5V supply from custom cape and USB & mv between grounds - BeagleBone Black

We have a custom cape that we power up the BeagleBone Black with through J9_6 (VDD_5V) and J9_2 (GND). This was originally used flawlessly with the BeagleBone Black Wireless. Since they have been discontinued we’ve started using the Black and it requires connecting to laptops using USB, which is also providing power to the board.

We’ve started to see some strange issues with DC offsets in signals and more noise than we used to have and occasionally our 5V linear regulator (LM7805 in TO222 package with a heatsink) gets very warm. I may have erroneously thought the PMIC would handle these multiple 5V sources. One could very likely be a slightly different potential than the other which if not handled I believe would cause some strange behaviors (but not enough to shut the Beagle down).

So, is this a potential issue?

If so, how do we connect the USB w/o providing power to the Beagle through it? It seems that would be our best method anyway and the machine we’re developing does not need the laptop. We just need it during R&D.

I think what you are doing should be fine.

Yes the PMIC should switch between using the barrel jack/VDD_5V and the USB power.

From a quick perusal of the PMIC datasheet it does look like it would be possible to turn on both the AC input and the USB input at the same time.

The reset state for the power path register does in fact enable both AC and USB power inputs, but looking at the block diagram it would appear the inputs are protected from driving back to each other.

I don’t think slight voltage differences would cause an issue. Have you got one of those handy USB power monitor plugs ? Would be interesting to see if the BBB is drawing any power through the USB.

I assume your 7805 only gets hot when the USB is plugged in ?

Thanks! I’ll have to get one of those handy USB power monitor plugs.

There’s something weird going on with our board and I’m working on figuring it out. Very strange. Getting ~18-~32mv between various ground points on the board and so far I have not found any bad solder joints, etc. We’ve made this board before without any issues like this but since using the Black and not the Wireless, I think we’ve had this problem. Do you know if it ok for GNDA_ADC and GND to be connected? We’ve always had it this way based on some posts I had read a few years ago. Maybe the Black care more than the Wireless did?

Connecting GNDA_ADC and GND won’t hurt, however if you are using the ADC probably best not to or you are likely to get more noise of the ADC’s. The 2 grounds are separated by a ferrite bead.

Is your cape pcb single sided / double sided ? 18-32mv is not very much. What does it look like on a scope ? Some of that is probably ripple/noise.

The big spikes are true signals. There is definitely noise and we have fought that for a long time and have knocked a lot of it out. But the DC offset ~45mv is new and you can see how it (and the noise) drop just past 39.18s. It’s a mess.


That does look a bit strange. What is different in the system before 39.18s and after ?

Without a lot more information about the signal, its source, what is driving it and what is connected to it, it would be impossible to guess at the cause.

Does this only happen when the USB is connected ?

If it does only happen when the USB cable is connected you could try something like this -

USB CARE01 USB isolator, galvanic isolation

Sorry for the slow response…life got in the way this week.

So, here’s what’s weird. Typically, we use a trimpot to tune the signal to it’s baseline value - around -5mv - which depends a bit on the material in the sensor’s ‘view’. When the Beagle is plugged in to the software developer’s Asus Laptop’s USB AND the laptop is plugged in to its’ AC power adaptor, the baseline signal rises slowly to about 0.3V. If it is not plugged in to the AC power adaptor, it does not rise. If the AC power adaptor is disconnected, it goes down to the baseline value slowly.

Now, if instead, we plug in to my Lenovo Yoga laptop’s USB, the baseline does not change at all.

Very, very strange.

In all these situations, I measure the 5V pin to GND voltage on our custom board and the 1.8V we supply from a linear regulator separate from the Beagle, but having the same ground, and the values measured never change as we plug in and unplug these computer connections mentioned above.

No idea what is going on but I suspect something with the Grounds. Always a safe guess, right??? :slight_smile:
Needless to say we are incredibly puzzled and frustrated by this.


It is certainly puzzling.

Obviously the issue is with the Asus laptop AC adapter.
Is the AC adapter earthed ? If so is the DC GND output tied to earth ? It shouldn’t be but you never know.

If this is a product you aim to ship then potentially it is going to be an issue.
If it is just for yourself, then I would suggest either trying the above galvanically isolated USB adaptor or an opto isolated USB adapter assuming they work at the required USB speeds.

The alternative of course is not to use USB. You said you used the black wifi.
Can you not use a usb wifi dongle to provide wifi ?

The system is on a mobile platform that is not connected to any mains and not connected to earth but the AC adaptor for the laptop is plugged in to mains - two prong. I have no way to know if the DC GND output of it is tied to earth really.

This is a prototype of a product and we do not intend to have it tethered to a computer via USB. We will have it connect via wi-fi. We had always used it with the Black Wireless via wifi but only had a few of those left (why, oh, why did they stop making these???). We are novices at Linux, drivers, kernel, etc. but I did follow some on-line information and got a wifi dongle to work with the USB. But all that does is allow me to connect to an existing wifi network. We had been using the wifi network that the Black Wireless brings up on its own and connected to that with a tablet, phone or laptop and it worked beautifully. (Then, no more BB Wireless.) During this R&D phase we were satisfied with plugging in with the USB while another team member gets us up with the same wifi network service we had with the wireless. He’s just part-time and its taking a long time.

So your question caused me to decide to use the last unused BBB Wireless in stock here and put it in this prototype so we can eliminate that USB connection to this Asus computer. We did that and the DC drift caused by the AC power connector is gone. Thanks for asking that question! Sometimes we get so close a problem that we don’t see the simplest solution anymore.