Merged Topics on Ceramic CPU Recovery with Yields

I see a couple different types of CPU's for sale from 486 to gold topped Pentium Pro's. I realize these all have different amounts of gold but does anyone have any rough numbers for amount of gold per processor.

As an example, does the average Pentium Pro Gold topped processor have $8 worth of gold. And does the average Pentium 60 to 200mhz (non-gold tops) have $2 worth of gold, etc.

Also, same for memory. If you get 10 lbs of older memory are there any general rules. Such as 10 lbs of gold tipped older 4mb/8mb/16mb/32mb types of memory would yield 1/2 ounce of gold.

The problem I believe a lot of newbies (including myself) have is you don't know how much gold content there really is.. So people guess and run up ebay auctions and related reclaims. And something that is really worth $50 gets ran up to $100 because no one has a clue on how what they are buying.

Gold Digger,

Welcome to the forum and thank you for the post.

This is an excellent question, but not one that is easily answered. The best method is to purchase each type of scrap and accurately weigh the input scrap verses the output gold. I keep a spreadsheet of the various items I process to help me with my purchasing decisions. This questions answer is further complicated by several factors including who processed the scrap (how well they did at cleaning the scrap), what process you use to recover the gold, and how much of a scrap items weight can actually be processed.

As an example:

If I buy 160 slot 1 processors and trim the edges (fingers) myself, I know they are free of excess board material and surface mounted devcies (SMD's). They will consistantly yield 1 oz of fingers per 10 cpus or 1 pound of fingers. From this pound of fingers I can accurately say that if I process them with acid-peroxide I'll get 2.5 grams of gold. If someone else trims the fingers with excess board material or if you use a different method of recovery the yields may vary considerablly.

I hope this helps,

Steve

Regarding trimming those 132 pin computer memory sticks I use the following method which took me several hours to figure out. I was first trying to use clippers, then heavy duty pliers to bend and break off the gold tipped memory edges. This was slow and labor intensive.

Now I just use a big heavy vise that I got at harbor freight for $30. It's jaws are 6" across (match the length of these memory sticks). I place the gold tipped edges into the vise and tighten the vise. I then take a heavy rubber hammer and with 3 or 4 swings these sticks separately nicely right where the clamp is located. I found this method to be just as good as trimming (o.k. 95% as accurate) but it's much faster and easier..

I use a dust mask for this technique as I can see junk coming off these memory sticks when I hit them with a hammer.

A better technique would be to use a tabletop metal shear but these cost $400+. And it may not be that much faster or accurate than my technique using a $5 hammer and $30 vise.

Does anyone else have some good feedback on techniques they are using to quickly separate the gold parts from the boards?

The method I currently use, though time consuming, pays great dividends in clean product and accurate assesment of yields per pound. I use a scroll saw, also known as a band saw. It's takes some geting used to and your have to use the right blade for the job, but the fingers are of impeccable quality. These are the highest yield fingers you can find.

To cut to the chase I've done some experimenting on direct chemical removal of the gold fingers edges only from the memory stick and it's showing great promise. I'll get some photos posted of the items mention herein later tonight.

Steve

does anyone know anything about the older non-gold memory? what is it plated with??? I have about 20 lbs of it that i'm sitting on. It's non-magnetic, but i have no idea it it's silver, stainless steel, or some other base metal...

It may be silver or not... You sould test it.

Hi Gold digger:

I am kinda like you, some what confused. I am on ebay quite a bit and see folks bidding several dollars for pentium pro chips. I have done many pentium pro chips in the past and never took the time to really get serious about the exact gold content in a pentium pro cpu chip.

Well I just finished two unoffical tests on refining pentium pro cpu chips. I crushed two (2) pentium pro chips very fine and ran them through an Aqua Regia of Nitrate soda and hydrocloric. I then neutrilized the nitric acid to .60 PH and then precipitated with Sodium Met. Let one batch of two sit for eight hours and then decanted off liquid and washed the mud with distilled water and the aqua ammonia then rewashed again for three times with distilled water, dried mud on hot plate and then put it in my oven. Cooked for 40 minutes and then dumped in a sulphric acid to knock off the flux. I also used my special homemade flux to make sure I burn off most unwanted metals. Gold comes out pretty good. The nugget came out at .7 grams.

I then repeated the same test again, except I let this batch sit for 24 hours before decanting, I then poured off the settled gold and then I ran the entire spent aqua regia through my Buchner vacumm filter, using a fine filter. I dried the filter and burned it to ashes and then dried the gold using the above technique. I then melted the gold with the burned filter paper ashes and beleive it or not I got a nugget of .7 Grams.

My calculation is gold at $661.00 per oz. I get 98% spot price for all my gold. This represents about $14.58 per nugget. or $7.29 per chip. I dont understand why some of these folks are paying 5 to 8 dollars per chip. They must know something that I dont.

Catfish

catfish said:

I crushed two (2) pentium pro chips very fine and ran them through an Aqua Regia of Nitrate soda and hydrocloric. I then neutrilized the nitric acid to .60 PH and then precipitated with Sodium Met. Let one batch of two sit for eight hours and then decanted off liquid and washed the mud with distilled water and the aqua ammonia then rewashed again for three times with distilled water, dried mud on hot plate and then put it in my oven. Cooked for 40 minutes and then dumped in a sulphric acid to knock off the flux. I also used my special homemade flux to make sure I burn off most unwanted metals. Gold comes out pretty good. The nugget came out at .7 grams.

Click to expand...

Wow! I need to adopt your technique! I've just been breaking up chips with a hammer and running them through aqua regia. I think I'm averaging about 10 grams recovered per 100 chips. Works out to a tad over $2.00 per chip. I get a bit more by reverse electro-plating the gold plated squares.

There must be a lot of gold concentrated in the silicon matrix that I'm not getting to with the old "smack with a hammer" trick!

How are you fine crushing your chips Catfish?

Hi Tarvis:

Plesae note these were Pentium Pro CPU chips. The 386 and 486 regular ceramic chips has much much less gold than these chips do. In fact you would be very lucky to get 1.0% gold per weight of 386 & 486 Caramic cpus. If you refined the crushed chips and then ran the gold tab through your cell and recovered the gold plate material. 386 & 486 Ceramics and Pentium Pros are like comparing apples and oranges.

Catfish

PS I built my crusher. A 6" long piece of 2" diameter galvanized pipe with a cap srewed on one end. then I took a 10" piece of 1.25 galvanized pipe and placed a cap on one end and filled it full of lead, then attached a cap on the other end. You may want to use electical tape on the large piece of pipe to keep the blisters down on your hands. This method works much better tha the hamer crushing and you dont loose any of the gold.

catfish said:

Hi Gold digger:

Well I just finished two unoffical tests on refining pentium pro cpu chips. I crushed two (2) pentium pro chips very fine and ran them through

I then repeated the same test again, except I let this batch sit for 24 hours before decanting,

My calculation is gold at $661.00 per oz. I get 98% spot price for all my gold. This represents about $14.58 per nugget. or $7.29 per chip. I dont understand why some of these folks are paying 5 to 8 dollars per chip. They must know something that I dont. Catfish

Click to expand...

Yes, it appears others do know something--it appears gold yield varies by manufacturer.

I just had my lab assistant repeat your first experiment. We do every quant testing 3 three times around here.

We used pentium pro chips...2 AMD, the 2 INTEL per test. Crushed all to minus 200 after removing metal. Then recombined metal with ground silica for the final digestions.

The AMD ones yielded 1.1479 grams +/- .001% in the 3 tests.

The Intel's yielded slightly more at 1.2873 grams +/- .001% in the 3 tests.

Then they ran ICP on the remaining filtrate and noted the recovery rates of 98.8% on each type.

Interesting experiment. Thanks for the "intell"
It would appear your processing is leaving a statistically significant amount, especially if calculated in dollars and cents per at your $661 price of gold.

a man named Sue

Sue said:

We used pentium pro chips...2 AMD, the 2 INTEL per test

Click to expand...

The AMD ones yielded 1.1479 grams +/- .001% in the 3 tests.

Click to expand...

Where did you find 6 'AMD Pentium Pros'? A pentium pro cpu is a socket 8 device and to the best of my knowledge AMD never produced any socket 8 devices.

You have a very talented staff indeed to recover so much gold from six cpus that don't even exist.

Steve

Chipset Models - Today there are many chipset models in the marketplace. The most popular for mainstream desktop computers are Intel's 810, BX, LX, and ZX. There are also "third party" chipsets available from Acer Labs (ALi), Silicon Integrated Systems (SiS), and VIA Technologies (VIA). The latter are quite similar to their Intel counterparts but may add features not available in the Intel chipsets. The third party chipsets may also support non-Intel processors (like those from AMD and others that have a 100MHz data bus and use the "Socket 7" processor-to-motherboard socket). Intel also produces chipsets that support dual processors. [At the time this FAQ was prepared, only Intel was producing chipsets which support multiple processors.] The following are brief descriptions of the key features of the most popular primary Intel and third party chipsets as of the date this FAQ was prepared (mid-June '99).

Click to expand...

Socket 7 instead of 8. Big deal. Basically, Intel hasn't done anything that AMD hasn't come along and ripped off or tweaked and issued their own "version." It's called capitalism.

I'll check later for more accolades for my fine staff. :lol:

a man named Sue

The biggest difference is the dual core nature of the socket 8 processor, meaning a greater number of hair fine gold wires connecting the cores to the legs. Not to mention that a socket 7 only has 321 gold plated pins versus the 387 gold plated pins in the socket 8. I think that when you are quoting accuracy to the nearest ten thousandths or better of a gram in the yields that these values are a significant scientific differences.

One other important figure you and the original poster left out of the yield data was the purity of the gold produced. We all know that 1 gram of 100% gold is a more valuable yield than 1.5 grams of 50% gold.

These omissions make it impossible to scientifically confirm the yield of a pentium pro cpu by either method.

Steve

Well, geez whiz, you now admit AMD made pentium pro type chips, so I guess that means we're making progress.

As far as incomplete according to you, Yes, I did forget to type in the gold results purity was X-Ray 9995 on our tests, for both types. Did you really expect my guys to come up with less purity than that. Afterall, it is our livelihood. I'll stick with my lab guys. They're some of the best in the business.

You have a 3rd set of results to contribute or are you going to continue to dispute my results.

I look forward to your test results on a similar pentium pro challenge or anyone elses for that matter.

The original question was why some ebay bidders are paying more than $5-$8 per chip. I tried to answer that question by duplicating the test performed by the newbie with the data they provided.

a man named Sue

There is no such thing as an AMD Pentium Pro, only Intel Pentium Pro.

It's pretty obvious you don't know the difference between a Pentium Pro and a Pentium I or II so we will leave it at that. Truthfully, I'm more impressed with the yields you quoted if they came from pentium one (socket 7) class cpus.

As for my results I don't have a team of scientist to process my gold for me, but I know your results are NOT from Intel Pentium Pros (at least the AMD figures you quoted) which was the question asked.

catfish said:

I crushed two (2) pentium pro chips very fine and ran them through an Aqua Regia of Nitrate soda and hydrocloric.

Click to expand...

You are comparing apples and oranges when you compare socket 7 cpus to socket 8 cpus with respect to the gold yields of each.

Regardless of the type of cpu processed or the yields you can attain from them in the lab, it doesn't increase the value of the cpus for a home experimenter. As hobbyist we must go on the yields we can obtain, not what can be attained in a lab somewhere.

As you stated

Sue said:

--it appears gold yield varies by manufacturer.

Click to expand...

The value is derived from this yield, which is different for each of us.


Steve

The relationship between the 7 socket and the 8 socket number of legs and wires is a qualitative one, or to put it another way, is a mathematical difference.

The difference between the results of my lab guys and the two types of chips tested is also a mathematical one.

There ARE AMD pentium pro "class" chips 100MHZ, which I've already proven, and which my guys say they tested. They lie to me, they get fired. It's real simple. We have millions riding on correct answers around here. This was a simple experiment to offer an answer an earlier question.

I've looked in the ebay ads this morning. The chip sellers offer no details. Are you trying to convince this forum you only process Intel Stamped socket 8 Pentium Pro class chips for gold yield?

The major difference besides math and how many pins in a 7 socket vs an 8 socket is dollars of pure gold for sale per chip processed.

My guys got more dollars per chip than the original questioner. That is the significant part. i would have thought this forum would be happy to learn there is more gold available than they are apparently harvesting. I only offered it so that it might set the guys here to looking for how to get at that unharvested gold more effeciently.

You are correct about one thing, however. I don't care about clock speed, cpu manufacturer, etc. I do care about monetary results from test showing interesting differences and theorizing about why.

One reason for the difference I suspect is my guys took their ground up chips to minus 200 before adding back in the metal pins, and the original questioner took his to "fine" whatever that definition is to him.

Isn't it all about money in the wallet after the acid is disposed of?

a man named Sue

First of all I am very impressed with the results that "Man named Sue" came up with in a lab. enviorment as to the "exact" amount of gold is in a pentium pro chip.. I have gained quite an insight about the depth and vast knowledge of the contributing members of this forum.

I envy you on having a first class laboratory at your disposal to make these kinds of tests. Unfortunately I work out of my small electronics repair shop and only work with gold and silver as hobby.

I want to quote Lazersteve if I may.''This forum is all about learning for all of us and having a good time while doing so. There are no stupid questions and each have a unique perspective on the subject. That is what the driving force of this forum is.

I am still in mild shock after learning just how much knowledge and experience you folks bring to this forum. Keep up the good work and continue to create good healthy dialog.

My hats off to both Steve and Man named sue.

Catfish

catfish,

I work in a garage also. Where Sue has an formal education in Metallurgy, mine is in Electronics Technology. I'm a humble hobbyist as you are when it comes to my gold 'refinery'.

Sue's results are astounding to say the least. By his descriptions of the bus speed and chip types, he is referring to newer (relative to Pentium Pros) generation (100 Mhz FSB) ceramic pentium cpus that produce with two chips what I had believed was only possible with a pound of these cpus! More over, as he stated, mathematically speaking the older Pentium Pro (larger, dual core, 60-66 MHz FSB with external gold plate) chips must produce way more than what you or he has suggested. I will indeed have to do some test of my own.

I have a stockpile of cpus and don't really mess with them except for experiments in reverse assembling them. I like to carefully remove the cores with the gold wires still intact, sort of like artwork. Now that I know this information about the increased yields of gold, I may process the stock I have on hand. I will need a good ball mill first.

Ultimately, the value any one experimenter can derive from a cpu varies greatly with technique and equipment. The only advantage I see to home refiners is that we don't have to pay assisants, buy expensive test equipment , and pay the bills to keep a full refinery up and running. In the long run our lower yields of gold may cost us less to process than the big guys higher yield when looked at on a overall cost per gram to produce.

This has been one superb post topic. I have had a great time discussing this interesting and enlightening topic with all of you.

Steve

Pretty rowdy in here. :lol:

Do we all agree that it is not worth buying any cpu for scrap unless it's cost is under 1.50 U.S. Don't forget about the cost of chemicals and labor, even if we are doing this as a hobby it seems to me that we still want to come out doing better than just breaking even.




-Bill

I am thinking about investing in a decent ball mill for exactly this reason- mashing cpu's into a very fine dust and then processing in AR for the gold values. Is this the very best method? I've gleaned that it is from the debate above.

Well done gents. Great exchange of incredibly detailed info!

Fever