Thank you for all the comments. Good to know what is happening out there. I won't try and defend our testing in another email. We do indeed test 100 percent of our products and from day one.
Found this on Kingston site http://www.kingston.com/company/videos/mfg/module.... to see production and testing on video. Vid shows machine testing, with that I think the 100% testing is believeable. You should have included this in the link Wesley
I do not think it has anything to do with logic. Unless it was specified on package that the ram does not work on Asus P4P800 board at CAS 2, they had no reason to refuse RMA. Sounds like they dropped the ball.
Depending on when Kingston started testing 100% of their rams, this article may well be a PR campaign.
They worked at slower speeds. Or at 400mhz with cas 2.5.
The tech support said that they would only work at cas3 on my motherboard. The modules said they would work at cas2.
Kingston refused the RMA since they said there could be no guarantee that they would work at rated speeds on a Asus P4P800 board. I really don't understand their logic, since Ive ran lots of different Cas2 memory on this board.
My recent experience with Kingston hasn't been too good either. Out of the 4 Kingston Value Ram modules I have personally had in my hands, one was DOA and another died in under six months. The other two seem to be working OK right now.
The author made two references in his article and one in the comments section regarding Kingston manufacturing wafers. To the best of my knowledge Kingston has no wafer manufacturing capability itself. They are an assembler who purchases the memory chips from Samsung, Micron, Infineon, Winbond and the other DRAM manufacturers.
I think the article is a little misleading in that regard. The only link I can see between Kingston and wafer manufacturing is a $50 million investment they made in Elpida (formerly Hitachi and NEC semi operations)
Do you remember exactly what model Hyper X memory you threw in your trash and when the throwing incident took place? Do you also remember when you bought them?
"Bought 2x512MB Hyper X. It would not even work Cas2 at 400Mhz (rated speed),..."
Did they work at any other speeds or did they not work at all?
"They said it would only work at CAS3,"
Was that what the tech support say or was that the specification that was specified on that particular Hyper X modules?
"Anyways, since the stick wouldnt work I couldn't sell it so I just threw this crap memory in the trash. I am not lying. My roommate saw me."
So Kingston refused RMA? If so, for what reason?
And for KTCSkins, when did Kingston start testing 100% of all their memory modules?
Kingston is used by some big corporations for all their memory needs/upgrades - the quality of their regular RAM is quite good. I've encountered very few problems with Kingston RAM in OEM-type machines.
I also don't believe you test all modules. My only experience with Kingston was terrible. Bought 2x512MB Hyper X. It would not even work Cas2 at 400Mhz (rated speed), even with the voltage at 2.8V. On top of that, when I asked for an RMA, the tech support said it was not compatible with my Asus motherboard, which was just B.S. They said it would only work at CAS3, but I've run tons of other cas2 memory on this board, and similar models (p4C800E) without a problem. Anyways, since the stick wouldnt work I couldn't sell it so I just threw this crap memory in the trash. I am not lying. My roommate saw me.
I bought some 2x512MB Corsair XMS and it worked perfectly straight out of the box. Much happier with them. Never buying Kingston again and all of my friends know about their "great" support.
Don't buy Kingston - they can't compete with other memory manufacturers like Corsair or Mushkin.
#35
"Interesting thread on the ICs. We don't offer this type of information."
that's the thing -- corsair do (on their forums, they have people providing exact IC's based on serial number and people willing to speculate on the amount of headroom)
If you work for kingston and want to capture more of the enthusiast market, then you may want to take a look at what corsair are doing
also being able to offer "free" headroom is what the enthusiast value market is all about and the rave thing right now is winbond UTT's which kingston dont currently offer...
I guess people just want some highly overclockable ram, as cpu's can nearly do 50% but most of your ram line cant nearly even approach that (or that of some other memory makers)
oh - forgot to ask, are your chips speed binned by hand or by machine?
sorry if this sounds like a big diss for you guys - I like kingston, its just not cutting it in the enthusiast market........
Interesting thread on the ICs. We don't offer this type of information. We qualify different components depending on which line they are for. that is what is great about Kingston. We have a flexible model. We have relationshipos with allt he top DRAM suppliers, not just one. We don't specify ICs because they often can go dicontinued or we may use a different IC based on qualification or supply.
Our ValueRAM line uses different components that our HyperX line. We only promise the memory at the qualified speed (this goes for HyperX as well). We don't promise particular ICs.
ValueRAM is based on JEDEC spec, while HyperX are usually outside of JEDEC standards. Because of the different types of motherboards, etc. we can't promise what type of headroom a module will offer. It is based upon bios, overclockability etc. We only promise it at a particular speed that HyperX is tested at. There is a limit for sure. Users of ValueRAM aren't the overclockers, but those that are looking for industry standard modules. HyperX is used for those that are enthusiast. ValueRAM is not designed for that. HyperX is also designed for enthusiasts, but there is a point to where that obviously goes out of warranty like any product. If a product doesn't have the highest of headroom that doesn't make it a bad module. It depends on how much someone wants to push a module. Is there ever enough headroom?
HyperX PC 3200K2 (2-2-2-6-1T) (KHX 3200K2) --chip--> Winbond BH-5
HyperX PC 3200AK2 (2-3-2-6-1T) (KHX 3200AK2) --chip--> Winbond CH-5
HyperX PC 3200ULK2 (2-2-2-5-1T) (KHX 3200ULK2) --chip--> Samsung TCCD
HyperX PC 3500 (rated 2-3-3-7 @ 433Mhz, 2-2-2-6 @ 400Mhz, old revision) (KHX 3500) --chip--> Winbond BH-5 (maybe relabeled)
HyperX PC 3500K2 (2-3-3-7) (KHX 3500K2) --chip--> Winbond CH-5
HyperX PC 3500A (2-3-3-7) (KHX 3500A) --chip--> Winbond CH-5
HyperX PC 4000K2 (3-4-4-8) (KHX 4000K2) --chip--> Samsung TCCC, Hynix D43, D5
HyperX PC 4300K2 (3-4-4-8) (KHX 4300K2) --chip--> Hynix D5
I got this list from a forum of ram IC listings (http://www.xtremesystems.org/forums/showthread.php... just shows that there are not many products available where there is much headroom left other than what is speed rated.... eg. when u buy 3200, dont expect to get much beyond that (exceptions:TCCD + BH5)
Many tests show that the value ram line also has little to no headroom at all either..
Its not a bad thing - but dont expect to get something for nothing :)
Well, I can answer some of your questions. We purchase only the top premium components for our modules. During any SMT process there can be always be potential issues no matter what company you buy from.
We don't have a high failure rate, and we would like to keep it that way. Having our modules are 100% tested ensures we have done everything we can to make sure our product is reliable. With things outside of our control, like shipping, user error, static, etc. things do happen. We aren't perfect, but we strive to always provide consumers with a quality product.
Lots of people may be skeptical in our testing processes. We also visually inspect 100% of our modules as they go through the process. People are skeptical because most company's don't do this. I can only prove it to you with our word. It is up to you to decide. Some don't mind buying a cheaper, less quality product. It may work. But it may not... You take your chances. Some consumers don't care. The ones that purchase Kingston know that if there ever is a problem, we are 1.) financially strong with zero debt 2.) we have a lifetime warranty and 3.) free technical support and have the service to support our customers.
Remember, you have the freedom to choose. But we hope you choose Kingston.
I have nothing against Kingston, in fact I have bought plenty of their ram before and I'll keep buying it in the future, but I won't believe that you test every module you sell by hand in a computer test.
As for reliability, Kingston may be a bit more reliable as I've never had issues with their memory (compared to others) then again, Kingston is usually a bit less agressive with their timings than the competition.
Heather, you imply that you have a rather high failure rate when you say that the cost of RMAs would be insane [if you shipped without screening everything].
>>Any insight as to why failure rates would be so high?
>>What happens with the failed product ... do they get binned down to less speed grades (not unlike how Pentiums get turned into Celerons when their cash is flawed)?
>>How much margin do you give the memory? Meaning, if it's rated at PC3200, how much faster do you test it to ensure that it's not operating right at the edge of ability -- i.e. fail at PC3202?
I know you likely cannot answer some of these questions, but anything you can offer is welcomed.
Testing DIMMs? Are those testers former Microsoft employees? I've had only 3 ValueDimms DDR400 CL2.5. Two of them were just broke, and the last wasn't able to run at DDR400.
This "testing" is why I'll never buy again Kingston.
i have to visit some friends in Taiwan this year. any links or review of non-trance/non-top 40 music nightclubs would be sincerely appreciated. thanks for the article. btw. really fun in light of Sunday + Monday
Hello - I can guarantee that the testing procedures done at Kingston are not B.S. I have worked at Kingston for over thirteen years and have never seen it done any other way. Yes, it does take a lot of time and effort, but it is worth it. Sometimes that does involved a slight premium, but we are very competitive in our prices. We don't want to have returns and value the quality that we have in our products. Sure, we could cut corners and make things easier, but the rate of Returned Merchandise would be insane and that price is a lot higher to pay.
Some of our motherboard testing is done by hand depending on the quantity and some automated. We have an exceptional test engineering team who have patented many of Kingston's testers to make Kingston memory the utmost in quality.
Again, I guarantee you that we don't just put on a show just because we have a visitor. I know that other manufacturers might do this, but it will come back to haunt them. We want to show people what we really do well.
As far as Kingston being in the Enthusiast market - we are definitely there. We don't launch products until we find that they are not only just useful in the market, but can be done on the quality level that we expect and done not just for a handful of people, but to the market as a whole. Any company can come out with releases that show a higher, faster performing cards. But often speeds and performance are only validated on an engineering sampled board with beta bios etc. We want to make sure that outside of one board and one person that the masses reach these numbers. We launch products that aren't vaporware/paper launches. That is what makes us unique. It may look like we are behind sometimes in regards to speeds for enthusiasts, but we are definitely not. We just launch when we feel that it is useful to the market and we can reproduce more than just a handful.
#15 & #18 - I have seen wave solder machines used for mixed motherboards with some SMT as well, but you are probably correct that the machines I saw were reflow soldering. I have changed the article to say "SMT-soldering stations" which should be a correct description in the absence of more info.
I can't believe they don't use a automated test machine. Load a stack of dimms in and the machine loads them, tests them, records results, kicks out bad ones. Loading by hand into a pc is so 70's :)
I would be surprised if they actually use wave solder to "glue" the parts down. It is more likely they use some form of reflow technique such as infrared, convection, or vapor phase.
Wave solder is used for leaded parts, but reflow makes more sense for SMT parts -- and as far as I can tell, Kingston doesn't have any leaded parts (parts with legs which poke through the circuit board) on their DIMMs or flash products (SD, CF, et al).
#11
they said they test them -- not run memtest for 24 hours -- they also said that now their competition is also doing the same thing....
its feasable but kingston is not much of a player in the enthusiast market theses days
I smell BS. There is no way they test millions of manufactured memory sticks by hand in a computer. Yeah, they may do some testing of random sticks but there is just no way that they do this to all of them. Maybe they just test the ones that are suspect after the machine test, but to test them all would require more man power than it takes to make them. Their prices then would be much higher than their competition and that's not true either.
yeah i guess you're right Wesley, we all need a break from theinquirer and such sometimes...
i still think that they can let you in some stats like how much do the machines cost, how often they’re replaced, mtbf, throughput ...
maybe a video of how the machines do their job next time and why not tweak them (while the daemon guards with the lightning whips are not looking) to make them dance?
#6 - Without wafer-making at Kingston Taiwan the process is pretty straightforward. We kept looking for the laser Samurai warriors, but they just weren't there. There are already enough sites who invent or embelish the truth. That's why you come to AnandTech, yes?
well i'm a bit disapointed. where are the videos and at least some technincal details or some interesting back story (with laser-sword wielding samurai warriors and hot princesses perhaps, no?).
raw materials are brought in.
raw materials are glued together.
glued together raw materials are labeled, tested and packaged.
the end.
"...but Kingston is the only memory maker that we know whose name is recognized in every corner of the globe as the world’s largest memory maker"
Maybe because Kingston sells its products under the same name in the whole world. Other producers (mainboards mainly) have different names for US market and European market.
Anyway, very interesting article. I remember seeing (in Romania) an ad about 5 years ago, for Kingston memory for Sun workstations.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
48 Comments
Back to Article
KTCSkins - Wednesday, June 22, 2005 - link
Thank you for all the comments. Good to know what is happening out there. I won't try and defend our testing in another email. We do indeed test 100 percent of our products and from day one.OCMAN - Tuesday, June 21, 2005 - link
Found this on Kingston site http://www.kingston.com/company/videos/mfg/module.... to see production and testing on video. Vid shows machine testing, with that I think the 100% testing is believeable. You should have included this in the link WesleyOCMAN - Tuesday, June 21, 2005 - link
Jbog - Monday, June 20, 2005 - link
I do not think it has anything to do with logic. Unless it was specified on package that the ram does not work on Asus P4P800 board at CAS 2, they had no reason to refuse RMA. Sounds like they dropped the ball.Depending on when Kingston started testing 100% of their rams, this article may well be a PR campaign.
MrMuk7 - Saturday, June 18, 2005 - link
They worked at slower speeds. Or at 400mhz with cas 2.5.The tech support said that they would only work at cas3 on my motherboard. The modules said they would work at cas2.
Kingston refused the RMA since they said there could be no guarantee that they would work at rated speeds on a Asus P4P800 board. I really don't understand their logic, since Ive ran lots of different Cas2 memory on this board.
cryptonomicon - Saturday, June 18, 2005 - link
i wonder what they use to test their ram. probably some proprietary testing program?Zepper - Saturday, June 18, 2005 - link
My recent experience with Kingston hasn't been too good either. Out of the 4 Kingston Value Ram modules I have personally had in my hands, one was DOA and another died in under six months. The other two seem to be working OK right now..bh.
Eskimo - Friday, June 17, 2005 - link
The author made two references in his article and one in the comments section regarding Kingston manufacturing wafers. To the best of my knowledge Kingston has no wafer manufacturing capability itself. They are an assembler who purchases the memory chips from Samsung, Micron, Infineon, Winbond and the other DRAM manufacturers.I think the article is a little misleading in that regard. The only link I can see between Kingston and wafer manufacturing is a $50 million investment they made in Elpida (formerly Hitachi and NEC semi operations)
Jbog - Friday, June 17, 2005 - link
Do you remember exactly what model Hyper X memory you threw in your trash and when the throwing incident took place? Do you also remember when you bought them?"Bought 2x512MB Hyper X. It would not even work Cas2 at 400Mhz (rated speed),..."
Did they work at any other speeds or did they not work at all?
"They said it would only work at CAS3,"
Was that what the tech support say or was that the specification that was specified on that particular Hyper X modules?
"Anyways, since the stick wouldnt work I couldn't sell it so I just threw this crap memory in the trash. I am not lying. My roommate saw me."
So Kingston refused RMA? If so, for what reason?
And for KTCSkins, when did Kingston start testing 100% of all their memory modules?
Icehawk - Friday, June 17, 2005 - link
It all depends on what you want to get out of it.Kingston is used by some big corporations for all their memory needs/upgrades - the quality of their regular RAM is quite good. I've encountered very few problems with Kingston RAM in OEM-type machines.
MrMuk7 - Friday, June 17, 2005 - link
I also don't believe you test all modules. My only experience with Kingston was terrible. Bought 2x512MB Hyper X. It would not even work Cas2 at 400Mhz (rated speed), even with the voltage at 2.8V. On top of that, when I asked for an RMA, the tech support said it was not compatible with my Asus motherboard, which was just B.S. They said it would only work at CAS3, but I've run tons of other cas2 memory on this board, and similar models (p4C800E) without a problem. Anyways, since the stick wouldnt work I couldn't sell it so I just threw this crap memory in the trash. I am not lying. My roommate saw me.I bought some 2x512MB Corsair XMS and it worked perfectly straight out of the box. Much happier with them. Never buying Kingston again and all of my friends know about their "great" support.
Don't buy Kingston - they can't compete with other memory manufacturers like Corsair or Mushkin.
xsilver - Friday, June 17, 2005 - link
#35"Interesting thread on the ICs. We don't offer this type of information."
that's the thing -- corsair do (on their forums, they have people providing exact IC's based on serial number and people willing to speculate on the amount of headroom)
If you work for kingston and want to capture more of the enthusiast market, then you may want to take a look at what corsair are doing
also being able to offer "free" headroom is what the enthusiast value market is all about and the rave thing right now is winbond UTT's which kingston dont currently offer...
I guess people just want some highly overclockable ram, as cpu's can nearly do 50% but most of your ram line cant nearly even approach that (or that of some other memory makers)
oh - forgot to ask, are your chips speed binned by hand or by machine?
sorry if this sounds like a big diss for you guys - I like kingston, its just not cutting it in the enthusiast market........
unclebud - Friday, June 17, 2005 - link
hmmm! some good discussion going on here. thanks for the fortrightness... it will be rememberedKTCSkins - Friday, June 17, 2005 - link
Interesting thread on the ICs. We don't offer this type of information. We qualify different components depending on which line they are for. that is what is great about Kingston. We have a flexible model. We have relationshipos with allt he top DRAM suppliers, not just one. We don't specify ICs because they often can go dicontinued or we may use a different IC based on qualification or supply.Our ValueRAM line uses different components that our HyperX line. We only promise the memory at the qualified speed (this goes for HyperX as well). We don't promise particular ICs.
ValueRAM is based on JEDEC spec, while HyperX are usually outside of JEDEC standards. Because of the different types of motherboards, etc. we can't promise what type of headroom a module will offer. It is based upon bios, overclockability etc. We only promise it at a particular speed that HyperX is tested at. There is a limit for sure. Users of ValueRAM aren't the overclockers, but those that are looking for industry standard modules. HyperX is used for those that are enthusiast. ValueRAM is not designed for that. HyperX is also designed for enthusiasts, but there is a point to where that obviously goes out of warranty like any product. If a product doesn't have the highest of headroom that doesn't make it a bad module. It depends on how much someone wants to push a module. Is there ever enough headroom?
quidpro - Friday, June 17, 2005 - link
Hey how cool is that? Get invited on more tours please! Kinda like ol' Mr. Rogers showing us how crayons are made. (I mean this in a respectful way.)Videos would have been a treat.
xsilver - Thursday, June 16, 2005 - link
HyperX PC 3200K2 (2-2-2-6-1T) (KHX 3200K2) --chip--> Winbond BH-5HyperX PC 3200AK2 (2-3-2-6-1T) (KHX 3200AK2) --chip--> Winbond CH-5
HyperX PC 3200ULK2 (2-2-2-5-1T) (KHX 3200ULK2) --chip--> Samsung TCCD
HyperX PC 3500 (rated 2-3-3-7 @ 433Mhz, 2-2-2-6 @ 400Mhz, old revision) (KHX 3500) --chip--> Winbond BH-5 (maybe relabeled)
HyperX PC 3500K2 (2-3-3-7) (KHX 3500K2) --chip--> Winbond CH-5
HyperX PC 3500A (2-3-3-7) (KHX 3500A) --chip--> Winbond CH-5
HyperX PC 4000K2 (3-4-4-8) (KHX 4000K2) --chip--> Samsung TCCC, Hynix D43, D5
HyperX PC 4300K2 (3-4-4-8) (KHX 4300K2) --chip--> Hynix D5
I got this list from a forum of ram IC listings (http://www.xtremesystems.org/forums/showthread.php...
just shows that there are not many products available where there is much headroom left other than what is speed rated.... eg. when u buy 3200, dont expect to get much beyond that (exceptions:TCCD + BH5)
Many tests show that the value ram line also has little to no headroom at all either..
Its not a bad thing - but dont expect to get something for nothing :)
KTCSkins - Thursday, June 16, 2005 - link
Sorry, I don't understand your question.scott967 - Thursday, June 16, 2005 - link
I'm wondering if the receipt QA inspection on the chips include any sort of speed testing?scott s.
.
KTCSkins - Thursday, June 16, 2005 - link
Well, I can answer some of your questions. We purchase only the top premium components for our modules. During any SMT process there can be always be potential issues no matter what company you buy from.We don't have a high failure rate, and we would like to keep it that way. Having our modules are 100% tested ensures we have done everything we can to make sure our product is reliable. With things outside of our control, like shipping, user error, static, etc. things do happen. We aren't perfect, but we strive to always provide consumers with a quality product.
Lots of people may be skeptical in our testing processes. We also visually inspect 100% of our modules as they go through the process. People are skeptical because most company's don't do this. I can only prove it to you with our word. It is up to you to decide. Some don't mind buying a cheaper, less quality product. It may work. But it may not... You take your chances. Some consumers don't care. The ones that purchase Kingston know that if there ever is a problem, we are 1.) financially strong with zero debt 2.) we have a lifetime warranty and 3.) free technical support and have the service to support our customers.
Remember, you have the freedom to choose. But we hope you choose Kingston.
fishbits - Thursday, June 16, 2005 - link
Does the RAM fresh off the assembly lines taste better than the RAM we buy in stores? I always thought it would. So jealous.gibhunter - Thursday, June 16, 2005 - link
I have nothing against Kingston, in fact I have bought plenty of their ram before and I'll keep buying it in the future, but I won't believe that you test every module you sell by hand in a computer test.As for reliability, Kingston may be a bit more reliable as I've never had issues with their memory (compared to others) then again, Kingston is usually a bit less agressive with their timings than the competition.
Houdani - Thursday, June 16, 2005 - link
Typos, sheesh....binned down to *lesser* speed grades...
...when their *cache* is flawed...
Houdani - Thursday, June 16, 2005 - link
Heather, you imply that you have a rather high failure rate when you say that the cost of RMAs would be insane [if you shipped without screening everything].>>Any insight as to why failure rates would be so high?
>>What happens with the failed product ... do they get binned down to less speed grades (not unlike how Pentiums get turned into Celerons when their cash is flawed)?
>>How much margin do you give the memory? Meaning, if it's rated at PC3200, how much faster do you test it to ensure that it's not operating right at the edge of ability -- i.e. fail at PC3202?
I know you likely cannot answer some of these questions, but anything you can offer is welcomed.
KTCSkins - Thursday, June 16, 2005 - link
Sorry to hear that. Would be intersted to know where you purchased these from.patrick0 - Thursday, June 16, 2005 - link
Testing DIMMs? Are those testers former Microsoft employees? I've had only 3 ValueDimms DDR400 CL2.5. Two of them were just broke, and the last wasn't able to run at DDR400.This "testing" is why I'll never buy again Kingston.
unclebud - Thursday, June 16, 2005 - link
i have to visit some friends in Taiwan this year. any links or review of non-trance/non-top 40 music nightclubs would be sincerely appreciated. thanks for the article. btw. really fun in light of Sunday + MondayKTCSkins - Thursday, June 16, 2005 - link
Thank you very much.yacoub - Thursday, June 16, 2005 - link
I run Kingston in my current system and after this article and Heather's post, I'll likely buy them again when I upgrade my rig in the coming weeks.KTCSkins - Thursday, June 16, 2005 - link
Hello - I can guarantee that the testing procedures done at Kingston are not B.S. I have worked at Kingston for over thirteen years and have never seen it done any other way. Yes, it does take a lot of time and effort, but it is worth it. Sometimes that does involved a slight premium, but we are very competitive in our prices. We don't want to have returns and value the quality that we have in our products. Sure, we could cut corners and make things easier, but the rate of Returned Merchandise would be insane and that price is a lot higher to pay.Some of our motherboard testing is done by hand depending on the quantity and some automated. We have an exceptional test engineering team who have patented many of Kingston's testers to make Kingston memory the utmost in quality.
Again, I guarantee you that we don't just put on a show just because we have a visitor. I know that other manufacturers might do this, but it will come back to haunt them. We want to show people what we really do well.
As far as Kingston being in the Enthusiast market - we are definitely there. We don't launch products until we find that they are not only just useful in the market, but can be done on the quality level that we expect and done not just for a handful of people, but to the market as a whole. Any company can come out with releases that show a higher, faster performing cards. But often speeds and performance are only validated on an engineering sampled board with beta bios etc. We want to make sure that outside of one board and one person that the masses reach these numbers. We launch products that aren't vaporware/paper launches. That is what makes us unique. It may look like we are behind sometimes in regards to speeds for enthusiasts, but we are definitely not. We just launch when we feel that it is useful to the market and we can reproduce more than just a handful.
I welcome questions.
- Thanks,
Heather Skinner
Kingston
Wesley Fink - Thursday, June 16, 2005 - link
#15 & #18 - I have seen wave solder machines used for mixed motherboards with some SMT as well, but you are probably correct that the machines I saw were reflow soldering. I have changed the article to say "SMT-soldering stations" which should be a correct description in the absence of more info.Backslider - Thursday, June 16, 2005 - link
@15I'm sure the author just used the wrong term. for the process. There is no way they could use wave soldering on SMT parts, it has to be reflow.
Son of a N00b - Thursday, June 16, 2005 - link
great article...thanks!Brazos - Thursday, June 16, 2005 - link
I can't believe they don't use a automated test machine. Load a stack of dimms in and the machine loads them, tests them, records results, kicks out bad ones. Loading by hand into a pc is so 70's :)Houdani - Thursday, June 16, 2005 - link
I would be surprised if they actually use wave solder to "glue" the parts down. It is more likely they use some form of reflow technique such as infrared, convection, or vapor phase.Wave solder is used for leaded parts, but reflow makes more sense for SMT parts -- and as far as I can tell, Kingston doesn't have any leaded parts (parts with legs which poke through the circuit board) on their DIMMs or flash products (SD, CF, et al).
Regs - Thursday, June 16, 2005 - link
How exactly do all those one's and zero's copy themselves into the ram? That's what I want to know.yacoub - Thursday, June 16, 2005 - link
awesome article!Anyone else want to grab a case of completed DIMMs and turn it into a giant RAM disk? :D
xsilver - Thursday, June 16, 2005 - link
#11they said they test them -- not run memtest for 24 hours -- they also said that now their competition is also doing the same thing....
its feasable but kingston is not much of a player in the enthusiast market theses days
gibhunter - Thursday, June 16, 2005 - link
I smell BS. There is no way they test millions of manufactured memory sticks by hand in a computer. Yeah, they may do some testing of random sticks but there is just no way that they do this to all of them. Maybe they just test the ones that are suspect after the machine test, but to test them all would require more man power than it takes to make them. Their prices then would be much higher than their competition and that's not true either.semo - Thursday, June 16, 2005 - link
yeah i guess you're right Wesley, we all need a break from theinquirer and such sometimes...i still think that they can let you in some stats like how much do the machines cost, how often they’re replaced, mtbf, throughput ...
maybe a video of how the machines do their job next time and why not tweak them (while the daemon guards with the lightning whips are not looking) to make them dance?
Brian23 - Thursday, June 16, 2005 - link
I got excited seeing the QA Engineer in action.Wesley Fink - Thursday, June 16, 2005 - link
#6 - Without wafer-making at Kingston Taiwan the process is pretty straightforward. We kept looking for the laser Samurai warriors, but they just weren't there. There are already enough sites who invent or embelish the truth. That's why you come to AnandTech, yes?wien - Thursday, June 16, 2005 - link
Logic wouldn't give you the pictures though. :Psemo - Thursday, June 16, 2005 - link
well i'm a bit disapointed. where are the videos and at least some technincal details or some interesting back story (with laser-sword wielding samurai warriors and hot princesses perhaps, no?).raw materials are brought in.
raw materials are glued together.
glued together raw materials are labeled, tested and packaged.
the end.
logic alone can tell me that
faboloso112 - Thursday, June 16, 2005 - link
neatarfan - Thursday, June 16, 2005 - link
Good Article, and Oh... they use Dell Computer too....Calin - Thursday, June 16, 2005 - link
"...but Kingston is the only memory maker that we know whose name is recognized in every corner of the globe as the world’s largest memory maker"Maybe because Kingston sells its products under the same name in the whole world. Other producers (mainboards mainly) have different names for US market and European market.
Anyway, very interesting article. I remember seeing (in Romania) an ad about 5 years ago, for Kingston memory for Sun workstations.
jm20 - Thursday, June 16, 2005 - link
I always enjoy reading on the companies that manufacture technology like this. It feels comfortable knowing they tested my USBkey before I bought it.Rapsven - Thursday, June 16, 2005 - link
Interesting look on how our RAM is being made.Good article.