Or how I just love to overcomplicate things
Lights, buttons knobs and dials have always fascinated me. In fact, it says so in my baby book pretty much exactly. I think that was genesis of my love for technology, electronics, sci-fi and general science and nerdom. After all, what could be cooker then “techy shit” as gmunk puts it?
With every project I do, I tend to dream up ways of making the interface and functionality as flexible (read: complicated) as possible. I want to feel like I’m flying the frickin’ Enterprise when I do my thing.
Others, like Dave Jones, take a simpler approach. Sure, he loves his nerd porn as much as anyone else, but looking at his home-built electronic load, he just whacked it up out parts he had laying about and uses it bare-board. Put it in a case? Nah! who could be bothered with that? I appreciate this simplicity and certainly have no problem using things that are bare boards and such. For him it’s justified, he’s designed a million things doubtless, and his own products like the µCurrent look pretty snazzy I must say.
I haven’t made anything good yet. So I’m puzzling over case designs and interfaces for my dummy load. My intention is to finish my projects looking a professional as possible, just so I can say I made something.
Thoughts on dummy load interface, function and design
The bus and in bed at night is where I do my thinking. Apart from obvious things like the panel meter, the control pot and a power switch, what does this project need?
What first comes to mind is some sort of safety feature. If the thing is set to its max dissipation (about 3.2A) and I try to plug in a beefy voltage source, it’s going to be dramatic and probably a little dangerous. I really would love to have my connectors firmly seated before I start drawing current. Simply remembering to dial the pot down to zero is iffy at best and I could see myself screwing this up with some regularity. Worse still, the panel meter will only read what current I’m drawing, not the set current, so it will say 0A but that is irrespective of where the pot is set. Being a 10-turn pot, it could be anywhere and the line on the knob is likewise useless.
I could solve both problems by using a double-pole switch. A simple ENABLE/DISABLE labelling would help and I could use it as a safety switch. In DISABLE mode, one pole could tie the set voltage to the ground rail, effectively forcing the MOSFET off whilst the other pole swaps the voltage measurement from across the shunt to the set voltage. That way, I could set the current beforehand before drawing anything, and connect my voltage sources safely. In ENABLE mode, the set voltage will come from the small trimmer to drive the gate and the panel display would be set back to measuring the voltage across the shunt (the current in other words).
That was easy. Wouldn’t be a bad idea either to have a LED indicating which mode I’m in either, which is rather trivial. I might skip the power LED at that point, since it’s obvious if the display is on the unit is on.
Other thoughts on design
The more I think of it, the more I really want the heatsink inside the case. I saw a few candidates for an appropriate case when I was downtown last that would easily house the board, the heatsink, everything that needs to be on the front panel etc. I’m not so keen on having a heatsink outside the case where it could possibly be touched, even mounted at the back. It does however mean that I will need a fan just to be sure it’s being cooled properly and to ensure temperature stability of the circuit (particularly the shunt, which I have no temperature data on).
Apart from that, I’m starting to dream of the panel printing. It’s pretty much a no brainer, just print on adhesive transparency and whack it on the front (with some care of course). The graphic designer in me is already thinking of a nice clean font and what warnings and labellings I will have to put on it. Though it will only be me actually using the thing, it is best practice to design in case someone else uses it, for safety if no other reason.
I could imagine the greybeards both approving of my thoughts on safety and flexibility, whilst simultaneously laughing at me putting lipstick on what is essentially a piece of home-brew test equipment. I believe it is the prerogative of the hobbyist.
It seems a simple matter to wire a switch, but sometimes I find it oddly confusing. In examining the above idea to switch out the voltmeter and base lead from the MOSFET, I’ve really been puzzled to how it’s going to work well. At first, I thought “oh no worries, I’ll use a DPDT switch and she’ll be right”. Not quite so. It’s important to look at what needs to be switched and where to determine the number of poles needed. Yes, experienced reader, I fully realize I could use transistors and add circuitry to allow this switching to happen using a SPDT switch, but I really do not want to for a few reasons. Many of the lines I am switching are very sensitive and I do not wish to load them accidentally which could lead to inaccuracies of the readings/settings at best, or a dangerous accident at worst.
So here’s what I need to do:
- Switch the output of the 10kΩ trimmer between the second opamp non-inverting input, and the + input of the panel voltmeter
- Switch the non-inverting input of the second opamp from the 10kΩ trimmer to the ground rail (holding the MOSFET off)
- Switch the voltmeter + input from the top of the shunt to the output of the 10kΩ trimmer to measure the voltage of Iset
- Switch on an LED (or two LEDs, mutually exclusive) without loading the other parts of the circuit.
It becomes apparent, since I have four items to switch, I need a 4PDT switch. How caveman . Other solutions, such as using transistors or switching ICs would work and eliminate the need for a fancy switch, but also may load parts of the circuit where I need to measure/transmit precise voltages. The safety factor also demands the simplest, 100% guaranteed to work solution which is a manual contactor. I *could* eliminate the second item, which would be tying the input of the second opamp to ground, and theoretically with no input to the non-inverting input it should have a LO output (keeping the MOSFET switched off). Something tells me this is bad practice and to be doubly sure, tying it to ground will force it off and stay off.
I could have also used a combination of multiple switches for the voltmeter and safety but then I have to keep track of two switches and who wants to do that? It makes perfect sense that if the load is switched off, the voltmeter should measure Isetload.
Yet another trip to the shops is required. Story of my life. Whenever in doubt, draw it up. So I did:
This tells me my wiring plans are sound and I merely need to buy the switch and wire it up. It also tells me I will need a 6-pin PCB connector for the switch, which I have.
Speaking of which, this is a good time to determine what sort of connections I will need to the PCB. I do like to use PCB connectors whenever possible rather than having soldered flying leads running everywhere. Sure, it’s extra expense, but it prevents me accidentally compromising a solder joint in wiring or if I have to subsequently take it apart for repairs or modifications. Also easier to deal with a board with no wires flying off of it for later work. Everything can be kept nice and neat and pluggable. So, what connections do I need? I’ll list them.
- 2-pin low-current connector for the DC power IN from the barrel connector/power switch
- 2-pin low-current for the fan (optional, I may tie the fan directly to the power switch/barrel connector)
- 3-pin HIGH-current connector to attach the MOSFET since it is mounted on a heatsink off-board
- 4-pin low-current for the panel meter (+input, -input, power, ground. may tie -input and ground to use a 3-pin connector)
- 6-pin low-current for the mode switch (see above)
- 2-pin high-current to mate with the banana jacks on the front panel for the voltage input
I hope I haven’t forgotten anything. The indicator LEDs do not need their own PCB connectors as I an wire them directly from the switch. Soon it will be time to lay out the board. Still have to acquire the panel meter and 4P switch which I will do soon. In the meantime, I do have a couple more tests to run. I can check out the mode switch as outlined above just to confirm function. The most important bit will be – if I tie the input of the 2nd opamp to ground, will it hold the MOSFET well and truly off? It should do, but pays to be sure. A quick back-of-the-envelope calculation shows that I will need a ~1.4kΩ resistor to limit the current to the LEDs so I will use a standard value of 1.5kΩ and that can just be added to the PCB between the unregulated power rail and the 6-pin connector to the switch (shown as power for LEDs on the diagram above). Easy.
Oh yeah, mustn’t forget I’ll need a range switch for the panel meter. Not the biggest deal in the world, a simple SPDT switch will do and the addition of the appropriate divider resistors. So I’m looking at three switches: power, mode, and meter range; two connectors: DC barrel and dual banana jack, one 10-turn pot for Iset; and the meter itself and that about covers everything needed to mount on the case itself.