Booting the RASPI with no keyboard?

[Dave_G_2]

Apologies if I have posted in the wrong section or this topic has been covered before.

I have just read in a review written by pcpro http://www.pcpro.co.uk/reviews.....pi-model-b (see page 2)

that the RASPI will not boot without a keyboard:

Quote:

"Indeed, our system simply refused to boot until we had a keyboard connected."

Not having a Pi (yet), can anyone confirm if this is true?

[Gert van Loo]

Simple and plain: False.

[RaTTuS]

it can and does boot without a keyboard

[Dave_G_2]

Fantastic, thank you.

Someone from the Foundation should send pcpro a quick mail informing them of their inaccuracy/ies, there is probably more then one.

[Joe Schmoe]

Note that the quoted text didn't say that the board won't boot without a keyboard - just that they (the testers) couldn't do it.  I.e., this is in the same category as someone saying that the HDMI output doesn't work.  We all know that it does, so there is obviously something else going on in the tester's situation.

[Dave_G_2]

It may be that the testers didn't know how to do it, although I can't see it needing any coding to set that option.

Either it works or it does not.

Which brings me nicely back to the original reason for my post as experience has taught me that people that write reviews generally know very little about what they are reviewing and consequently just confuse readers.

[asb]

Dave_G_2 said:



Not having a Pi (yet), can anyone confirm if this is true?


Interesting. I've certainly booted my Pi plenty of times with no keyboard, and it makes no sense that lack of a keyboard should be a problem. I didn't test the particular image that PCPro received with no keyboard...but it's hard to think of a logical reason why it would fail. I'll have to take a look.

[Dave_G_2]

Thanks asb.

Apparently they received the unit for review directly from the Foundation so one would assume that it came with the Foundations preferred distro, Debian.

[Joe Schmoe]

Re #6:

1) It sounds like we agree - that it was tester incompetence at work.

2) The core issue that seems to be emerging from all the posts I've been reading from new Pi owners - is that the power supply is key.  Bad power seems to explain everything and anything - and this could well be the downfall of the Pi - that just any old phone charger just doesn't cut it.

Though, what we've been hearing is more along the lines of:

Bad power + keyboard = not work

Rather than:

Bad power + no keyboard = not work

So, there ya go...

[Dave_G_2]

I think you have hit the nail on the head, it may well be the power issue although surely adding the keyboard would only add to the already overloaded power supply's load.

While there is no doubt that power supplies are definitely responsible for some of the problems, it would be interesting to investigate exactly what causes the problem/s.

Is it a volt drop caused by the overloaded supply, or extra ripple on the supplies output, perhaps the voltage regulator/s used on the Pi"s board are not as " low drop out" as the designers thought or even all of them.

Once I get my Pi, I will purposely be using an inferior power supply and then hauling out the oscilloscope and measuring supply ripple and voltage.

[asb]

Dave_G_2 said:


Thanks asb.

Apparently they received the unit for review directly from the Foundation so one would assume that it came with the Foundations preferred distro, Debian.


Yes, I prepared the demo image that has been sent out. It's basically Debian with XBMC, Quake, the Qt graphics demos, Scratch etc pre-installed and easy to launch. Hope to integrate some of that (in a cleaner way) in the reference image that's available for download.

[Dave_G_2]

Just had a look at the Pi's schematics and I see the main regulator is a ncp1117 (3.3v).

Having a look at it's datasheet, it quotes the drop out voltage (@800mA) as:

Vin – Vout = 1.1V

( http://www.onsemi.com/pub_link.....1117-D.PDF )

This means that 4.4V is needed as a minimum input for the device to maintain regulation.

Now 4.4V is only 0.6V below the rated 5V recommended input and it's very likely that the cheap power supplies are dipping below this and causing the regulator to drop out of it's regulation range and causing the numerous problems.

Just a thought.

Perhaps one of the designers could confirm/dispel/add to this.

Edit:

Just noticed that the 5v supply also goes to the BCM2835 where it then generates some other needed voltages.

Not knowing the Broadcom chip set, perhaps it has a built-in voltage supervisory circuit which holds the processor in reset until the supply (5V) is up to spec?

Again the lower supply given out by the cheaper supplies are causing the problem.

[Gert van Loo]

If the supply drops below 4.75 volts your out of spec for everyhing. So 4.4 is far below that. And no that input to BCM2835 can live from 4.4 volts.

[Dave_G_2]

Hi Gert

Thank you for clarifying that the "headroom" on the low side is in fact 0.25v as opposed to 0.6V

Certainly cheaper power supplies will get pulled down to 4.75V and even lower.

[jbeale]

I tried 4 different USB supplies rated for 1 amp, testing under load (using a 5 ohm load resistor, to draw 1 amp at 5 V). All four of them could actually do it and still stay above 4.75 volts (current and voltage details marked in below photo). But no doubt the lower-spec'd supplies, and the super-cheap ebay specials, do not do as well.  I also tried a 500 mA supply from SparkFun (not shown) which could only do 2.5 V into a 10 ohm load (500 mA @ 5V).







From Electronics

[Dave_G_2]

Thanks jbeale, that is good info which will help people make better choices as regards their power supplies.

As regards the Sparkfun PSU, surely you mean 5V into 10 ohms (500mA) and not

2.5V into 10 ohms (250mA) as you wrote.

Do you by any chance happen to have an oscilloscope at hand?

That way you could check the better power supplies under load for ripple content.

There is a small chance that the ripple is a tad high and the multimeter (I assumed you used one as opposed to a scope) could be "averaging" the ripple and the DC together.

If you don't have access to a scope, no big deal as the fact that the power supplies managed an output in excess of 4.75V, probably means the ripple is pretty low.

[asb]

Tested, my Pi boots no problem without a keyboard connected.

[Tass]

asb said:


Tested, my Pi boots no problem without a keyboard connected.


Same here - boots up fine without a keyboard on the latest Debian image (19-04-2012).  I'm running it headless (just Ethernet & power) and using SSH/VNC to connect - would the lack of a screen help the situation?  As a side, I'm powering it off my laptop USB.

[Dave_G_2]

Thanks asb for confirming (along with Gert, RaTTuS and Tass) that the RASPI will boot with no keyboard connected.

Makes one wonder what the person doing the PcPro review was up to.

Tass, consider yourself very lucky that you have such forgiving USB ports on your laptop as they normally switch off when the load exceeds 500mA.

[Tass]

Dave_G_2 said:


Tass, consider yourself very lucky that you have such forgiving USB ports on your laptop as they normally switch off when the load exceeds 500mA.


Yes - I've been warned by several people about that   I suppose I might hit problems as I start doing more with it?  Right now I'm just testing out various configs.

[jbeale]

Dave_G_2 said:

As regards the Sparkfun PSU, surely you mean 5V into 10 ohms (500mA) and not 2.5V into 10 ohms (250mA) as you wrote.
That would make more sense, yes? But in fact, no. The Sparkfun supply labelled "TRAVEL CHARGER  MODEL: BDL<IC> INPUT: 100-240VAC 50/60HZ OUTPUT: 5V= 500mA" when connected to 10 ohm 1% resistor (through a cable & two connectors, so really a bit more than 10 ohms) gives me a nasty unregulated waveform averaging about 2.5 V. It looks like this:







From Electronics



On the 1 A supplies I mentioned, I also checked with the Vac setting, which was near 1 mV (RMS) or below, so I considered that good enough to consider well regulated, without checking on the scope.

[jbeale]

Ah well, the 1 mV ripple is only considering low frequencies. I believe the Vac(RMS) setting on the multimeter doesn't have much frequency response above 1 kHz. It turns out there is definitely noise around 60 kHz on the Adafruit "5.25V +/- 5% 1A" supply, with some higher frequency spikes as well. The scope says 56 mV(rms) in a 20 MHz BW and 372 mVpp. Scope plot below is in AC coupled mode, while supply is delivering 500 mA to 10 ohm load. Probably the other 1A supplies are similar, they're all cheap switcher designs.







From Electronics



(and, maybe a moderator could move at least some portion of this thread, which has wandered away from "Absolute beginners" I think

[Dave_G_2]

Thanks very much for that jbeale.

That sparkFun PSU is no "fun" at all, in fact it"s down right horrid.

Now we know which ones to definitely avoid.

EDIT:

Just out of interest, what kind of resistor did you use as the load?

If it's a wire wound type, perhaps it's too inductive at the frequency of the supplies operation and causing it to go a bit "bonkers".

END EDIT

As regards the high frequency spikes/noise on the supply, it is very common in SMPSU units like you say.

I don"t know if that amount of "crud" on the supply is sufficient to negatively impact on the proper operation of the Pi (or even bleed thru to the audio output and for that matter video) but considering that most SMPSU units operate at between about 50 and 300KHz, it may be wise to incorporate an inductor in series with the +V input line which would then form a low pass filter with the capacitors on the Pi PCB.

The Pi has a 220uF cap on it"s supply input, just don"t know if it"s a low ESR type.

You are also correct that this thread has gone beyond the realms of "Absolute beginners".

My apologies for that.

[jbeale]

I'm using a DALE RS-10 10ohm 1% 10W resistor, which is a wirewound type. For the 5 ohm load I used two in parallel. I don't know the actual inductance.

You're quite right though, the RasPi should look like a capacitive load instead, so a wirewound resistor isn't the best matched test load for measuring ripple voltage.

[Dave_G_2]

Not to worry as it still shows that not all power supplies are created equal and have very different outputs under the same load conditions.

I for one will be buying two power supplies, a good quality 2 Amp model (even if it does not have the correct connector) and another very cheap one (with the right connector on a lead) which I will then cut the lead off, throw the cheapy away, put a high current RFC in series with the good supplies +V terminal and use the lead from the cheapy to connect to the Pi.