Slot1 core voltage revealed?

Resources, guides, discussions for modding your system.

Slot1 core voltage revealed?

Postby atang1 » Sun Aug 11, 2002 1:32 am

Slot1 motherboards are designed with PNP bios, all infomation about the motherboards are hidden. Sometimes, the updated bios are noted to have the cpu recognition updated. But that is only the id printout and wait state adjusted. Multiples are also correctly printed out on the bios. FSB, however, is totally dependent on the PLL IC used on the motherboard. If you search the PLL IC number on the internet, by its specifications, you will know whether software adjustments in Windows will do you any good.

By the same token, if you look for the low voltage dropout regulator ICs, you will find US3004, KA34063A, etc., you will also find the specifications and circuit design to have the core voltage(1.3-3.5v), then two other voltage additions to take care of trace resistance built-in on the motherboard to the core and I/O of the cpu. That trace resistance allowance may be insufficient for slotkets. Hence, some slotkets have to have core voltage selection jumpers(vid pinout change) to make cpus work better. On slotkets, addition of 0.9volts to the cpu specification will usually work better. On motherboards, 0.4-0.9v addition depending on trace length and whther they turn corners or not. Of course if you do overclock experiments, the isolated vid pinout voltage selection is necessary.

More on voltage regulation and modifications to take advantage of trace resistance to tame power hungry cpus, later.
atang1
 

Postby atang1 » Mon Aug 12, 2002 12:24 pm

One of the most interest thing that I discovered when I design a simple hair pin turn heater for an epitaxial furnace was that current seeks the path of least resistance. The hair pin turn, caused the current to stay in the path of the shortest distance. The inside radii gets red hot and the outside radii was cold black graphite element.

You would think that this has something to do in the trace resistance leading to the core voltage of the cpu or the I/O vooatge of the cpu? And you would be right.

As the pitch and the cross sections of the traces of the circuit are printed foils getting smaller and smaller, its resistance became significant when cpu draws 14 watts at what ever voltage level. Newer tracess requires 0.9 voltage drop to reach the cpu from the mosfet transistor to the cpu pins. To cool down the mosfets, 7 amp current requirement demands a 25 amp mosfet.

If you look at the layout of the traces, circuit designers try to equalize the resistance by using"corner turns" A 45degree turn or a 90 degree turn of the same cross section of copper foil has different resistance at the corners.

Knowing the resistance, between the mosfet and cpu, means the resistance can be used as a current limitor. You can increase the core voltage and remain cooler by having the resistance take up the extra wattage. Some cpu works better even when they don't work at all without the extra voltage.

When I modified an Amttron PM8600B, the traces require 0.9 volts offset to get 2.2v core voltage. But at 2.5 volts, all the rejected K6-2 would work, using a small socket 7 heatsink/fan. On the otherhand, using socket370 heatsink/largest fan on other motherboards, 2.2 volts get other k6-2s very hot still..

Current limiting resistors have some uses.
atang1
 

Postby len444 » Mon Aug 12, 2002 8:49 pm

The inside radii gets red hot and the outside radii was cold black graphite element


are there magnetic fields created, which helps to draw the electons inward?
len444
Black Belt 3rd Degree
Black Belt 3rd Degree
 
Posts: 3987
Joined: Thu Nov 08, 2001 1:01 am

Postby atang1 » Tue Aug 13, 2002 1:53 am

Issac Newton never considered the friction factor in the Newtonian law. Because it was negligible in the fall(gravity pull) of the apple.

So in the fluid flow or current flow, all molecules travel at the same speed and the shorter distance gets more flow. More flow means more collision(electrical resistance in quantum physics), and it heats up in a hair pin heater. Making that part of the heater thicker will even out the heat. The mention of the fluid flow is to easily see it in the currents flowing near river banks, especially in rapids.

But if you study the traces on motherboards, mosfets are always put very close to sdrams, and cpu. Collectors of motherboards, really appreciates these little details in the design. Why some of the smallest details do fantastic engineering feats.
atang1
 

Postby atang1 » Wed Aug 14, 2002 1:19 am

If you search for Cherry CS5151 low drop out voltage regulator specifications, you get the specs for AMD k6-2 voltage requirement. Surprisingly you get all the manufacturers and their votage regulators and circcuit diagrams. All of them show pinouts of vid0-4. And you can change your core voltages if you want to hard wire them on the voltage regulators.

Since AMD publiched this k6-2 power requirement specifications some time ago; many old 430TX motherboards(DTK or Gemlight PAM-0057I) have CS5151/5156 designed for the intel 233 cpu. If you ever want to put k6-2 cpus on the socket 7 motherboards look for the low voltage dropout regulators mentioned in the AMD literature.

Have fun collecting these slot1 and socket 7 motherboards. They are fun to modify and improve the speed and power by subsituting cpus not intended, initially.

More on slot1s later.
Last edited by atang1 on Tue Aug 20, 2002 3:50 pm, edited 1 time in total.
atang1
 

Postby atang1 » Fri Aug 16, 2002 7:37 am

CS5151 revisited.

Gemlight being in HongKong had British trained engineers. When they design motherboards with designated chipsets, they wanted to use the latest technology IC components. On the PAM-0057i E0 motherboard they used CS5151 to supply the core voltage and I/O voltages. But instead of giving the jumpers on all the vid pins. They give up on the vid3 and vid4, leaving them not connected, which means hung high. And with only 3 jumpers to switch the vid0, vid1 and vid2 pins, they limited the core voltage to 2.8 volts. This philosophy of not built in all the benefits on one motherboard, eventually caused their demise. The next motherboard E1, they added one more jumper and can switch vid3 to go down to 2.2 core voltage. If they proceded to enable the switching of vid4, they could have capability of supplying 1.25 volts. Some day, if AMD or other licensee goes forward with 0.13 micron lithography K6-3 or k6-2+, Gemlight motherboards can handle them at ease. As it is, we can modofy the motherboard by soldering two wires to two jumpers to get all the voltages of slot1 motherboards.

This, of cause, points to the voltage limitations of some slot1 motherboards, that the pins on the low voltage dropout regulators may not have the vid3 connected and thus no core voltage below 2.0 volts? Do you have such a slot1 motherboard? You can correct it, of course, by soldering to the vid3 and the vid4 pins on the slot1 motherboard to ground to get all the voltages way down to 1.25 volts.

Just a notation on my way to the forum.
atang1
 

Postby atang1 » Thu Aug 22, 2002 5:13 am

It sounded like the end of the subject matter, but it is only because we have not discussed all the low voltage dropout regulators, people have designed into the slot1 motherboards. And whether they have taken advantage of the regulators and give those motherboards all the benefits of 1.25v-3.5v core or I/O voltages either by jumpers or by bios selections?

More on this later, when we discuss motherboards and their designs?
atang1
 

Postby atang1 » Sun Sep 01, 2002 4:26 am

Newer Pcchips slot1 motherboards used KA34063A low voltage dropout regulators, which has less pin outs. It is still regulated by vid pin outs, but the address registers are set by a serial port input. Now the data to control the core voltage has to be controlled by the bios.

The vid pin outs from the cpu is scanned by the bios, then the bios sets the registers in the KA34063A and the cpu gets the correct core voltage. You would wonder how the cpu is started and at what core voltage?

Well, the bios is like a small computer, it first tests the cpu registers, and sets the values(boot screen data is later displayed) . So the core voltage is starting at the lowest voltage(depending on the date of the bios, you update your bios, you can use newer cpus) and moved up to the correct voltage. Someday, with the bios size increased, the bios can determine the overvoltage it needs to run the cpu at its maximum frequency(overclocked).

Will that day evercome? Submicron lithography will enable us to put more transistors in the flash semiconductor rom. And then away we go. Computer desiners have not yet have the mentallity to redesign the bios rom augurhythem yet, so far.

But we can wait. What about all those other voltage regulators, how do they work? Coming up, next.
atang1
 

Postby atang1 » Sat Sep 07, 2002 3:16 pm

The beauty of knowing your low voltage regulators is the ability to change core voltage or I/O voltage and the add on voltage to overcome trace resistance. We have to change the voltages in order to overclock. Automation is not available on older motherboards. So, we have to do some hot rodding.

You can have a bios update to change voltage, You can have pinouts tied to ground or hung high to change voltages. and you can sleeve the cpu vid pinouts or tape the slot1 vid gold fingers. Three levels of hot rodding, whichever you think is easier?

And I am going to help you do your own research. Most of the info is at INTEL developers' website.

More on low dropout voltage regulators' design, and how to hot rod them.
atang1
 

Postby atang1 » Sat Sep 21, 2002 8:48 am

When you look at the voltage and vid pinout chart, you have to wonder, how they have projected down to 1.25 volts for core voltage. When the cpus were 3.3 to 3.5 volts. Core voltage and I/O voltages being the same, when they introduced slot1 motherboard.

Those of us who were in the semiconductor inductry, of course, can project the core voltage by the submicron lithograaphy size that will be used in production of cpus.

It turned out that O.35 micron lithography can sustain 3.3-3.6 volts by isolation of a certain line width. 0.25 micron ran into production errors due to mask and optics tolerances. The core voltage had to be 2.0 volts, much lower than the anticipated 2.5 volts(k6-2 runs 2.2-2.4v).

Then the 0.18 micron production came with 1.65-1.75 volts for core. The 0.13 micron cpus had 1.475 volt core but 1.25 volt for I/O. These latter group of cpus work with a larger opening on the mask. The dye laser illumination source uses phase shift to regulated beam size. If the beam size is smaller but the mask opening is larger, isolation voltage becomes higher. So. 1.475 core voltage can be adjusted by phase shift of the laser beam. The slower I/O transistors sustained only 1.25 volts.

The phase shift control of the dye laser beam width enabled changing line width for the production of cpus. This is evidenced by both Intel and AMD slowly going faster and faster gradually. Instead of sudden change of speed by changing to smaller masks. Then they have to make the gate oxide thinner to accomplish faster cpu speeds. Also you will notice that changing submicron size did not change the flip chip silicon outlined size.

Well, more power to them. Next generation 0.10 cpus however, will have to change to smaller outlined size. But slot1 motherboard owners will be left behind. We don't have to worry anymore, core voltage will not be compatible at all.
atang1
 

Next

Return to System Modification

Who is online

Users browsing this forum: No registered users and 2 guests