EDITS: 17/07/18 - updated links
It’s been some time since I last posted on this project because I’ve been busy with the museum and away for a few days. It only took about 4 days to get the new PCBs manufactured and delivered from China, which is pretty amazing. As it turns out, it was a bit of a blessing in disguise making the mistakes on the PCB as it gave me time to create the 4Duino interface which has proved very useful here.
I soldered up the Control board:
But note the circled component - a Schottky Rectifier. That sucker is soldered on reversed - not sure what I was thinking (actually I am - I wasn't thinking at all!) but definitely my fault for not taking enough care. The upshot was, the Mosfet and Sense resistor blew. See the thread here for more information. Fortunately, I was able to get the Schottky, Mosfet and resistor off without damaging the board, and I had spares - the Schottky survived as it happens.
Here's the control PCB on my prototyping board, nearly ready to go. Output is wired to the dummy load - a 10 Ohm, 100W resistor, actual reading of 10.1 Ohms. At the front, left to right there is the load on/off switch, Current Control and Voltage Control.
And with the 4Duino plugged up to all sensors and power - naturally I tested the 5V supply first! The thermistor waving in the breeze is reading ambient; the other two are kapton-taped to their respective component.
At this stage, I wanted to test basic functionality off the supply:
- Can it supply 15V and 1.5A through the dummy load (I really need to get a 5 Ohm power resistor as well to test the full 3 amps)
- If I turn off the load supply, will volts and current drop to 0
- If I turn the Voltage control, can I adjust the voltage down to 0; will current follow in line with Ohms law
- If I turn the Current control, can I adjust the current down to 0; will voltage follow in line with Ohms law
- Can I set current/voltage limits.
- Is the 4Duino display working and displaying readings correctly; how accurate, compared to a DMM, are the readings
The quality of the supply voltage and current - i.e. ripple, overshoot, noise, soak etc - I need a scope to do, which is on it’s way, so those tests will be a follow up post. In the meantime, watch the videos to see the results of these tests. Sorry about the quality, but I had to handhold and twiddle at the same time. The second video covers the voltage/current limit setting which I forget to do when recording the first video!
(in the video above, when I talk about iMon, I see the LT3081s generate a small current for conversion - I meant a small voltage!)
I made a number of adjustments to the 4Duino code as a result of testing (which are already in the video):
- reduced the display refresh time of temperatures down to 15secs. It didn't seem necessary to have a second-by-second temperature reading.
- reduced the display refresh time for voltage, current and power to 500ms. This was to prevent it swinging around and making it hard to read.
- changed the sampling for voltage and current from the sense chip (INA260) to average 16 readings over 8.244ms for both. This fits within a 500ms display refresh window. I also tried 64 averages over 2.116ms but it didn’t make any noticeable difference to the readings. It means that the meter display is more stable as well and I may play around further with the settings.
- the iMon current sample reported voltage for just one LTC3081 component. I’ve doubled this to get a ‘comparator’ for that read by the INA260. In future, I will drop the iMon sense and free up the Analog port on the 4Duino for something more interesting. At this time, I was purely interested for comparison purposes and it sort of serves that purpose, but the conversion (uV / 5000 * 1000 * 2) means it isn't that accurate.
- depending upon the voltage supplied, the reading by the INA260 is approximately x millivolts higher than that supplied at the output. Where x = amps read, so at 15V at the INA260, it reads approximately 15mV down at the output, as measured by the DMM. If I measure the voltage with a DMM at the INA260, the readings shown on the 4Duino and DMM are the same (adrift by 1mA) so the INA260 sense conversions are accurate. Amps are similarly higher by 15mA - 20mA. I’ll investigate this further in the next round of testing but I suspect some of it is down to lead lengths, perhaps trace length (although that's only 2cmx1.5mm)
- temperatures of the components are under control: the LTC3081s are heating up, but only in line with expectations - it’s around 50c for approximately 22.5W of provision; I think it will be important to test to the full 3A/45W. The Bridge Rectifier and Mosfet are running cool.
I’m really pleased with these results so far. Functionally at least, it is doing what I want, although obviously that’s not the same as providing a quality output. The next posting will be a bit delayed as I take delivery of a scope, play around with it and understand how best to use it. Perhaps I’ll write that up as well!