induction heater but end up making a capuccino machine.
 

Attachments

extropic

Active User
Registered
Joined
Nov 24, 2014
Messages
557
I know that you started the design with the AL heat exchanger shown in the OP but, why not just change to a copper/brass radiator rather than add the additional complexity and maintenance item of an ion filter? The heat sink could easily be made of copper also. How about brazing copper tube to a thinner copper plate?
 
Last edited:

Norppu

Active Member
Registered
Joined
Jun 15, 2018
Messages
74
I have also thought to eliminate the aluminum from the cooling system. However, all components will then have the decimal point in their price tags moved one place to the right. In addition finding a radiator of suitable size is a little bit hard. There are some suitable radiators which are intended for car air conditioning (heating element) but those have usually very strange shape and are not widely available.
It is my intention to make this project doable for everyone so the radiator has to be affordable and available.

So, how about eliminating the copper ?

The conductivity of aluminum is about 60% of conductivity of copper. With these frequencies the skin effect plays a minor role on current distribution in the coil so making the coil of aluminum with thicker wall might just work. Bending the aluminum tube would also be easier if the tube wall is thicker.

With aluminum the challenge is to get the connections reliable. It is possible to solder aluminium with correct flux but one has to be really careful with temperatures and the vapors of that flux are nasty.

On use aluminum would behave better in the coil as it would not corrode because of the oxide layer. Copper will get quite nasty look after some use even if it is cooled.

One more option would be to plate the copper tube inside with some suitable metal. The trick here is to get the coating complete so that every bit of the inside of the tube is covered. Here a suitable metal would be zinc. Tin might just do it and further more, the tin can be applied chemically without electrolysis or excessive heat. Still this method will be nothing but easy as the tube is rather long and slender.

Last but not least machining copper and achieving an acceptable surface quality is something I have found to be really hard to do. So probably the heat sinks would need to be made out of brass instead of copper.

Edit: Introduced Current aka Cathodic corrosion prevention

Actually the ion filter is far simpler than I originally thought. In this case it will be an aluminum tube connected to positive voltage through a suitable resistor. The resistor limits the current to a suitable value so that the sacrificial aluminum tube corrodes at acceptable speed. If this tube is thick enough (5 mm wall thickness) it will practically last forever. Further more it is easily replaceable.

So how does this work. Instead of letting the copper ions to escape from the coil, the positively charged sacrificial anode forces them back into the copper surface. The sacrificial anode will release aluminum ions into the cooling system which are not harmful for the aluminum parts. The voltage applied on the sacrificial anode could be 5 volts (readily available) which should be high enough to keep the copper ions where they belong to.

The aluminum tube(s) are placed between the copper coil and the rest of the cooling system. The radiator and other aluminum cooling part still need to be isolated in order to minimize corrosion when the device is powered off.
 
Last edited:

Norppu

Active Member
Registered
Joined
Jun 15, 2018
Messages
74
Aaaahhhhh .... finally Vacation - Urlaub ... Endlich - Viimeinkin Loma !

IMG_5092.JPG

Mounting surface mount components requires some specialized equipment. Needle-sharp pliers, soldering paste in a syringe and magnifying glass. First one squirts a small drop of the soldering paste on each soldering pad on the PCB. Then the components are inserted on top of the paste using the pliers. Quite delicate task requiring good arm-rest. This is the reason why my electronics table is quite high. Better working position.
IMG_5093.JPG

For the mcu and other components having many tightly spaced feet, a worm of paste is squirted and then it is assumed that the flux in the paste removes the solder between the legs.
IMG_5094.JPG

When all the components are placed on their respective places, the PCB is heated gradually until the solder in the paste melts and forms solder joints. When the amount of paste is correct and the pads have correct geometry the components float on top of the solder and align themselves automatically. In my case this did not work out. The soldering paste has a best before date and the date in mine was two years ago. The PCB and all components were wasted. This is not a problem as I have four more PCB's and I am not going to run out of electronic components during the next decade or so....
IMG_5095.JPG

Hot rolled steel block 100 x 150 x 20. This is the base for the milling attachment. I do not like to drill holes in my lathe saddle so I made this plate to accommodate the vertical slide.
IMG_5096.JPG

The slide arrived in the post today. It is made of Chinesium and had quite rough sliding surfaces. Cleaned it up and filed all burrs and now it is useable.
IMG_5097.JPGIMG_5098.JPGIMG_5099.JPG

Had to solder the components to the PCB the old fashioned way. This is not a problem if one has a very steady hand, soldering iron with a 0.3 mm tip and VERY thin solder. I have a Metcal soldering station and also a hot-air rework station. Both can be used for this task. In the picture the PCB is already finished lacking only the heat sink. All voltages are there and works as planned.
IMG_5100.JPG

Here is a tip tip. Having a steel needle in one of the measuring probes allows me to penetrate through varnish or other material that could impact the measurement. The steel needle buries quite nicely into tin.
IMG_5101.JPG

Darn. I am not perfect ! There is an error in the PCB (actually two). The fure component in my library has an error in it resulting soldering barrier on top of the pad. Lot of scraping of the solder resist and then that was fixed. Another error was a short of bias circuit with a ground plane. Milling the copper out by Dremel ... free hand with carbide tip (0.4mm).
IMG_5102.JPG

Now this needs the heat sink and a program.
 

Norppu

Active Member
Registered
Joined
Jun 15, 2018
Messages
74
Bought a piece of aluminum stock. I hate this material as it is gummy and it is easy to scratch it. Nevertheless, this is a 70 mm piece out of the 1000 x 100 x 10 stock of marine grade aluminum. I do not have a motorized saw so I had to hacksaw it. I really hate this material.
IMG_5103.JPG

Squaring out the edges. The milling attachment came with a vise. This unit had flimsy 6mm bolts and t-nuts which were a joke. I will make a decent base for this vise and a bunch of t-nuts to attach it to the milling device. Here the vise is attached with two 8mm bolts. This is as temporary as it gets. Milling aluminum with sharp end-mill does not generate a lot of cutting forces. Worked well.
IMG_5104.JPG

Does this qualify as a deep hole? Bored the two water channels inside the heat sink. To my surprise the drill did not come out from either side and when I bored the water inlet and outlet holes it appeared that the water channels were exactly where they are supposed to be. Happy with that.
IMG_5105.JPG

The heat sink is now in place and next step will be some programming. Doing this with GNU avr toolchain. Looking forward to see real action on the diverse outputs.
IMG_5106.JPG
 

Norppu

Active Member
Registered
Joined
Jun 15, 2018
Messages
74
I got the microcontroller to communicate with the programmer. I also noticed that the UISP program I was using is now stone-dead. Not only does windows get impacted by updates - it is linux as well. Had to install a new program for uploading the code to mcu. It is no mode uisp ... now it is called avrdude and the command line syntax is from an another galaxy far-far-away.

I also noticed that it would be easier to do the program-test-correct-coffee-program-test-correct loop if there would be the actual user interface on place. So this will be my next task. Collected the components and did some MAD. (Manual Aided Design). It is nice to have the actual interface elements on hands. Pictures are so 2-dimensional.

IMG_5107.JPG

For the buttons/lights the components are military grade switches with ordinary lamps inside. These are nice as one can draw the picture on the button which then is backlighted with the lamp and covered with tough plastic cover (polycarbonate).

The three potentiometers for attack-sustain-decay timing are logarithmic so it will be easy to adjust short times. They are also stereo which means that I can connect two potentiometers in parallel thus increasin the reliability of the system. These are quality potentiometers from Piher and are at least 30 years old. The power potentiometer is a 2-turn linear wire wound pot which will not break easily. This user interface will have a life expectancy of at least 10 years when properly installed and shielded.

edit: I already hate the 5 mm thick aluminum plate which I will use to create the support for these.
 

Norppu

Active Member
Registered
Joined
Jun 15, 2018
Messages
74
As much as i hate aluminum it has some properties that makes it excellent material for user interfaces. It does not need any kind of protective layer as it forms the layer itself. At the moment it is a little bit too shiny for my taste but i know this is just a temporary shine. It will become dull and that is exactly what I want.

Next task is to set up wiring for the controller. The PCB has a dedicated connector for this and it should be a no-brainer to do the wiring. Being a no-brainer carries it's risk ... it is really easy to misconnect something....

The combination square is in one picture for a purpose. This one is a Starrett and I had to do quite some negotiations with my financial supervisor to be able to get that. My recommendation: Get one of these instead of the cheap chinese ones. I bought also the chinese and that was not good. The 90 degree angle was actually 92 degrees so with the chinese You get bonus-degrees. It was advertised as stainless .... it was actually stainless aluminum and plastic. The only steel part was the ruler which I kept. Rest went into bin. With a bad tool it is all too easy to spoil perfectly good material and waste a lot of time. With the Starrett i can only blame myself if something is not exactly as it should be.

IMG_5108.JPGIMG_5110.JPGIMG_5112.JPGIMG_5113.JPG
 

Norppu

Active Member
Registered
Joined
Jun 15, 2018
Messages
74
Temperature sensors were constructed out of good quality coax-cable and the sensor element itself. The sensor is about 1.2mm in diameter.
IMG_5121.JPGIMG_5122.JPGIMG_5123.JPG

The rectifier consists of six 15 ampere units which are connected in parallel to have enough current capacity. The 15 amps per unit is a continuous rating. This totals in 90A continuous (average) current.
IMG_5124.JPG

Connecting six rectifiers in parallel is not a trip in the park. If they are connected in parallel without any precautions the result will be a tragedy. If one of the rectifiers gets hotter than others it will, due to silicon rectifier properties, carry most if not all of the rectification current. This will result in catastrophic failure of the unit. There is a remedy for this problem (thermal runaway) - each rectifier is connected to the (+) raiul with a small resistor. Suitable value for this is 0.08 ohm. The problem is that these kind of resistors are not easily available with high current capacity. These must be shop made.
IMG_5125.JPG

The resistor wire cannot be directly soldered. The easiest way is to use a copper tube and then silver solder the resistor wire into that. Then the copper can be soldered the normal way. There exists a solder (ALUSOL) which could be used but it makes a very brittle joint and smells awful.
IMG_5126.JPGIMG_5127.JPGIMG_5128.JPGIMG_5130.JPG
 

Norppu

Active Member
Registered
Joined
Jun 15, 2018
Messages
74
My vacation is now over and I have tyo do at least two months virtual machining only. Here are the highlight of this project and it's current status.

Layout of the components is actually quite critical. The high frequency parts must be kept as far as possible from the controller. I think I succeeded quite well with this.
IMG_5203.JPGIMG_5204.JPG

It is not a good idea to make sharp bends to water pipes as they may collapse and block the coolant flow. So I used a suitably sized copper tube, had the hose on top of it, bend it to suitable form and then applied heat to it. When it cooled it persisted the new shape. The hose is some mysteryt material suited for butan/propan torches. Hopefully it does not take back the original form when it heats up with the coolant.
IMG_5207.JPGIMG_5208.JPG

So my vacation is over and I had to put everything on hibernation while I am away. Most of the wiring is done.
IMG_5209.JPGIMG_5209A.jpgIMG_5210.JPG

IMG_5206.JPG
 
  • Like
Reactions: rwm
[5] [7]
Top