Don’t ever start a project thinking: “This is gonna be a quick build”. Because Murphys Law will *beep* you hard.
What I wanted
Cobble together a simple current source, stuff it into an old case that was lying around and make a quick 3D print to just hold a very simple front panel.
A few other things that later bit me in the ass was the idea to use rather old components (“old” as in vintage, not “used” components) and the plan to use one of my cool old (vintage) operational amplifiers.
The technical specification are:
- Input Voltage: about 2 to 100 Volts
- Sink Current: max 9 Amperes
- Continuous Power: 20W
- Short period power: 100W for 3min max (30min cool down)
What didn’t work right
The transistors. As much as I wanted to use a couple of my old(isch) bipolar transistors (BUX98A), they’re just not the right choice for the task. Mainly because I wanted to go down to fairly low voltages.
In order to compromise the current through the sense resistor as little as possible I had to keep the base current very small. Which in turn meant to use a multi stage darlington circuit. That however required to raise the base voltage pretty high. That made the whole current controller pretty unstable. I then decided to use a MOSFET (20N60C3) which doesn’t introduce any additional currents at all.
The operational amplifier. Originally I wanted to use a vintage Motorola MC1748 opamp. But apparently opamps from that era weren’t as good as modern parts. This made the whole system even more prone to oscillations. The “solution” was to just use a modern opamp (LM741). The keen eye might have noticed the 74 in both type numbers. This is because the LM741 is the modern equivalent of the MC1748, duh.
The power supply. As great as opamps are to use in a circuit, there’s always the problem of having to use a symmetrical power supply. At least in most applications. I was trying hard to get around that. But even with modern components, which I had already switched to at that time, “high power” applications tend to require more solid engineering. I ended up using a DC/DC converter module (+12V to +/-15V).
Luckily, I had chosen a way-too-big case for the device, just because that was what I had lying around. Also, I didn’t think I could use that case for anything else. This gave me the opportunity to make the circuit board as large as it would just fit into the case to have plenty of room for any kind of power supply.
Feature creep creeps in
During prototyping and testing I realized that it would be a good idea to add at least a current meter to the device. Measuring the voltage would later be easy by just hooking up a voltmeter in parallel to the electronic load. But in order to measure the current I had to hook up another voltmeter to the sense resistor of the load. This in itself wouldn’t be too big of a problem. But measuring a current indirectly across a 0.39 Ohms resistor would be rather cumbersome and error prone.
Having a small panelmeter lying around I thought this would be a simple addition.
I was wrong.
Partly because I went overboard and decided that I wanted to be able to measure the voltage with that same panelmeter by switching between current and voltage measuring back and forth.
Partly because I didn’t expect that “just” measuring a current or a voltage could be a big deal.
Problem 1: Trying to display a current ranging from 0.05 to 9 amps with just 3 and a half digits couldn’t be done with just one range. At least not properly. The same is true for the voltage measurement. Measuring 2 to 100 volts with reasonable resolution required switching ranges.
Problem 2: Switching ranges also requires to switch the decimal point around. And – since Murphy doesn’t like me – the voltage and current ranges required completely different decimal point positions. So I had to come up with a fairly convoluted arrangement of switches and relays to get that to work. (Note: the schematics contain errors. Also, changes have been made that didn’t get changed in the plans.)
Problem 3: Independent power supply. As many cheap panelmeters do, it required that the ground of the supply voltage is separated from the ground of the voltage that you want to measure.
So, not only did I need a symmetrical power supply for the opamp, I also needed an isolated DC/DC converter for the panelmeter. As usual, I stole that from an old network interface card.
But wait, there’s more.
The DC/DC converter for the panelmeter has a 5V input while the DC/DC converter for the opamp has a 12V input. Therefore, in addition to all that fuss I also needed a 5V regulator to step down from the 12V supply Voltage.
|The black block (GTS) is the isolated DC/DC converter.
Left hand side is the 7805 with its heatsink.
In the front, mounted vertically, the +-15V DC/DC converter.
The final Problem with that: No space on the front panel. I didn’t change my choice for a case, even after adding all the “features” I talked about before. Since the panelmeter used up most of the space on the front panel I had very little room left to place the rest of the controls and plugs.
A simple front panel … NOT!
After I had added so many stuff to the device it was clear that I coudn’t get away with just 3D printing a simple frame and closing it off with a small acrylic panel.
So I decided to design the whole front panel for 3D printing.
Being at it, I thought it would be a good idea to model the holders for the circuit board into the front panel object. While this doesn’t sound like a bad idea, it is when you have to change stuff later in the game.
With all the problems, changes and additional requirements I started to lose track of all the components and modules that had to play together properly. So I gave in and drew a proper schematic. (Note: the schematics contain errors. Also, changes have been made that didn’t get changed in the plans.)
I still refused to design a circuit board with CAD. Instead I used Photoshop and a graphic tablet to draw by hand. Well technically – and literally – this is still Computer Aided Design. But not in the same sense.
You got to be kidding me !!!
After cramming all the components, the now ridiculously involved power supply, the range switching and the voltage dividers for the panelmeter onto the circuit board, I realized that I had forgotten to shorten the circuit board to accommodate for the additional stuff on the front panel that also takes up space on the inside of the case. In short: my circuit board was too large.
My only luck was, that I still had a way-too-big case to work with. So I embarked on the task of redesigning the front panel in a way so that it could hold the circuit board at an angle. This also required me to redesign the back panel which contained PCB holders as well.
The Final Sprint
After all that fuss, I was finally ready to put everything together and make the final test of the whole shebang. This went surprisingly smooth. I did notice that the potentiometer, which was a vintage 10 turn pot, didn’t quite work as smooth as I’d liked. So I replaced it with a modern one. That dropped my maximum current to 9 Amps, because the new pot is just 20k instead of 22k. I could have replaced the dropper resistor that’s in series with the pot, but at this point that resistor was pretty hard to reach, let alone soldering in a new one.
In the spirit of this website, there are of course a few things unfinished.
The panelmeter: During my initial tests I discovered that the decimal point is a lot stronger if I connect it to the supply GND rather than the GND of the input voltage. This turned out to be a bad idea in later tests. However, I forgot to fix that before the final assembly. So now I guess I’m going to ruin the display over time, probably, at some point all the decimal points will stay on constantly. But, whatever, the panelmeter was basically junk in the first place.
Input protection: Although I had it in my mind, over the course of developing this I totally forgot to add some level of input protection. There’s also no protection for the panelmeter at all. The 100V maximum input voltage that I state in the spec are – well – optmistic.
Knob and labels: As of writing this, there’s still no knob for the potentiometer. I had printed a rough design that didn’t survive the testing process. Since I replaced the poti I have to change the design, anyway. I also have to put a bunch of labels on the front and back of the case.
Fan: The main limiting factor for the power that can be dumped is the heatsink. I got that heatsink from a PentiumIII slot processor. That means, there was a fan on there before. I did think about using it, but I got lazy and now my thinking is, I can put it on when I need it.
Let’s see if those things will actually happen at some point in the (near?) future.
Never, ever, ever, ever assume something will be “just a quickie”.