Wednesday, July 21, 2010

Surface Mount Board Reflow

I have been doing hardware design and prototyping using surface mount devices for quite some time.  I have tried many method of reflowing boards but I've found both the oven method and "hotplate" method to work quite well.

In the past I have been managing the oven and hotplate via a thermocouple and PID controller but I decided to move to a full ramping controller to make the process less labor intensive.

Old Reflow setup; digital controlled hot plate

I purchased the reflow oven controller kit from sparkfun, overall it works ok; the software is good but the hardware has some flaws.  Most notably is the driver for the relay; they're using a BJT which doesn't seem to want to drive the relay coil reliably (even after modding the base resistor).  I solder a mosfet in its place (note the uglyness in the pic below) I used a n-channel mosfet I had kicking around in my junk box.  Another flaw is their choice in relay; the one that was shipped to me didn't have blade connectors on the top (as pictured on the site) so I had to solder wires on the bottom (not very elegant).

The controller didn't work well for the hotplate; the hot plate has too much thermal mass and the control loop on the temperature controller is too simple to compensate.  The controller does work quite well for the oven (for which it was designed).

Oven with temperature control

The plan is to mount the controller in the oven chassis and run the LCD and buttons to a new front panel.

Results using this system have been good; I have been soldering the processor I'm using for a design (LM3S6965) in a TQFP-100 package; results have been good.  Minimal reword required; just wicking a few solder bridges.

TQFP 100 solder with the oven and controller

Powder Coating

Update2: Description, Schematics and boards

Update: added some more details

So over the last few months I have built a number of powder coating guns as well as a powder coating oven. I'm trying to make something better than the $60 harbor freight gun, from what I have read the HF gun tends to not give consistent powder output and struggle with multi coat.

Mark I:
This gun was inspired by this design
The overall design was pretty much the same; I used an old toaster oven to bake it. It worked ok, but the voltage was too low to get good powder adhesion on additional coats, and it was hard to clean and didn't put out a consistent powder flow.

Mark-I ghetto gun

BMW E36 caliper support brackets (before and after)

Mark II:
The second prototype moves the high voltage electronics into a separate unit; I also added previsions for a variable voltage supply, solenoid controlled air supply, and a separate powder chamber. Results were pretty good.  Its still very much a prototype, for the next revision I'm moving things around and sending PCBs to be made.

Main Power Supply and Control system

The system as I use it; lots of hours on it now.

Here is a picture of the power supply; There are three stages; a generic 110VAC to 36VDC 3A power supply; Provisions for an adjustable stage (36V to 10-32VDC 3A); High voltage step-up stage (10-32VDC to ~25KV done in the switching power supply section)

The AC-DC input stage is built around a standard toroid step down transformer; full bridge rectification and capacitor bank. The supply is controlled by a switch on the front.

The adjustable stage is currently not implemented; instead I have simply added a 12VDC regulator used for supply of the switching section of the HV supply and for the actuator control. In the next revision I will be adding the variable voltage control using a standard adjustable regulator (LM317) with pass transistors to provide additional current. This stage is controlled via a relay which is controlled by a switch on the gun.

The high voltage step up stage is basically a switching power supply driving a flyback transformer out of a CRT computer monitor. The frequency and pulse-width of the drive circuitry is tuned to match the transformer. I set the duty cycle to 20% and adjusted the frequency until maximum output was observed (by measuring arc length). A 555 timer was used for pulse generation; it feeds a IRF240 power mosfet which drives the primary winding of the transformer. The primary winding was made by wrapping 8 turns of 14 gauge transformer wire around the ferrite core of the transformer.

The flyback transformer is directly connected to the tip on the gun via 50KV high voltage wire bought off ebay.

The air supply is set via an adjustable regulator on the front of the chassis; after the regulator the air is connected to a solenoid which is activated by a switch on the gun.

Flyback and air control

The regulator air supply is fed to a powder container; it feeds at the bottom of the container in order to fluidize the powder; the powder then travels to the gun.

Third itteration of the powder pot; this one is quite solid and works very well

The gun its self is very simple it is simply PVC fittings which allow for attachment of various nozzles. The nozzles were purchased from powder by the pound (they're cheap ~$12 so It was easier to buy rather than build).  I've added a switch to the gun (not pictured) which switches on the HV and air.

When it all comes together I have had some pretty good results; powder attraction is much better than with the Mark-I gun and powder flow is more consistent. Multiple layers work well (tested up to three layers so far).

Powder coatingis great; but the hard part is getting a big enough oven to cook the parts in. I ended up building my own oven from sheet metal and steel studs. Since I wanted to keep costs low I used left over bits from various projects and friends and family. It turned out pretty good; it doesn't look like much but it heats up quickly and holds its heat well.

Oven; inside dimensions 2' x 3' x 4'
240V 30A

30A contactor for controlling the elements

Auto Tuning PID temperature controller for driving the contactor

Some formerly brass trim from my fireplace (ready to be cooked)
For temperature control; the elements are fed by 240V at 30A; they are connected via a electrical contactor which is controlled by a auto-tuning PID temperature controller.


Powder Coating System Mark-III:

I've started building the MK-III version of my powder coating system. Here is the plan.

- Move to larger 19" rack mount chassis
- Provisions in chassis for second regulator and PSI gauge for future addition of powder pump.
- Adjustable low voltage DC supply in chassis, all HV moved to gun
- Switching supply and AC fly back mounted on gun
- 5x Voltage multiplier on gun to bring output voltage up to 100KV for large number of coats
- Use custom PCB for switching supply;
- New Gun cabling using circular connectors

Monday, July 5, 2010

e36 M50 Manifold Swap

A few pics from the swap I did on my 1996 BMW 328is. I modified the M50 manifold such that all of the existing M52 piping works without modification.

This was accomplished by cutting the base of the M50 manifold such that it will accept the M52 ICV and oil separator mounting plate. I made an aluminum bracket (powder coated flat black) to allow the M50 manifold to mount to the M52 motor.

Posted by Picasa

Coffee Roasting

So I've been roasting coffee on and off for a few years. Below is a picture of my original method using an air-powered popcorn popper. I modified it such that the fan can be controlled independently of the heater. To make things simpler I experimented with different methods of temperature control. The picture below shows the Sparkfun electronics reflow oven controller connected to the popcorn machine. I programmed the reflow oven controller with a profile for roasting beans. It works well and allows for consistent hassle free roasting.

The big drawback of the hot air popper for a coffee roaster is the limit on the amount of beans you can roast at once. I rigged up a larger capacity roaster as seen in the pics below. It is essentially the pan and drive of a bread machine, an exhaust vent to take care of the chaff, a heat gun to provide the heat, a thermocouple and temperature controller to run the show. Results were good; consistent roast; no user intervention required.