For testing the capabilities of the step-down modules (KIS-3R33S) I have recently acquired on Ebay I needed a dummy load that could take 4A. So I made one.
The concept is pretty simple. It is a constant current sink controlled by 2 opamps. Two paralleled logic level MOSFETs do the heavy lifting (heatsink!). Another pair of paralleled P-MOSFETs + a Zener diode act as reverse polarity protection. A capacitor is used to keep the first opamp from oscillating. A simulation done with LTSpice works pretty well. The low cut-out voltage is determined by the body diodes of the P-MOSFETs that need to be forward biased.
The reverse polarity protection using the P-MOSFETs has one issue though, the V(gs) voltage cannot be too low, or these things get hot. Exactly that happens with low DUT input voltages. Fortunately I had added an option to replace this method with something else: a thick wire or a beefy Schottky diode. The latter is a bit of a desperate hack and needs an additional heatsink (small), but it does work. The voltage drop of the MBR1660 is quite low, so not too much heat is created in it.
Initially I used the IRLIZ34N MOSFETs. They come in TO-220 FULLPAK. Unfortunately the maximum power they can handle is rather low. Although I connected them to a big heatsink with the proper clips I managed to terminate 2 of the 3 I bought. Now I’m looking for a beefier replacement. The IRL3083 looks promising. I only hope these don’t give me any of the “let’s make the opamps oscillate” BS.
Today I received the circuit boards from Elecfreaks. This time I chose to have the boards shipped with DHL and it only took 3 days instead of the usual 2 weeks.
So far it seems to work quite nicely. A few things had to be “fixed” on the V0.1 boards though. All these fixes are part of the latest design files. Navigate to the YauDL project page to find them.
V0.1 errata / bugfixes:
• The voltage regulator NEEDS Tantalum caps to be stable. Ceramic ones WON’T work.
• It also requires a minimum load current to be stable.
• Adjusted the status LED current to make the regulator happy.
• Added about 20µF capacitance between the DUT binding posts to prevent oscillations.
• Input voltage: 0-20V
• Input current: 0-5A
• Power: 20W continuous (depends on heatsink)
• Reverse polarity protection (*)
• Current calibration with a 25-turn trimmer
• 10-turn wire-wound pot for current adjustment
• Binding posts for easy DUT / multimeter connection
• Runs with a 9V battery
• Power LED + on/off switch
Enjoy the images!
Power resistors and 0.1% tolerance 1:4 voltage divider (V1.xx board!)
Keeping the TO220 cool while soldering
Picking up some noise
Left-over printed circuit boards might be found here!
This currently only covers PCB version 1.00, as pretty much nothing else should be out there in the wild anyway. At least nothing that came from me.
You’ll find all files, schematic, gerbers, KiCad… in these git repositories: ,. Make sure to choose the version that matches your hardware! Some versions are tagged, so go look for ‘tags’ in the repos.
The latest ‘ERRATA’ file will always be found in the ‘master’ branch.
Always keep your iron’s tip clean, use plenty of flux.
Step 1: solder all surface-mount parts on the top side
There is ONE exception: DO NOT populate C5! It interferes with the zip-tie. Skipping C5 does not affect the proper operation of the device.
You’ll also have to add a small bodge-resistor of 1k across the output of the LDO voltage regulator. This ensures stable operation.
Step 2: solder all surface-mount parts on the bottom side
Step 3: solder all small through-hole components
Special care must be taken for the ‘I_SET JUMPER’. Make sure it is absolutely flush with the bottom!
You DO NOT want it to look like this:
This is what it should look like, no protruding pins at the bottom:
For extra insulation and piece-of-mind, place a bit of plastic sticky-tape over the pads.
Step 4: Solder the 10-turn potentiometer
Step 5: Solder 5W resistors + binding posts
The 5W resistors must be free-standing – i.e. not touching the PCB. They may get HOT.
Step 6: Test-fit the heatsink and TO-220 packages
DO NOT use thermal grease yet. Mount the heatsink so it can slide side to side a bit and insert the TO-220s. Insert the mounting clips into the heatsink above each TO-220 (there’s a little groove) and firmly press down towards the heatsink until they snap into place. Now adjust the heatsink so everything looks good, pins are straight, nothing bent… Then tighten the screws at the bottom.
Now it is time to solder the MOSFETs and Schottky diode!
Once soldered, remove the clips and heatsink. Now apply a small amount of thermal grease to the TO-220 packages, put the heatsink back on, tighten the screws and finally add back the mounting clips.
You’re done with soldering! As usual, please check for solder bridges.
Step 7: Connect the 9V battery and fix it with a zip-tie
NEVER operate the device with a floating control voltage!
If you use manual control via the potentiometer, ALWAYS MAKE SURE the I_SET jumper is connected in the “HAND” position.
If you use remote control (0-5V signal), set the I_SET jumper to “REMOTE” and ALWAYS MAKE SURE you have a good connection to your voltage source. In case you use a micro-controller that could go into reset-state, best add a weak (50k or so) pull-down resistor!
NEVER EVER connect the device to AC mains!
Whatever you do, please keep the heatsink below 85°C.
The MOSFETs and Schottky diode are very beefy, so they should survive a bit of abuse (high current spikes and so on), but please watch the heatsink temperature!
DO NOT let the device run unattended when connected to a high-current-capable source.
Always turn the current-set-potentiometer all the way to the left before you turn on the device.
For calibration you will need several things. One multimeter (#1) to read the output voltage (“Multimeter” binding posts), one multimeter (#2) to measure the load current (in series with one of the “DUT” binding posts), a small blade screw-driver to adjust the calibration potentiometer, a power supply that can deliver enough current and a bit of patience!
Once everything is hooked up, turn the current-set-pot all the way to the left and turn on the dummy load. Then turn on your power supply. The current reading should be zero, no magic blue smoke should rise. It would be wise to set and use the current limiting of your power supply. A suitable value for calibration might be 2A.
The following procedure is a single-point calibration. For most applications this should be enough. Feel free to use different points and optimize.
Step 1: Set the reference voltage to 1V
Turn on the dummy-load, the yellow LED should come on.
Use multimeter (#1) to measure the reference voltage, while turning the 10-turn potentiometer clockwise until you reach 1.000V. Place the multimeter probes to the pads shown below.
The dummy load should now draw some current as well. Depending on how far off the calibration is, multimeter (#2) might read 500mA or your power supply has reached the set current limit.
Step 2: Adjust the gain with the I_CAL trimmer
Use multimeter (#1) to measure the output voltage on the binding posts labeled “Multimeter”. The goal is to measure 1.000V there as well.
Now adjust the I_CAL trimmer until the “DUT” current reading of multimeter (#2) is 1.00A and the output voltage on multimeter (#1) reads 1.000V or thereabouts.
Step 3: Iteration
Play with the 10-turn potentiometer and the I_CAL trimmer until the readings of both multimeters match as closely as possible: (1.000V | 1.00A).
Once you’re satisfied with this data-point, check if the calibration is good enough in other places.
You may have to re-calibrate for vastly different current ranges if you need more accuracy. E.g. for a low current range: 0-250mA and a high current range: 0-5A.
The MOSFETs and the Schottky diode can take a lot more than just 5A – given that they are cooled appropriately!
If you need more current, alter the calibration. To get a maximum of e.g. 10A, calibrate it to (1.000V | 2.00A).
But be careful, you will be dumping a lot of energy into the heatsink. It will get hot very quickly. You may want to think about a more advanced cooling solution. Adding a fan might help, maybe not. A liquid-cooled heatsink would be best, but keep condensation in mind.