Introduction
Before I start I’d like to get a small message across. I’ve fiddled around with BX chipset boards and 100 MHz front side bus Pentium II CPUs for quite a few weeks now. Playing around with all these components is one thing, but making you able to understand how it works and making myself feeling confident to write a meaningful article also requires a lot of reading. The basic literature for these new Intel products is luckily provided by Intel itself, in the shape of several data sheets.
The information in this article will enable everybody who fancies it to run any Slot 1 CPU at 100 MHz FSB in any motherboard with Intel’s BX chipset, even if the CPU should only be designed for 66 MHz FSB. I am completely aware of the fact that this information provided by Tom’s Hardware Guide will very soon be spread as common knowledge all over the Internet. I am asking everyone who is planning to publish this information as well to please refer back to Tom’s Hardware Guide. I would also like to remind you of the fact that running a CPU beyond its official specifications can damage this CPU. Tom’s Hardware Guide will not take any liability for damages that should occur to your system.
The Background
Yesterday I was again wondering if there shouldn’t be an easy way of running any Intel Celeron CPU at 100 MHz in any BX motherboard, not only the ones which would let you choose the FSB between 66 and 133. Most motherboards with BX chipset are following the Intel specification, which demands an automatic adjustment of the FSB, depending on the CPU that is plugged into the system. This adjustment is done via detecting the 66/100 MHz FSB specification of the CPU by checking the logical state of CPU pin ‘B21’. If B21 is logically ‘high’ the 100 MHz bus is chosen, if it’s logically ‘low’ the motherboard chooses 66 MHz FSB. Intel’s data sheet of the Pentium II 350 and 400 shows how this is one in particular.
You can see the resistors ‘R5 and ‘R6’ inside the Slot 1 CPU, which I mentioned in my article ‘Intel’s Slot 1 CPUs Uncovered’. In case of a 100 MHz FSB CPU R5 is 1 kOhm, R6 is 3.3 kOhm and a 66 Mhz FSB CPU as the Celeron and the Pentium II 233-333 have two 0 Ohm ‘resistors’.
Now you can see that Intel is asking for a ‘pull up’ resistor of 200 Ohm on the motherboard, so that the ‘B21’ pin can be read by the clock generator. If R5 and R6 are 0 Ohm, the signal at the clock generator is logically ‘low’, because the pin B21 is inside the CPU directly connected to ground. In case that R5 and R6 add up to 4.3 kOhm, the signal that the clock generator gets is of course ‘high’, due to the 200 Ohm pull up resistor. The beauty of this situation is that due to this pull up resistor, which you can find on any BX motherboard, the clock generator will get a logical ‘high’ signal just as well, in case that pin ‘B21’ is not connected to the motherboard at all. This is all we have to take care of and the motherboard will ‘believe’ that a 100 MHz FSB CPU is plugged into the system.
Now how is it done?
All we want is that CPU pin ‘B21’ is not connected to the motherboard after we plugged in our Slot 1 CPU. A super easy and very efficient way is to simply cover pin ‘B21’ with a small bit of adhesive tape. To do that, have a close look at the CPU contacts first and check how wide they are, luckily ‘B21’ is one of the contacts in the lower row. Then cut a small maybe 1 inch long stripe from the tape which is roughly as wide as the contact. It’s pretty easy if you’ve got a Celeron, because it hasn’t got a cover, otherwise it’s a bit more difficult finding the correct contact. On the Celeron it says ‘B1’ and ‘B121’ on its backside, the side where there’s no chip on. In case of a Pentium II you need to look at the side which carries the hologram. In both cases it’s the side of the CPU which shows the longer row of contacts to the right, the shorter row to the left, both counted from the little notch and seen when holding the CPU the way that the contacts are pointing down. Now find ‘B21’ by starting to count to the ELEVENTH of the lower contacts starting on the RIGHT side of the CPU. Stick the tape onto this contact and cut off the remaining part of the tape. Make sure that you are using a tape that’s sticking very well and check if the tape is fixed properly on the contact so that it won’t move when you plug in the CPU. If you have got an automatic motherboard and it ran at 66 MHz so far make sure that you turn down the multiplier. Now plug in the CPU and you will see that like a miracle it’s now running at 100 MHz FSB. Voila!
This is how it looks on a Celeron, the Pentium II has the cover over it, but you can still access ‘B21’ very well without removing the cartridge.
That’s what it should look like after using red adhesive tape. You can afford covering a bit of the thinner contacts (B20 and B22) left and right of B21, because the metal clips of Slot 1 only touch the upper part of these two, where the contacts are wide. Be careful that the tape doesn’t go up too far, but make sure that it covers all of lower B21.
It’s the easiest and most efficient way of getting your Celeron or Pentium II 333 to run at 350 or 400 MHz in *ANY* BX motherboard.
Feedback
In the first hours after posting this article I received several mails suggesting nail varnish instead. Indeed it may be an even better idea, since it’s easy to apply and also easy to remove. Harvey Rubens also pointed out that adhesive tape can contain aggressive chemicals which could damage the contacts. He suggested the following:
One possible solution is “anti-corona dope,” that red paint you see in
high voltage electronic equipment as an insulating paint. You could
try fingernail polish or a paint used for model aircraft. Any such
paint should be allowed to dry completely to prevent scratching of the
paint or accidental contamination of adjacent terminals on insertion.
Another possibility is the tape used to splice analog recording tape and
cassettes. This tape is very thin and is formulated specifically to
prevent the “bleeding” I mentioned, because, in the early days of
recording, this was a very big problem.
Tom:
I have read your section on isolating the B21 lead on the Pentium II carTRidge. Suggestions you have listed are: nail polish, elecTRician’s tape, anti-corona dope” high-voltage insulating paint, model airplane paint, or splicing tape for analog recording cassettes. I hesitate to use any of these methods because the constituent materials were not specifically designed for this type of use. Nail polish and model airplane tape may have small capacitances due to their dielecTRic behavior thereby creating an unstable elecTRical joint. Insulating paints, on the other hand, may have abrasive surfaces that will eventually wear down the Slot 1 mating connectors. In addition, thermal cycling and pressure may cause the interface between paints or nail polish and the B21 lead to fracture, causing flaking of the material. This result could be a short between neighboring leads.
While none of these problems have yet been observed, I have a simple and cheap solution that may eliminate all these potential issues. The answer? Teflon tape. Teflon is a unique material that has an exTRemely low friction constant, excellent dielecTRic sTRength (no leakage current), and is resilient to temperature fluctuations. In tape form, Teflon is relatively soft and can be easily sTRetched with ones hands. The smooth, soft surface eliminates abrasive problems and will conform to the B21 lead/Slot 1 contact. The tape can be applied and removed very easily unlike nail polish or paint. The operating temperature of Teflon is roughly -100 F to + 500 F (-70 C to + 260 C) so thermal cycling should not pose a problem. For an example of Teflon’s use in isolating elecTRical components, one need look no further than Granite Digital (www.scsipro.com), a premier SCSI cable supplier. Their cabling is considered the best by many and uses Teflon instead of the standard PVC.
To purchase Teflon tape (in the US) contact McMaster-Carr at www.mcmaster.com. I have listed tape rolls and their corresponding features/price below. The tape comes with an 1.5 mil thick (1 mil = 1/1000th of an inch) silicone adhesive, and for the product numbers listed below, the width is 1/4 inch. Other dimensions are also available from McMaster-Carr. According to their catalog, there is no order minimum so if you want just one roll, you can get just one roll.
In case you decide to post this, Tom, please add this:
Users attempting to use Teflon tape on their systems do so at their own risk. I will not be held responsible for any problems or damage resulting from this suggestion.
Brent Clothier
P.S. Great job on the BX Review! Thanks for your hard work. It is sincerely appreciated.
Tape Thickness | Product Number | Price(per roll) | Dielectric Strength | Tensile Strength |
2 mil | 76475A11 | $5.73 | 8000 V | 25 lbs./in.-width |
3 mil | 76475A16 | $6.53 | 10000 V | 25 lbs./in.-width |
10 mil | 76475A17 | $13.84 | 19000 V | 50 lbs./in.-width |
For those interested:
The dielecTRic sTRength should be the voltage applied through the thickness of the tape at which the insulating properties severely degrade and current starts to flow. I say “should be” because no procedure was given in the catalog on how these numbers were obtained.
The tensile strength is the stress per width of the tape required to make permanent deformation in the tape. The “per width” should be obvious since wider tape is stronger. Thickness is also important as can be seen from the 50 lbs./in.-width for the 10 mil tape as opposed to the 25 lbs./in.-width for the 2 and 3 mil tapes.
You can also try and break the line to B21 on the CPU PCB (as I didn’t dare to write, but as suggested by David Klepes), this would obviously be irreversible unless you want to re-solder it later, which will at least be visible on Celeron CPUs. Whatever you want to use, it simply has to isolate ‘B21’ from its Slot1 connector.
Good Luck !