Apple/Daystar Dual 200MHz 604e CPU card - core voltage -

[stynx]

I have checked the core voltage of my dual 200Mhz card and found that there was about 2.8-2.9V on the output pin of the LT1580.

Checking the resistors (R1/R2) in the data sheet*, there was a recommendation of 110Ω for both resistors when 2.5V is to be set.

The resistors on the card were 118Ω for R2 and 100Ω for R1 on both LDO for the two processors. This would result in 2.725V core voltage instead of 2.5V (recommended for 604e CPUs). The higher measured voltage may be due to the measure point being directly on the LDO output, essentially measuring the voltage between the LDO-output and common ground.

I then swapped the 4 resistors for 110Ω resistors and checked the LDO-output.
It is still too high with about 2.75V and i am a bit stumped why that would be?

I measured other LDOs used on 604e-cards and they typically use LT1580-2.5 Versions, which has a fixed output of 2.5V.
This i could measure as about 2.6V, where some cards were having on point 2.5V but most were around 2.6V.

The card seems to run a bit cooler now and has shown no problems so far (rendering C4D scenes).
Good cooling is still needed, since the card will heat up a lot when it just sits there.

*(the data sheet is attached)

Does anyone have an idea, why the LDO output on the Dual 200MHz card is so high?

-Jonas

 

[stynx]

Just to be sure, i took another voltmeter and rechecked the voltages.
The result is 2.566V on the upper LDO and 2.6V on the lower LDO.
This means that my other voltmeter is defective and shows about 8-10% higher values.

Therefore the mod was successful and the CPUs will now run at about 8% lower core voltage.

 

[stynx]

After letting the computer run for a few hours, i am certain, that the core voltage is high enough and the CPU-card works normally. The Heatsink is a bit cooler, although not much. I am still a bit baffled, why Apple/Daystar did have the core voltage at 2.725V instead of 2.5V. The CPU does not need a higher core-voltage and the higher voltage only produces more heat...

On another note: checking older measuring devices is always a good idea. The defective (out of calibration) voltmeter was bought in the early 90s. I was typically using several different multimeters, depending where i was testing. The other multimeters are seemingly ok and show the correct voltages. The measurements of the other CPU-cards were doe with a good multimeter while the measurements of the dual-card were done with the bad one. The dual 200MHz cpu is installed in a 9600 and the measure points are hard to reach. the defective multimeter had special clips to hook onto the LDO more easily. I have never removed them from the defective multimeter and therefore most likely often measured the wrong voltages. I may have to recheck some PSU i repaired where i set the 5V to typically 5.1V which would have been around 4.7V in reality...

...no wonder, that some of my repaired computers have been extremely buggy in the last few months.