Upgrading the soldered-on RAM of a Dell XPS13 7390 laptop | Hacker News Reader

Upgrading the soldered-on RAM of a Dell XPS13 7390 laptop

gregdavill.github.io

479 points_Microft209 comments

Abbreviations and some comments for those who are not quite familiar with electronics:

BGA - Ball Grid Array[0], a way to mount integrated circuits (IC) to a printed circuit board (PCB)[1].

TSSOP - Thin-shrink small outline package[2], a form factor of ICs

8L design - this means the PCB has eight layers, with some being ground planes (that usually are just complete layers of copper and thereby also spread heat which complicates heating up a component for (de)soldering).

SMD - surface-mount device[3], the small form factor for electronics parts that go directly onto a PCB, instead of e.g. bulky resistors with wires coming out of them that are then threaded through holes before being soldered on.

[0] https://en.m.wikipedia.org/wiki/Ball_grid_array

[1] https://en.m.wikipedia.org/wiki/Printed_circuit_board

[2] https://en.m.wikipedia.org/wiki/Small_outline_integrated_cir...

[3] https://en.m.wikipedia.org/wiki/Surface-mount_technology

IPA - India Pale Ale, best thing to clean a PCB with, especially after you've had a meal break right on top of it :D

I actually spilled a full glass of IPA in a mechanical keyboard once. Ironically, used IPA 99% to clean it with the "straw on switch stem" technique.

IPA - Isopropyl alcohol, 99%

I’m from Michigan, 100% thought this was a beer.

I've been to Shenzhen and seen (been taught) how they reball BGAs for phones in the markets, none of that futzing about with solder balls - they just clean the pads on the chips, re-tin them, lots of flux, then place the steel pad stencil over the top of the pads, spread solder paste over and through the holes, wipe off any excess then hit it with some hot air until you see the balls form, if you do it right the stencil will just pop off (don't push it)

Any video of this particular trick?

Here's an example:

https://www.youtube.com/watch?v=alQq74uWJJA

Thank you, very elegant and nice no nonsense video as well. Great tip!

For smaller numbers of connections, you don't even need a stencil if you how to use surface tension and have a fine tipped soldering iron.

Letting people know you could pull this off eventually leads to the inevitable assignment in which you need to solder a dozen of NAND chips in some obscure package that nobody seems to have a stencil for :) Oh and the deadline is 24 hours away.

Exactly why I'm only saying that anonymously here on HN :)

FYI, I also do some SMD by hand. It's not very hard with leaded paste. It's a very practical skill to have when you only use it for yourself and a few select friends (read: not colleagues!)

Good thing we can reach you through GitHub ;-)

Half joking, if you have forgotten that you added that link here's your chance to remove it B-)

And... it's gone now! (Thanks for reminding me BTW)

I prefer keeping a low profile. Because I like speaking my mind freely, which I might not do if my account was linked to my name.

You can never be too careful, you can get cancelled for thoughts that go against groupthink.

If you dealing with these kinds of jobs you either get a generic stencil and use a kapton tape as a mask or you invest in a laser cutter capable of cutting stencils in a half/quarter of a mil stainless sheet.

Also RoHS goes out of the window and you use a leaded tin which makes your life so much easier, as well as usually guarantees a longer life time for the fix because it’s far more ductile after soldering.

Unleaded solder is awful, sadly.

If you see this late reply - what if you use a fume extractor turned up to "headphones with music required"-dB, connected to an optimal exhaust configuration?

Sounds profitable, whats the problem?

https://www.youtube.com/watch?v=I4m0LoGx7Qo

you can also go one ball at a time https://www.youtube.com/watch?v=LrXL1uGs0vU

Yup, the trick though is you want some way to get all the pads balled to roughly the same height - otherwise when you heat it to solder it and the chip floats on the pads one or two may be too high to make contact ... or there's too much solder and it squeezes out and causes a hidden short.

The process of filling each hole in the stencil with the same amount of solder and carefully scraping any excess off of the top effectively adds the same amount to each ball

> Yup, the trick though is you want some way to get all the pads balled to roughly the same height -

I see you are familiar with the method :)

Here's another "ghetto BGA reball" trick if that happens: lower the temperature of your iron to make the "pseudo balls" that are too big thinner by drawing some paste away into a vertical line extending up : let it cool down and just trim it to length with nail clippers.

You can get balls again by rewarming a bit - ideally with a special IR heater to avoid pushing the balls away by the flow of air coming from a hot hairgun, but in a pinch a regular oven can do the job (prebake to temp of course)

If you see this late reply - what sorts of IR heaters would I be looking for?

that reminds me of when I tried to soldier an extra chip on my PS1 and instead pulled up an original one oops

Thanks so much for this, I always assumed I was bad at stenciling but I see I was just doing it wrong. I was spreading the paste then pulling it up, not heating it first as I assumed the sldier would melt onto the steel stencil. Thanks!

It may if you get the stencil hot enough, you have to watch the solder

Funny I just saw this thread, today I learned that my integrated (not removeable) XPS 15 9570 laptop battery just died just a few months after my warranty expired on my Dell XPS laptop... The Internet is littered with complaints about dell PCs failing after warranty, and how recalls are covered up on them frequently.

And my XPS desktop had faulty connections on it's mother board (That were known by Dell even before I bought it) for the Hard Drive controller that I discovered just after the warranty replacement expired on it... 2 problems that lead to costly repair and buying expensive replacement parts.... I have a 10 year old Dell XPS 16 that still runs (granted only windows7), but I never had any problems with it.... XPS quality has declined greatly since my first one.

The standard 1 year warranty isn't enough for the way Dell computers have become shoddy, I'm deeply disappointed in Dell, and will not be buying any more of them... I don't have time to build my own, does anyone know a rock solid/powerful brand I can buy instead of Dell? I kind of miss Micron PCs.

Dell quality and practices have been questionable for some time.

Back in they day, they went out of their way to use non ATX power supplies.

My first xps desktop was utterly incapable of having all of its drive bays used without said drives cooking to premature failure.

My current work XPS laptop will downclock to 800mhz when chromium has hardware acceleration enabled. Other colleagues have had multiple laptops replaced because theirs downclocked when plugged into AC.

Dell is no better in quality than most of the MFGs who died in the 90s and 00s.

(As a random aside, I loved how HP would use a nice ASUS mobo and then trash everything else back in the day. We used to joke those little cases were what an Antec 1080 pooped out)

My current work XPS laptop will downclock to 800mhz when chromium has hardware acceleration enabled.

That is at least partly Intel's fault for creating CPUs with "configurable TDP-down" and the whole power limit thing. It happens to Lenovos too. There are utilities that can stop that from happening and let you use the full potential of the hardware. There's some discussion about that here: https://news.ycombinator.com/item?id=18425687

When did they stop using nonstandard PSUs?

Also, their motherboards are not a standard size.

At least they aren't pulling the crap HP does on the enterprise side - requiring service contracts to access firmware updates. This means there's financial incentive to include bugs in component firmware now, to "fix" over time.

Oh yeah, and they also DRM'd drive caddies by integrating a bunch of useless capacity indicator LEDs on the front of the drive tray, because they want people to buy their stupidly overpriced drives instead of buying caddies at $5/pop off ebay and putting bulk-purchased drives in.

I've owned several XPS laptops.

For less than two days each.

All returned. The quality control is a joke and I more wouldn't touch Dell with a bargepole.

One XPS 15 powered off every 5 minutes. One had a high pitched whine. In one I had the keys were loose and not fitted properly. In another the hinge was screwed and another became inexplicably hot at idle to the point it was painful to touch.

When I asked to return these they didn't even care why I was returning them.

I had one with the whine and a trackpad issue together! I returned that crap so fast. I don't know how that whine was left out of all the reviews I had seen. All you had to do to hear it was cause the laptop to turn on its fans.

HA! I forgot i also had a bulged trackpad in one.

How dell makes money out of this is beyond me. Every post to the dell redit is about returning an xps

you could give the framework laptop a shot, it's designed to be easily repairable. otherwise maybe a thinkpad?

I was just about to recommend a framework. I've been running Ubuntu on one from batch 1 and I couldn't be happier. Funny enough, I also came from an XPS, and switched after realizing that dell intends for their machines to be disposable. Now I can upgrade and replace anything and everything and boy do I ever love it.

Only if commenter is from USA.

If he's from EU, he might try Pcspecialist as mentioned in another comment.

Or Canada, actually.

The 9560/5520 battery was perfectly removable. I didn’t know they did away with that the next year.

I have a 9570 - it's still a removable battery, just no longer external to the system.

another crappy experience with a Dell xps15 9575... the feeling of a laptop built to quickly auto desintegrate, with crappy components, bad BT/wifi module, very unreliable touchpad, butter screws... everything sucked ass. I remember also not being able to use the laptop 90 degrees on its side to see a movie for example because the keyboard would stop working and the fan would start to run at max speed. that was the worst laptop I've ever had and the most expensive too. never again dell for me.

after that I bought a custom 14 inch laptop in pcspecialist back when they still had amd ryzen models (Ryzen 7 4800H 8 cores, 64 Gb ram, no preinstalled OS...) which is the best laptop I've ever had in my entire life. It's been complete opposite experience in terms of enjoyment and reliability, with very few issues using Linux (my only complain the inability to put to sleep, because it messes up the video after waking up even with the latest kernel).

Thanks for mentioning Pcspecialist. They seem to be making nice laptops which are aren't more costly than brand names such as HP or Dell or Lenovo.

Fellow 9570 user here; I upgraded my RAM, but my biggest complaint with it by far is around power management. Even running Windows 10 and nothing "weird" as far as VMs or whatever, it feels like I get completely unpredictable battery life— sometimes it's 6 hours no problem, other times it's at 50% after just an hour. Most troubling still is that I've had multiple instances of bringing the computer out of sleep after it was sitting on the charger, and then a moment later it goes into a battery panic and shuts itself down. But then after rebooting (on external power), it shows the battery as full again and doesn't charge any further.

I basically can't depend on the computer to be usable on battery unless I fully shut it down when I know it's full.

I've read/heard enough Internet anecdata to be >50% confident you're experiencing issues associated with "modern standby", most current implementations of which are horribly broken (both, IIUC, in terms of bad/nonexistent Linux support, and broken Windows experiences).

A quick google found https://www.notebookcheck.net/Dell-backtracks-on-XPS-15-9570..., which notes that the BIOS removed S3 sleep support (the standby method used since the dinosaurs) in version 1.3.0. The suggestion might be actively harmful, impossible, void your warranty, irrelevant for reason(s) I'm not aware of, or any combination thereof, but I wonder what would happen if you downgraded the BIOS to that version. (Once again, this suggestion is the equivalent of blindly reaching into a Pandora's box full of piranhas, pulling the first thing out I touch, and going "hmm, would this work?" :D)

In any case, googling around for "modern standby" and "s3" (ie, "9570 s3") will reveal the debacle behind the curtain ._.

Edit: Found https://www.dell.com/community/XPS/XPS-15-9570-BIOS-1-3-0-sl... which adds some interesting insight ("The issue is the following"), along with https://www.dell.com/community/XPS/XPS-15-9570-BIOS-1-3-0-sl... (BIOS downgrade blocked starting at 1.4.0) :(

I could not get my Latitude 5515 to sleep properly after weeks of playing with it. I settled on a song and dance of setting hibernate on and making sure I pull the power cord out and wait a few seconds before closing the laptop. Works every time and 'bootup' takes a few seconds.

About a year ago, I wanted to work around BIOS limitations by writing patched ACPI tables, like what was done for a long time for Hackintoshes.

So I documented in more detail my finding for the 9250/7275 on https://www.reddit.com/r/Dell/comments/kolf5f/patching_the_d...

I got in touch with a few other people, but nobody seemed very interested, so I gave up and got myself a Thinkpad instead.

If you want to work on that, on a BIOS advertising S3 support, IIRC all you need to do is keep the EC powered so it can proceed the GPE from the power power button. On S01x it's powered off to avoid spurious wakeups.

Get the schematics, disassemble the BIOS, and you'll the logic would be about as complicated as:

if (S01x) { EC_poweroff; }

else { EC_powered; EC_monitor_GPE_something; }

Interesting! Thanks for the pointer on that. This is a work-issued computer, so I'm not sure I officially even have BIOS/boot access on it, however the Dell Update utility is telling me that I have a 1.21 BIOS update waiting in the wings, so perhaps I'll install that and see what happens. It would be hard to picture it getting any worse, though I suppose the people updating to 1.3 thought that too.

Huh.

Before updating...

I was under the impression 1.3.0 was The Update Of Doom, not prior versions. Your BIOS might still have the S3 option available, and furthermore, based on the anecdata I found about downgrade being blocked starting at 1.4.0, you might be able to go to even lower BIOS versions too if you need to.

Even if the option does not appear to be available, it might be an interesting idea to ask around wherever this is being discussed, before updating, in case there's any interesting insight to be had there.

In any case I'd definitely have a thorough look around the current BIOS version first.

If onsite IT has their heads screwed on correctly :) it shouldn't be too difficult to demonstrate the benefits of a) checking the BIOS (if you don't have read-only access) and b) enabling the option if it's available (if you don't have read-write access).

I've had to change the battery of every laptop I ever owned after a year or two after warranty expires. Why would anyone buy a laptop with a non removable battery? It's crazy.

Because the thin ones are all like that. They typically use prismatic cells glued in place.

Same goes for soldered SSDs, both components degrade with use.

It’s like buying a car with tires that last 50k miles, but if you want to replace them you have to cut off the entire drivetrain and weld on a new one. Technically possible but only with excessive effort.

>It’s like buying a car with tires that last 50k miles, but if you want to replace them you have to cut off the entire drivetrain and weld on a new one.

I think that car manufacturers can borrow a lot of bad ideas from IT: not being able to own a car, instead just renting it from the company, artificially gimping it and requiring a sum of money to "unlock the features".

>requiring a sum of money to "unlock the features".

Tesla already did this

I feel like the more egregious example is BMW's subscription model for heated seats. I think once you buy the self-driving package, you get to keep it, along with future updates. Sure, it's pricey but it's also an incredible feature. The idea of being able to have your car do the work on interstates sounds amazing, not that I can afford it. If it required yearly payments and cost extra to upgrade, that would be pretty bad, in my opinion.

For Tesla FSD you can pay a lump sum to unlock it permanently, or you can pay a monthly fee. When I did the math I recall it would take ~4.5 years to equal the lump sum, although they say to expect price increases on the monthly plan.

Tesla also has a monthly fee to unlock rear heated seats in the Model 3.

>Tesla also has a monthly fee to unlock rear heated seats in the Model 3.

I didn't realize. This kind of thing is really lame. In my case it wouldn't matter at all but for some reason it really rubs me the wrong way.

> Why would anyone buy a laptop with a non removable battery? It's crazy.

For me personally, it's because I never use my laptop without it being plugged in. There's nowhere I would want to use my laptop that I can't find an output (home, work, planes, airports, friends' houses, etc.)

The only exception I can think of is when actively riding public transit, but I never use my laptop in those situations.

A lot of laptops are also bought by people's employers and are a tiny cost that no one cares about (~$1,000 every 3 years for an employee making $300k+ in that amount of time). They don't care if the battery degrades.

As long as the battery lasts long enough to last an hour or two in a meeting room, its good enough. But now I work from home permanently so that doesn't even matter. I'm glad MacOS recognizes this and only charges up to 80% to preserve the battery health if I ever did need it

The only time I ever cared about battery life on a laptop was in highschool where it had to last all day without being charged.

When the battery fails (or decides it has failed) you probably won't be able to use the machine anymore unless you can disconnect the battery.

I bought a $1000 plus gaming laptop for my son and exactly a year and a month later the hinges bust off the thing. I am so ticked off and will never buy another one. Too cheaply made and the fact that it happened a month after warranty was the kicker. I keep hearing my dad talking about companies designing parts to last just long enough. They don't care if some people claim warranty they have done the numbers and if it can last a year they don't care. Sad.

> The Internet is littered with complaints about dell PCs failing after warranty, and how recalls are covered up on them frequently.

Technology has advanced to the point where they can get some components to fail soon after warranty period expired with decent accuracy. Power unit failure tends to be a common one.

In terms of PC, your best bet is probably Framework laptop.

> Technology has advanced to the point where they can get some components to fail soon after warranty period expired with decent accuracy. Power unit failure tends to be a common one.

Unless the hardware is designed to be damaged (or restricted) by software, there's no way this is true. Technology can be great, but you can't "program" hardware to die on a certain date without knowing how it will be used, how long it will be used, what temperatures it will operate in, etc.

And you're not even considering that the manufacturer has no idea when the hardware will be sold and therefore doesn't even know the date that the hardware "should" fail.

Unless you have some evidence that this is both possible and happening, it's just a conspiracy theory.

It's not really a conspiracy, just the natural outcome of design optimization - any part that still works reliably at the end of the design lifespan is a candidate for cost cutting in the next iteration.

> any part that still works reliably at the end of the design lifespan is a candidate for cost cutting in the next iteration

This assumes a positive correlation between quality and price for parts. That isn't always the case. There's no reason a better part will be more expensive.

And again, you're ignoring my primary argument: "lifespan" is not measurable in calendar years for electronics. Some people I know use their laptops 360 hrs/mo, and some people (like me) only use them when traveling, ~20 hrs/mo.

After a calendar year of that, my laptop has only 240 hrs of life lost, while another laptop might have 4,000+ hrs of life lost.

Please provide a source about how a hardware manufacturer (without the aid of software) could plan for a component to fail after either 240 hrs of usage or 4,000 hrs of usage.

It could be from the standpoint of thermal design and having a reasonable idea of how long components will last at the temperatures they will typically run at. Realistically, all you need is a capacitor or two to fail in the ballpark to achieve that end with most customers. Granted, one wouldn't be able to to say 'after exactly X years of expected usage it will fail' but more likely be able to say 'after no more than X years of expected usage most units should fail'. Some heavy users would have theirs fail early and the company might have to correct via a warranty claim but that could easily be more than made up for by the masses of other customers that arrive for replacements roughly on schedule.

One doesn't even need to be that devious: just make 'consumables' like batteries and flash drives not easily replaceable nor easily sourced. Then just sit back and wait for them to start failing. As long as most manufacturers did this, you wouldn't have to worry about losing business to your competitors due to the practice. Pretty sure you can put a check mark next to most phone and laptop manufacturers on that front these days.

This isn't unique to tech. A couple of real world examples of planned obsolescence via designed-in failure modes: incandescent lightbulb filaments were designed to fail ~1000 hours of use (even though they could easily have lasted far longer) and most clothes washers are designed to fail after ~10 years (IIRC, manufacturers design a seal in the tub to slowly disintegrate via your laundry detergent.)

> One doesn't even need to be that devious: just make 'consumables' like batteries and flash drives not easily replaceable nor easily sourced. Then just sit back and wait for them to start failing. As long as most manufacturers did this, you wouldn't have to worry about losing business to your competitors due to the practice.

Based on the lack of profits from selling hardware in the first place, it makes more sense to me that these lower quality products are made with tradeoffs that result in a lower price for their target audience. The grand conspiracy idea, even if it were true, clearly did not pan out considering the loss in market value for basically every laptop brand other than Apple.

> This isn't unique to tech. A couple of real world examples of planned obsolescence via designed-in failure modes: incandescent lightbulb filaments were designed to fail ~1000 hours of use (even though they could easily have lasted far longer) and most clothes washers are designed to fail after ~10 years (IIRC, manufacturers design a seal in the tub to slowly disintegrate via your laundry detergent.)

Source? Unless there was only 1 or 2 washing machine manufacturers, it seems kind of trivial for a competing company to sell the tub with a better seal with a longer warranty. Once again, it is probably lower quality products aimed for people who want to pay lower prices. The speed queen or Miele washer with 5 year warranty is available for those who want to spend for it.

Personally, I would rather buy a washing machine that costs half as much from Costco which comes with a 4 year warranty, and trash it when it breaks. If another manufacturer wanted to sell a washer with a 5 year warranty that cost the same, or a 10 year warranty that cost less than double, I would be game for that.

The same situation for the light bulb. If there are competing manufacturers, people would rather pay more to not have to change light bulbs. Although, this depends on how easy it is to change the bulb. For a parking lot light needing a truck with a boom, I would spend double to get a quality bulb with a better warranty. For a house bulb I can change with a step ladder, I would rather pay $2 and gamble on it lasting longer than a $4 bulb, because even if it does fail after 1 year instead of 2 years, I am not going to spend my time warrantying a $4 item.

Light bulbs lifetime was open collusion because longer-lifetime lightbulbs are less efficient electrically. Lifetime electricity cost for incandescents is much more than purchase cost, so buying longer-lived bulbs costs you money.

I do not see how that is collusion in the usual sense of the world?

That is simply consumers opting for the better value, making the sale of longer lifetime bulbs an uncompetitive option.

It was the manufacturers that agreed not to sell non-standard 1000-hr bulbs, not consumers who stopped buying.

> clothes washers are designed to fail after ~10 years (IIRC, manufacturers design a seal in the tub to slowly disintegrate via your laundry detergent.)

Worse :( Seal can be easily replaced. They cast Drum Support/Spider part from specially selected alloy with curious property of easily Dissolving in washing powder :)

https://www.google.com/search?tbm=isch&q=corroded+spider+was... https://www.automaticwasher.org/cgi-bin/TD/TD-VIEWTHREAD.cgi...

Drum spider corrodes and falls apart, incidentally it is often the Only internal part not covered with anti corrosion paint :), and drum itself is obviously stainless. For comparison something like Miele is all stainless.

Audi (VW group) 2.0 TDI BLB Engine is made to fail at ~200K no matter what you do, from oil pump to head micro fractures. Crankshaft sprocket at 200K http://lh6.ggpht.com/_m_33vQTtsxM/SyuSlFIeSXI/AAAAAAAAG9o/rt... http://lh3.ggpht.com/_m_33vQTtsxM/SyuSn5YALLI/AAAAAAAAG9s/ul... http://lh3.ggpht.com/_m_33vQTtsxM/SyuSp1yUs4I/AAAAAAAAG9w/mO... VW answer: this part is non serviceable, whole crankshaft needs to be replaced :D Why is it made out of putty while smaller sprocket (the one taking more force) is still 'working'? Why is it designed not to be replaceable? Good 'old timer/not by the book' mechanic is still able to replace it (heat whole crankshaft, use big hammer). 200K is usually first crankshaft turn/grind, engine block itself is good for >500K easily.

> incandescent lightbulb filaments were designed to fail ~1000 hours of use (even though they could easily have lasted far longer)

This is repeated a lot but not correct. All the "forever incandescent" bulbs have one thing in common: they're dim, and they have a very red color temperature. They live long, because their filaments are much colder. If you want whiter light, and more of it (because lower filament temperatures are less efficient, since they emit even more infrared, as the entire spectrum is more red-shifted and flatter), the only way with incandescent is a higher filament temperature, which inherently lowers lamp lifetime.

What changed, besides lifetime, when lamps went from around 2000-3000 hours to 1000 hours is that they got way brighter and more white. Which is what customers wanted.

although it certainly is true these days, that the average LED based bulb is pushing those LEDs hard, which incidentally also makes them less efficient, rather than just using more of them at the same wattage.

Check out the dubai lamp to see a lamp designed to be efficient and last:

https://www.youtube.com/watch?v=klaJqofCsu4

The Dubai Lamp is not that special. There are similar lamps with only slightly more wattage, but also slightly more lumen. Osram Professional LED Retrofit Classic, or somesuch for instance. In Europe. Don't know about the Americas, Asia...

I googled that model, it doesn't really seem to have all that many filaments for its wattage rating, so I doubt it's underdriving those LEDs like the dubai lamp. The special thing about it isn't the lumen/wattage, but the longevity by reducing the amount of stress on the LEDs

When this made the rounds first a while ago, I wondered what the fuss was about and compared. At that time I had only the 'candles' with E14 socket, non-dimmable, clear, at 2.5W, 230V/50Hz:

[1] https://www.osram.com/ecat/PARATHOM%20Retrofit%20CLASSIC%20B...

which seemed comparable, though running at about double wattage.

Anyway, they last and last, and I'm perfectly happy with their light.

Meanwhile I upgraded everything else also, and have a few of these

[2] https://www.osram.com/ecat/PARATHOM%20Retrofit%20CLASSIC%20A...

also non-dimmable and clear, for lamps with a single E24 socket, which should be bright, at 7.5W & 10.0W, 230V/50Hz.

Still very happy with the light, but can't say much about lasting, because only about 2 to 3 years in use, while the candles are now 5 to 6 years old.

All 2700k btw. I like it that way.

edit: I should add, before that I followed the fad with the CFL stuff, none of which lasted longer than 6 months,

made ugly light, and that not even instantly(warm up), sometimes with noisy chirping.

So for me it looks like I should have skipped the CFL phase until these were available.

To be honest I wouldn't want to light living spaces with CRI 80 LEDs, no matter how long they last or how efficient they are.

Yah, Yah. I hear that so often. Maybe I'm colorblind, and you are all tetrachromates, but I don't see any color-misrepresentations here. And I've tested that.

But it has been proven formerly that the https://en.wikipedia.org/wiki/Phoebus_cartel existed.

OTOH, given how all consumer electronics contain complex microcontrollers running some RTOS and usually a buffered RTC, would it be inconceivable to have some periodic check running there, and after 2 years of time, changing some voltage on some path to ever so slightly higher, spinning some fan just a little bit slower, doing other weird pwm stuff, and have it cooked? Who would ever know with obfuscated firmware, deeply embedded and encrypted?

I'd say it would be less complicated to implement than the stuff which caused Diesel-Gate.

>Unless you have some evidence that this is both possible and happening, it's just a conspiracy theory.

It's quite easy and I think most manufacturers of electronic devices, electric appliances and even cars are doing it.

You just have to calculate what the average working time during the warranty period and adjust the quality of materials and the build quality so the MTBF is just a bit higher than that time.

> even cars are doing it

Cars are not doing it. Like I said: provide a source that isn't your own suspicions and anecdata.

I am a shareholder of an auto insurance company, and our actuaries would know if cars were doing it. There is a lot of very high-quality data about vehicle reliability, and major trends are very visible.

Car companies are often stupid and/or evil, but they can't (and don't) precisely plan obsolescence. Car usage is a huge bell curve, with the average American driving ~15k miles/year, but large groups of people driving far less or far more. You couldn't possibly plan a car to fail at X years.

Car manufacturers usually set the warranty as 36 months/36,000 miles, for example. I'm not claiming that this is the case, but it'd be far easier to plan for X miles than years.

It’s fairly easy you calculate the MTTF when you design anything. That said it’s probably not as malicious as people might think their goal isn’t to artificially lower the MTTF to just beyond the warranty period but rather that they don’t put any effort in extending the MTTF beyond that.

And really it’s not that companies in the past did that much better, it’s just that in the past getting replacement parts was easier and you had fewer integrated components. Today everything is pretty much in a single package even basic electric circuits like a bridge rectifier. So things got smaller and are more likely to fail and they are harder to replace.

> It’s fairly easy you calculate the MTTF when you design anything.

Where did I disagree with this?

What I'm saying is that MTTF is measured in usage time, whereas warranties are measured in calendar time.

Cycles under normal usage conditions aren’t that hard to calculate.

The range of "normal" is massive for things like cars and laptops.

Again, provide a source please. Otherwise this is just unprovable speculation on your part for something that is theoretically difficult in the first place.

>Unless you have some evidence that this is both possible and happening, it's just a conspiracy theory.

we are close to 2022, this shouldn't really be a conspiracy theory anymore

As some have mentioned Lightbulb is one of those prime examples. Where vast majority of LED lightbulb offer say 10 years of warranty, and yet because manufactures know most people dont bother sending them in, they cheap out on quality. Most LED ( if not all LED ) dont fail at all, it is the controller that fails, it is expensive and makes the difference. As long as it last 3-4 years, no one is going to complain. People will attribute this failure to bad luck / QA and just buy another one.

As well as other mention of capacitor. One easiest way is to get have Power Unit fail after 3-5 years, which seems long enough for Consumer Electronics. Most dont go and figure out it was X failure and just buy a new one. Synology J series NAS power unit is an example of this.

Consumer Electronics has been doing this in China for over a decade. There are reason why things are cheaper. And most internet user are still stuck with "Spec" comparison as if spec means everything.

>And you're not even considering that the manufacturer has no idea when the hardware will be sold and therefore doesn't even know the date that the hardware "should" fail.

Electronics get used when they are sold. MTBF since usage in many cases can be extremely predictable.

There is another point to look at it, a lot of consumer actually prefer to have CE fail within reasonable time so they get an excuse to buy something new. The market for some people like me who thinks something should last at least 10 - 20 years is ridiculously tiny. Although the buy for life internet following is certainly helping a bit.

> As well as other mention of capacitor. One easiest way is to get have Power Unit fail after 3-5 years, which seems long enough for Consumer Electronics.

You're moving the goalposts here. You said that:

> they can get some components to fail soon after warranty period expired with decent accuracy

But then you specifically mention a window of 2 years. From a business perspective, 2 years is not "decent accuracy".

I am pointing at normal circumstance with example for Synology, which in itself is a decent brand. Power unit is also the easiest to game when the devices is constantly on. There are cases where the accuracy is in months. Digital TV STB sold that last only 9-12 months with a 6 months warranty. And all it takes is using low quality components and capacitor in a special batch.

These dont ever reach the shore of US or EU unless you order them via Aliexpress. Consumer protection and expectation sets things apart. But doesn't mean vendors dont do it.

I also gave the LED light bulb example where evidence are given with respect to warranty and lifetime usage.

And no one implied that all of these product, every single batch of it will specifically failed at the same time. Which I think is what you are getting at. My original post only mention it is technically possible, and in many cases actively done so.

>does anyone know a rock solid/powerful brand I can buy instead of Dell? I kind of miss Micron PCs.

I've had good experiences with HP Elitebook x360.

Depending on model, I guess. My 2 year old x360 has so bad airflow for cooling that I've had multiple "meltdowns" if I forget to put a spacer underneath it. Basically the whole thing shuts down and won't boot again for an hour, and the underside is painfully hot. A pencil underneath the screen hinge works best for keeping enough airflow to avoid this.

The first time it had a meltdown, my touchpad stopped working completely, so I went with an external mouse. Second time, the touchpad melted itself back into place I guess, so now it works again.

The newest Macbooks (2021) are looking pretty good tbh. They reworked the design to make the battery trivially replaceable and by the looks of it, they will be selling the battery to end users soon.

But battery is about the only thing you can replace. If your trackpad or keyboard fails, you have to swap out the whole top assembly. If your camera or display cable goes, you have to replace the lid.

Part of that is unavoidable as laptops become ever more thin, but I am reasonably sure Apple could make the keyboard replaceable.

I recall on the older machines that they had upwards of 60 screws holding the keyboard into the top panel. I’ve not checked out a teardown in a long time, is that still the case or is it riveted in?

I always suspected that 60+ screw solution was very much over engineered but I guess they wanted to eliminate any potential for flex of any sort.

> The Internet is littered with complaints about dell PCs failing after warranty, and how recalls are covered up on them frequently.

Not to divert from the topic, but I sometimes wonder how companies on Twitter have their staff monitor tweets to get on top of social media controversies started by random people involving their brand (race, discrimination, etc.), while giving a cold shoulder to paying customers complaining about broken product features or design (with staple replies such as "contact support" or "see warranty info").

I stopped buying Dells ~10 years ago when my last one stopped charging from the AC adapter. It used some hardware check to determine whether it was using an “authentic” Dell AC adapter, but if there was a failure with the check you were SOL.

> Funny I just saw this thread, today I learned that my integrated (not removeable) XPS 15 9570 laptop battery

takes about 10 minutes and the correct screwdriver to replace the battery on most every dell XPS laptop i've seen.

"But next time I think I’ll just buy the 16GB variant upfront."

Next time don't buy from a company that forces you to spend the maximum if you want to use your laptop after 3 years.

I am hopping but not expecting that this chip shortage will force vendors to make systems more modular again as they can't get enough ram/disk out of the get go. Wouldn't that be a nice world where you can upgrade your laptop or desktop PC yourself?

There aren't many options for Windows laptops.

There are HP laptops with user-replaceable RAM, such as the Pro and Elite Books. The 13" ones are comparable to the XPS line in terms of size / weight / performance. I don't have experience with the other sizes.

The top panel being metal is also nicer in my opinion than the weird sticky thing on the XPS.

The company I work for uses those, and they don't seem to have any issues with them.

Honestly it's quite his fault by choosing the XPS 13, an ultra portable. Take the XPS 15, it's the best compromise between portability and repairability. I have a 9750 or something and chose it because it was just the most repairable laptop at the time. You could replace ram and disk for sure, but also buy new fans directly from Dell, or new hinges, a new battery, a motherboard, you name it. Remove a few screws and you have access to everything, easy to mount and dismount.

So no, the company didn't force him, he just went for portability.

It's his fault for buying a product being sold with defects that are apparently known to insiders?

This kind of shit makes me hope Framework succeeds.

Actually, the trend is to put the RAM inside the main chip itself so as to avoid those pesky latency and area-costly PCB traces. The performance improvements this allows are hard to resist and some manufacturers are already falling in (e.g. Apple). And I guess the side effect of making upgrades impossible is nice, too.

Is there actually noticably lower latency when going from sodimm/on MB RAM to memory integrated in SoC? I would expect bottlenecks to be somewhere else, not in those few extra centimeters of traces

Look at Apple's or HBM benchmarks. Also, it's not just the length, but the extra contact points, the "safety" margins for it all (to allow some extra routing distance for lazier/complicated mobo design), etc. all force you to a lower frequency, which also reduces bandwidth.

(IMHO, I would not trade off modularity for that. But the public disagrees.)

Apple uses the exact same LPDDR4x that others route through the mainboard, in stock JEDEC spec, so they don’t really get performance advantages from memory on package.

It's the chipset & CPU that have to slowdown to account for the extra cost to reach it, hidden in whatever stage you call it -- even if the actual DRAM chips may be the same. They're the dumbest part in this entire diagram.

Also, they benefit from larger bus widths, lower power, etc. https://www.anandtech.com/show/16252/mac-mini-apple-m1-teste... ( despite this being DDR4 I have to go to DDR5-systems to find similar bandwidth & latency results ) , https://www.anandtech.com/show/17024/apple-m1-max-performanc... and https://www.anandtech.com/show/17024/apple-m1-max-performanc... ( I have to go quad-channel DDR5 or HBM to find similar results elsewhere)

Time cost is dictated by the clock frequency and timings, not the physical distance itself. Lower distance may help with achieving better specs but Apple is not overclocking the memory, they run JEDEC spec. Maaaybe they train secondary and tertiary timings a little bit tighter, that’s about it.

> despite this being DDR4

It is not! LPDDR IS NOT DDR!! It’s a very very different standard despite the similar name.

> It is not! LPDDR IS NOT DDR!!

And is that what explains the apparent increase in memory bandwidth?

Yes, LPDDR4X in particular is rated for 4266 Mbps (at least that's the only spec available from https://www.samsung.com/semiconductor/dram/lpddr4x/ right now) while the top official JEDEC spec for DDR4 is 3200 Mbps.

4266 Mbps * 8 channels * 2 (bytes in a 16-bit channel) = 68.25GB/s theoretical maximum. (Or 4 channels * 4 bytes in a 32-bit channel, same deal.)

See AIDA64 bandwidth benchmark results: https://storage-asset.msi.com/global/picture/news/2021/nb/lp... — these are from MSI's Intel Tigerlake laptops. There are AMD Renoir laptops with the same memory configuration too. This is not unique to Apple, this is what everyone has now. I guess the new big MBPs might have the largest version of this, but what the regular M1 has is not special whatsoever.

They get power savings from reduced capacitance by reducing the distance from the memory to the CPU from a few cm to a few mm.

Yes, when you get to the very high frequencies of modern RAM, physical distance is significant. For years overclockers have been using mini-ITX motherboards to get the RAM closer to the CPU for higher frequencies.

ITX usage was never about trace length, it’s about having a 2-DIMM board for running 2 sticks – Unoccupied DIMM sockets are terrible!

These days we have the luxury of overclocking boards that are 2-DIMM in ATX form factor.

> Is there actually noticably lower latency when going from sodimm/on MB RAM to memory integrated in SoC?

No.

But it's easier and lower power to run higher bus frequencies for higher bandwidths. Same goes for bus width.

There isn't any latency reduction on the M1, Apple use the exact same 4,266 MT/s LPDDR4X that can be found soldered onto the motherboards on a few window's laptops, and they are just using the stock timings.

I think the main reason Apple are doing it is to simplify their motherboard PCB design and make the design more compact.

Is Intel making SoCs like this yet? All the Core, Pentium, etc. CPUs I've seen are still just CPUs. True Apple is doing this, and ARM Chromebooks probably do it, too.

Intel also makes it, kinda: https://www.anandtech.com/show/16195/a-broadwell-retrospecti...

You bought into Apple marketing. Speed of signal in copper ~20cm per 1ns. 5-7cm of extra track length to a slot would not hurt anything.

Oh my god, did you just discredit the entire field of EDA place & route in just one sentence? This is not your usual 20MHz bus. These are signals that switch at 2, 3, 4 GHz, and there are a shitton of effects that get in the way and limit how fast you can toggle a (longer) wire. Or at least significantly alter the power vs latency vs speed vs noise tradeoffs.

(and no, I most certainly did not buy into apple marketing, since I'm pointing that as a negative)

Did you miss the part where Apple is clocking those chips at same speeds as normal DDR4? Closer placement did not result in lower latency nor higher speeds. Apple M1 is same situation as RPI foundation using POP ram. Requires fewer pcb layers, simplifies layout, lowers cost and as a bonus gives assured future sales due to forced obsolescence. I didnt dismiss anything, I simply know physics and have some experience in high speed layout.

I can understand the claim the M1 has the same timings as off-the-shelf DDR4 (thought this was a nice way to explain the apparent increases in memory bandwidth shown in benchmarks), but here you are dismissing the concept of HBM entirely.

Afaik M1s dont use HBM (which is 1024bit per chip and are their own thing), but ordinary POP ram like RPI. Anything up to 512bits is just cost cutting. Look at Xbox - no problem fanning out 320bit bus in addition to pcie lanes, same for hiend GPUs. 4096bit is where you can claim someone actually uses HBM for performance, like on Radeon R9 Nano/Fury.

LPDDR4 RAM is literally not available in any socketed form factor. The signal integrity becomes too degraded over physical connectors. The solution is to increase the voltage (at which point it is no longer Low Power DDR4) or bypass the socket entirely and solder the chip to the board.

Everything has a tradeoff, including repairability.

I dont entirely buy that considering stock DDR5 goes down to 1.1V using normal DIMM sockets.

Great write-up, went to 11 at

> Lastly, and this is definitely overkill. I had a dentist take an xray of the new and old parts.

I did not know dentists offer such a service but i'm still a bit hesitant to ask at my next checkup.

Depends on your relationship with the dentist, and the specific one. I'v heard of others getting their dentist to xray small things for them, either for free or a nominal fee.

I think this is more "my neighbour and buddy who happens to be a dentist" versus "my dentist who happened to ask about my hobby".

I've been reading through comments and many are complaining about planned obsolescence of laptops, lackluster build quality and lack of upgradeability.

If law makers don't want or can't stand up for consumers, it is for consumers to stand up for themselves.

Vote with our wallets, make bad publicity for badly behaving manufacturers, support good manufacturers.

Use crowdsourcing and make a Wikipedia like resource with good and bad experience regarding devices.

HN community is very small. We are conscious but average customer isn't. They will still buy HP printers, super expensive Apple products that can't be fixed easily (yeah yeah the new Self Repair), or anything they sell at BestBuy. We just need to be careful, we can't save everyone.

I am voting with my wallet and buy the right stuff. If other's don't learn from other's mistakes, so be it.

I hate to say this, but “Vote with your wallet” assumes that consumers are rational and companies are honest, which isn't happening.

In the current scenario, voting with our wallets wouldn't get us far and lawmakers should step in real hard or else we would keep getting “paper/metal straw” solutions for complex problems like endangered aquatic life.

Not an electronics engineer, but have enjoyed the read. If you can pull this off, more power to you.

With soldered storage and RAM onto the boards becoming nowadays fairly common, the author can probably open up a business and do this as a side-job or full time!

Add Chromebooks into the mix, and you'll probably have customers.

>> With soldered storage and RAM onto the boards becoming nowadays fairly common, the author can probably open up a business and do this as a side-job or full time!

Especially if they can figure out how to upgrade certain Mac models.

Nice, I really like how they left the 'Dell, proprietary' in place :)

It's interesting that a board schematic has apparently left Dell, is that sort of thing common?

It also would be quite hilarious to return the machine to Dell for servicing, I think they would be quite confused.

It's quite usual for laptops if you know where to look for.

Before buying I usually check schematics, as they are often more precise than the specs (say, about PCIe lanes assignment, to avoid bad surprises)

Where would one go to look? Would love to review old schematics, specifically the XPS 9350, 9343... and macbooks!

XPS 9350 - Compal LA-C881P

XPS 9343 - Compal LA-B441P

Macbooks - there have been a ton of different models, search the exact model number (e.g. A1181) and "schematic".

so, what terms are there for, say a Lenovo T440s? While these search terms work, I have a hard time to come up with them (same goes for mobile phones, TVs,...)

STING UMA / VIVL0 NM-A102, most likely Wistron like the rest of thinkpads.

Tons of (I guess they are leaked) schematics are available for laptops, phones, and related electronics. Dell didn't do the motherboard design; the OEM is Compal and as you can see from the motherboard, its actual model is LA-H931P.

>Dell didn't do the motherboard design; the OEM is Compal and as you can see from the motherboard, its actual model is LA-H931P.

That begs two questions:

1. You can find that mainboard in other manufacturers laptops or it is just made for Dell?

2. It is Dell just integrating bits and pieces designed by others?

No / Yes

This is a normal thing. The brand makes a spec sheet and the supplier (OEM/ODM) builds something to that spec. For some reason in automotive circles the brands are called OEMs, which is kinda the opposite of the usual meaning.

The schematics for almost all popular laptops are floating around in the internet.

Yes, indeed most of the popular brands of laptops people are familiar with don't do their own electronic design (except Apple, since their schematics do actually say Apple on them; but then everyone knows they use Foxconn and others for the manufacturing). More details here: https://ctacs.weebly.com/news--articles/will-your-next-lapto...

The OEM model numbering isn't segregated, so e.g. the LA-H931P is the Dell in the article, but a LA-3481P is a Tohsiba Satellite A200 and LA-8133P is a Lenovo Thinkpad E530.

I found myself wondering what is actually lost if schematics like this get out. Are companies worried about counterfeit/copycat product? Are there major differences in layout skill across the industry, such that Compal would want to prevent people learning from them? It looks like Dell was entitled to the schematics, so it shouldn't be that a customer could cut out the vendor.

I'm sure it's not about preventing competent people from doing this kind of service on hardware they've bought. Is it more about making warranties work? Or something else?

My guess is that it's due to NDAs on the reference designs (from Intel) that these were originally derived from, and due to legal bureaucracy and such that no one wants to challenge, so they're "confidential" by default.

If you've looked at other schematics from Compal, Inventec, Quanta, Wistron etc. they all have their own uniquely distinct style, but otherwise for a given platform they are all electrically very similar.

How does it benefit Intel to keep reference designs confidential?

Intel is super secretive about everything, just because.

They used to be far more open; I guess that was before the lawyers and manager-types started taking over the company. Here's the famous 440BX reference schematics:

https://web.archive.org/web/19991010005207/http://www.intel....

Here are all of them, it looks like they started being closed with the 900 series chipsets; the 800 series (P4 era) were the last ones they offered plenty of documentation openly:

https://web.archive.org/web/20040823181302/http://developer....

Because they only give you the design if you plan to buy a bunch of chips and pay them for support. It's a way to make money. Additionally, keeping the reference design under NDA reduces the risk of someone building an chip that is drop in compatible with Intel boards.

Keep data sheets and reference designs confidential and your only customers will be large-volume companies able to shell out serious dollars upfront and with enough competent staff on hand.

Hobbyists and small shops who require more assistance in getting up to speed are left to the scraps that the Raspberry Pi Foundation throws at us.

Sorry for the sarcasm, but the situation really REALLY sucks and no one seems to care about it.

The soldered RAM dissuaded me from buying a XPS13. My 6 year old T460s is dying after one too many drops. The XPS13 and Thinkpad X1 Carbon are the final choices for replacement, but all of them have soldered RAM and cannot be upgraded. Also the XPS13 keyboard is worse than the Thankpad (why don't Dell fix the PgUp/PgDown layout?). Finally I settled on the AMD Ryzen 7 Pro based T14. The RAM is expandable. And the CPU performance is double of the i7! It's cheaper even after max out on 4k display and ssd drive.

It's not as simple as soldered versus socketed. There are no DIMMs for low power LPDDR4 RAM, so if you buy a machine with DIMMs, the battery life is slightly worse. This is especially noticeable when the machine is sleeping. I just buy twice the RAM I really need, to help future proof it. Although the disadvantage of this is that one is often forced into buying hotter and noiser Intel chips.

Very good information to know. The model I picked has 16gb soldered RAM (hopefully LPDDR4?) and the expandable socket, allowing up to 96gb. Most xps13 and x1 models have 16gb soldered. Only one model of either has 32gb max.

The T14 has DDR4 RAM instead of LPDDR4, in order that it can support DIMMs. Even the soldered portion will still be DDR4. The X1 has LPDDR4.

It passed through 4 passes of all it’s tests without issue!

Personally I would go for at least 10 or more --- sometimes intermittent memory errors show up after an extended amount of time. I usually leave memory tests on new (to me) hardware running over a weekend. I wonder if the Rowhammer tests find anything? This is LPDDR3 so might be affected, but I don't know how to read Hynix datecodes. (<2010 is usually not, 2010-2011 is questionable, 2012 and newer definitely rowhammer-vulnerable.)

Tried to do the same with a Toshiba AC-100. Have thrown it out at the moment I have realized that they just did not route the extra address line which would made the chip swapping possible. 512MB RAM was not enough for anything back then, even my phone had more...

Back when I was a kid, we un-soldered the original Mac 128 memory and upgrading to 512k...

As described in Dr.Dobb's Journal Volume 10, Number 1, January, 1985 http://dserver.macgui.com/Dr%20Dobbs%20Fatten%20Your%20Mac%2...

Correct! One of my college friends was an EE with good soldering skill, but would not do it for me. But he did coach me and it was successful. Still have the Mac tucked away in one of those old padded carrying cases.

Good to see that there is some 21st century craft skills in the west still (have only seen similar things from Asia on YT)

So, general question: where do you go to get the schematics? (would come in handy so many times!)

>similar things from Asia on YT

Got any pointers?

https://www.youtube.com/watch?v=bloCpO7baO8 - I might have seen something similar on HN actually. Anyway, this is basically the thing done above, just sold to repair shops for a set of very common devices!

>Lastly, and this is definitely overkill. I had a dentist take an xray of the new and old parts.

Is this a friend? Or would any dentist consider this for a fee?

I'll be broaching this subject at my upcoming clean and polish.

MSP for a bunch of dental offices here.

The number of times I've xrayed my car key fob or cell phone when testing out the digital xray sensors after a new PC install at a client site can't even be counted.

If they have a digital xray system it'd take literally seconds and cost them nothing, so if you have a good relationship with them you can probably do it.

I wonder if some day we will get a DRAM board specification thats smaller then SODIMM.

It could be done simply with Pogo pins, or like for cpus a long time ago, ZIF sockets

I believe one issue with all of those is they go through all layers of the PCB. Not sure if surface mount sockets exist?

Think in 3D: I never said the Pogo should be under the chip in one 2d plane.

They could be on the sides at an angle, and the chip in a "hole" cut in the PCB

That would probably not work, those chips have ground planes under them for cooling purposes.

RAM remains passively cooled to this day, so some thermal pads diffusing the heat to the metal case might be enough

I'm having difficulty picturing this. Can you point to an example?

What's the reliability like on a socket like that to a decade of bumps?

I don't think they are available as standard components, but there are companies out there that make sockets for mounting DRAM chips to allow for debugging.

ZIF sockets likely would be extremely challenging in terms of mechanical integrity at high pin density and EMF radiation hardening. Not worth the effort, sadly.

That was my question as well when looking at how much space SODIMM requires on the Framework laptops. Although I am not entirely sure if there is a market for it. Part of the reason ( or benefits ) for going soldered Memory and NAND is assembly automation in production line.

I feel like the RAM quantity on laptops hasn't really increased much over time, anyone know if that is true and what is driving it?

Very few people need any more. There's also chip shortages of various durations.

Laptop makers are making sure people are locked in with less RAM so that they’ll buy another laptop soon.

Both my last laptops were made obsolete only by less RAM.

I very much doubt this is the reason. Soldered on ram lowers the cost of the device, makes it thinner, and improves the reliability. At the cost of upgradability. Since the average user never upgraded the ram when they had the ability to, its purely a win for most users.

It also allows the manufacturer to charge users who want more RAM higher prices. They often charge much more than the street price of DIMMs.

same here...

Or one can look at the anemic and declining profit figures for laptop makers (other than Apple) and come to the conclusion that they are desperately trying to compete by trying to meet their customer’s price points.

I recently bought a new laptop for my son. One of my main concerns was upgradeability, so I had to discard a lot of models with soldered-on RAM.

It really struck me how many laptops are still coming out with 4GB of RAM, which is simply not enough nowadays. Also, many vendors hide or make it difficult to know how much RAM you can add.

Yep true, although it's not just laptops. I often see 'high end editing PCs' with 16GB of RAM, which is possibly sufficient for 1080p editing, but nowhere near enough for 4K editing for multiple programs open.

I guess it's similar to how some laptops/budget desktops still have hard drives, even though an SSD wouldn't cost much more but it'd offer great performance benefits.

Because contrary to popular belief, RAM prices haven't really declined in the last decade [1]

1. https://secureservercdn.net/166.62.107.55/ff6.d53.myftpuploa...

Where did they find the schematics and it looks like Allegro design files - not even gerbers.

Where can we grab those from??

Search for "Compal LA-H931P". I didn't look too hard but there are some places selling them, although you'll probably find a free download (possibly needing to register on a forum) if you look enough.

Either way, does it sleep or hibernate effectively? https://news.ycombinator.com/item?id=28639952

I have the 9310 and it sleeps perfectly. I'd close it and toss it in the felt sleeve I have for it for a while before that warning came out, mine Just Works.

Same experience here (9310, Kubuntu 21.10), but I really miss S3. I hope I can go back on my next laptop...

I love this. I wish I had enough faith in my solder skills (or all the equipment on hand) to attempt anything like this, but I love seeing other people do it successfully!

What kind of jig is that? That thing is beautiful, I love the copper color, but I don't even know what to google for ('copper colored jig' doesn't do the trick). This is the thing I mean on the right: https://gregdavill.github.io/posts/dell-xps13-ram-upgrade/im...

Its probably a HT-90X or similar reballing kit.

They are generally made out of aluminum and anodized various colors.

I mean jig is a pretty generic term. i searched for reballing jig and found it pretty quickly on aliexpress:

https://de.aliexpress.com/item/1005001617899659.html?spm=a2g...

on ddg images there were also several results for a site called dhgate.com that were cheaper, but no idea if that site's legit

> I mean jig is a pretty generic term

Yeah I realized, but I didn't know a better term for it. Not a native speaker, but I also couldn't think of anything in Dutch so... Thanks for the link!

> So as a final step I fired up memtest86.

Doesn't Dell have a built-in hardware tester in the BIOS (press F12 for the one-time boot menu, then choose its option), complete with a memory tester? Or is that only on the more professional Latitude/Optiplex lines? The built-in tester would have the advantage of also testing other parts of the board.

This writeup is excellent. I wish I could work my hot air gun... I'll need to watch some videos, I think.

> Remember how I mentioned that the schematic specified 2133Mbps? Well the LPDDR3 from my old mainboard was only rated at 1866Mbps

Desktop DDR4 can be overclocked by a lot, surely the 1866 rated ICs would work completely fine at 2133…

Say you’re inexperienced and have a 20% chance of screwing the pooch each time you attempt this. I wouldn’t risk it.

The only risk in memory overclocking is that it’s not stable and your data gets corrupted in memory sometimes. Which, well, who gave you a guarantee that it won’t happen on stock specs? Nobody without ECC ;) So run a memtest and you’ll be reasonably sure it’s good enough.

No, I’m referring to the surgery. While I’m only a hobbyist, my success rate is about 80%.

Practicing beforehand was definitely the right thing to do.

It's interesting to see that nowadays reballing has the same exoteric feeling as desoldering your C64 chips back then.

Or: get a dell latitude 7390 instead of dell the xps 7390 and upgrade to 32gb by just swapping a ram stick.

I personally would avoid soldered ram. If you ask a software engineer this is against Open/Closed principle.

It is an unecessary bad practice against the benefit of the user no matter what companies claim.

We've slowly been removing socketed components in laptops for 30 years. Soldering irons aren't DRM, they're just require some tools and skill.

Why is RAM special and not CPU cache, floating point coprocessors, integrated graphics?

They moved FPU onto the CPU 30 years ago and that wasn't a big deal to FSF people.

> unecessary bad practice against the benefit of the user

Are there not noticeable battery life consequences of socketed RAM? From what I can tell, and I’m sure others here would know more, this often comes up as a reason for the battery life of the Framework laptop (or other socketed RAM portable laptops) compared to similar laptops with soldered RAM.

If the majority of your sales are intended for businesses that plan to constantly buy new laptops, it makes perfect engineering sense to just solder the ram to reduce costs. You also have to balance MTBF, the costs of rework, and the length of the warranty, these are all decisions engineers make when building products. You could make a laptop with components that should last 15 years, a modular design, and a plan to build replacements for years after but you'd end up with really expensive laptop that few would be interested in buying. Obviously engineers should strive for sustainability and not being hostile to diy repairs but I know who signs my paycheck and it's not XPS or Thinkpad enthusiasts on the internet.

This is genuinely impressive and a wonderful write up! Dear lord, what a Herculean task

That was as a great read!

> So as a final step I fired up memtest86.

I held my breath as I vaguely remember something about memtest86 causing hardware issues back in the day but I am not sure if I am remembering it correctly.

memtest86 doesn't cause issues but it sure finds them.

You can have a computer that is nominally "working fine" and memtest86 can still find issues.

And, to be fair, memtest86 was always right.

Reply on news.ycombinator.com