S-XPS 1645 AC Power Throttle Issue Investigation

The efficiency level for PA-3E(90w) is V .Here u could check what that means.

It(90w adapter) needs a efficiency of 92.6 % at 89 w load .

 

Mitchel your faith in engineers is a bit disturbing lol. Throttlegate is more of a policy they seem to have implemented probably dreamt, conceived and implemented by management (who are usually engineers themselves). Its a shady deal for lowering costs, forgoing a proper redesigns, like hp seems to have done with the envy, to crippling a system of its potential. Its your call to say what you will, but either they didnt have the foresight or were lazy and didnt expect this 1645 to throttle so blatantly(so yes good engineers i dunno about that). But still to continue on with their shadyness they let it through testing and hoped it wouldnt be noticeable. Which is why i dont think there will ever be a full resolution(as they seem to have an acceptable level of throttling in mind), and explains a lot of what we see from their actions. Which is why im returning the laptop.

I may have a method on measuring the amperage going into the laptop, without cutting things up. Ill see if i can safely do it tonight. The main thing is heat though, i would be interested how hot the 90w gets after an hour of being at 100watts at the wall. Gaming could be hours of high loads.

 

That seems to confirm what testers are seeing with bios A05. It still throttles, just not as bad.

If the PA-3E has a V rating, where did the 70% efficiency number come from earlier in this thread? Most small adapters are more efficient than a typical desktop power supply so the >90% efficiency number makes more sense.

The only thing that doesn't make sense now is that if it is 90% efficient, then you shouldn't need oven gloves to move it. An efficient power adapter shouldn't be putting out a lot of excess heat like some users are seeing. Maybe not all of these made in China specials are as good as the label claims.

 

if im using the A05 bios do i still need to use throttle stop. and if i use throttle stop should i go for the 130 or the 150w because i have the RGBLED screen

 

Thanks, though I wasn't assuming we'd ever see 1.5v at 220/240v, giving 330/360w. I was assuming that the adaptor would pull whatever amperage it needed to deliver 90w on the output side; and if that was 150w, then it would need to take 1.5A at 100V, but only 0.625A at 240V.

But like you, I'm open to suggestions from anyone with some proper knowledge; and I guess it's quite possible that the adaptor will behave differently on 240V in other than just the straight amperage reduction.

Get hotter over here than in the US, maybe? Or vice versa?

But what I can say is that I see/hear on 240v here exactly what you see/hear on 100v in the US; same sort of Kill-a-Watt behaviour, same stuttering audio on when using the adaptor (making Bioshock undecipherable) but fine on battery (for the 20-30 minutes it lasts).

 

When I used A05 it was hard to really tell the difference, I could feel it a little, I found that with A05 the BIOS tries ALOT harder to throttle down from 100% so even with TS, it still would hit lower than 100, and then TS would force it back to 100%, I never found this with A03.. So I switched back to A03.

 

Thanks, very useful reference.

But I read the PA-3E as a V adaptor with a Pno of > 49 to 250 watts, Standard, and so needing an efficiency of >= .870 (87.0%), at any load.

How did you get the 92.6%?

 

I don't know about the 70% - I think that was empirical - but I got 60% by observing that the PA-3E says 100-240v 1.5a and 19.5v 4.62a. Maybe that 1.5a is some sort of max current under real stress, though (no load or short?) and is not what it pulls when under regular load conditions.

In which case, if the 87% or better is to be believed, 90w in means 78.3w or better delivered to the 1645.

But like you, I'm sceptical. Less than 12w heat dissipation is nowhere near enough to fry the delicious eggs I've griddled on my PA-3E :smile:

 

Yeah, I'm a bit crazy And I do see your point, but my point in short (I'l try short ) is that any engineer that gets to design a notebook should (would?) try to make it the best he can, out of professional pride, or call it what you want. If he doesn't, it's probably due to external factors like heat from above. Or from the PSU

Interesting!

I just did two sessions of Dragon Age at 1920x1080, everything @ max.

One without TS, where I'd get ~25 FPS and ~89W at the wall. CPU & GPU temps were in the low 70s. The adapter gets warm, but could easily still touch.

The second session I locked ClockMod to 100%, and multi to 12x, everything else the same. My FPS climbs a bit, but not much, to ~30FPS. PSU draws ~98W at the wall, CPU & GPU temps roughly the same, maybe 1 or 2 degeees higher. The brick gets warmer, but not much. Can still touch without oven mits, but not for very long. So it's not blazing hot, more like borderline uncomfortable.

This I understand, but if it is rated for 1,5A @ 220V, then 110W (= 0.5A) is so far under its rating that it should easily be able to handle this, is my point.

I'd assume less heat @220v, because less A for the same W. But that could be wrong... Maybe it doesn't matter.

Not sure if you were talking to me (picture Robert de Niro ), but I'm also on 220v as I'm right across the little pond...

I have just experienced the total opposite. on A05 it tries hard to stay at 100% clock and as close to 12x multi as possible. Like I wrote above, Dragon Age (which might not be very demanding, I don't know) is now perfectly playable where I'd get anywhere between 5 and 15 FPS before, I now get an almost consistent 25FPS without TS. The difference in W with and without TS for this game is only about 10W at the wall, when locked to 100%/12x.

 

@Mitchell2.24v

I could be wrong, but I picture Watts like water flowing in a tunnel; height Volts, width Amps. So if you have more Volts, you need less Amps for a given Wattage. And vice versa. So you need 1.5A at 100V to get 150Watts, but only 0.625A at 240v. So it's actually in the UK/Europe that the adaptor runs at much less that its max(?) rated amperage, not in the US.

Re the heat, it's the Watts that go into the input, but never come out of the output, that appear as heat. And it's the same amount of Watts (Volts times Amps) in the US and the UK/Europe (assuming the same efficiency, if that's valid); though the Volts are up, the Amps are down to match.

And no, I didn't at first notice you were this side of the pond - but that 'you' was a collective 'you-all'. Travis Bickle isn't the only one here, this time....

 

Run the TS log while playing, it’s not enough to really notice a difference, but while watching TS I found it report 87% clock modulation even though its forced to 100% and then jump right back to 100%, never found it do this with A03

What I’m saying is that the A05 BIOS is much more aggressive with the clock modulation settings it wants to impose on the system.

 

BTW, with BIOS A05 on a 820 processor the system forces the multiplier to 9 with a clock modulation of around 75%... which means the 1.73Ghz processor I paid for is "designed by dell" to run at 897.75Mhz while under load... WOOOT!

/sarcasm

 

Lol, did maths. If u check the quote in my post it says
"110 W X 75% = 82.5 W DC"

so, i just calculated the value for 89 watts power output. Since 89 w in not the peak load , it won't either be peak voltage or peak current or both. Losses won't be maximum. That why i said it needs to be 92.6%. Its a class V after all.

Power factor of 0.9 at full load. Makes it 90% efficient at 90w , though there might be variables but that won't change the value by considerable amount.
Well another thing 110w might also be rated class V. So calculating values after losses seems a waste of time.

Dell's got pretty good standards for the adapters and we could say, they put a lot of faith in them, supplying a 90 watt with a machine requiring 110+

Edit:

by considerable amount i mean +,- 4%.

 

Dragon age i think is a bit special. It seems to load up 4 cores half way and not at 100% load. Leaving to a 13x multiplier and a half loaded cpu. Doesnt really draw that much power i dont think. When i play TF2 or l4d2 the power is up in the 100-110 range easily. Just thought i mention that.

It is true watts = volts * amps.. so up the voltage ya gotta lower the amps.

Wish we could actually get some hard numbers on these adapters. Although the competition all having big adapters kind of tells a lot right there.

 

Dragon age is really bad for throttle testing lol, sometimes I get throttling sometimes I don't (my settings never change)

 

If the adapter heats up the efficiency should go down, thats usually how these things go. Who knows if that has a measurable effect or not. But maybe if you cool the thing you'll get more output.

 

Hi,
I still have the A03 bios installed and a 150watt adapter that I have not connected yet.
Is it worth installing the A05 bios and using the 150watt for normal use? I hardly ever game but do use Virtualization software with multiple OS's running.

 

More like Amps being height and width, and volts being how much force is pushing it across.

One coulomb (64 trillion if I remember) of electrons moving across a point is volts, and one coulomb of electrons in width is amps.

This is just from memory, so don't quote me on any of this.

 

@SlyNine

you are correct.
lol, I also noticed this but didn't commented. It would be off the topic.
any how cross section(height) and length are related to the resistance.
volt(related to potential ) is the force causing the flow of water. and ampere (current) is the rate of flow of water in the pipe due to this P.D .
and Power would be rate of flow into potential(P=Volts.Ampere) . and also flow would need to be turbulent to mimic AC supply, but still it can't acquire the same velocity as electricity.

This is way off the topic.

 

Electricity confuses me , I still get confused by watt hours and measuring electricity as silly as that sounds. As soon as I feel a grasp on it, someone will ask me a question that will throw my thought process for a loop. Probably questions based on faulty premises but still.

Hijack ended.

On another note, I've posted some things on www.anandtech.com comment section with some interesting replies by JarredWalton http://www.anandtech.com/video/showdoc.aspx?i=3720

So I am doing a write up on this. No idea if it will be published at Anandtech. But I will give it a try. If anyone has any ideas or tips give me a shout.

 

I was playing with the 1647 at Best Buy yesterday and one interesting thing in noticed is that when I resumed the laptop from sleep the hard disk was vibrating like mother.. under the right hand palm rest. it was churning and making all noises then quieted down. I never had this incident with my, returned, 1645. But then again it could be a defective hard disk. just something interesting to note.

 

unable to delete this

 

Irrelevant post.

 

Coulomb: 6.242 x 10^18 electrons per second...also is an amp a second. We just did electrochem, lol.

~Ibrahim~

 

ThrottleStop 1.89
http://www.fileden.com/files/2008/3/3/1794507/ThrottleStop.zip

Just a couple of minor things. Windows was butchering the upper left and lower right edges of the system tray icon so I found a work around for that.

I also added a new feature where the tray icon will change colors when you have ThrottleStop disabled. Some users only use TS when gaming and otherwise have it disabled so now it will be easier to tell by looking at the system tray whether TS is enabled or not.

Thanks to John (kviri) for these suggestions.

 

good stuff thank you!

 

Try testing the throttling issue at a Bestbuy store with the 1647 and tell me how that goes

Im outside the US, if not I would sacrifice myself and do it!

 

Im on it.

 

Cool

Keep us posted!

 

Could not open winring0.dll ????

even my old versions say it now.

now this laptop really is garbage lol

 

I've been trying to read through these posts since i returned my system, but can anyone tell me if the magical BIOS update has done anything significant to boost performance? I have a pending offer from my dell rep to upgrade to a precision m6500 with the same CPU and RAM, but it's going to cost me an additional 700 USD. What do you guys think? I don't mind the machine being bigger as it will mainly be a DTR, and likewise, battery life is totally irrelevant for me.

 

Do you know that the Precision M6500 also has throttling problems? Make sure after you upgrade that you have the right to return this new laptop if you are not 100% satisfied. At least the M6500 comes with a decent power supply so if you are willing to use ThrottleStop, you shouldn't have too many issues.

The leaked A05 bios for the XPS 1645 has been a mixed bag. It fixed some things and made some things worse. The biggest fan of A05 on the forums decided to go back to A03 and wait for the next official bios.

Some other user was having that same problem and I think he found some GPU software on his computer that was interfering with the WinRing0 driver and preventing ThrottleStop from opening up. I wish I didn't have to depend on an open source driver but I can't afford a certified driver to keep Windows x64 happy. WinRing0 might also be blocked by an overly aggressive antivirus or anti-spyware type program.

 

M6500 has a much faster gpu. If it also comes with WUXGA RGBLed screen, I'd take it. But it seems a little over priced comparing to lenovo w510 and hp elitebook 8540w

 

Here is our latest update on Direct2Dell.

 

I dont know why they cant just have a BIOS that will do:

IF you have a 90W adapter :Throttle to work within power limits

If you have a 130W Adapter or Greater : Draw as much power as needed for optimal performance.

And then start shipping 130W adapters to those who request it. In reality a lot of 1645 owners don't even know of this problem. They just think the computer cant handle that game or the gfx card isnt good enough, etc etc.


Edit:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

@ BradatDell : Hopefully you guys will come to the CORRECT conclusion which is that 90W will never be enough to power this laptop for it to run @ 100%. Unless you want customers to spend 1000+ USD and have to "settle" for what they get.

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

 

I wouldn’t call me a fan of A05, my main disappointment was they raised the voltage of the processor with it, and the computer just runs better with A03 and TS than with A05 and TS due to more aggressive throttling of the A05 BIOS.

 

This solution has been introduced before and I think its the best so far as far as pleasing the customers and financially not too burdening to dell.

 

Honestly... Just tweaking the bios would be just a placebo. As another user (sorry, can't remember the name), the best solution is to give us a "double sided" bios that can work both with 90W throttling, and at full power with 130/150W.

By the way, I just spent 2300€ for this machine which is, otherways, really at top. Opposed to many users (that I really don't blame for their position, of course), I wouldn't mind to spend another 50€ for a new supply. I just want my machine to perform at its full power, nothing else.

 

@unclewebb

Thank you for the heads up, i had no idea that the precision was suffering from the same problems, but i'm hoping that the throttling isn't as severe as the 1645s, and yea, the 210w power supply might just be the most important upgrade if i go for the m6500

@yejun

I agree with you. Eventhough the HP and Lenovo both come equipped with similiar HW, i think the RGBLED tips the scale in favor of the precision for me too. But thanks for the tips!

 

One guy was using a slim 210 watt adapter on his Precision M6500 but when fully loaded, the adapter kept crapping out. He'd have to unplug the adapter to get it to reset itself and start working again. I think he's waiting on a replacement now.

Some of the business class customers that have spent $3 G on a laptop don't like talking about issues like this publicly. Hopefully a few more will contact me via PM or e-mail so I can share their results.

 

I'd like to restate that tests done with furmark should be ignored, FurMark is considered to be a power virus.

Taking that in to consideration and the efficiency loss 90Watts could end up being 130watts out of the wall. These 90Watt adapters MIGHT* not need to be replaced.

We need more tests done without the use of FurMark.

Also a helpful tip (Not a fix) is to turn the power play in the CCC to maximize battery life even when on A/C. This will help prevent throttling of the CPU and lead to a more consistent gaming experience (at much reduced settings of course). without doing that it didn't matter how far I turned down the graphics my CPU would throttle like mad leading to gaming at below 20FPS at lowest settings.

 

Would you be comfortable gaming on the 90w adapter, using more watts then it currently is, in the warmer temps of the summer?

 

Exactly, its just pure math.

 

Actually I found out, playing a hd video will force GPU run at 450MHz, which is the middle of GPU frequency. It may improve some games performance.

 

That's like asking if you'd be comfortable running the CPU max out using prime and using something considered to be a power virus like FurMark.

If its performing within its specification then yea, I'd be fine.

The math shows a double standard by many here.

 

Can we force it to run at those speeds without playing a video. That might be a good middle ground for now.

But I would not be happy with it long term or mid term *looking at you Dell.

 

@unclewebb

I actually considered taking the 210w "power brick" just to be safe, but i had not seen anyone mentioning this before, so ofcourse i had to choose the slim adaptor! Because i ended up accepting the offer, pffffffff. I guess i'll be doing some testing when i get this machine. Again, thanks for the info.

 

Finally someone that chose the 210W... Please tell us as soon as you can, I'm still considering it

 

Math is my forte, but when people shows pictures of the plastic melted on their adapters its a concern. Under normal operation, the adapter for me got quiet hot, barely on the edge to where i couldnt touch it. Others have posted saying couldnt hold onto it as it was so hot. to expect more wattage would increase that heat.

Perhaps experiences do vary. That was just my reasoning/quip at a joke.

 

@Lorthirk

It's for the precision m6500, my replacement for the SXPS 1645, just so you know. But i'll gladly post some test results once i get it. In the meantime I won't take up any more space with the precision, so others might be able to find info on the 1645 instead.