Home › Forums › Hardware discussions › Info: Power supply via barrel connector.
As I found the LED to be extremely bright and quite hot, I was very close to desoldering R48 and replacing the 1K resistor with a 4K7, 6K8 or even 10K resistor. (12V through a 1K resistor means that the LED will use 12mA, which is quite a lot these days; especially for SMT LEDs).
But when I saw it’s a 0402, I hesitated; I only have 0603 stocked.
-So I started looking at the schematic regarding the power.
First of all, I wanted to see what the acceptable voltage range was.
So I looked at the schematic and searched for all occurrences of “12V”.
There were only 6, so they were easily checked.
First thing I saw was a 13V zener. This means the voltage should not exceed 13VDC.
Next thing I checked was the absolute minimum and maximum voltage for each of the connected components.
That turned out to be the RT6220; the datasheet is easily availble for this component.
I matched the EspressoBIN schematic up against the 5V application circuit and it is very close; there are only slight changes.
The datasheet says that the RT6220 will run on a minimum voltage of 5.2V.
As there are no other components that are connected before the RT6220, I thought I’d try a 5.9V / 3.5A power supply.
I checked the polarity of the DC barrel; center = +5.9V, outer ring = GND. Size was alright too: 2.1mm.
Time to try it; plugged it in with the microUSB connected; my terminal gave the usual output. The EspressoBIN runs fine.
As I read elsewhere in the forum that some people had problems running the board without 5V connected on the microUSB, I decided to test if I could run with only the 5.9V PSU connected. I ensured that I could SSH into the board and rebooted the device without the micro-USB cable.
Again a success; it works well.
Doing this means you cannot power a 3.5″ SATA harddisk. I will not be using a 3.5″ drive anyway (I’m going for the 2.5″ WD Red drives from Western Digital, as they consume very little power: only 1.4 Watts when active and 0.2 Watts when sleeping).
There is one more thing that I need to try when it becomes possible for me:
I’ve seen that people mention that it’s not possible to power the board via the micro-USB socket.
This is actually a real bad idea anyway; I won’t recommend doing it, because micro-USB is not suitable as a power connector and if the device is using more than a few hundred milliamps, then the connector will get very hot. Some people had problems with their SBCs due to the micro-USB connector powering the boards.
The reason I want to try it anyway, is that I know that most people would use a switchmode PSU for the micro-USB power.
I’ve had very bad experience with putting buck converters after a switchmode PSU (eg. switchmode after switchmode isn’t really a good idea).
-So I’d like to try a conventional PSU (yes, the “good old” heavy power supplies that are very inefficient). If that can’t power the board, then I’m pretty sure it’s settled. 😉
Hmm.. I might just desolder that R48; it has no function anyway.
-The only thing the on-board LED is good for is to confirm that the DC barrel connector isn’t broken.
(it’d be much better to connect a LED to +3.3V on the GPIO block – or even extract the 12V from the Molex connector if needing the same functionality).
A small update:
I just tried a switchmode power supply rated 12V/2A for a 2GB EspressoBIN with a 3.5″ WD Blue disk.
The WD Blue disk draws 0.65A on the 5V supply and 0.5A on the 12V supply.
The board boots up fine, but if I log in via SSH and log in via the “serial” console, then as soon as I log out from the “serial” console, the board freezes.
Various other combinations result in the same thing. Logging in via SSH twice causes the board to freeze.
If disconnecting the harddisk this does not happen.
-So I expect that my 12V/2A switchmode power supply does not really make 2A available to the board.
It’s very likely that there is a voltage drop as soon as I start drawing a little extra current.
Thus I’d recommend either an old power-hungry iron-core PSU or a switchmode PSU rated for much more than 2A.
One more thing: I will not recommend powering the board via the LP4 Molex connector.
The reson is that the DC-jack works both as a power-inlet and a switch. It will switch the 5V DC power supply from the micro-USB connector off, and use 12V on the same circuit instead. This power is fed into the buck-regulator that generates 5V. -So do not use an ATX PSU directly with the LP4 connector or you might fry components on the EspressoBIN and maybe even your USB port on your computer if you’ve connected the serial cable.
The LP4 connector is really a mistake!! First of all, male plugs must NEVER supply power.
-Imagine that you have 110V or 240V prongs sticking out of the wall everywhere in your house.
Aside from the risk of electric shocks you’d get very often, the risk of short circuiting two prongs would be very high.
Male plugs are made for receiving power, female plugs are made for supplying power.
A few other reasons that the LP4 plug is a mistake:
The entire board contains very fine components. The LP4 connector will often require a lot of force to disconnect (and even to connect). Aside from that, the LP4 connectors are usually of such poor quality that they’re difficult to mate and it can result in damaging either the male or the female connector.
My recommendation is to not use the LP4 connector at all, but power your harddisk from an external PSU completely.
For instance, Mean Well RPD-60A would be great for a single 3.5″ drive – use a separate PSU for the board.
RPD-60A can be built into a 1U rack casing and can be purchased from Farnell/Newark/Element14: 2815574 or RS: 644-7203. It’s medical and IT-Equipment approved, plus it has all the certifications for being used world-wide.
I’ve even considered desoldering it (like I desoldered R48 on both boards to get rid of the too-bright LED light).
One connector that *could* have been used instead of the LP4 connector, would be a female SATA power connector.
Those are made for PCB-mounting and come in SMT variants. They’re widely used for backplanes and USB docking stations.
Did you do anything regarding those (bright) LEDs?
I unsoldered the resistor, no more annoying LED light. =)
One could solder a replacement 1K resistor in there, but it really has absolutely no function.
It’s just as good as having a LED on the power supply, as the LEDs are before any electronics; even before the power regulator.
But I’m not fond of the EspressoBIN. It’s been a real disappointment over the years. Never was stable.
This is not Marvell’s fault; they produce high quality microprocessors; the problem is with GlobalScale’s poor quality.
It really looks like I should have bought a Clearfog (Armada 388) instead; they offer much more goodies and the quality is a lot better.
Technical specification tables can not be displayed on mobile. Please view on desktop