24V to 5V 10A 50W DC DC Converter Regulator Car Step Down Reducer 24V to 5V 10AMP Daygreen CE Certificated (WHL #46)
Alright, part two of the DC-DC converter tests that’ll lead to a small project. This time we got a 5V/10A converter from Daygreen, model number B10-1224-05, shipped together with the 12V/10A unit from WHL #44. This item was 2.41€ excluding shipping, as described in that blog post.
It’s a brick style unit as well, a tad smaller, but also has two mounting holes left and right which are flush with the case. Contrary to the more powerful 12V one, this is only a plastic case that was potted, but it has four screw terminals and therefore offers more flexibility when it comes to attaching your own wiring. When in free air, the case is just fine. I guess when mounted inside a badly ventilated unit that could be different, but only for applications that get close to the 10A rating. If you’re below that, it’s absolutely fine. IP68 rating – not sure how that works with the exposed contacts. These by the way are held in by regular M3 screws that can be removed entirely. It’s a bit fiddly to get them back in.
Again Daylight does strange things with the input voltage rating. This one is advertised as 10 to 30V input, however, the device has “12/24V” printed on it. The other one had a rating of 15 to 32V but got “24V” on a label on the back. Why?
Other specs include over temperature, short circuit and reverse polarity protection, from which the last one is instantly dismissed by the very last sentence in the item description: “Please connect the polarity correctly, or the converter will be burnt.”. Well, I didn’t try… Same goes for the short circuit stuff. Over temperature doesn’t kick in until well in overload conditions, and that might even be an over current protection instead of a temperature thing.
As for hard tech specs, it’s said to have 0.2% line regulation, 0.2% load regulation, 1.5% voltage accuracy and 120mVpp ripple. I once again would like to compare this to ATX power supply specs where there’s a 5% voltage tolerance for the 5V rail (4.75V to 5.25V) and 50mVpp ripple. Unlike the 12V converter, this one has a ripple rating that is significantly larger than the ATX spec.
Fortunately, my 80W power supply from the last time does offer enough oomph this time, so I was able to test the entire range of this DC-DC. It has an idle consumption of 24mA which I think could be significantly better, but given you wouldn’t buy/use/install a 10A converter for low power applications, that’s basically alright. I did leave it running for ten minutes at full 10A load and temperatures went up to 37°C in the lower right corner (near the output terminals) and 64°C at the top left. That should be clearly visible in a thermographic image – too bad I do not own a Flir camera or the like This was tested at 21°C room temperature, so that’s 43K above ambient.
I won’t bother you with more text, here’s the regulation graph:
Starting at 5.2V, ending at 4.86V under full load. 11A is something like 4.82V and it will drop below 4.5V eventually, at around 13A or 130% load. There’s a cutoff around 15.5A, but I do not know if that is induced by current or temperature. No matter the cause, you probably shouldn’t run a power supply at 155% rated load anyway…
So output regulation isn’t really that dependent on input voltage, but we’re very much off the specs posted above. Still that’s inside ATX specs which is good enough for me.
The efficiency graph is also no surprise, but the figures are generally lower than last time.
This unit is advertised as 95% peak efficiency, and it got to 94% this time at 1A/10% load. It drops below 90% at 6A and reaches 86% at full 10A load, meaning about 8W of dissipation in the unit. I don’t think there’s much influence of my wiring, as input voltage is measured on a second pair of wires at the terminals that do not carry power, current is measured via current clamp, and output voltage also has a secondary Kelvin-style pair of wires that do not carry any current. I have to say that the readings of the ADS1115 were jumping around a bit (say 2%), so I tried to catch alike voltage-current pairs every time. I also tried the same setup with a fixed ~6A load that did not show this behaviour, so the 350ish kHz switching frequency (spec: 50 kHz!) might interfere with the circuit of my electronic load.
Due to the amount of graphs in both 1µs/div and 10µs/div resolution (different to last time due to different switching frequency!) I’ll just add these pics as a gallery. Click through them as you like – the sinusoidal ripple from the no-load beginning doesn’t really get any better with increased load
Ripple starts out pretty high at 70mV (0A) and quickly jumps to the spec – but not much higher. There’s 115mV at 5A, but only 125mV at full 10A load. Let’s factor in that I did not enable the 20 MHz bandwidth filter and we’re basically exactly on spec in the worst case. A suitable low-ESR cap could help if you need better performance than that.
One more thing about the growing number of spikes within each switching cycle: This is not instant crap of failed regulation, this is oscillations on top of the oscillations! It’s inside the single-digit MHz range, and thanks to the deep memory of my Rigol, this comes out nicely when zooming in. Any ideas what’s causing this? I really would need nicer gear to rule out effects of my electronic load causing this, I think the MHz regime is still in reach for the OP amps used. But why the irregular patterns?
I also tested behaviour when turning on and off the input power, with the 6A pure ohmic load attached. Pretty unspectacular, very gracious on-ramp and also nothing wild on a sudden power loss. Ramps up in 5-6ms, discharges a bit quicker with a little bump at the very end. No overshoot at all, excellent!
Well, that’s it for this baby. Works fine, not in total violation of the manufacturer’s specs, certainly adequate for the task that I bought it for. Will report once it’s in operation!