Homemade PSU?
#1
It turns out the charger I have been using is subject to a certain amount of voltage drop, which explains a few random lockups. I don't have any other feasible chargers at hand, but what I do have, is a 12v 1A DC power source, and a TI PTR08100W switching regulator. That regulator can output up to 10A, accepts input voltage from 4.5V to 14V, and outputs 0.6V to 5.5V, and is up to 96% efficient.

Now, I have used that regulator many times with Li-Ion batteries, and it is rock solid stable even under a decently heavy load. I have not, however, used it attached to a plug-in type DC adapter.

Based on my rudimentary electronics knowledge, I believe that it should work. Power in must equal power out, and 12V@1A is 12 watts(obviously), so with the regulator it should be able to output [email protected]. Even if I assumed the regulator was only 80% efficient, it should still be able to handle [email protected].

So, am I thinking about this entirely wrong? Is there something about using a DC adapter that will make this not work?
Reply
#2
The Pi requires a 5V input to its micro-USB power connector.
This connector has only the power pins wired into the Pi.

The phone and tablet chargers you can buy (from reputable brands like ASUS and Samsung etc) with a fixed cable will probably work very well with the Pi because they have a suitably designed wire thickness in the cable.

But if you use a charger with a USB connector on the charger and a regular USB cable between it and the Pi then you are in for bad surprises like I was!
The reason is not the charger but the cable!!!!

"Anybrand" USB cables are very often designed for a minimal power feed on the red/black wires so the makers have used a very small wire, which means that when the Pi draws its 0.7-1A (depending on overclocking level amongst other things) then there is a voltage drop in the cable itself that makes the pi unreliable!
This drop can be 0.5V in some cases and brings the Pi supply out of range.....

So in your situation I would first check the cable and only after all else fails go for a self-designed power supply.
There have been recent discussions here on this very topic:

http://forum.xbmc.org/showthread.php?tid...pid1744412
http://forum.xbmc.org/showthread.php?tid=198820
http://forum.xbmc.org/showthread.php?tid...pid1742235

These have also suggested the following information pages:
http://elinux.org/R-Pi_Troubleshooting#T...r_problems
http://elinux.org/On_the_RPi_usb_power_cable
http://www.righto.com/2012/10/a-dozen-us...le-is.html

All point to the badly designed USB cable as the real culprit.
Bo Berglund
Sweden
Reply
#3
Thanks for the info. I made sure to get a high quality usb cable right from the get-go, because I had read about them being a weak point. My Pi rarely freezes up, but it does happen, maybe a couple times a month running 24/7.

I'm mostly wanting to try this idea with the PSU for the sake of trying it, but I don't want to risk burning out my Pi if I'm thinking about this wrong.
Reply
#4
OK,
I understand.
If you make a PSU of your own the only thing you need to make sure is that the output to the Pi is not going to exceed the max value in the Pi specs, which probably is something like 5.3V or so. Of course you want to stay as close to 5V as you can get in the Pi end of the cable.
The Pi itself has a protection fuse on its Micro USB power connector input, so it can handle some irregularities of incoming power.

Note that you should start out this endeavor by measuring the actual voltage between TP1 and TP2 on the Pi board using your current setup.
This is the internal 5V supply line after the input fuse so it shows what the Pi is actually fed.
If it is anything less than 4.8V then you should think of investigating the cable again.
But you should also try to measure the output (under load) of the power supply. This is not so easy because the connectors do not give access to the wires in the cables. You probably need a USB break-out connector to do this.
This test will reveal the voltage drop in the cable, though. It is whatever is measured on the power supply end minus what you measure across TP1-TP2.

Once that is done you can proceed with your own PSU design and make sure that you set it for something like 5.2V output and with a current limit of 1.5-2.0A
Then connect it up and repeat the measurements above with the pi running.
Then you will have all your needed data.
Bo Berglund
Sweden
Reply
#5
Adviced by people already running this setup i'm going to give this a go:

https://www.fasttech.com/p/1270310

with this cable

https://www.fasttech.com/p/1758001

hooked up to a normal IEC connector (with on and off switch + fuse support) with a 20A fuse in it.

Seems to be doing the trick for others and I can't say it's really expensive Smile.
Reply
#6
This is really overkill, a 5A supply is not needed by the Pi (it has an internal fuse that will operate at about 1A).
And the specifications for the cable seems not to say anything about the wire size for the power leads so you might have a bad cable right there.
But since the power supply does not have a USB connector you have anyway to cut this cable so your best bet would be to cut it about 50 mm from the micro-USB connector and then extract the red and black leads and solder on a fair size wire to the power supply (like 20-22 AWG).
This would give you a rock solid power feed to the Pi.

But beware of the safety issues here, this power supply is meant for use INSIDE an enclosure of some kind to protect the exposed high voltage connections. You have to get it out of children's reach if it is to be used in your living room...
Bo Berglund
Sweden
Reply
#7
Oh yeah sorry I knew that, I should have specified it properly.

Indeed you can just cut the cable to length making sure you get full power from it, that would avoid any bad cable problem.
As for the connection from PS to IEC I'm going with AWG 20 myself, seemed like the best choice imo.
The cable itself is just a https://www.samsung.com/in/consumer/mobi...data-cable
You can google it if you want, it's just a samsung cable, I'm not sure if it's any good but reviews of global reviews of people (that are most likely not technical at all, they just see their stuff get transfered and their phone charge) seem good.

On the plus side of the fact that this is overkill is that you can power multimple things with it simultaneously if you want to, if not, it CAN reach it while still using little to no extra power itself (if I recall correctly it's about ~75% accurate making it use something like 8 watt tops, most likely less but I don't have a proper fluke to test it with).

As for the enclosure, yea, you're screwing around with DC voltage so you need to properly insulate everything and, if you want, make a little enclosure of sort with enough airflow to keep it cool or just build all of it (including ur pi) into a little box with a 5v cooler in it Tongue.
I wouldn't suggest leaving any form of electronics laying around with small children running around. I have my Pi totally open floating somewhere inbetween my tv and the ground, that's not suggested for small children either.

Then again, involving ur kids with electronics, or any kind of scientific field for that matter, on a very young age drasticly improves their ability to surpass you in the future, or just give them the basic understanding of said field if they end up not caring Smile.
Reply
#8
(2014-07-03, 10:18)Bosse_B Wrote: The Pi requires a 5V input to its micro-USB power connector.
This connector has only the power pins wired into the Pi.

This is not entirely true. The pi can easily be powered directly by GPIO, though there are some implications such as the bypass of the polyfuse meant to protect the pi. If you fuse the input yourself, I don't see why there would be any issue. I power 2 pi's this way myself, never with any issues.

http://bit.ly/TQiDar

My GPIO connection:
Image
Reply
#9
That is where I hooked up my DVM during testing, but it is not a good way for powering since it bypasses the fuses. IMO using the externally accessible Micro-USB connector is more convenient since it is easily pluggable.
Bo Berglund
Sweden
Reply
#10
Works great for me, been like this for probably a year now. I avoid drops caused by the polyfuse itself and with my own inline fuse, it's still protected..
Reply
#11
(2014-07-08, 17:42)ActionA Wrote: Works great for me, been like this for probably a year now. I avoid drops caused by the polyfuse itself and with my own inline fuse, it's still protected..

What would be the best fuse to use for this, and what rating? I have to imagine if it's rated exactly the same as the polyfuse, then there's really not much of a point.
Reply
#12
Sure there is, it's been noted over and over that the quality of the polyfuses used (possibly depending on what specific generation of pi you have) had issues dropping voltage. I really don't want to recommend what would be best as I'm not an electrical engineer but, I'm certain that since this is a hardware related issue, it will have been covered somewhere in the raspberrypi.org forums in some depth.
Reply

Logout Mark Read Team Forum Stats Members Help
Homemade PSU?0