I'm glad I dont have to make another thread about this. I have alot to say, so top off your drink, and get comfortable.
First a little about me. This is not bragging, I just want people to know where I'm coming from. I have been working with computers since 1983. I had the first tandy, commodore, amiga, 286, 386 etc etc. I have degrees in electronics, and auto tech. I have been taking things apart since I was 2, and fixing them since I was 8. My motto is "If I can't fix it, it ain't broke" a little pretentious, but its kind of a joke. In reality, I can count the number of things I could not diagnose or fix over the years. I also ran a computer business for several years back in the 90's, which gave me access to a wide variety of hardware and software, and at that time Micron was the only RAM I would regularly reccomend. Now you know I'm not some 1st grader with no experience with computers.
Anyway, I bought two sticks of Ballistix last fall. Instantly I had memory errors. Since this was a memory upgrade (opposed to a new system build) the obvious problem was the memory. Old RAM in: Stable, New RAM in: Unstable. So my first thought is a config problem in the bios. After many, many, many hours of config setting changes I could only acheive moderate stability in 'compatibilty' mode, or the lowest speed settings. I had a brand new Asus board coming in for someone else at the time, and since the mem was compatible, I opted to put the old RAM back in, and wait to test the Ballistix on that board. Same problem on the Asus, unstability, and memory errors.
I started poking around the net, and to my suprise, alot of people were having problems with Ballistix. I really got flushed, because In years past I would tell people over, and over Micron was the cats meow of memory makers. So of coursee I RMA'd the RAM, and had assurances that it was likely isolated. Ok, no problem, I have RMA'd hundreds of things.
Get the new RAM, same problem. Unfortunately I didnt have another board to test it on, but I have a friend that runs a shop, and has access to a memory tester. Tests concluded that the memory would not perform at advertised speeds. So, my descision at the time was to run underclocked, and I should be stable right? Well, sort of. What I hadnt expected was that the memory would degrade further, which is what happened. Unbeknowenced to me I have made many bad CD's because of this. Yes I can corelate this to the memory. From checking pre, and post memory installs, I can see that about %40 of the disks I made have errors. One drive was pre and post memory, the other was post memory. Since the discs are dated I can see when each disc was made. It is probably this single fact that infuriates me more than anything. The crashes, and lockups aside, some data was irreplaceable, other stuff was very difficult to come by, but sifting through 200 CD's to see what was bad or not priceless.
Now I know the SOP for the industry is to blame the customer, and to be fair, as a person from the industry, the first rule in troubleshooting is 'check the nut behind the keyboard'. But there is no doubt in my mind that using your common sense is just as important as skill and knowledge. That said, It's difficult to understand how there could be such widespread problems, and not be a recall. Oh, right, it's too expensive to do the right thing anymore.
Now, this poor guy has replaced his entire computer, and still the finger gets pointed at him. Oh, right, it's statistically impossible?
Crucial_Katana wrote:
The problem is that it's extremely unlikely to receive two bad sets of RAM in a row. Five in a row simply does not happen, so going only on simple logic and statistical probability it is far more likely that something is causing the RAM to go bad.
Maybe, under normal circumstances I would agree, but here we have some aggrivating factors aswell as alot of factual evidence that your statistics are wrong. New RAM installed in new systems, in static free enviroments have instantly and repeatedly failed. All replacement RAM is refurbished. this was told to me by customer service, if this is not true, then they lied to me. Seems like a very unlikely lie. They said this was SOP.
Crucial_Katana wrote:Keep in mind that hardly anyone ever makes a forum post to tell the world how happy they are with a product, the vast majority of forum posters will be the ones who are unhappy... they may only be 1% of all the people out there using the product, but they will be the ones who make a point talking about it everywhere, to as many people as possible. The vast happy majority usually remains silent.
Dont know where your getting your %1 stats from. I would like to see documentation. In fact many people with problems do make posts about the problems, but very often, others with the same hardware without problems, post in those same threads saying they dont have problems. So I strongly disagree with your %1 'guesstimate'.
Crucial_Katana wrote:Have you tried running the generic modules consistently for a few months,
YES!! I have read several times that people with Ballistix issues have gotten fed up after several RMA's, bought another brand of memory, and *poof* problem solved. This is alot more reliable than guessing that 5 RMA's couldnt be all bad. I'll tell you why in a minute.
Crucial_Katana wrote:Some motherboards are rather picky about the RAM sticks and/or its timings/subtimings.
True, and this is not something I would quickly overlook, but keep the scope of the problems. They are too widespread to use this attribute.
Crucial_Katana wrote:Off the top of my head, and since you asked, with the information you have given me about your system the first explanation that comes to mind is that the modules over-heat, and over time that kills them. You said you added active cooling to the RAM, what kind of cooling specifically? What temperature is the RAM running? Not your CPU or your ambient case temperature, but the actual modules themselves.
So, its back to blame the customer. The easy thing right? Truth is, even sensitive chips should handle 71C without frying. Thats about 160F for people still using the Imperial system. Basically the rule of thumb is: If you cant hold your finger on it for 2 seconds, its too hot. I have been through this many, many times with people, and I'm the kind of geek that gets his jollies off by reading engineering datasheets, schematics, and help files. Now I do agree with you 'heat kills'. In fact heat and dust (which contributed to more heat from poor cooling) are the number one and two killers of semiconducters.
Crucial_Katana wrote:The purpose of the heat spreader on a Ballistix module is to spread the heat evenly over the part, which prevents constant "hot spots" that can damage components.
Poorly stated. The purpose of a heatsink (spreader if you prefer) is to distribute heat from the device or part, into the air. It is a thermal exchange.
Crucial_Katana wrote:The heat then radiates away from the heat spreader, and if two modules are close together pockets of heat can get stuck between them and over time cause the memory to fail.
Rarely possible in some configurations. If you had no fans, and the case was sealed. Maybe. but there is this little problem with your 'theory' called convection. Convection is simply put: thermal circulation. The effect is seen in liquid, gas, and plasma. The idea tha the heat would pool between the memory sticks is absurd. Of course it would be a little hotter there because that where the heat is coming from!!
Now lets get down to brass tacks. Having been in several manufacturing plants, I have a slight more instight than just a consumer, so let me guess at more than likely whats going on here.
Manufacturers have whats called bining. Some use different language, but heres the skinny. When stuff comes off the line, it gets tested. If it passes, its bin A1. The good stuff. If it fails its, say, bin B1. Now there are several things that happen to 'B' bin items. If your making cookie dough and the cookie doesnt get pressed right they take the reject, and dump them back into the next batch of batter. Basically recycled. But in electronics they only have a few options. You can pay a bunch of guys like me $40 an hour to repair these, but thats not cost effective. You will pay much more for the techs than you will for the return on the product. On large and more complex Items like TV's, automobiles they do this regularly. Another popular way is to sell off the 'B' bins in lots (usually) with the original manufacturers labels removed. They usually come in shrink wrapped pallets that are auctioned or sold off, and the person can do whatever they wish with them. Fix them or sell them. Thats alot of the really cheap stuff you see on that popular auction website (the rest are clones)
But in the case of memory, alot of memory fails only at certain speeds, like mine did. In that case they may just change the SPD from say 1333 to 1066 and repackage it. This is one possibility for the widespread failures, and seems to be a popular theory in this case scenario.
Another likely possibility is the exact opposite. That the chips are overclocked in the first place. There are many examples of this practice over the years. Dont expect any manufacturers to admit to anything like this. The obvious problem with this is of course too much heat for the design specs. But hey, if you can sell the same thing tomorrow for more money than today......
A third possibility is that all the replacement RAM is RMA'd from someone else.99 times out of 100 the manufacturer blames the customer (maybe not to their face), and there have been many instances of manufacturers testing under 'ideal' conditions which differ greatly from real world applications. So they take in some allegedly bad RAM that didnt work right in a real world setting, put it in a memory tester in a chilly room thats 10C, and run a generic test, and call it good. Then it gets repackaged and shipped back out to some poor sap who had an RMA themself. So, after a couple of times of this, and its real easy to blame the customer. Heck, after awhile it makes you second guess yourself.
The soulition is simple, if you see this pattern from a manufacturer, avoid their products, and let others know how you are treated by the company. I have watched for many years as manufacturers slipped deeper, and deeper into lower quality products, and unethical, if not illegal, practices.
Today my Ballistix finally died. It was the RMA'd pair I recieved. You could say well at least you got 9 months out of it. Yeah, I did, but it never ran at advertised speed, and was only stable in compatibility mode. I guess that 26C ambiant was just too hot for it, or maybe the heat pooled up in the 5cm gap between the sticks 
Good luck Micron. I wont be buying anymore of your products, nor any other products that carry your parts and/or name.
Colt
