Raspberry Pi Forums

IN short, the rumour here is that somehow the power supply in the Pi is faulty.

It is not. Hardware engineer with $4000+ in electronic test equipment with thirty years using it, and photos below, say it isnt.

Many are experiencing problems with kbd and mouse, stuck key, pointer etc and are taking the easy scapegoat to blame others who designed and built this device, who, BTW, at least as far as Sony goes, are MASTERS of the question.

My other post details mouse pointer problems and of course, persons are eager to hold their opinions that power supplies are the problem, so I dug out the test equipment, schematics for the Pi and datasheets on the PS regulator.

The primary power regulator (RG2, primary because its the largest bulk regulator) is made by ON Semiconductor, not opinions of persons on Forums.

Their datasheet gives - input voltage range 4,1 -4,4 VDC, my bulk switch mode power supply is 4,95Vdc.

Thus the OPINION of a poster here that 4,75V is insufficient, is wrong.

Regulated current is rated at 1 amp, so the 3-G modem will probably not work as the one I have (Novatel) is rated at almost 0,7 amp. Too close for comfort. That is a high power, high current device whereas a kbd/ mouse are not, they are just data devices and a low power microwave link back to the Pi.

As to heat, the Pi circuit board is correctly designed as to package heat sink tab copper area. IOW, the green area under and around TP1 is a heat sink, its correct at their spec of 400 mm-square.

The Polyfuse is dropping 0,1 volt, insignificant, well within the differential voltage of RG2, so theres nothing wrong with it. Its also an extremely high tech, well tested and documented product of a giant high tech manufacturer and not opinioned Trolls on forums.

Notice all the complaints are about one power supply circuit (RG2) but there are two other similar regulators and no one is complaining about them, they are the same type made by the same company.

People are quick to pick up easy and convenient rumours and run with them, its referred to as the Avaiablity Cascade in psychology. Wiki it.

Those measurements are DC, this is not a DC device and these are not DC power supplies, they supply high frequency AC.

This is why the digital volt-ohm meters used by probably all of you are useless. No DVM worth $15 US is worth anything but for a door stop in a very complex AC device such as this.

O-scope traces photographed below, the AC noise floor on the 3,3 RG2 output is so low I can barely measure it, 10 mV-P-P. Thats zero, compared to 3,3V.

The 5V bulk input from the switched mode PS is about 50 mV noise, and RG2 eliminates all of it, so its not relevant. That means there is also no significant noise or power brownout to the USB port, which uses bulk 5V. That noise was measured at around 70mV which is also zero. USB connector 5V supply is 4.95 also, so theres no power drop from some mythical Polyfuses.

There is a modest increase in noise when the mouse is moved or some program is launching, but its on the order of 5's of mV, and that is for all practical purposes, zero.

In short, unless the bulk power supply is about 4.5VDC or less, damaged, connector on the Pi damaged, the Pi damaged from improper ESD handling or some USB peripheral such as a high powered 3G/4G modem is used, theres nothing to blame in power supplies.

Certainly nothing that Trolls and code hacks on a Forum are qualified in.

NOW Im gloating.

As to errant kbds and mice, there was some correlation between RFI and pointer problems, but its not consistent.

The traces below and measurements above were with the newer Chinese wireless KBD/mouse that was causing the severe pointer crashes. An older Logitech set with a receiver unit a distance from the Pi across a cable worked much better. Not proper, pointer works but is in visible until 'Customize look and feel' is run, which appears to be a code problem,

Linux is known to be flaky. At least under Windows I can get a mouse pointer, it just wont shut down (Windows haters humour)

This set of phantom problems points to either RFI or flawed code, and with the correlation to moving the keyboard, then moving the microwave receiver away (that made the GUI work better, the Taskbar icons were screwed up with the new KBD/mouse receiver plugged directly into the USB port) then its probably RFI.

WHen its a bare board and an extremely high tech electronic device without shielding used in an environment where lots of RF noise is about, the first suspect, again, is NOT a readily avalable scapegoat that the average Forum user understands (power supply), the first suspect is RFI.

Many Degreed EEs dont understand RFI.

Theres no excuse for a wireless KBD not working as far as the USB port goes, we have USB standards and theres no reason to not design to that standard, especially since those who designed the Pi did not specially design the comms chip for the Pi. The USB port should be OK digitally. It is from a power standpoint, the test equipment proves it.

It is NOT power supplies and Polyfuses, unless someone has abused the board, and again, being a very complex high tech device used by amateurs, thats all but guaranteed, when electronics get damaged by ESD, they very often draw large currents.

As to Wi Fi, my Belkin N card has no meaningful effect on power supply. IT works fine. Since the SW wasnt available to launch the 3G, and it might crash anyway, so much the better it didnt work. That would have to be used via a powered Hub, and at nearly an amp current, would be good practice for a PC also, those signal connectors in the USB socket are not very heavy.

Knowing that I didnt check the SHA1 code on the first OS burn, Im off to do it again, failing that, MCM will get this junk back for a refund.

it might be made by a qualified contractor, but if it wont work, whats the point?

PS the catch with the modern wireless KBD/mouse is that the microwave receiver is built into the USB connector, that stands to couple RF energy into the Pi that it may not be able to handle.

PSS As the USB connector receives bulk 5V from the external power supply, the supply should be a switch mode regulated supply and not an ordinary AC-line transformer type.

The first Pi's weren't assembled at the Sony factory in Wales. They were made in China. Being made there, it wouldn't suprise me that the manufacturer uses some cheap Chinese clone linear regulators instead of the original ones from which you checked the datasheets.
Further on, it seems like you have missed the 5V connection that goes directly to the Broadcom soc.
So, there is either a switching or linear regulator in the soc itself, or there are circuits in it that operate on the 5V directly. It would suprise me if you know the specs for that chip supply, and if it's known to work down to 4V4 as well?
I have read tests done with an adjustable supply. Why the Pi core indeed keeps working at lower voltage, usb devices loose connection and the video output dissapears.
There are also reports of people that changed the power cable by a thicker one and got the Pi working with it. They did measure a slightly higher voltage with the new cable.
So, it really happened several times already that ensuring a supply voltage of at least 4.85V (TP1 - TP2) made the Pi work, when it didn't worked with a lower voltage. Maybe it's because the soc can't handle EMI equal well on the lower voltage, who will tell. We all know that the time of nice square digital pulses on digital equipment is long gone. The frequency has become so high that every inch of pcb track is basically a coil that forms a resonance circuit by the capacitance with nearby tracks.
Besides that, you are probably right they are doing a great job at the sony assembly factory in Wales, but they shouldn't be honoured for the carefull design of the pcb, since it's not their design. Maybe they did the rerouting that changed the rev1 board into rev 2. But the initial design and routing of the board is not their work. Honor who deserves honor.

Interesting I like your approach.

But I fell like I follow the board fairly well. And your post is a bit confusing. I don't see a lot of people blaming the Pi it falls into a few categories one they have a power supply that claims to be one thing and isn't this has nothing to do with the Pi or it's ability a cheap power supply or one that isn't made properly will cause problems. The other trying to get too much out of the device no an overpower hungry keyboard and mouse won't work flawlessly even with a good power supply let alone mixing in a power hungry wireless card.

There will always be those that want to believe what they believe and no amount of convincing will change their mind. The device is solid and if those that purchase the Pi understand that the power going in is key they will encounter less trouble. The ultimate solution will be when they package a power supply with the Pi that is as solid as the Pi itself. Even then over powered keyboards and peripherals will cause those to come saying hey my Pi is broken. And there are a few that have actually received a defective Pi I would wager it's less than 1% of those manufactured but a few have been through the ringer and were left with no other option but to try and get a replacement. I have only seen this maybe once or twice in all the posts but don't keep count.

The 3.3 volt supply (and the others derived from it) are not the issue. The "power problem" is the 5 volt supply which has no onboard regulation and which (after passing through polyfuses) is used for the 5 volt supplies on the USB sockets and possibly in the LAN and USB interface sections of the onboard LAN/USB chip.

Having said that, it is indeed the case that deficiencies with that supply are not the sole reason for all RPi problems.

(Hardware engineer with 47 years experience of using electronic test equipment, many items of which cost more than $4000 each.)