Expected PowerBRICK positive voltage output

[lemoneer]

First - forgive me if this is the wrong forum or someone else has already asked this.  I did a search, but it didn't return any results.

I've purchased a 5V powerBRICK, and while the negative voltage output is steady at ~ -4.996V the positive output voltage fluctuates quite a bit between ~ 5.29 and 5.17V as can been seen in the screenshot I've attached from my scope.  I have tried powering it from both the USB connection as well as Vin via my bench supply.  With the bench supply I varied Vin from 4.5 - 5.5V with no change in the output.  Is this the normal operation for the powerBRICK? I was expecting the positive output to be like the negative output, close to 5V without the voltage swings.

[JColvin]

Hello @lemoneer,

I don't recall that being normal operation for the PowerBRICK when I used it; I don't have one available at the moment, but I'll be able to test it out when I get back to my desk on Monday.

Thanks,
JColvin

[Bianca]

Hi @lemoneer

I tested my brick and I couldn't obtain that. Mine is quite steady at 5.128V from what I saw with the Analog Discovery 2. The devices are different from one to another and might give different values but not such fluctuation. @JColvin Unfortunately I have just one of 5V here at the office. If you could test yours to see if there is much of a difference, would be great.

Thanks,

Bianca 

[lemoneer]

Hi @Bianca and @JColvin,

In the off chance that the brick needed to run for some time to stabilize, I left it running powered by my bench supply for a few hours.  Unfortunately, it did not make a difference.  I took a couple more screenshots to hopefully add some more insight.  The first is the +5V rail with cursors added to show the length of the swing is about 60 ms and the swing in voltage is about 145mV (including noise).

And the second is the output from the -5V rail which is quite flat.

[JColvin]

Hi @lemoneer and @Bianca,

I tested a 5V PowerBrick. I'm getting a small amount of fluctuation, but depending on the power supply I use I'm getting between 30 mV and 37 mV of swing in a ~40 ms range, which isn't rock solid to be sure, but also at least 3x less than what you are seeing. Let me do some more testing to see if I can abuse this a little more (and check out some specs).

Thanks,
JColvin

 

[JColvin]

So I finally got the opportunity to try to test it some more and more or less got the same results. I changed the power supply input from the 5V that I was using to different voltages to see if the PowerBrick was just having difficultly with the generating a 5V output from a 5V input, but changing the input voltage to VIN only changed the timespan of the triangle wave fluctuation that I was seeing from 30 ms to slightly over 40 ms. The voltage difference stayed at about 40 mV (which is a little higher than what I was seeing yesterday). The negative rail solidly sits between -5.055 and -5.054 V.

Regardless, it's not as large of a swing as you're seeing, but also not what Bianca reported for her module. @Bianca, do you have any further recommendations to try out, or do you agree with my opinion that @lemoneer's particular module is not functioning correctly?

Thanks,
JColvin

[lemoneer]

I was able to get something that looks stable on the +5V rail by putting a 1000 uF cap across +5V and GND.  At least it doesn't have the 60 ms cycle.

Changing the volts per division for 78.0 mV to match my previous captures gives

Both captures were done with 5V input from my bench supply.  I haven't had the time to do any further testing.  It is definitely better than before, but the peak to peak is still 80 mV with an average voltage of 5.23V.  There isn't a schematic available on the documentation page for the powerBRICKS, but I wonder if perhaps mine doesn't have a bad cap/missing cap on the output.  That or the 1000 uF cap is just smoothing the symptoms of some underlying issue.

[Bianca]

From your screens it looks that you don't have a load connected to the brick. Can you try to see if you can get something better with a load? You can connect a resistor, that should help reduce it. 

Bianca

[lemoneer]

Hi @Bianca,

You are correct, the first captures I posted are open circuit.  I had originally tested with a 100K resistor which showed this.

That didn't seem right with the negative rail being essentially flat, so I removed the resistor to see what was going on and those are the captures I posted.  With the resistor in place, I am still getting 150mV peak to peak albeit with a shorter transition.  Can you tell me what the minimum load is?

I have noticed a couple other interesting things when testing this morning with the resistor load on the +5V rail and a DMM measuring voltage across the resistor:

1. After I turn the brick off, the voltage across the resistor begins to drop.  Once the voltage across the resistor drops below about 4.8V, applying power to the brick does not turn it on.  The voltage across the resistor continues to fall.  Once the DMM reads less than 100 mV across the resistor, if I apply power to the brick, the voltage across the resistor will come back up to 5.25V.  When I apply power and the brick fails to come on, there is a brief spike to about 5V when I turn the power off again.
2. If I place a discharged capacitor of larger than about 47uF across the +5V rail before the resistor load, the brick will not turn on.  It would seem the earlier test I did with 1000uF was with a capacitor with some degree of charge.  Turing off the brick with the resistor load discharged the capacitor.  When I later tried to turn it on, nothing happened.  Speculation is always troublesome, but perhaps the inrush current to charge the capacitor trips some over current protection or the design of the brick does not like so much capacitance on the output.

I did not monitor the -5V rail during any of this to see what it was doing, but I will this repeat the above with a second DMM and oscilloscope channel monitoring -5V this evening.

[lemoneer]

I was able to do a little more testing this evening.  First up, the +5V rail with a 2.2K resistor.

If we reduce the time scale to 1ms, we see our familiar waveform.

It seems increasing the load shortens the period, but the 150 mV peak to peak is persistent.

I did a little testing on the -5V rail and found:

1. It stays essentially flat under both test loads
2. It has no problem coming up with 1000uF across it
3. When the +5V rail shuts down (assuming due to in rush over current) it shuts down as well

Please let me know if there is something else I should do, look at, or test.

Thanks,

Matthew

[JColvin]

Hello,

I have sent you a PM about this.

Thanks,
JColvin