RG1 1.8v regulator

Thanks for that Tooms, I was just thinking about trying to get hold of a v2 to repeat some of my earlier measurements.

From your thermal images it certainly seems that the next target is to replace RG2 with some form of switching regulator.

Hi

yes i am working on getting an replacement for the RG1 and RG2 and maybe the RG3.

but as i write i have still not found a good one but i am in the process of testing the TI mano switchmode modules but this is an 8 pin QFN chip therer is 2x3mm in size and i have not done so small smd things before so it will take some time and i hope it as good as they say it is, but time will tell...  i have got the boards (they are 14.4x9.3mm) and i am in the process of mounting the parts.

Tooms

Great article Toom, thanks for sharing.

As I mentioned before, my opinion is that the entire power architecture has to be redesigned for the Rpi, the idea of using phone chargers with micro USB was not a good one and it is the source of many many problems and frustration, it also leaves out other and better ways to power the board.

What switching regulators are you looking at ?  TI has a very nice and vast line of them, particularly the SWIFT ones are highly efficient.

I had very good results with this tiny one from TI (actually an original National part later acquired by TI) which switches from 5V to 3.3V and can handle up to 1A, it is QFN and has the inductor incorporated on the chip (soldering and rework has to be done carefully to avoid removing it)

http://www.ti.com/product/lmz10501

Here is a pic of one mounted on a SMT to DIP adapter for testing (not the best or recommended pcb layout, but it worked).

One alternative I'm using on other designs is the ISL9104 from Intersil, it is high efficiency with low Iq running at 4.3MHz, so the inductor is just a small ferrite bead. Up to 500mA.

http://www.intersil.com/en/products/power-management/switching-regulators/integrated-fet-regulators/ISL9104.html

Here is a pic of a small evaluation board:

For the main power input I decided to get rid of the idea of using a USB charger and the microUSB connector and power the Rpi directly via the P1 header bypassing the main polyfuse. Then for the main 5V I'm using a SWIFT switcher from TI which I had very good results on several commercial products, the TPS54231 (http://www.ti.com/product/tps54231), I can feed up to 28V and can provide up to 2A, output voltage is adjustable via a pair of resistors. Here is a pic of a quick prototype board I put together to power the Rpi with it (at the right there is a 3.3V LDO to feed power to the extra parts on the protoboard and not drain power from the Rpi).

Let me know what part you have in mind, I can easily get samples and eval kits from TI or other vendors.

Cheers

Jorge

jamodio wrote:

 

Great article Toom, thanks for sharing.

 

As I mentioned before, my opinion is that the entire power architecture has to be redesigned for the Rpi, the idea of using phone chargers with micro USB was not a good one and it is the source of many many problems and frustration, it also leaves out other and better ways to power the board.

 

What switching regulators are you looking at ?  TI has a very nice and vast line of them, particularly the SWIFT ones are highly efficient.

 

I had very good results with this tiny one from TI (actually an original National part later acquired by TI) which switches from 5V to 3.3V and can handle up to 1A, it is QFN and has the inductor incorporated on the chip (soldering and rework has to be done carefully to avoid removing it)

http://www.ti.com/product/lmz10501

 

Here is a pic of one mounted on a SMT to DIP adapter for testing (not the best or recommended pcb layout, but it worked).

 

One alternative I'm using on other designs is the ISL9104 from Intersil, it is high efficiency with low Iq running at 4.3MHz, so the inductor is just a small ferrite bead. Up to 500mA.

http://www.intersil.com/en/products/power-management/switching-regulators/integrated-fet-regulators/ISL9104.html

Here is a pic of a small evaluation board:

 

For the main power input I decided to get rid of the idea of using a USB charger and the microUSB connector and power the Rpi directly via the P1 header bypassing the main polyfuse. Then for the main 5V I'm using a SWIFT switcher from TI which I had very good results on several commercial products, the TPS54231 (http://www.ti.com/product/tps54231), I can feed up to 28V and can provide up to 2A, output voltage is adjustable via a pair of resistors. Here is a pic of a quick prototype board I put together to power the Rpi with it (at the right there is a 3.3V LDO to feed power to the extra parts on the protoboard and not drain power from the Rpi).

 

Let me know what part you have in mind, I can easily get samples and eval kits from TI or other vendors.

 

Cheers

Jorge

Hi

yes your spot on, it is the LMZ10501 that i am trying to use and i have now got one of them mounted on my boards.

it is set for 3.3V ouput with the resisitors as per the PDF files but i am getting 3.27V so i guess that i have to change the values abit so i can get it a little higher.

here is an analyze of my module (it is in microsoft xps format as i dont has a pdf writer installed)

http://www.tooms.dk/Download/element14forum/raspi/LMZ10501-3.3V-V1.20120923.1652.xps

as you can see it is first hitting 90% eff. at 100mA and seems to top with 93,9% at 260mA, but it is alot better then the LDO.

Now i just need to build an 1.8V and 2.5V board and then do testing with the RPI board.

If you have so easy access to the parts then i will love get a couple extra of the LMZ10501 as i only have a couple of them via the sample service and maybe 3 Intersil so i can set one to 3.3V and 1.8V and the last to 2.5V

Thanks

Thomas

Tooms wrote:

 

Now i just need to build an 1.8V and 2.5V board and then do testing with the RPI board.

personally I'd not bother with the 2.5v one, going by the schematics it's only used for the composite video and doesn't look like it'll ever have significant current requirements, so an LDO will be fine.

jamodio wrote:

As I mentioned before, my opinion is that the entire power architecture has to be redesigned for the Rpi, the idea of using phone chargers with micro USB was not a good one and it is the source of many many problems and frustration, it also leaves out other and better ways to power the board.

The USB charger idea is not too bad. The main problem is that the polyfuse and cheap USB cable have too much resistance. This causes voltage sags even if the charger can provide enough current. So the foundation should've put more effort in promoting good cables, and the fuse is not ideal to say the least....

All this could've been mitigated a bit if there was some hardware or software that could monitor the power.

Roger Wolff wrote:

The USB charger idea is not too bad.

Yes is not too bad, it is extremely bad.

USB chargers are designed and intended for well, exactly that, charge a battery, not something else than in particular will be in turns providing power to attached devices.

Another important issue is that the Rpi is designed and intended to be connected to other stuff, like a monitor or TV, which has an independent power source, an aspect not taken into consideration in the design requierements of chargers for mobile devices since it is not expected you will have something else connected in turns to the mobile device.

Then some USB charger manufacturers can cut some corners to reduce cost and provide minimal isolation between main power and the device under charge, and you have seen already the reports of people having problems about blowing up the Rpi or their TV, which most probably are due to poor isolation and ground loops.

It is good that some distributors are now "bundling" or offering a power supply so users have to avoid the trial and error test with their Rpis to decipher which one works.

For a proper or better power supply you can take as a reference what was done on the Beaglebone or in a cheaper way on the OLinuXino.

-J

jamodio wrote:

 

Roger Wolff wrote:

The USB charger idea is not too bad.

 

Yes is not too bad, it is extremely bad.

 

USB chargers are designed and intended for well, exactly that, charge a battery, not something else than in particular will be in turns providing power to attached devices.

 

Another important issue is that the Rpi is designed and intended to be connected to other stuff, like a monitor or TV, which has an independent power source, an aspect not taken into consideration in the design requierements of chargers for mobile devices since it is not expected you will have something else connected in turns to the mobile device.

 

Then some USB charger manufacturers can cut some corners to reduce cost and provide minimal isolation between main power and the device under charge, and you have seen already the reports of people having problems about blowing up the Rpi or their TV, which most probably are due to poor isolation and ground loops.

 

It is good that some distributors are now "bundling" or offering a power supply so users have to avoid the trial and error test with their Rpis to decipher which one works.

 

For a proper or better power supply you can take as a reference what was done on the Beaglebone or in a cheaper way on the OLinuXino.

 

-J

So.. You're stating that USB chargers are BAD because they are designed to charge batteries and not provide a stable 5V power.

In fact, for charging a battery, something that would vary wildly between say 4.5 and 5.5V would provide adequate power for charging a battery. But in fact, nobody has found a charger that is lacking in this respect. So far even the cheap $1 usb chargers from china provide adequate power for the raspberry pi.

You're stating that usb chargers are bad because they do not allow the raspberry pi to deliver power to downstream USB devices. In fact, a charger that provides enough current will easily allow downstream devices to function properly. Most of the failures in this respect are due to the polyfuses. And non-compliant devices that only work if the power is above say 4.7V, while the spec requires them to work down to 4.2V.

So you are stating that mobile chargers might not expect other devices to be connected to the device being charged. If that were the case, they might not provide an isolated output. However that would be extremely dangerous. Even the $1 USB chargers from china are not that bad. In fact, I have not heard of any failures because of non-isolated powersupplies. (Even if say connecting the rpi to a TV would lead to trouble, then BOTH the TV and the rpi-powersupply would be at fault. If just one of them was properly isolated, there would be no trouble).

So you are stating that some usb chargers are not properly isolated. Some people are taking apart the chinese USB chargers and finding "too little separation" for their taste. In "electronics hobbyist" magazines, they sometimes have a piece about safety. There the real margins are exagerated a bit to make things safe, even if things go wrong. Then someone reads that and multiplies the margins by two just to be safe. Now we have someone walking around who shouts that such a charger is unsafe. Fine. That's what happens on the internet. Lots of people with big egos who think they know stuff are talking about things they don't know enough about. Let them talk.

You claim I have heard about people blowing up their pi or TV. I have not.

I restate my claim: the idea to power the RPI via a micro-USB connector wasn't too bad. What gives the impression of powerproblems (not related to the actual powersupply!) is:

- the polyfuses

- bad cables.

- noncompliant devices.

I'll agree that not all powersupplies are "good enoug". IIRC I've seen an USB powersupply that claims 0.5A, and its output drops to 4V or thereabouts when you load it that much. Enough to charge a battery (maybe slower, but still the battery will charge), but not enough to power a 'pi. First of all because the 'pi has always been advertized to require 0.7A. In fact the 'pi works fine on voltages near 4V. The BCM will downconvert any voltage between 3.5 and 5.5V to the 1.2V that it uses internally. Next we have a 3.3V regulator that should work just fine downto at least 4.3V, but most likely works properly down to 4V. Otherwise there are very little users of the nominally 5V on the 'pi. Again noncompliant USB devices will stop working before 4.2V.

After reading so much fanboism about the Pi, we finally get one about the usb chargers.

The Pi only works reliable between 4.85 and 5.25V. Even the foundation won't deny that.

A cheap usb charger will provide a voltage of 5V usually with a tolerance of 10%

If the Pi can't cope with these specifications, it's a bad design to use them.

Look in the Forums if you don't believe that.

You claim I have heard about people blowing up their pi or TV. I have not.

http://www.raspberrypi.org/phpBB3/viewtopic.php?f=28&t=7650

http://www.raspberrypi.org/phpBB3/viewtopic.php?t=5632&p=91242

There have been many similar reports, but it's easy to miss things due to the poor signal to noise ratio in the RPF forums.

In my case, the quality of the usb charger was never a question. The 0.5v drop across the crappy micro usb cable was the real problem. For that reason alone I agree with jamodio that the idea was a bad one.

Note that they didn't make any effort to promote better cables or better power supplies. It seemed much more like "everyone has a micro usb charger and it'll work with any old crappy charger and cable you have lying at the back of the cupboard". It's the combination of a cost constrained design with the flippant attitude that makes it a bad idea.

Hindsight is always 20-20, but in hindsight it's difficult to argue that it was anything other than the wrong choice, even when it's understandable why the choice was made.

Thing is that a crap charger and even worse cable probably works just fine as a phone charger simply down to them spending a few pennies extra on a decent power system... The same would likely be true for the Pi...