Well, the use of the adapter is certainly essential in motherboards with linear voltage regulators, I think that in that case selecting 6X or 400MHz, it would be quite hot, perhaps hot, the problem would not exist with switching regulators, from experience the linear ones without adequate cooling can create problems already at 200MHz, while with switching it goes smoothly at 450MHz and probably even beyond.
I don't know what MB you use, but a k6-450 can be easily installed in almost all MB S7 (not super), if they don't support the required 2.2V, several can generate voltages lower than 2.8V, usually 2.4V or 2.5V , but in case the minimum was 2.8V a simple modification could be made to lower the voltage to about 2.2V, it's not a modification I've ever done but I believe it is possible to get that voltage using one or two diodes.
Unfortunately I am not a technician but I know that these changes or other similar ones have been made in the past, unfortunately not all cards are the same, so the change must be made specifically for the card, afterwards if you have the right voltage, you have to understand if the BIOS recognizes the CPU correctly, if it is not recognized, you need to make a change to the BIOS.
Unfortunately I can no longer find the sites where there were the modifications for the socket7 VCore.

snufkin wrote on 2021-12-20, 20:46:

Ah, missed that, I'd already downloaded a pinout that had them labeled BF2, BF1 and BF0. […]
Show full quote

Sphere478 wrote on 2021-12-20, 20:39:

Nc is the label for the right most bf pin on the schematic I uploaded. […]
Show full quote

Nc is the label for the right most bf pin on the schematic I uploaded.

says 2x/6x is BF[2..0] = 110b.

I don’t understand

Edit, ohh! So we want bf 0,1 on and bf2 off?

Ah, missed that, I'd already downloaded a pinout that had them labeled BF2, BF1 and BF0.

Not sure how to describe this, but when giving the setting of a bus of several lines, like BF[2..0], the setting will be in the same order as given in the square brackets.
So BF[2..0]=abc means BF2=a, BF1=b and BF0=c. So BF[2..0]=110b (b for binary, would also make sense to give it in octal/hexdecimal) mean BF2 and BF1 set to '1' and BF0 set to '0'.

Okay, so then this should be be 6x 2.020v
That should be bf0 off, and bf1 and bf2 on

Apparently it’s sometimes a issue getting cpus back out of these things. Keep in mind. Might be a good idea to not fully push it down or file a radius on the socket edge before installing cpu or something.

Another idea: the fet under the cpu might be able to use one of those rubber thermal pads like on gpu ram to make contact with the cpu for better cooling

Let's see. SW[1..3] = BF[2..0] (from your earlier tests).
SW[4..8] = VID[0..4] (using VID[0..4] from the 5057 datasheet).

So SW[1..8] should be '11000000'b for x2/x6 @ 2.05V (officially, from the datasheet). I think. I frequently get things back to front, so best to double check.

Sphere478 wrote on 2021-12-20, 21:06:

Apparently it’s sometimes a issue getting cpus back out of these things. Keep in mind. Might be a good idea to not fully push it down or file a radius on the socket edge before installing cpu or something.

Another idea: the fet under the cpu might be able to use one of those rubber thermal pads like on gpu ram to make contact with the cpu for better cooling

Give it a harness of dental floss, for the yoink.... if something like this catches you unprepared though, ISA/PCI slot blanking plates can be used either side (onboard SMDs protected appropriately) to raise opposite sides at the same time. You can probably tape 4 to an aerosol can, find a ring or a few elastics to close the "claws" and make a crude chip puller like that.

snufkin wrote on 2021-12-20, 21:09:

Let's see. SW[1..3] = BF[2..0] (from your earlier tests).
SW[4..8] = VID[0..4] (using VID[0..4] from the 5057 datasheet).

So SW[1..8] should be '11000000'b for x2/x6 @ 2.05V (officially, from the datasheet). I think. I frequently get things back to front, so best to double check.

I’m just gonna have to plug this in to be sure, but in double checking What you said I came up with 10ff 2on 3on 4off 5off 6off 7off 8off

It’s possible the confusion got the better of me though.

At least the voltage is figured out 🤣. Multipliers can be guessed at.

It seems that configuration results in 300 mhz so somehow It is set at 4.5x🤔

But good news! It works! Lol

That plug on the side of it is like a floppy but not a floppy, can anyone confirm what’s going on there? Worried that the 5v and 12v maybe are supposed to be reversed because the way it’s wired now it looks like it’s regulating the 12v?

Hmm, looks like I got things back to front. When the switch is on that ties that line to '0', when the switch is off then it gets pulled to '1', either through internal pull ups in the 5057, or via those 330 ohm resistors for the BF lines.

Just to check I've got that right now, does setting SW[8..4] to : on,on,off,on,on set the voltage to 1.85V? That's what I think should be VID[4..0]=00100.

If that's right then SW[8..4] all set to off sets the input to the voltage regulator to all '1's, which from the datasheet is 2.0V (pretty close to the 2.02V, but then so would all on, which would be 2.05V).
SW[3..1] set to on,on,off. That's BF0=on='0', BF1=on='0' and BF2=off='1'. So BF[2..0]=100b.

Still doesn't match up though. The table I linked before has 100 as being 2.5x. 4.5x should be 000, which only happens if BF2 is being ignored. 2x should be BF[2..0]=110, so BF2=SW1=1=off, BF1=SW2=1=off, BF0=SW3=0=on.

Switch 3=on, 2=off, 1=off.

snufkin wrote on 2022-01-07, 22:30:

Hmm, looks like I got things back to front. When the switch is on that ties that line to '0', when the switch is off then it ge […]
Show full quote

Hmm, looks like I got things back to front. When the switch is on that ties that line to '0', when the switch is off then it gets pulled to '1', either through internal pull ups in the 5057, or via those 330 ohm resistors for the BF lines.

Just to check I've got that right now, does setting SW[8..4] to : on,on,off,on,on set the voltage to 1.85V? That's what I think should be VID[4..0]=00100.

If that's right then SW[8..4] all set to off sets the input to the voltage regulator to all '1's, which from the datasheet is 2.0V (pretty close to the 2.02V, but then so would all on, which would be 2.05V).
SW[3..1] set to on,on,off. That's BF0=on='0', BF1=on='0' and BF2=off='1'. So BF[2..0]=100b.

Still doesn't match up though. The table I linked before has 100 as being 2.5x. 4.5x should be 000, which only happens if BF2 is being ignored. 2x should be BF[2..0]=110, so BF2=SW1=1=off, BF1=SW2=1=off, BF0=SW3=0=on.

Switch 3=on, 2=off, 1=off.

So which one should I try for 2/6x? You lost me.

I hate having to take the cpu out of these things..

We need a community project for new powerleap style interposers anyway, one that uses a actual socket.

From the K6-2 (is that what you're using?) datasheet it seems the BF[2..0] setting of 110 (so off-off-on on SW1,2,3) is either 2x or 6x depending on the stepping of the CPU. From the datasheet: "The ratio selected is dependent on the stepping of the Model 8. The 2.0x ratio is supported on the Model 8/[7:0], whereas the 6.0x ratio is supported on the Model 8/[F:8]"

Thing that bothers me is how it's getting 4.5x, that doesn't make sense give your last photo showing the switch settings. If you do take the CPU out, could you check if I've got this diagram labelled correctly:

snufkin wrote on 2022-01-08, 17:20:

From the K6-2 (is that what you're using?) datasheet it seems the BF[2..0] setting of 110 (so off-off-on on SW1,2,3) is either 2x or 6x depending on the stepping of the CPU. From the datasheet: "The ratio selected is dependent on the stepping of the Model 8. The 2.0x ratio is supported on the Model 8/[7:0], whereas the 6.0x ratio is supported on the Model 8/[F:8]"

Thing that bothers me is how it's getting 4.5x, that doesn't make sense give your last photo showing the switch settings. If you do take the CPU out, could you check if I've got this diagram labelled correctly:
BF_Pinout.jpg

this is how it is wired

Btw, this also needs to be talked about 🤣

Here is a actual floppy style connector installed on the same pins.

Where is that 12v line going?😑

Anyway, I confirmed the red wire is going to the mosfet, so that’s good. And since I didn’t flip the connector all should be the same as before which did work.

Sphere478 wrote on 2022-01-08, 17:43:

snufkin wrote on 2022-01-08, 17:20:

From the K6-2 (is that what you're using?) datasheet it seems the BF[2..0] setting of 110 (so off-off-on on SW1,2,3) is either 2x or 6x depending on the stepping of the CPU. From the datasheet: "The ratio selected is dependent on the stepping of the Model 8. The 2.0x ratio is supported on the Model 8/[7:0], whereas the 6.0x ratio is supported on the Model 8/[F:8]"

Thing that bothers me is how it's getting 4.5x, that doesn't make sense give your last photo showing the switch settings. If you do take the CPU out, could you check if I've got this diagram labelled correctly:
BF_Pinout.jpg

this is how it is wired

Ok, so those are back to front to what I thought from the earlier measurements. So SW3=BF2, SW2=BF1 and SW1=BF0. That makes things a bit less confusing.

Do you know if the other side of the switch all go to Ground, and if the via in the middle of those 330 ohm resistors goes to something +ve (probably +5V)?

[edit: That +12V probably goes to pin 8 of the RC5057M, it'll be the VccPower input to drive the MOSFET gates.]

I forget if we finally solved this or not but here are the settings for 2x = 6x

1 on
2 off
3 off

Once we got the order of SW->BF correct, that makes sense. The one that doesn't make sense is the 4.5x multiplier. From the photo, you had SW1 off and SW2&3 on. Which (now we know the correct order) should have been 5x, but you got 4.5x.

Maybe your SW1 is stuck on, so BF0 is always connected to ground.

Using the AMD and 5057 datasheets I've put together what I think are the SW settings, would you mind checking a few to see if they're correct:

snufkin wrote on 2022-02-11, 11:43:

Once we got the order of SW->BF correct, that makes sense. The one that doesn't make sense is the 4.5x multiplier. From the ph […]
Show full quote

Once we got the order of SW->BF correct, that makes sense. The one that doesn't make sense is the 4.5x multiplier. From the photo, you had SW1 off and SW2&3 on. Which (now we know the correct order) should have been 5x, but you got 4.5x.

Maybe your SW1 is stuck on, so BF0 is always connected to ground.

Using the AMD and 5057 datasheets I've put together what I think are the SW settings, would you mind checking a few to see if they're correct:
TurboChip.png

Holy crap man, you took the time to make that?! Good job! You are a forum treasure!

I don’t think the switch is stuck as I had to set it to get from 366 to 400

It may be simply the board detecting and displaying wrong. Some bioses are like that.

I will try to confirm this next time I have the device out.

Btw here is where you can measure set voltage with no cpu installed,

Could try multiple bus speeds to see if it's the BIOS "guessing" the speed wrong or the jumper actually not setting what you think it should.

Sphere478 wrote on 2022-02-12, 00:48:

Holy crap man, you took the time to make that?! Good job! You are a forum treasure!

It was just taking the tables from the two data sheets, then adding how that maps on to the switch settings. I was going to write a thing explaining how I think each bit goes to each switch, along with how a '1' is switch 'off', then realised it'd be quicker for me and easier for anyone to understand if I just wrote out the switch settings.

I should say that I'm reasonably sure about them, but not certain. I know it's a pain getting a cpu out and back in, and there's always the risk of bending pins, but if you do get a chance to take the cpu out then as well as checking the Vcore you can also measure the voltage on each of the BF pins and make sure they at least match the binary values in the table.

Socket 5/7/SS7 (Motherboard) Tweaker (Released)

figured this link out mainly from this interposer shown in <this> thread

If it helps anyone else I had to set the jumpers as indicated to 2.7v to get a K6-3 400 working stable, and this was after replacing the caps. No post at the indicated 2.4 setting.