wiretap wrote on 2021-03-17, 14:06:

Looks neat. 😜

For the tracking number for things coming out of China, I usually use 17track.net -- they cover a lot of shipping companies.

Thanks!😃
JLCPBC also sent me to this site but it can't find the tracking nr. no matter which of the listed shipping companies I choose.
Waiting really isn't my strong suit!😅

Yea sometimes the tracking numbers don't show up. I've had that happen before. With my last order with stencils + PCB's, the tracking number didn't show up until the day before it was at my mailbox.

For my last order, the tracking on the JLCPCB website worked perfectly fine (also EuroPacket):

Once it had arrived at their EU distribution center, they provided a DHL tracking number.

It took 5 days from their facility to my doorstep.

Well, they're still well within the ten day window. Let's hope I just picked a bad day for having my order arrive quickly.😅

I have now ordered a batch of 20 PCBs and all necessary parts. Should be enough to meet the first wave of demand over here in Europe I guess. Just to clarify: I do not intend to make any profit, of course. The price for each GFD will be calculated to cover the cost of parts that I have had for this project, including the first (smaller) batch of Northwind GFDs. I will provide a detailled overview of the cost just like gbeirn did.

@Doornkaat: Regarding the placement of the switch, I'd prefer it to line up with the edge of the CPU card, so that would be Rev. 1.2. 😀

gex85 wrote on 2021-03-17, 20:48:

I have now ordered a batch of 20 PCBs and all necessary parts. Should be enough to meet the first wave of demand over here in Europe I guess. Just to clarify: I do not intend to make any profit, of course. The price for each GFD will be calculated to cover the cost of parts that I have had for this project, including the first (smaller) batch of Northwind GFDs. I will provide a detailled overview of the cost just like gbeirn did.

@Doornkaat: Regarding the placement of the switch, I'd prefer it to line up with the edge of the CPU card, so that would be Rev. 1.2. 😀

If the cost with shipping from you to the UK is cheaper than from gbeirn, I will buy from you. I expect shipping from the EU to be cheaper, but obviously don't yet know the cost to get PCBs shipped to you or the local components cost.

I ordered a batch of 10 aswell. It will be shipped via Global Direct Line. Hopefully it will work as good as EuroPacket. Big thanks to wiretap and gbeirn for engineering and sharing the data with us.

I assembled and tested another 6 and all are working well. I put these up on everyone’s favorite auction site if people feel more comfortable purchasing that way. No link but it’s an easy enough search I think.

4 sold and shipped, 1 reserved, 6 up there and 9 left.

PM sent!

I haven't followed this project much in the past weeks, but it has advanced beautifully !
Kudos to all of you I'm glad that we can finally have goldfingers without having to wait for them to show up considering how simple they are on principle.

Nicely done. I'd seen the header when I cracked open the case of my Athlon recently[1], hadn't realised what it could be used for. I suspect mine won't overclock much (early model and the actual chip is marked as 500MHz) but might give it a try. I got curious about what the other bits of the connector are for and spent a bit of time tracing it out. Most pins on the header only end up going to the actual chip (which I can't find a pinout for) not the slot connector. So I don't know what they're for. But still, could be useful. I've tried to make sure it's correct, but there may be errors. I got confused from time to time, what with the A side of the Slot connector being the B side of the debug header. Though in the end, all the interesting signals are on the A side. Of the Slot.

1Slot A Pin | Header Pin |  Use   |  Use  | Header Pin | Slot A Pin
2      A17        A1      Reset   | Gnd         B1
3     A114        A2      VID[2]  | VID[0]      B2       A112
4     A115        A3      VID[3]  | VID[1]      B3       A113
5     A119        A4      FID[0]  | Gnd         B4
6     A118        A5      FID[1]  | PullUp      B5       CPU_A19 ;Could
7  CPU_D20        A6      Output  | PullUp      B6       CPU_G18 ;be
8  CPU_A20        A7      PullUp  | PullUp      B7       CPU_E19 ;JTAG
9                 A8      Vcore   | Gnd         B8
10   CPU_D6        A9      Clk_Out | Clk_In      B9       CPU_R3 ;HighFreq
11   CPU_C6       A10      Clk_Out | Clk_In     B10       CPU_R2 ;Balanced
12   CPU_C7       A11      PullUp  | PullUp     B11       CPU_G8 ;Terminations
13                A12      Vcore   | Gnd        B12
14  CPU_G17       A13      PullDn  | PullDn     B13       CPU_F18 ;APIC?
15  CPU_D17       A14      PullDn  | PullDn     B14       CPU_C18 ;Maybe?
16                A15      Vcore   | Gnd        B15
17     A117       A16      FID[2]  | PullUp     B16       CPU_P19 ;Links to SMI?
18     A116       A17      FID[3]  | Gnd        B17
19   CPU_T4       A18   BP_FID[2]  | BP_FID[1]  B18       CPU_R6 ;Corrected from
20   CPU_T1       A19   BP_FID[3]  | BP_FID[0]  B19       CPU_T3 ;TomsH'w diag.
21  CPU_B20       A20      PullUp  | Output     B20       CPU_C19 ; (can't even guess)
22
23All pullup/down resistors seem to be 1k, except:
24-What I think are the clock lines are 54 ohm terminated.
25-Pull downs on the VID lines are 100 ohm.
26-Uses marked 'output' have no pull up/down resistors, so I assume are outputs.
27
28There are PCB pads to link B16 on the header to the SMI pin on the Slot connector.
29It's not fitted on mine, but I can see how that'd be useful for a debug header.
30
31Edit: Fixed BP_FID[2] and BP_FID[3], I think Tom's Hardware diagram has them
32the wrong way around, or at least disagrees with 
33https://www.ocinside.de/workshop_en/athlonresistors/

If those are JTAG pins, they could be interesting, if anyone has any kit for that, thought I can't identify which ones are which.

This all helped me get my head around how the goldfinger is working. The FID and VID really have nothing to do with the actual CPU, they're just information to the motherboard. In particular the VID requests a particular voltage based on the AMD spec (page 35):
https://en.wikichip.org/w/images/c/cb/AMD_Ath … _Data_Sheet.pdf
On my KA7 the VID lines are pulled up by 10k resistors to +5V, and the FID lines by 10k resistors to +3.3V. So I guess that's where the 5V and 3.3V requirement for the goldfinger comes from. Though I think TTL levels mean anything above 2V gets read as a '1'.

No idea how any of the goldfinger devices without external power can possibly work reliably. In particular how they can get a VID that's set as '0' (so with a 100 ohm pull down on the CPU PCB) up above 2V so it can be read as '1', when the only voltage on the debug header is Vcore (so around 1.6V). Worst case, it could end up pulling all 4 down, setting Vcore to 2.05V.

[1] This was part of the trying to fix memory errors escapade. Thought it could be a cache issue, so cracked open the case and found mostly daylight between the heatspreader and one of the cache chips. So I fixed that with some thermal pads. Whilst I was at it I added a couple of tantalum caps on the obvious pads on the PCB as that might help stabilise Vcore, although wouldn't help Vcache. All of which led to no change.

Hello community, i am also in favor for one device,if anyone has any of them left please PM me.Back in the days i had a smaller version than this Thank you .

Just a quick update: I have received parts and PCBs for the Ninja Micros clone and assembled the first one. Quickly tested it with my 550 MHz Athlon on my EP-7KXA successfully: At 600, 650 and 700 MHz it will POST (the latter with increased VCore), haven't tried anything higher yet because I have to get back to some "real" work. Will do some more thogrouh testing and share photos later.

Thanks for testing gex85. I have sold a few more on here and eBay I have 12 left currently.

amijim I can send one to you but I am in the US and shipping cost and time may be faster and cheaper from someone in the EU.

Assembled two more (enough soldersmoke for today).
I decided not to use the assembly service at JLCPCB because I found it to be too pricey.
Here are some crappy phone pics:

I have ordered PCBs and parts for 20 devices. A few of them are already promised, but most are still available. Shipping across Europe should be in the 5-8€ range, depending on whether you want tracking or not. I haven't calculated the price for the actual device yet, but it should probably be somewhere around 15€.
I can either ship kits (PCBs and loose parts to assemble yourself) or fully assembled units. For assembled units I will not charge any extra costs, but I would kindly request buyers to make a small donation (like 5-15€) to the German Maritime Search and Rescue Association (DGzRS) in addition to the price of the parts. Anyone interested please reach out to me via PM.

The only thing we need now is a nice 3-D printed case for these beauties!

gbeirn wrote on 2021-03-19, 23:03:

I assembled and tested another 6 and all are working well. I put these up on everyone’s favorite auction site if people feel more comfortable purchasing that way. No link but it’s an easy enough search I think.

4 sold and shipped, 1 reserved, 6 up there and 9 left.

Me and my friend got one each. Thanks again.

I completely forgot to post this:
The PCBs are here. They work! 😁

Tested on an 800MHz K75 running on a Gigabyte GA-7VX.
The CPU doesn't POST with the 10.5x multi but it goes up to 1000MHz and down to 300MHz.
Voltage controls also seem to work fine. I tested 1.55V-1.95V.
Sorry for bad lighting btw.

I attached the Gerber files for the final revision 1.2. It's only 21.3mm tall.
You will only need of course the TE Connectivity AMP Connector 1734037-4 and an 8 pos tri-state dip switch like E-Switch KAT1108E or Diptronics TII-08-V.
Here's what it'll look like:

Just imagine the grey block to look like the switches in the previous pic. 😉
The final revision is not verified yet but it's based on the same schematics, so it shoud work fine.
Edit: I have received the PCBs. Here's the assembled GFD:

It has been verified working. Here's the list of motherboards I tested it on:
Re: Making an AMD "goldfinger" clone ?
If you make your own please let me know what CPU/motherboard combo you used them on and how it worked out for you. 👍
/Edit
I have ordered ten PCBs to verify the design is working. I'm going to have some to spare that I'll sell at cost if anyone is interested and if that's ok with the mods.

If you want any, DM me or post here if you can't DM yet. 😀

Nice. Thanks for sharing the mini design. That's a super awesome budget model that should be easy for most people to assemble.