Computer Scrap Pentium Pro CPUs - VIDEO

[kadriver]

Here's a new video that I finished this morning.

I've done three batches of these Pentium Pro CPUs - each batch with four CPUs.

The first was a single refining and I got a dirty gold bead that weighed 1.6 grams (from four Pentium Pros) threw it in with a batch of karat scrap.

The second batch was also a single refining and it yielded a dirty gold bead that weighed 1.3 grams (from four Pentium Pros) also in with the karat scrap.

In this video I double refined the gold and I only got 1.0 grams from the four Pentium Pros - I listed it for sale on eBay.

It took just over 13 hours from start to finish includimg the time to shoot and edit the video;

https://youtu.be/7OKTgdXbcEM

As always, comments and suggestions are welcomed.

Thanks for looking!

kadriver

 

[Anonymous]

Hi there ka. Merry Christmas 8)

Great video- thanks for doing this.

Did you run the comparison of methods we discussed a while ago?

A couple of things to note though- your yield for the clean button of 1g is very light so you are losing gold, (these things yield .35g minimum each,) and there are a few things that you can improve upon in the process you have in the video.

Jon

 

[NobleMetalWorks]

Hi there ka. Merry Christmas 8)

A couple of things to note though- your yield for the clean button of 1g is very light so you are losing gold, (these things yield .35g minimum each,) and there are a few things that you can improve upon in the process you have in the video.

Jon

Actually, Pentium Pros can run anywhere from .33-.35 grams each on the high end. I believe it is the L-cache that makes the difference.

I have literally run hundreds upon hundreds of PPs. I received over 500 all at one time as payment t to process an ore concentrate alone. I have used many different extraction and recovery methods. I have extracted PP gold using BDG to Hydroquinone and more. I have never on any single or multiple run found more than .35 per CPU, and have averaged at worst .3 grams. But these were all double sided gold plated. I removed the heat spreaders to process separately and pins as well. There are Pentium Pros that yield less, these don't include the gold cap on top (heat spreader) I have never processed these types. Now days, if I come across Pentium Pros, I put them up for auction on eBay, you make more doing that instead of processing.

I feel where most people have problems is in not removing base metals prior to processing. If Kovar alloy is dissolved with gold, for example, it makes recovering gold more difficult. If you do process the entire chip in AR all at the same time, I would suggest using BDG as it does a wonderful job extracting Au from dirty solutions incredibly well, and if cleaning the BDG with HCl proof to precipitating with Oxalic Acid, you can obtain at least 99.99 purity and perhaps more with only one refining.

Great video, Married, you do an awesome job making instructional videos.

Scott

 

[Anonymous]

Hi Scott

That's fine, and by your own figures the yield is therefore light. 0.25g is still too low and you have alluded to that yourself, as the PPros here are the gold capped ones.

I cannot agree with your comments regarding the kovar though as I too have processed a lot of these and similar chips and not experienced any issues at all retrieving the gold.

The "dirty drops" that had between 1.4g and 1.6g were I believe closer to the true yield and just needed cleaning up with one of the cleaning methods posted on the forum.

Jon

 

[Eamonn]

Hi Guys

Do you know where the missing gold is more likely to be.
In kadriver's first filtrate or still with the cpus.

Eamonn

 

[NobleMetalWorks]

Hi Guys

Do you know where the missing gold is more likely to be.
In kadriver's first filtrate or still with the cpus.

Eamonn

If if it there, and it isn't an issue with the actual gold on the CPUs (I have seen Pentium Pros with gold paint on the top cap) then other metals are preventing gold from precipitating out of solution properly. Kadriver, from what I have seen on the forum of his posts, does an excellent job recovering precious metals from many different types of scrap. I have a hard time believing that he will not be recovering any other potential values from his solutions. It might be it comes out in his stock pot, or he might use a different method to recover the rest if it exists. But it very well might not exist. It's difficult to say because Pentium Pros are so old, and sometimes subjected to whatever external influences that cause they to loose gold plating. Remember, these CPUs were manufactured in 1995, that's just over 20 years ago. During that time, it's not unreasonable to suggest that some of the gold might have been worn/rubbed off, or maybe other things. It's high purity gold, which is very soft and can be rubbed off over time without much effort. You have seen karat gold rings that become thinner over time, this is the same process I am speaking of.

Scott

 

[Anonymous]

Actually, Pentium Pros can run anywhere from .33-.35 grams each on the high end. I believe it is the L-cache that makes the difference.

I have literally run hundreds upon hundreds of PPs. I received over 500 all at one time as payment t to process an ore concentrate alone. I have used many different extraction and recovery methods. I have extracted PP gold using BDG to Hydroquinone and more. I have never on any single or multiple run found more than .35 per CPU, and have averaged at worst .3 grams. But these were all double sided gold plated. I removed the heat spreaders to process separately and pins as well. There are Pentium Pros that yield less, these don't include the gold cap on top (heat spreader) I have never processed these types.

Scott

Hang on Scott.

Which is it? The long post you put up or the more recent one? I think your first post is correct and has far more validity and it also backs up my own experience from running a wet process. I have not experienced other metals preventing gold being precipitated and reading Ka's post he hasn't either if you look at it logically.

He added an extra step in for the lower yield process, a step that wasn't in the other two higher yielding runs.

 

[NobleMetalWorks]

Actually, Pentium Pros can run anywhere from .33-.35 grams each on the high end. I believe it is the L-cache that makes the difference.

I have literally run hundreds upon hundreds of PPs. I received over 500 all at one time as payment t to process an ore concentrate alone. I have used many different extraction and recovery methods. I have extracted PP gold using BDG to Hydroquinone and more. I have never on any single or multiple run found more than .35 per CPU, and have averaged at worst .3 grams. But these were all double sided gold plated. I removed the heat spreaders to process separately and pins as well. There are Pentium Pros that yield less, these don't include the gold cap on top (heat spreader) I have never processed these types.

Scott

Hang on Scott.

Which is it? The long post you put up or the more recent one? I think your first post is correct and has far more validity and it also backs up my own experience from running a wet process. I have not experienced other metals preventing gold being precipitated and reading Ka's post he hasn't either if you look at it logically.

He added an extra step in for the lower yield process, a step that wasn't in the other two higher yielding runs.

If you read his post, he states that he produced a dirty gold result, and then refined a second time to high purity. That is the difference, it's not that he used to different processes, it's that he double refined to higher purity. So the other two runs were not higher yields, Kadriver simply refined a second time to remove the impurities that were dragged down during the precipitation of gold.

I don't buy material I know has been altered. I also do not process CPUs whole, I always remove the pins, the caps, etc. When I first started processing, I did process CPUs whole, by breaking them up (Not Pentium Pros) and realized then that doing so reduces the yields when precipitating and required more work to recover the rest of the values. In the Pentium Pros I have seen where the gold was stripped from the top cap, and then a gold like paint added, I have not purchased them.

Anytime you have a dirty solution, you will either not recover all values the first time, or you will have substantial drag down. Personally, I prefer to not create dirty solutions as they are never as successful as parting the material into their component parts, and processing independently. When I first encountered this problem, I decided to do tests with double gold plated Pentiums. When compared you can see a difference in the yields.

As well, you can literally check the manufacturing information and discover exactly how much gold was used in plating, or the tolerances that are acceptable. Then you might compare that information with what you are able to recover. However, age, usage, and other factors might change the makeup of individual chips. I am sure I have processed a lot of chips like this, and would be the reason why I get .33 on average and why when running smaller lots, I average .35. The difference may be in the L-Cache, or also might be because of usage, wear, or other factors.

I also have a lot of respect for the work, and videos Kadriver posts. In reading his process I am sure he probably realizes better than those of us speculating, where he might change his process to generate higher yields, if the gold is there. But I want to be fair in what I post as well, and careful, and leave room for other things that might have caused a lower than average yield.

If you have ever attempting to mill anything gold plated, you will see that some of the soft high purity gold rubs off on your milling material, like steel balls, etc.

 

[g_axelsson]

Remember, these CPUs were manufactured in 1995, that's just over 20 years ago. During that time, it's not unreasonable to suggest that some of the gold might have been worn/rubbed off, or maybe other things. It's high purity gold, which is very soft and can be rubbed off over time without much effort. You have seen karat gold rings that become thinner over time, this is the same process I am speaking of.

Scott

Old CPU:s doesn't have the wear that old jewellery has. Normally you only stick it in the computer once and pull it out once and if you look at the video the print on top of the CPU didn't have a scratch even. So I don't believe that wear would lower the gold content measurable unless it's been tampered with.

One possible problem that's been discussed before is the heat spreader. It's a copper tungsten alloy and I tried to see from the video if it was fully dissolved but I couldn't tell.If there is free metal left then it could have cemented some of the gold. The tungsten copper alloy should dissolve in hot aqua regia but it could take some time.

Another variable could be the size of the dies. A larger die takes more gold to solder in place in the capsule. Without knowing the s-spec codes of the CPU:s it's hard to say what went into each batch. According to data the PPro was manufactured both in 0.35 and 0.6 micron process. Smaller geometry gives smaller dies. Differing cache sizes (256k, 512k, 1M) could also affect the gold usage in the chip.

This is of course speculations and I'm sure kadriver have better control of where any possible gold could be hiding.

Göran

 

[Anonymous]

Anytime you have a dirty solution, you will either not recover all values the first time, or you will have substantial drag down. Personally, I prefer to not create dirty solutions as they are never as successful as parting the material into their component parts, and processing independently. When I first encountered this problem, I decided to do tests with double gold plated Pentiums. When compared you can see a difference in the yields.

I would partially disagree with this. You can very easily recover all the values the first time from a dirty solution however your drag down can be substantial as you rightly pointed out. You can still quite happily achieve >95% purity even from such a solution using either SMB or Copperas. As a note I prefer to use Copperas for dirty solutions.

As well, you can literally check the manufacturing information and discover exactly how much gold was used in plating, or the tolerances that are acceptable. Then you might compare that information with what you are able to recover. However, age, usage, and other factors might change the makeup of individual chips. I am sure I have processed a lot of chips like this, and would be the reason why I get .33 on average and why when running smaller lots, I average .35. The difference may be in the L-Cache, or also might be because of usage, wear, or other factors.

As you point out yourself, your minimum average yield is 0.33g per chip from processing larger batches of these processors therefore Ka's yield IS low no matter how we look at it. You have provided your own data which I agree with regarding the yields and given the quantity that we have processed between us there is no reason to suggest that Ka has some randomly lower yielding processors. Especially as we can see how clean his product is from his video. Age and usage have absolutely no impact on these figures. Usage involves no wear on this kind of product because it is simply put into the processors slot and left there.

I also have a lot of respect for the work, and videos Kadriver posts .

As do I. It takes a lot of gumption to put up a video and stick your head over the parapet and open yourself up to potential criticism both constructive and otherwise. From my dealings with Ka, he's open to constructive criticism and guidance and wouldn't be upset just because he is told that he could improve a couple of things. That's the double edged sword involved with posting up processes, and if something can be bettered, then it should be perfectly alright to offer that help.

In reading his process I am sure he probably realizes better than those of us speculating, where he might change his process to generate higher yields, if the gold is there. But I want to be fair in what I post as well, and careful, and leave room for other things that might have caused a lower than average yield.

Put simply we are not speculating. Both you and I and many other forum members have yield data for these processers. You have publicly stated yours on more than one occasion on this thread (which I broadly concur with) therefore the reported yields are low and I think we can offer help to prevent that.

If you have ever attempting to mill anything gold plated, you will see that some of the soft high purity gold rubs off on your milling material, like steel balls, etc.

This is not relevant to these processors and would not account for the difference in yields reported.

Personally I believe that first "drop" is always treated as recovery, not refining. You can clean up gold from that first recovery very well and to be honest it should be cleaned before putting it into a proper refine. I don't use filter papers for doing this because it presents an opportunity for powder to be retained between the processes and potentially cause a headache when reporting yields. Cleaning the gold between recovery and refining has the added advantage of making that refining step cleaner and less likely to present base metals.

I applaud Ka's use of a Buchner vacuum setup. Put simply it is definitely the way to go if you are looking at doing this "properly" and nothing Ka has shown us has been performed in a slipshod manner. He quite clearly takes this seriously and has put the effort in to make it as "pro" as possible.

You've got three yields - two of which are very similar and one which is much lower than the others. This isn't due to drag down or not getting the gold out of a dirty solution, or wear on the chips, or differences in plating. Scott I am sure that on watching the video you can see where the problem lies and it is rectifiable.

What do you think?

Jon

 

[nickvc]

One point I picked up from the excellent tutorial video was that in the first dissolution the AR was still reacting when removed from the heat, brown fumes always indicate to me that dissolution is still occurring and I only ever remove from the heat when no more brown fumes are been seen and that is after extra acid additions and more heating, I'd rather have excess nitric than trapped values.
Another point is that personally I wouldn't have filtered that first gold drop just rinsed in the beaker and then re dissolved and in honesty I would have done the same with the second drop only then either dried in the beaker or washed into a drying vessel.
All small points but with small amounts it's very easy to lose big percentages on little things.

 

[maynman1751]

in the first dissolution the AR was still reacting when removed from the heat

Nick, I also noticed that. It may have been the tungsten still reacting, but it could still be tying up some of the gold from dissolution. There was considerable fumes and reaction present. I agree that further addition of nitric and heat should have been added to complete the process.

 

[Anonymous]

Hi Scott

Hope you had a great New Year. Have you got time to take a read of my post of 31st December and let me know your thoughts?

Jon