I’ve been testing the U3V70A regulator and I’ve noticed that its output isn’t consistent over time when given battery power and a purely resistive load.
The setup is as follows:
U3V70A regulator set for 5 V output
Regulator input is connected to two Li-ion 18650 batteries that are essentially in parallel
Regulator output is connected to a 1 m, 20 AWG USB-C cable
USB cable connects to a small USB-C breakout board
USB-C breakout board attaches to ~4 Ohm resistor network, giving a 1.25 A load
I charged the batteries briefly - maybe 10-15 minutes. The brief charging time was to keep the test short, as I didn’t want it to run for hours. I then connected the resistive load by plugging the output USB-C cable into the breakout board.
The screenshots below are the regulator output voltage as measured at the load resistance. You can see the voltage jump up when the load is first attached to the regulator, as expected. Unfortunately it doesn’t jump directly to 5 V, but moves up & down. These movements aren’t brief, but tens of seconds long, as you’ll see from the oscilloscope time scale.
The two waveforms are from the same test. You could concatenate them and consider them one long trace. The scope was in roll mode because of the long time scale. I grabbed the first waveform, and then after a few of its features had moved off screen, grabbed the second.
Does this look like normal behavior from this regulator? I ordered two of them and I believe I’ve seen it on both, though I’ve only investigated it on one.
Thank you for including so much detail in your post. I would expect the regulator’s output to be more stable than that. The load you mentioned might be too much for your partially charged batteries (it’s around 1.75A on the input side when stepping up from 3.6V). If you still see the same behavior after fully charging your batteries, could you try monitoring the voltage directly at the regulator’s VOUT and VIN pins and post those waveforms as well?
Over the past couple days I was able to run more experiments and I now believe it’s the regulator itself that is the problem.
I’ve basically run 2 experiments: one using batteries to power the regulator, the other using a benchtop power supply. Both exhibited the same problem: the load voltage isn’t stable with time.
Expmt #1: Same setup as described in the initial post, but with fully-charged batteries instead of partially-charged. Monitoring voltages at all 3 places of interest: reg input pins, reg output pins, load. Scope is in roll mode due to the long time scale. Screenshots overlap in time - they are snapshots of a single continuous test. Time scale: 300 s/div.
Expmt. 2: Same as previous, but replaced the batteries with a benchtop power supply set to 4.2 V output. Display read ~1.6 A. Test leads from PS to regulator were identical to the leads from the batteries to the regulator in the previous experiment (2 @ 16 AWG, ~6" each). As before, screenshots are overlapping & make up a single continuous test. There were no batteries to die, so you won’t see the device shut off. Time scale: 60 s/div.
To summarize, the problem occurs whether the batteries are partially or fully charged, and it even occurs when the batteries are replaced with a benchtop power supply. This leads me to suspect the regulator itself is the problem.
It’s worth re-emphasizing that the long time scales on the oscilloscope show that these are very long-lasting events. They are on the order of seconds, tens of seconds, or even minutes - they aren’t milli- or microsecond glitches.
It’s also worth noting that the wiggles were not caused by bumping the test setup or otherwise messing with it. I was literally just sitting there staring at it. And in the fully-charged-battery experiment, I was out of the room for a good chunk of the time.
Yup - I found the culprit. It was my resistive load.
I began to wonder about this because it seemed very strange that the regulator output pins themselves showed a stable voltage, yet the load voltage was moving around. I figured this could only happen if the load current was changing, which would cause a changing voltage drop across the USB-C cable’s resistance. It’s this cable that feeds the load. I think that theory turned out to be correct. But how could the load current be changing??
Well, I had used a solderless breadboard to connect my load resistors together. I kept the connections VERY short because I knew I was pushing the amperage limit of a breadboard, and I also knew that my load resistors got hot.
To test the theory, I decided to solder my load resistors together, and then solder them to the USB-C breakout board. In other words, completely remove the breadboard from the setup.
The result? The load voltage was perfectly stable - both with batteries and with the benchtop power supply. See the screenshots below. I only captured ~10 minutes each because this should easily have been long enough to see the problems that show up early.
So I think what was happening was that shortly after plugging in the load, the resistors/breadboad would heat up locally, causing connection resistances that fluctuated with temperature. Maybe the leads were even moving slightly as the breadboard responded to the heat - not sure.
And in the fully-charged-battery experiment, I was out of the room for a good chunk of the time.
