I've been developing myself a bench PSU, and finally reached the point where I want to apply a load to it to see how it performs. After watching Dave Jones' excellent video and looking at a few other internet resources, I came up with Tiny Load. This is an adjustable constant current load, which should be able to handle about 10 amps. The voltage and current are limited by the ratings of the output transistor and the size of the heatsink.
It has to be said, there are some really clever designs out there! Tiny Load is really basic and simple, a slight modification of Dave's design, but it will still dissipate the power needed to test a psu, so long as it doesn't get more juice than it can handle.
Tiny Load doesn't have a current meter attached, but you can connect an external ammeter, or monitor the voltage across the feedback resistor.
I altered the design slightly after I built it, so the version presented here has an LED to tell you it's switched on and a better pcb pattern for the switch.
The schematic and PCB layout are presented here as PDF files and also as JPEG images.
Step 1: Principle of Operation
For those not well versed in electronic principles, here's an explanation of how the circuit works. If all this is well known to you, feel free to skip ahead!
The heart of the Tiny Load is a LM358 dual op-amp, which compares the current flowing in the load with a value you set. Op-amps can't detect current directly, so the current is turned into a voltage, which the op-amp can detect, by the resistor, R3, known as the current sensing resistor. For every amp that flows in R3, 0.1 volts is produced. This is shown by Ohm's law, V=I*R. Because R3 is a really low value, at 0.1 ohms, it doesn't get excessively hot (the power it dissipates is given by I²R).
The value you set is a fraction of a reference voltage - again, voltage is used because the op-amp can't detect current. The reference voltage is produced by 2 diodes in series. Each diode will develop a voltage across it in the region of 0.65 volts, when a current flows through it. This voltage, which is usually up to 0.1 volts either side of this value, is an inherent property of silicon p-n junctions. So the reference voltage is around 1.3 volts. Because this is not a precision instrument, there is no need for great accuracy here. The diodes get their current via a resistor. connected to the battery. The reference voltage is a little high for setting the load to a maximum of 10 amps, so the potentiometer which sets the output voltage is connected in series with a 3k resistor which drops the voltage a bit.
Because the reference and the current sensing resistor are connected together, and connected to the op-amp's zero volts connection, the op-amp can detect the difference between the two values, and adjust it's output so that the difference is reduced to near zero. The rule of thumb in use here is that an op-amp will always try to adjust its output so that it's two inputs are at the same voltage.
There is an electrolytic capacitor connected across the battery to get rid of any noise which finds it's way into the op-amp's supply. There is another capacitor connected across the diodes to damp down the noise they generate.
The business end of the Tiny Load is formed by a MOSFET (Metal Oxide Semiconductor Field Effect Transistor). I chose this one because it was in my junk box and had adequate voltage and current ratings for this purpose, however if you are buying a new one there are much more suitable devices to be found.
The mosfet acts like a variable resistor, where drain is connected to the + side of the supply you want to test, source is connected to R3, and through that to the - lead of the supply you want to test, and the gate is connected to the output of the op-amp. When there is no voltage on the gate, the mosfet acts like an open circuit between its drain and source, however when voltage is applied above a certain value (the "threshold" voltage), it begins to conduct. Raise the gate voltage enough and its resistance will become very low.
So the op-amp keeps the gate voltage at a level where the current flowing through R3 causes a voltage to develop which is nearly equal to the fraction of the reference voltage you set by turning the potentiometer.
Because the mosfet is acting like a resistor, it has voltage across it and current flowing through it, which causes it to dissipate power, in the form of heat. This heat has to go somewhere or else it would destroy the transistor very quickly, so for this reason it's bolted to a heatsink. The maths for calculating heatsink size is straightforward but also a bit dark and mysterious, but is based on the various thermal resistances which impede the flow of heat through each part from the semiconductor junction to the outside air, and the acceptable temperature increase. So you have the thermal resistance from the junction to the transistor case, from the case to the heatsink, and through the heatsink to the air, add these together for the total thermal resistance. This is given in °C/W, so for every watt that is being dissipated, the temperature will rise by that number of degrees. Add this to the ambient temperature and you get the the temperature your semiconductor junction will be working at. (At least I think so - please correct me if I'm wrong!)
Step 2: Parts and Tools
I built the Tiny Load mostly using junk box parts, so it's a little arbitrary!
The PCB is made out of SRBP (FR2) which I happen to have because it was cheap. It is coated with 1oz copper.The diodes and capacitors and mosfet are old used ones, and the op-amp is one of a pack of 10 I got a while ago because they were cheap. Cost is the sole reason for using an smd device for this - 10 smd devices cost me the same as 1 through hole one would have.
- 2 x 1N4148 diodes. Use more if you want to be able to load more current.
- MOSFET transistor, I used a BUK453 because that is what I happened to have, but pick what you like, so long as the current rating is over 10A, the threshold voltage is below about 5v and the Vds is higher than the maximum you expect to use it at, it should be fine. Try to pick one designed for linear applications rather than for switching.
- 10k potentiometer. I picked this value because it's what I happened to have, which is one I dismantled from an old TV. Ones with the same pin spacing are widely available, but I'm not sure about the mounting lugs. You may have to modify the board layout for this.
- Knob to fit the potentiometer
- 3k resistor. 3.3k should work just as well. Use a lower value if you want to be able to load more current with the 2-diode reference shown.
- LM358 op-amp. Really, any single supply, rail-to-rail type should do the job.
- 22k resistor
- 1k resistor
- 100nF capacitor. This should really be ceramic, though I used a film one
- 100uF capacitor. Needs to be rated to at least 10V
- 0.1 ohm resistor, minimum rating of 10W. The one I used is over-sized, again cost was the overwhelming factor here. A metal cased 25W 0.1 ohm resistor was cheaper than more appropriately rated types. Strange but true.
- Heatsink - an old CPU heatsink works well, and has the advantage that it is designed to have a fan attached if you need one.
- Thermal heatsink compound. I learned that ceramic based compounds work better than metal based ones. I used Arctic Cooling MX4 which I happened to have. It works well, is cheap and you get lots!
- Small piece of aluminium for bracket
- Small screws and nuts
- small slide switch
Step 3: Construction
I built the tiny load out of junk box or very cheap parts
The heatsink is an old pentium era CPU heatsink. I don't know what it's thermal resistance is, but I'm guessing it's about 1 or 2°C/W based on the pictures at the bottom of this guide: http://www.giangrandi.ch/electronics/thcalc/thcalc... although experience would now suggest it's rather better than this.
I drilled a hole in the middle of the heatsink, tapped it and mounted the transistor on it with MX4 thermal compound and screwed the mounting screw directly into the tapped hole. If you don't have the means to tap holes, just drill it a bit bigger and use a nut.
I originally thought this was going to be limited to about 20W dissipation, however I have had it running at 75W or higher, where it got pretty hot, but still not too hot to use. With a cooling fan attached this would be still higher.
There's no actual need to bolt the current sense resistor to the board, but what's the point of having bolt holes if you can't bolt something to them? I used small pieces of thick wire left from some electrical work, to connect the resistor to the board.
The power switch came from a defunct toy. I got the hole spacings wrong on my pcb, but the spacing on the pcb layout given here should fit if you have the same type of miniature SPDT switch.
I didn't include an LED in the original design, to show that Tiny Load is switched on, however realised this is a foolish omission, so I've added it in.
The thick tracks as they stand aren't really thick enough for 10 amps with the 1oz copperclad board used, so it's bulked up with some copper wire. Each of the tracks has a piece of 0.5mm copper wire laid around it and tack-soldered at intervals, except for the short stretch which is connected to ground, as the ground plane adds plenty of bulk. Make sure the added wire goes right to the mosfet and resistor pins.
I made the pcb using the toner transfer method. There is a huge amount of literature on the net about this so I won't go into it, but the basic principle is that you use a laser printer to print the design onto some shiny paper, then iron it onto the board, then etch it. I use some cheap yellow toner transfer paper from China, and a clothes iron set to a little under 100°C. I use acetone to clean the toner off. Just keep wiping with rags with fresh acetone until they come clean. I took plenty of photo's to illustrate the process. There are much better materials available for the job, but a bit beyond my budget! I usually have to touch up my transfers with a marker pen.
Drill the holes using your favourite method, then add the copper wire to the wide tracks. If you look closely, you can see I messed up my drilling a bit (because I used an experimental drilling machine that is somewhat imperfect. When it works properly I'll do an Instructable on it I promise!)
First mount the op-amp. If you haven't worked with smd's before, don't be intimidated, it's quite easy. First tin one of the pads on the board with a really tiny amount of solder. Position the chip very carefully and tack the relevant pin down to the pad you tinned. Ok now the chip won't move around, you can solder all the other pins. If you have some liquid flux, applying a smear of this makes the process easier.
Fit the rest of the components, smallest first, which is most likely the diodes. Make sure you get them the right way. I did things slightly backwards by mounting the transistor on the heatsink first, because I used it initially experiment with.
For a while the battery was mounted to the board using sticky pads, which worked remarkably well! It was connected using a standard pp3 connector, however the board is designed to take a more substantial type of holder which clips in the entire battery. I had some issues fixing the battery holder since it takes 2.5mm screws, which I have in short supply and no nuts to fit. I drilled out the holes in the clip to 3.2mm and counterbored them to 5.5mm (not real counterboring, I just used a drill bit!), however found the bigger drill bit grabs the plastic very sharply and went right through one of the holes. Of course you could use sticky pads to fix it, which in hindsight may be better.
Trim the battery clip wires so you have about an inch of wire, tin the ends, thread them through the holes in the board and solder the ends back through the board.
If you are using a metal cased resistor like the one shown, fit it with thick leads. It needs to have some sort of spacers between it and the board so it doesn't overheat the op-amp. I used nuts, but metal sleeves or stacks of washers glued to the board would have been better.
One of the bolts which fixes the battery clip also goes through one of the resistor lugs. This has turned out to be a bad idea.
Step 4: Putting It Into Use, Enhancements, Some Thoughts
Tiny Load is designed to draw a constant current from a supply, no matter what the voltage is, so you don't need to connect anything else to it, except an ammeter, which you should place in series with one of the inputs.
Turn the knob down to zero, and turn Tiny Load on. You should see a small amount of current flow, up to about 50mA.
Slowly adjust the knob until the current you want to test at is flowing, do whatever tests you need to do. Check the heatsink isn't excessively hot - the rule of thumb here is that if it burns your fingers, it's too hot. You have three options in this case:
- Turn down the supply voltage
- Turn down Tiny Load
- Run it for short intervals with plenty of time to cool in between
- Fit a fan to the heatsink
OK okay that's four options :)
There isn't any input protection, so be very careful that the inputs are connected the right way round. Get it wrong and the mosfet's intrinsic diode will conduct all the current that is available and probably destroy the mosfet in the process.
It quickly became apparent that Tiny Load needs to have it's own means setting the current it draws. There are three ways to this.
- The simplest option is to fit an ammeter in series with the positive or negative input.
- The most accurate option is to connect a voltmeter across the sense resistor, calibrated to that resistor so that the voltage shown indicates the current.
- The cheapest option is to make a paper scale which fits behind the control knob, and mark a calibrated scale on it.
Potentially the lack of reverse protection could be a big problem. The mosfet's intrinsic diode will conduct whether Tiny Load is switched on or not. Again there are a number of options to resolve this:
- The simplest and cheapest method would be to connect a diode (or some diodes in parallel) in series with the input.
- A more expensive option is to use a mosfet which has built in reverse protection. OK so that's also the simplest method.
- The most complex option is to connect a second mosfet in anti-series with the first, which conducts only if the polarity is correct.
I realised that sometimes what is really needed is an adjustable resistance which can dissipate a lot of power. It's possible to use a modification of this circuit to do that, much cheaper than buying a big rheostat. So look out for Tiny Load MK2 which will be able to be switched to resistive mode!
Tiny Load has proved itself to be useful even before it was finished, and works very well. However I had some issues constructing it, and realised afterwards that a meter and "on" indicator would be valuable enhancements.