S-XPS 1645 AC Power Throttle Issue Investigation

So are there any Dell notebooks shipping with the new 5000/DirectX 11 AMD graphics cards? I guess what I'm wondering is if the 1645 is shipping with the 5770 yet. The X29 BIOS changelog mentions the Madison-XT, which is either the 5750 or 5770. The 1647 is apparently only getting the 5730.

 

The TDP is going to be that chip by itself without any other factors. since its crammed in with a bunch of other heat sources, the actual TDP could be higher based on the collective TDP.

I.E. with a hot GPU the TDP requirements (which again is for cooling purposes) could be higher on the CPU.

 

Right, that makes sense with the collective bit.

Cooling, right.

So there is absolutely NO scientific basis that the TDP is some mathematical function (or equivalent) of power requirements, correct? Besides the fact that hotter things usually take more power?

~Ibrahim`

 

AFAIK, TDP was termed for the sole purpose of saying "Hey, that cooler you're designing for this CPU needs to dissipate xxx watts of heat"

 

Sweet. Thanks, I appreciate the clarification. Oodles of reviews I've seen quote TDP as some power requirement, which I know now to be total BS.



Still no word from the Dell guys. Bit worried. :/

 

I was talking with my friend rge on XtremeSystems and I asked him to do some testing for me. He uses a Kill-a-Watt meter and has compared that to Everest and gets consistent power readings between the two.

I asked him to drop his Core i7-950 Desktop CPU down to mobile specs to see how many watts it would pull.

At 1600 MHz (12 x 133MHz) where the Core i7-720QM usually runs when fully loaded with Prime95, he was only pulling 32 watts. He was able to get the multiplier up to 18X for a total of 2400 MHz and he was still under 45 watts.

The Core i7 Desktop CPUs have a TDP of 130 watts so I didn't think 45 watts was going to get you very far but it looks like that number is very reasonable for a Core i7-720QM depending on what the core voltage is set to. His Desktop CPU was able to run 2400 MHz with only 1.02 volts.

 

And the plot thickens!

So Everest is to be trusted, then.

And, interestingly, it matches up nearly exactly to the TDP. Hmmm....coincidence?

 

might be just the cpu perhaps. Some reason though weve measured 50-60 watts being taken up when loading up the cpu through the kill-a-watt.

check the other post, it has siphens numbers in it. ive repeated a lot of them myself to verify it as well. perhaps some things come out of low power mode or something, but its still a big increase when loading the cpu

 

Right, like Zlog said, that would probably stem from P95 also stressing the chipset, memory, etc.

 

Which version of Everest is he running? I've downloaded a 5.3 Ultimate Trial, and I do not get all those numbers. They're probably also dependant on the type of motherboard. On my desktop (Intel Sharpsberg motherboard) I get more numbers than on my 1645, but still not everything. On the 1645 I only get temperatures and the Fan, which is always at 30%, no matter whast I do (in every program that displays the fan %).

On aside note, in the System Stability test, they also show CPU throttling, and that shows no change, even when the CPU does throttle

Only quick tests so far, I'll try and do some more comprehensive tests later this weekend.

 

The above screen shot was not of a socket 1156 CPU so we would need to see one of those for a more accurate comparison. I forgot about the changes from socket 1366 to socket 1156. Everest works on X58 but it might not support newer motherboards or the mobile chips.

 

Unless you are putting a meter on components the killawatt is going to always show you overall power draw. Any increased load on the CPU is also going to increase the load on the RAM, Chipset, GPU, Video Ram, and so on.... basicly, you cannot measure a single component's draw with a kill-a-watt.

 

My dekstop is also socket 1156, but probably the mainboard (with an P55 chipset) doesn't support al the readings, just like the 1645 doesn't support those readings.

Too bad.

 

Was not dell supposed to come ut with a new bios latest 16 jan ?

 

I have been watching these throttle threads for a while. When do Dell historically refresh their systems as in a new design?

 

The A05 bios that has been leaked has some improvements but when combined with a 90 watt adapter, doesn't solve the problem. They might have to call in Albert Einstein to get him to redo a couple of the laws of physics like how to get 110 watts from a 90 watt source.

Risco: This is a new design. They switched from the Core 2 chipset motherboards to the new P55 mobile chipset boards that support the new Core i7/i5/i3 line up. You might have to wait a while for the next major redesign.

 

Thanks, seems I am not invisible after all ( people keep ignoring my posts in other threads! )

Seeing as you programme this throttlestop, from what I can gather it only throttles when plugged in, but not on battery? Secondly the issue can be fixed by bios, however the laptop needs a 130W adapter. If Dell decided to start shipping with a 130W adapter, this would mean a major power supply recall and expense to Dell. Hence the reason why they will not fix it, correct?

 

Today BestBuy released the 1647 for sale and it comes with a 90W adapter. Very disappointing Dell.

 

yup Very disappointing!

 

They are still under the impression that it can run on 90W's of power?

Sigh, where is the voice of reason in this company?

 

If Dell reps are still saying that this laptop is designed at its optimal performance then what do expect from Dell? What kind of reasoning can we expect from them

 

My Motherboard is damaged [one of the RAM Sockets] so I had ot remove that RAM Stick for the laptop to function correctly. So I have to arrange for a technician to come and replace my motherboard....at the rate things are going this will be the last Dell laptop I purchase.

 

I'm not sure about A05 but I'm pretty sure these can still throttle on battery power. They don't throttle as much but they can still throttle.

Most users don't like to run high power consumption tests when on battery power to help protect their batteries. Hopefully someone can do some Crysis gaming or similar on battery power and then post a ThrottleStop log file to prove this.

An adapter recall would be a huge expense. Dell also plans months in advance so it's not as easy as it looks to switch to a more powerful power supply. If the new 1647 came with a 130 watt adapter, that could be seen as an admission that 90 watts is not adequate. They can't admit that without it costing them some serious $$$$ so they are forced to continue to use the 90 watt adapter. Sadly, most of their customers don't seem to know any better.

 

@unclewebb - that HP Envy has only a 6-cell battery. Does that mean only 2/3 the capacity of the 1645 battery, or are they different sized, or different technology, cells?

Re TDPs, Kill-a-watts, etc. My PA-3E is rated for 19.5v 4.62A output - so 90w, fine.

But we know the PA-3E isn't 100% efficient - and that it's sufficiently inefficient to run not just warm, but uncomfortably hot.

So - how inefficient exactly? Well, the input is rated at 100-240v, 1.5A. Now, assuming (big assume?) that it pulls 1.5A only at 100v, and less pro rata at the 240v which we have in the UK, that's still at least 150w I might be expecting to see the adaptor pull, in order to feed 90w to the 1645.

So, 90/150 giving 60% efficiency then. At best.

But I've only ever, momentarily, seen the PA-3E pull over 90w input at my Kill-a-watt-alike. And at 60% efficiency, that would mean the 1645 never sees much over 54w. And throttles to that number, not to 90w or anywhere near it.

But further, as we are not seeing Kill-a-watt readings over 130w (much) with the 130w adaptor and Throttlestop, that seems to me, based on the above, as still within the apparent operational parameters of the 90w adaptor.

Apart from the fact that currently, the adaptor gets uncomfortably hot just dissipating 36w as heat; if it was ever allowed to run at 150w in, 90w out, and dissipate 60w, it would get even hotter. But according to the label on the back of the PA-3E, it's rated for it.

Or is there a flaw in my logic? (And no, I'm not about to test for that by experiment!)

 

Hope you don't get a defective motherboard replacement....like I did!

~Ibrahim~

 

midnight there is no flaw in your logic. Infact someone did a tests while back and showed the 90w adapter as 70% ish effecient. But there is more to it, and rating is just a rating techinically.

The 90w is indeed rated for 90w output, but it doesnt say for how long it can safely sustain the 90w. And that length is well determined by heat and temps. It seems apparent to me if we pushed the output all the way to the rated 90w, and held it there, the results would not be good. And thats what our laptops looks to need a sustained high wattage.

the effeciency also varies to load, and isnt linear so that could leave room for margin of error. And the 130w adapter is going to have a different effeciency curve as well, so im not sure if you can safely correlate 130w adapter readings to the 90w adapter usage. That and the heat transfer that affects the two adapter is different, and you go back to what i noted earlier.

The 90w adapter seems to trip a current limiter at around 130watts, when we did testing with furmark and prime95, it would pull around 120-130 watts for a split second before being throttled.

Even if 90w was sufficient to run the laptop alone, the adapter sure isnt up to outputting that 90w. im pretty convinced as well, the output would have to be a bit more than 90w for those with the 820qm and rgbled screen, but no concrete evidence to support that though.

 

Indeed I'm still figuring out wether to head on a 210W or not, due to the efficiency value. What do you think?

 

Hey guys...

Where's the cheapest place to buy the 130 watt adapter? Are there any coupons? Thanks.

 

I ordered one off of ebay. It was only $25, which seems suspicious but the seller has 99.8% positive feedback and a 'top-rated seller' so I'll take my chances. http://cgi.ebay.com/JU012-NEW-Dell-ORIGINAL-PA-4E-130W-AC-Adapter_W0QQitemZ200420777968QQcmdZViewItemQQptZLaptop_Adapters_Chargers?hash=item2eaa025ff0

Ok I need to add that if you don't make a comment in your order asking for the part of the adapter cord that attaches to the wall outlet, it won't be sent. Apparently he usually doesn't send those out, but I asked him to send me one and he did.

 

i ordered mine off ebay too, $30 for a 150 watt and its huge, ugly, definitely not a dell one but i expected that since it cost less than $100

 

This is probably a ridiculous question, but I'll ask it anyway. I'm in the market for this kind of laptop (mostly for video editing) and I realize a lot of people have been burned by this problem, but I've been burned in the past by a junky HP and my current Vaio doesn't exactly smell like roses. So:

As current owners, and with the knowledge that you can use a bigger adapter combined with ThrottleStop to diminish the problem, would you still buy an XPS 1645?

 

Yes, I like the laptop and it performs well enough with the bigger adapter + ThrottleStop in any games I play. I personally wouldn't go for another brand/model.

 

My answer is no. At the drop of a hat, Dell could void your warranty for circumventing their "protection" that they've built into this laptop, or for using a 130watt adapter.

Theres no reason to spend money on this laptop when there are others from other manufacturers that have the same performance coming out or already out.

 

Hi

I recently got this system as a replacement for my xps m1730.

I've been told that all the 1645's have the problem. From 2 tech supports they said that RGBLED LCD effecting gameplay (Differences in brightness) will not be fixed by a bios update.

The manager said that the 130W does seem to make it run better but the system is built for the 90W. And that the manager said that they won't know how long your system will last if a 130W is used.

My specs of the 1645 are
i7 720QM, 6DDR3 1333Mhz, 1GB ati 4670, RGBLED LCD

I am getting it replaced with a 1640 with the same specs.
And the manager said that it does NOT have the same problem as the 1645. It runs better with a 90W.

 

I've got confirmation from my Dell guys that this won't be brought up on anybody. Impossible to prove (they admit it).

But, right, there are more trouble-free laptops available. HP and Sony seem to be the big replacements.

 

The 1640 also runs about 20 degrees hotter. Better order up a Zalman NC2000 cooler! Also demand a 9-cell battery, it will prop the laptop up a bit like the 1645 has and help with the temps.

 

Thanks. So I should treat the '90w' more as I treat the quoted top speed of my car, rather than how I treat its gas mileage.

To confirm this, I did a quick couple of very ad-hoc tests. Firstly, as most AC adaptors don't have the luxury of feeding a device whose power draw will vary much, I tried my Sony portable DVD player. 100-240v at 0.8A, supplying a nominal 9.5 V 2A - 19w.

I couldn't ever get it to pull more than 8W at the wall - full volume, full brightness, watching, winding, whatever. Say 5w at the device?

Then I tried my Toshiba Equium A100 - the one this 1645 is supposed to be replacing. 100-240v at 1.5A, output 15v 5A, for 75w.

Did a big video render, got it pulling 50w from the wall. Warm but not hot; and I guess the laptop was seeing 35w of that, max.

So, both devices with a lot more headroom in their respective usages than the PA-3E has when trying to drive the 1645 - bears out what you say.

NB: I have the RGBLED screen, though just the 720M, not the 820M. Got all the other toys, though, including a BluRay burner.

I've just written to Dell, reserving my right to return the machine if the new BIOS doesn't fix it. In the UK, the Sale of Goods Act covers this ('goods not as described' or 'goods not fit for purpose' - choose either), so I'm not bound by any steenking 7- or 21-day return limitation that Dell might like to impose.

 

I recently got this system as a replacement for my xps m1730

I have tried playing a couple of games like wolfenstein and OF ragon Rising, the games work for 3-5 mins then i get terrible stuttering sound wise and a very very long delayed reaction when shooting or trying to move??
This makes the games totally unplayable.
I spoke to Support and they have told me they will replace the graphic cards, i said isnt the graphic cards soldered on, then he said oh right i will replace the motherboard and that will fix this problem.
I mentioned that the forums have concluded that there is a throttling issue please could you send me a 130w power supply as that can alleviate the problem.
He laughed and said that the xps16 cannot and will never support the 130w power supply.
Normal internet browsing and watching a blu Ray here or there is fantastic on this machine, but to not really being able to play these games even on most basic settings and low screen rez is ridiculous.
Any advice.



I have the 820M and blu ray burner 4gb ram rgbled screen

Technician was supposed to come today and i just received a call that he will be here only tomorrow. Another day of work gone, bloody amateurs.

 

Your train of thought is quite interesting.

@Anyone with some proper knowledge of electronics and power:

Would my PA-3E adapter behave differently on 220v than on 100v? according to the label and Midnight_Voice's theory, mine should allow for 220v * 1.5A = 330W of power draw. That's a bit more then the 90W it used to pull.

And I say used, because I'm right now fiddling about with my 1645 a bit, and on A05 it pulls around 100W at the wall, without the help of ThrottleStop. That is at 7x and 100% ClockMod. It used to hover around 95W, with 7x and <100% ClockMod. So they have definately changed something in the BIOS. My GPU clocks are still @ 675Mhz / 795Mhz, so they didn't change those either.

Forcing the multi to 8x, it still only throttles down to 87.5%.

I might try and find some sort of limit for the PA-3E using ThrottleStop, if I'm brave enough. Running @ 100W for almost ten minutes now, I can still comfortably touch the power brick.

I planned on doing some tests with the watt-meter, but got pulled away by my work. I need to find some time, because I think this is worth investigating a little more...

 

I just configured the new studio 15 with the i7-820 and 1080p display and it also comes with a 90watt power adapter. I don't get why they are still shipping the 90watt with new systems? Will it still have a throttling issue?

The Studio 15 I configured come with the 512MB ATI Mobility Radeon HD 4570

Edit: Dell.ca by the way.

 

Multiplier X Clock Modulation = Actual Multiplier
7.0 X 100.0% = 7.0
8.0 X 87.5% = 7.0

It looks like the CPU performance of your computer is exactly the same whether you are using the 7 or the 8 multiplier.

A 7X multiplier at full load is the equivalent of your computer running at 58% (7/12) of the performance that Intel built into the Core i7-720. I know that's an improvement compared to previous bios versions but it's still nothing a user should be satisfied with. Maybe phone Dell and ask them for 42% of your money back in lieu of the reduced performance.

kienah: Welcome to the club. The fact of the matter is that a 90 watt power adapter can not adequately power a Core i7-720QM let alone a Core i7-820QM. It looks like the new A05 bios is helping with less severe throttling schemes but it's still throttling. Dell has left users to fend for themselves. If you buy a 130 watt adapter on your own, your laptop enjoyment factor will likely increase 100%. You may or may not require ThrottleStop depending on the games / applications you use but it's available if you do need it. It does not look like Dell is going to approve a 130 watt adapter for this application so I don't recommend waiting for them to solve this problem. The technician coming out to play musical motherboards will not solve anything if head office prevents him from replacing the defective part.

What GPU does it use? If it has an ATI 4670 or something that draws a similar amount of power then it will throttle with a 90 watt adapter.

 

The Studio 15 I configured come with the 512MB ATI Mobility Radeon HD 4570

 

That's right, I didn't realise this, but on A03 it would often throttle (way) below 87.5% in similar tests.

Again, that's true. The thing I'm interested in, is the fact that the A05 BIOS allows my adapter to pull more W from the wall. In light of the whole efficiency and in-W / out-W issue, I'm wondering what my adapter is truly capable of. If it really only delivers between 50 and 60W to the laptop, there is still some hope. Even at 110W my adapter still doesn't get really hot, and it is the regular PA-3E.

I'm not quite clear on the whole efficiency thing. Does being roughly 70% efficient means it only delivers 70% of the rated 90W, or that it uses 30% extra power to deliver the rated 90W?

That would be something.

 

the efficiency would be power output/ power input = 70% ish, meaneing the other 30% ish is going to heat. The efficiency curves are also non-linear with respect to load. Technically as long as the power brick doesnt overheat, you could pretty safely pull well over 100 watts from from the output(as long as current limiting etc doesnt kick in). If you ever see the difference between a 300w and 400w computer power supply, a lot of time the latter just has bigger heat sinks.

It stays at a consistent 100w? how long did you test it at 100w draw before seeing how warm it was. When the battery was charging or playing games for a long time on the 90w it would be blazing hot, It took a long while to warm up though.

I dont know to much about 240 vs 120. I heard the 240 etc is more efficient and would mean less waste heat is generated. It is a bit different though.

I told midnight earlier a rating is just a rating, doesnt always mean how long it can sustain a high output like 90w(that is most likely limited by heat). A lot of little power systems i use for my microcontrollers have max ratings with a temperature, and or duration it can sustain next to it.

 

Mitchell2.24v: I think you're discussion about watts is uncovering part of the problem here. An adapter rated at 90 watts is based on what it can output to your computer. Dell adapters output 19.5 volts so it should be able to deliver approximately 4.6 amps (4.6A X 19.5V = 90 W)

An adapter that is 75% efficient needs to pull 120 watts from the wall to convert that into 90 watts.

The original throttling scheme seems to be based on only drawing 90 watts from the wall which looks to be too conservative. You might have one of the better PA-3E adapters. Some users report that their adapters get extremely hot which is a sign of an inefficient adapter. Power coming from the wall is being converted into heat energy instead of being converted into DC power for your laptop.

Edit: Here's some good background info.
http://en.wikipedia.org/wiki/Power_supply_unit_(computer)

 

@atlstang:

Thanks for the explanation. I did read your comment on the rating, and that got me thinking. If it currently only draws ~60W, it is no where near its rating, so it should be able to draw some more.

edit: I had it at 100w (at the wall, so ~70W at the 1645, assuming ~70% efficiency) for at least 10 minutes, probably more than 15 minutes. One thing that might matter, is that I had the brick on its side, so it has a maximum area exposed to air for cooling.

@unclewebb:

I was thinking about quality control. That might be an issue with the adapters. I have had a few of the previous model 90W (the bigger ones). Two of those died within weeks after getting them. I used them on an E1705/I9400 (depends on where you live what they were called). I didn't really stress the machine, but the bricks died really quick. I bought two spare ones as I like having an adapter in every room I work in, and those are still going strong after years of use.

It might be the same with the PA-3E's. Maybe some are good, and others are less good. This could also be what's taking Dell so long. They might be researching the limits of the 90W adapter. Could it be possible that they never had to find this limit before?

Tonight I don't feel like working or watching Prime/FurMark, so I will play some Dragon Age and keep an eye on the Watt-Meter and on Throttlestop. If I'm feeling lucky I might even force the multi or clockmod a bit.

Will let you guys know how that goes.

 

you would think mitchell. the 70% ish the one person found was only at part load etc. Its kind of an odd result as well, as the energy markings on there say a 80%+ average efficiency rating. I would feel more comfortable if there were more tests to confirm it and at full load. The hard part is getting a dmm between the adapter and laptop to measure amps and test this theory. I may have a way but not sure.

If the engineers are anything like their reps, it could be they were just idiots and didnt realize things like this. The heat thing worries me though.

 

My only guess would be to sacrifice an adapter, and splice the cord. But my knowledge of electricity is not good enough to try this. I have anough adapters laying around (mostly the older ones though), and I do have a (cheap) dmm.

Somehow I don't think the engineers are that stupid. I'm inclined to believe that a real engineer always wants to implement the best solution. My guess would be that the engineers are limited by either budget or marketing. This has happened to me as a software developer often enough.

Rules from above can be quite frustrating, and the folks at the top will not always listen to reason (more often they listen to the short-term balance between income and cost).

So until proven otherwise I fully trust the engineers, but not management. The engineers are probably bending over backwards to keep the boss happy, while trying to solve this issue within budgetary (or other) constraints.

 

Here's a company that does testing for the 80 Plus power supply efficiency program.

http://www.efficientpowersupplies.org/methods.asp

And here's a document about external power adapters like a laptop uses:

Test Method for Calculating the Energy Efficiency of Single-Voltage External Ac-Dc and Ac-Ac Power Supplies Revised August 11, 2004

Here's a nice quote about their testing methods:

Note: UUT is an acronym for “unit under test,” which in this case refers to the power supply sample being tested.

What that says to me is that if a power adapter is certified by UL and CSA and has a label on it that says it can output 90 watts, then it should be able to do so without it shutting down or melting the side of the case. If it can't deliver the goods then those two testing organizations need to be contacted and informed of this problem.

atlstang's testing showed that his laptop needs to pull about 110 watts from the wall for a smooth gaming experience with no throttling. If these adapters are 75% efficient then a person should be able to do that and still be under the safe 90 watt DC output rating of the adapter.

110 W X 75% = 82.5 W DC

Now we just need to know how efficient these adapters are at full load. Peak efficiency is usually just before full load on many adapters.

 

Another update: captured system was hand-delivered to Taipei (where I assume Dell's real engineers are?). BIOS engineers: BIOS will still throttle, but just at a different point. Don't ask me what that means. They know the new BIOSes don't fix the issue, still working.

~Ibrahim~