VOGONS


First post, by Great Hierophant

User metadata
Rank l33t
Rank
l33t

Most socket 3 processors after the Intel 486DX2/66 used 3.3v. Boards that supported 3.3v processors would usually have a voltage regulator onboard near the socket. This presumably would convert 5v to 3.3v. The AMD/Cyrix 5x86 processors and some of the Enhanced AMD Am486s used 3.45v for their signalling. Would boards include a second voltage regulator for this voltage level? Would such processors run stably on boards with explicit support for only 5/3.3v (using the latter). After all, the voltage is very close, not like a huge jump from 5 to 3.3v.

http://nerdlypleasures.blogspot.com/ - Nerdly Pleasures - My Retro Gaming, Computing & Tech Blog

Reply 1 of 2, by Old Thrashbarg

User metadata
Rank Oldbie
Rank
Oldbie

Normally there was a single voltage regulator, in one of two setups: On fancier boards, it was a variable type regulator, and moving a jumper would change the resistance on the adjustment pin to boost the voltage a bit. Less fancy boards didn't even have separate settings for 3.3 and 3.45V... they would just do somewhere around 3.3-3.6V and label the setting as '3.3/3.45V'.

The 3.45V processors will run just fine on 3.3V. And often the '3.3V' setting is actually around 3.4-3.5 anyway. Nobody really went for exact values in those days... it just didn't matter much.

Reply 2 of 2, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Great Hierophant, I recommend reading this PDF on modifying your 486's voltage regulator,
Modifying your motherboard's voltage regulator for overclocking

You will see that changing a discrete value resistor by means of jumpers is all that is required to change the regulator's voltage to the CPU.

Recently I was testing two SiS boards, a 4DPS and a PCI400-4, whereby I measured the voltages of the 3.3 V setting. On the PCI400-4, I found it to be 3.29 V, whereas the 4DPS board has it at 3.35 V. Many processors have a +-0.15 V tolerance on the core voltages, so a 3.45 V processor should be qualified at 3.30 V and 3.60 V for its rated speed.

I did find, however that Cyrix processors may have an even tighter operational voltage range. Only one of 5 Cyrix 5x86 CPUs tested would operate at 3.29 V, whereas they all worked at 3.35 V. Perhaps Cyrix has a +-0.10 V tolerance. The best bet is to get a motherboard which outputs 3.45 V. Alternately, you can follow the PDF guide above and modify your 3.3 V setting to be 3.45 V. You do not need to use a resistance trimmer like I used. You can use a discrete SMD resistor to maintain that clean look, and in fact, that is actually how I used to run it. I later found that using a trimmer for an adjustable voltage regulator was more beneficial for testing. Trimmers have been known to have issues with parasitic impedances, however I've been using my 486 board with a trimmer since I wrote that article and haven't had any stability issues, and that computer is on 24/7.

Plan your life wisely, you'll be dead before you know it.