VOGONS


Tillamook

Topic actions

Reply 60 of 76, by Tetrium

User metadata
Rank l33t++
Rank
l33t++

Paper tape is conductive? If it's not, it's very cheap anyway 😉

I might go test this with a regular MMX chip first before I put my Tillamooks to the test.

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 61 of 76, by prophase_j

User metadata
Rank Member
Rank
Member

I have also read the data sheets, and I get the same feeling, as far as bridging the two signals back to the intended pins for desktop motherboards. It will be very interesting to see how that pans out. Let us hope the the BIOS isn't programmed to go simply off the CPUID as Tillamook and disable the L2 anyway.

While that pot of stew simmers, lets us explore the I/O voltage issue. Reading in the data sheet, it says the embedded version should have an I/O voltage of 2.5, unlike the desktop version where it is 3.3v. There is a listed tolerance of 5%, and a stated limit of 3.2v. This issue is compounded when we use a motherboard like the P5A, since the base setting there is a factory over-volted(for stability) 3.5! I have read something about socket adapter with built in voltage regulators, but they certainly must be a rare find. It worries me to have the voltage rammed up so high, as I fear it will lead to a very short lifespan.

"Retro Rocket"
Athlon XP-M 2200+ // Epox 8KTA3
Radeon 9800xt // Voodoo2 SLI
Diamond MX300 // SB AWE64 Gold

Reply 62 of 76, by Mystery

User metadata
Rank Member
Rank
Member

Yes, I suppose the high IO voltage is slowly killing the "desktop" Tillamooks. I'm not sure if there's a reasonable solution to lower the voltage.
Which pins are affected by the IO voltage? Some? All of them? In theory it would be possible to reduce the voltage, but I'm not sure if it's worth the trouble.

Edit: Just prepared the Tillamook for the surgery to reenable the two pins. Problem is, I can't find my conductive silver. So it might be a couple of hours (until I find it) or a few days (if I've got to buy it again). 😉

Last edited by Mystery on 2011-06-12, 06:33. Edited 1 time in total.

::42::

Reply 63 of 76, by prophase_j

User metadata
Rank Member
Rank
Member

What I gathered is that several pins on the processor package supply power to the die. It is likely that we would need to approach this be effecting some sort of hardware mod, on the voltage circuitry itself. I will see if I can't discern the particular component used and find it's datasheet. I know a good deal of electrical theory, but engineering an actual solution is a stretch. It will at least be a start.

"Retro Rocket"
Athlon XP-M 2200+ // Epox 8KTA3
Radeon 9800xt // Voodoo2 SLI
Diamond MX300 // SB AWE64 Gold

Reply 64 of 76, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Altering the motherboard's voltage regulation circuit is not much trouble. I have succeed in the past. Here are some simple steps:

1) Find the 2 voltage regulators on the socket 7 motherboard and note the part numbers.

2) Do a part number search and download the PDF spec sheet.

3) The voltage jumpers on the motherboard typically short or open paths to other surface mount resistors. You can use an ohm-meter to find out which resistor corresponds to what voltage jumper position.

4) These resistors are just used as a reference to the voltage regulator to tell what voltage to apply (Iref). It is not like they are used in a voltage divider configuration to sink current/power; the voltage regulator does all the work. These reference resistors can be found on the spec sheet connection diagram for which a theoretical relation is mentioned to determine your desired regulated voltage as based on the reference resistor.

5) Either you can calculate the theoretical resistance from the spec sheet's eqn, or you can use more of a "technician's approach." I usually solve it theoretically, then take the technician's approach (below, in 6).

6) Say you want to modify the 3.3 V setting. Find the 3.3V jumper, then find it's corresponding SMD resistor. Carefully desolder the SMD resistor and solder in place two wire leads. Aligator clip a variable resistor to the wire leads that has a range from, say, 2Kohm above and 2Kohm below the resistance of the resistor you removed.

7) Wire clicp your volt-meter to Vcc/GND or to the specific voltage regulator output pins to measure the voltage. Tune the variable resistor until you hit the desired voltage, say, 2.2V.

8) Remove the variable resistor and measure its resistence.

9) Now, you can either ghetto-solder a larger resistor onto these wire leads, of if you want to be fancy (like me!) you can source the exact size and resistance of an SMD resistor you need.

Ultimate 486 Benchmark | Ultimate 686 Benchmark | Cyrix 5x86 Enhancements | 486 Overkill Graphics | Worlds Fastest 486

Reply 65 of 76, by Tetrium

User metadata
Rank l33t++
Rank
l33t++

Thanks for the VERY informative posts you 3 😉

I don't know anything about electrical engineering and stuff like that, glad theres people here around who know about that stuff from which I can learn a couple things! 😁

Mystery wrote:

Yes, I suppose the high IO voltage is slowly killing the "desktop" Tillamooks. I'm not sure if there's a reasonable solution to lower the voltage.
Which pins are affected by the IO voltage? Some? All of them? In theory it would be possible to reduce the voltage, but I'm not sure if it's worth the trouble.

Edit: Just prepared the Tillamook for the surgery to reenable the two pins. Problem is, I can't find my conductive silver. So it might be a couple of hours (until I find it) or a few days (if I've got to buy it again). 😉

I'm not sure if the lower i/o voltage is really much of a problem. I read in the datasheet Intel lowered their i/o voltage so they would consume less power (I'd guess for improved performance for use in laptops).
Theres a good chance the Tillamook will run with the much higher i/o voltage regardless and will function perfectly fine (compared to having it receive it's correct 2.5V i/o voltage) except it will use a bit more power, which is not a problem for our intended use as desktop processors anyway 😉

The cache is the only real problem I think.

Edit:The weird multiplier is another hurdle, but who cares about that if the chip won't even run with L2 cache in the first place 😜

Last edited by Tetrium on 2011-06-12, 09:40. Edited 2 times in total.

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 66 of 76, by SquallStrife

User metadata
Rank l33t
Rank
l33t

Hoo lordy, look what I've started! 😜

Just to update you all, my Tilly is sitting on the shelf for now. The fastest it will go is 266MHz because of the motherboard it's on. 266MHz would have been great if I could also attach my DB to the onboard CMI8330... But no.

So I'm just using my regular SL2S9 in my S7 rig for now, till the right motherboard comes along.

VogonsDrivers.com | Link | News Thread

Reply 67 of 76, by Mystery

User metadata
Rank Member
Rank
Member

Small update: A first test with conductive silver was not successful.

However, the connection between the ADS and ADSC pins isn't easy to make, at least for me, so there's still a possibility that my first attempt wasn't making a connection.
Maybe I'll solder some wire to the socket pins themselves, under the mainboard. I'll think of something.

At least I didn't fry my CPU or mainboard in the first attempt, so it's not a complete failure 😉

::42::

Reply 68 of 76, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Couldn't you just use an ohm-meter on the pin side of the CPU to verify your taped connection?

A simple solder "tap" and some copper wire strands should be fine. Better to be sure than to always wonder. I'll eventually get to this test as well, but not for a month or two.

Ultimate 486 Benchmark | Ultimate 686 Benchmark | Cyrix 5x86 Enhancements | 486 Overkill Graphics | Worlds Fastest 486

Reply 69 of 76, by Tetrium

User metadata
Rank l33t++
Rank
l33t++

I might go test this with 2 little copper wires (the RG100 method 😁) and one of my lesser s7 boards within about 2 weeks or so? Depends how long I'll need to finish my attic, but progression is good actually.

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 70 of 76, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Off topic, but... I just looked over some old posts and from the dates, it seems your attic has been 'under construction' for years. I find that old dressers work great for retro IC storage.

Ultimate 486 Benchmark | Ultimate 686 Benchmark | Cyrix 5x86 Enhancements | 486 Overkill Graphics | Worlds Fastest 486

Reply 71 of 76, by Tetrium

User metadata
Rank l33t++
Rank
l33t++
feipoa wrote:

Off topic, but... I just looked over some old posts and from the dates, it seems your attic has been 'under construction' for years. I find that old dressers work great for retro IC storage.

Actually, those pics were of attic 1.0 completed! 😜

When I say I'm chaotic, I don't mean "loosing my bicycle keys once a week-chaotic", but more like "leaving my bike unlocked at the train station-chaotic". or "leaving the front door wide open when I go get some groceries-chaotic" 😜

If my head wasn't stuck on my body, I'd need to order a new one on Ebay on a weekly basis! 😁

That's why attic 2.0 is going to be a lot more organized 😉

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 72 of 76, by prophase_j

User metadata
Rank Member
Rank
Member

I think this needs revisited.

ppga.png

Here is what I think needs to be done to fake a Tillamook as a desktop chip.

Essentially what I have here is the proposed mod for reconnected the ADS/ADSC and BRDY/BRDYC, plus one other nugget I have discovered, from the intel data sheets:

VCC2DET:

Differentiates between the Pentium® processor with MMX™
technology and the low-power embedded Pentium processor
with MMX technology

In the low voltage version, it is an internally not connected pin, but in the regular one it is an active low output! Apparently the purpose of VCC2DET is to distinguish between single and split voltage processors. So in the case of the low voltage version, even though it is split voltage it is not activating this pin. Now, from what little understanding I have, if one was to connect to a VSS source that it send the logic signal "low" to the motherboard, emulating they way a regular P55C works. Honestly, i'm not sure here, but I have not seen any other mention of the VCC2DET difference anywhere other then datasheets for the processors and voltage regulators.

So, what I need is for you guys to look at this this and see if I might be on to something, and if I'm on the right track as for as connecting the VSS to get the logic low signal. Let's kick of the new year with some serious nerd science!

Uploaded with ImageShack.us

"Retro Rocket"
Athlon XP-M 2200+ // Epox 8KTA3
Radeon 9800xt // Voodoo2 SLI
Diamond MX300 // SB AWE64 Gold

Reply 73 of 76, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Interesting discovery. I'll be able to provide some insight when I benchmark my Tillimocks for the Ultimate 586 Benchmark Comparison. "586" is subject to change.

It might be some time before I get to this though. I'm swamped with other activities at the moment.

Ultimate 486 Benchmark | Ultimate 686 Benchmark | Cyrix 5x86 Enhancements | 486 Overkill Graphics | Worlds Fastest 486

Reply 74 of 76, by feipoa

User metadata
Rank l33t++
Rank
l33t++

I've begun working on this just now. Did anyone try those modifications? I'd be happy to, but unfortunately when I plugged in my Tillimock 266, set the core voltage to 2.0 V, multiplier at 4X, the system didn't turn on. Possible causes:

1) I/O voltage is set at 3.5 V which is 1 V over the rated voltage, however I figure the system should at least turn on (all you guys were doing it like this fine).

2) My China-sourced 266 Tilly (SL2Z4) is DOA?

3) Tillimocks won't work on i430TX socket 7 boards?

I think we can probably rule out #1, but if that is the case, I will modify the voltage regulator. Has anyone tested a Tillimock in an i430TX motherboard and if so, did it work?

If those two criterions pass, then my Tilly is probably toast. In which case, does anyone have a known working 266 unit they would be willing to donate, or to trade for a non-working one, for the sake of the Ultimate 6x86 Benchmark Comparison?

I will also test my toasted Tilly in a non-TX board to confirm the issue isn't motherboard-specific.

Does anyone know if the 233 Tilly (SL2Z3) is any more likely to work with L2 cache enabled compared to the 266 Tilly (SL2Z4)?

Ultimate 486 Benchmark | Ultimate 686 Benchmark | Cyrix 5x86 Enhancements | 486 Overkill Graphics | Worlds Fastest 486

Reply 75 of 76, by SquallStrife

User metadata
Rank l33t
Rank
l33t

My chip is a SL28Q. The board it booted on, earlier in this thread, was a J-TX98B, with an SiS 5598 chipset.

It may have been that board I got it up to 300 on, I don't quite remember.

VogonsDrivers.com | Link | News Thread

Reply 76 of 76, by feipoa

User metadata
Rank l33t++
Rank
l33t++
feipoa wrote:
I've begun working on this just now. Did anyone try those modifications? I'd be happy to, but unfortunately when I plugged in […]
Show full quote

I've begun working on this just now. Did anyone try those modifications? I'd be happy to, but unfortunately when I plugged in my Tillimock 266, set the core voltage to 2.0 V, multiplier at 4X, the system didn't turn on. Possible causes:

1) I/O voltage is set at 3.5 V which is 1 V over the rated voltage, however I figure the system should at least turn on (all you guys were doing it like this fine).

2) My China-sourced 266 Tilly (SL2Z4) is DOA?

3) Tillimocks won't work on i430TX socket 7 boards?

I think we can probably rule out #1, but if that is the case, I will modify the voltage regulator. Has anyone tested a Tillimock in an i430TX motherboard and if so, did it work?

If those two criterions pass, then my Tilly is probably toast. In which case, does anyone have a known working 266 unit they would be willing to donate, or to trade for a non-working one, for the sake of the Ultimate 6x86 Benchmark Comparison?

I will also test my toasted Tilly in a non-TX board to confirm the issue isn't motherboard-specific.

Does anyone know if the 233 Tilly (SL2Z3) is any more likely to work with L2 cache enabled compared to the 266 Tilly (SL2Z4)?

Upon testing my SL2Z4 (Tillimock 266) further, I have determined that it may not work in 430TX boards, at least not in a HOT-591P. My ALi-V board also wouldn't turn on with it installed. The VIA MVP3 boards will at least turn on, but as others have pointed out, L2 cache needed to be disabled. Perhaps a SiS chipset is key? Oddly, when I set the multiplier to 4.0, CHKCPU and Speedsys both indicated the CPU speed was running with a 3.5x multiplier.

My regular Pentium 233MMX at least could run Speedsys at 266 MHz, so I may just stick with that.

Ultimate 486 Benchmark | Ultimate 686 Benchmark | Cyrix 5x86 Enhancements | 486 Overkill Graphics | Worlds Fastest 486