It seems the newer screens are less efficient :/


bismuthdrummer

Active Member
Joined
May 13, 2011
Messages
534
Something I'd wondered when I first received my 1 GHz was whether the newer screen used less power than the older one since the screen itself was darker. Today I had the opportunity to test after swapping the screens between the CircuitCo (CC) and 1 GHz Rebirth Pandoras.

This is admittedly a very primitive testing method. First, I turned the screens all the way off and checked the current draw off the battery. The 1 GHz was higher, as expected, probably about 170 mA versus the CC's 150 mA. Then I turned the screens all the way up, and the power usage was still higher, 320 mA to about 270 mA. But my memory reminded me that the former number used to be about 350 mA idle on full brightness for my 1 GHz, and about 250 mA idle for the CC. My memory might be failing me, so anyone with their respective units can confirm or refute as appropriate.

However, the final test is so irrefutable to my eyes that I feel obligated to report that the older screens draw less power for the brightness, and significantly so. Leaving the CC on full brightness I matched the Pandoras in brightness as best I could, and seeing that there wasn't a setting exactly the same, I set the 1 GHz slightly brighter. At these settings, the 1 GHz draws about 250 mA off the battery, while the CC draws about 270 mA (same as above). Below are the photographs for inspection, again perhaps not the most scientifically conclusive results due to the methodology but I think others' perceptions will agree.

There was no overclocking or underclocking on either unit, no sound coming out of the speakers, wifi/bluetooth radios turned off for both, USB host and networking turned on for both (but nothing plugged into either USB port), SuperZaxxon 1.53 on both, so I cannot imagine what else could cause the difference. Hopefully some of those here with the knowledge can elucidate on why or direct me on other tests to try, provided they don't involve me taking apart either Pandora again.

DSCN2222.JPG

DSCN2223.JPG

DSCN2226.JPG

DSCN2231.JPG

EDIT: Added the last shot in the dark to better compare the brightness.
 
Last edited by a moderator:
Well, the new screens have a different gamma, so even with the same backlight, it might look darker in some areas.

(the manufacturer increased the contrast, as black is now a better black as before)

Most of the power the LCD uses is the backlight, which should be exactly the same on both units.

So if you want to compare the power usage, the backlight has to be at the exact same setting.

To change the darker look of the new LCD, you need to change the Gamma settings (it's in LCD Settings).

Usually, the quick gamma setup should be enough for your test.

The gamma doesn't change the power the LCD needs, so with a proper gamma setting, it should look the same as the old one.
 
The only way to do an accurate test is to swap the LCDs, use a single Pandora board with both screens and see which uses more power.
 
The gamma doesn't change the power the LCD needs, so with a proper gamma setting, it should look the same as the old one.
My camera's sadly not suited for this kind of fine-grained comparison, but I think the results in comparing luminosity still show.

Both max brightness, new screen in CC unit @ 200 gamma, old screen in 1 GHz @ 100 gamma. I show the two units swapped in case the position or angle mattered.

New screen on the left

newscreen_left.JPG

New screen on the right

newscreen_right.JPG

You are correct that this increases the luminance, but now the blacks look terribly weak in comparison. I have to respectfully and nonexpertly disagree that the blacks are better, they seem to be worse even at default gamma. Using gamma to close the luminance gap is in effect cheating; the details of the image change. An easy example is the $ in the console: at default gamma on both screens the | through the S is only partially visible, and jacking up the gamma makes it fully visible (while lowering the gamma makes it disappear).  Unfortunately the photos can't demonstrate this because my camera sucks so badly. If there were contrast controls this could perhaps mitigate, but my feeling is that this would still be more of a band-aid fix.

Could we talk to the manufacturer to do further tweaking perhaps, maybe give them one of the older screens to see if they can match it?

The only way to do an accurate test is to swap the LCDs, use a single Pandora board with both screens and see which uses more power.
Good point, and out of scientific curiosity I decided it wasn't too much work to disassemble the base and put the 1 GHz board in the black casing. The current draw results are about the same, which substantiates ED's claim, but the brightness difference still remains.

DSCN2263.JPG

DSCN2264.JPG
 
Last edited by a moderator:
You are correct that this increases the luminance, but now the blacks look terribly weak in comparison.
Then the quick gamma isn't enough and you'd need to use the advanced gamma tools to make sure the blacks stay black.

Could we talk to the manufacturer to do further tweaking perhaps, maybe give them one of the older screens to see if they can match it?
Well... yes, if you plan on ordering about 100,000 LCDs, then the manufacturer would do that.

They don't throw on their machines just for a couple of LCDs... if you want a couple of hundred units, you need to buy them from stock.

It's like if you go into GameStop and ask whether they can create a 3DS in your own color if you buy 10 or so...
 
hum, hey ED, would you sell me a 3DS in clear case? if it please me I might even buy a second one :D
 
Sure. Feel free.

Fact is, the old screen is substantially brighter than the new one (at least the one I got). I think it's valid to ask for a possibility to get a brighter screen somehow. The OP even tried to find the root cause. If it's not possible, fine. I already found that GameStop comparison a bit childish. This is how you would explain it to a child.
 
Right, naive question, got it.

The advanced gamma correction is not the silver bullet, I tried it per your suggestion, ED. It's a nice interface once you figure out that you have to change the numbers across the bottom. Lowering the numbers on the left restores the blacks but drops the luminance. Turning everything up high still doesn't match the brightness and washes out the colors. If it is a silver bullet, then give me the settings to use because as Yoshi stated the older screen is brighter out of the box.

But since no one with the power to do anything sees the difference, since nothing can probably be done, and since most customers won't even be aware, nevermind.

hum, hey ED, would you sell me a 3DS in clear case? if it please me I might even buy a second one :D
Here is where you ask ED questions publically, please stay on topic.
 
Last edited by a moderator:
If you take a cheap camera with autofocus and auto-white level proceed to point it at a light source at random distance and angle, it will produce visually quite different results each time.

Is the current draw comparable for both new and old chipset?

I have a colorimeter and a light-meter, might be able to produce a better stock colour calibration for a better apples to apples comparison. However i only have one rebirth pandora at hand, gave my CC unit away to a dev.

Edit: I had an idea, but there isnt any lightbleed on my pandora screen at all. 3M uniformity tape helps with that in cases where the backlight isnt even. I suppose its some kind of reflective tape for led-edge-lit LCD screens. Could that be used to provide more effective backlight, by reflecting light sent into the case back onto the backlight? Provided the back and sides of it isnt covered of course.
 
Last edited by a moderator:
@comradekingu

Yep, I agree, the shots so far are pretty unreliable for a true test. A light meter would be really helpful. But the light difference is evident in person; I'm just trying to give people some idea of the difference with what little I can do.

I'm not sure what you mean by comparable? I only tried testing both screens with the new chipset, and per my 2nd post both screens draw the same current. If I understand ED correctly this is what we should expect. I cannot imagine the older chipset having different draws as a result, and I've lost the interest to try.

I was wondering if it's something maybe with less reflective backing inside the screen. I haven't noticed any significant bleeding out the sides on mine either, and I've had the unit turned on naked (no case).

I'd have sent you my old unit before but now my 1 GHz has the older screen, and I'm considering sending my old unit to a developer too so I'd be without a unit.

For the general record, I'm not buying another P1 so I don't have any agenda here. I'd have thought that telling the manufacturer "your screen is inefficient" and giving them empirical proof wouldn't be all that unreasonable, but apparently it is. I am pleased with the latest and greatest revision of the Open Pandora overall, this is one of the only "issues" I can really speak of.
 
Last edited by a moderator:
Thank you for your efforts, really interesting, just wondering what can be done about it. Its the same screen as a archos 605 AFAIK. With general testing i usually check all the usecases, and one could throw in a few different configuration of each type to make sure its reliable. All LCD screens aren't made equal unfortunately,  even the same panels on the same batch.
 
We have three different panels, btw (from three different batches)


The LCDs are also out of production for over a year now, so it's only possible to get them from remaining stocks, sadly.
 
Luckily, as a human being, my eyes have quite advanced automatic gamma correction tools built in ;)


My only minor gripe with the screen is response time - spotting dark grey bullets on a black background in Galaxian is quite tricky :(
 
^

I hear you there man.  Don't play with any background lighting or glares.  That's why I spoke up about Galaga, plus it switches things up.  But I'm drifting off topic...
 
Rather than brightness, I found the touch sensitivity more of an issue between batches.  My CC unit touch response was spot on. The 1GHz response is somewhat slower, sometimes I have to repeatedly touch to get a response.
 
Last edited by a moderator:
It might be related to a different touchscreenlayer, that the brightness is lower. The new touchscreenlayer feels thicker qnd a bit harder to press.


I only tried it at Gamescom, but the differnce was quite obvious.


Has qnyone compared the panels without the touchscreenlayer?
 
Back
Top