I did a quick little test today to see just how obvious Sony is being about chaning fan speeds.  I used some very basic equipment (a multimeter and an Ardunioscope), so the results are somewhat aproximate; they are based on repeated readings of the same sensors to confirm readings.  This was all done in a room that stayed 25c for the entire test.  Note that I am posting the voltages that the multimeter read from the fan control signal line.  This line was tested with the arduinoscope each time the firmware was updated, to make sure that the pulses were still 5.0v, and thus the average voltage read by the multimeter serves as a reliable reference point.

Starting firmware: 2.10:
0.64v = 25c-46c
0.84v = 47c
0.90v = 48c
0.93v = 49c
1.00v = 50c
1.03v = 50c
1.07v = 51c
1.09v = 52c
1.13v = 54c
1.16v = 55c
1.94v = 56c…this is a noticeable increase in fan speed; after it increases, the system quickly cooled to 50c, and did not get any warmer…but the fan speed did not drop when the system temperature did.  I did not do a stress test because I didn’t have any games that would start on this old firmware.

Second firmware: 2.17
0.64v = 25c-45c
0.84v = 47
0.90v = 48
0.93v = 49
1.00v = 50
1.03v = 51
1.07v = 52
1.10v = 53
1.16v = 55
1.94v = 56…then the same increase in fan speed, followed by a drop to 45c this time, but with the fan speed remaining at 1.94v.  Played GTA4 for a while, lots of explosions and fast driving, system never got past 52c.

Third firmware: 2.53
I skipped a few firmwares and went to 2.53; all readings were the same as 2.17, with these minor differences:
1.07v = 53c
1.14v = 54c
1.16v = 55c

Fourth firmware: 3.00
I skipped a few more firmwares to 3.00…where I started to see important changes…
0.64v = 25c-47c – the point where the fan starts to spin faster is 2c higher than the others
0.85v = 48c
0.90v = 50c
0.93v = 51c
1.00v = 52c
1.03v = 53c
1.07v = 54c
1.01v = 55c
1.14v = 56c
1.16v = 57c
1.94v = 58c – The temperature where the fan first gets 1.94v is always the highest temperature in the tests, even while playing GTA4.  For firmware 3.00, this maximum temperature is 2c higher than 2.53.

Fifth firmware: 3.15
0.64v = 25c-49c – Once again the starting temperature is raised 2c.
0.80v = 50c
0.85v = 51c
0.90v = 52c
0.93v = 53c
1.01v = 54c
1.03v = 55c
1.07v = 56c
1.10v = 58c
1.14v = 59c
1.94v = 60c – And again, the maximum temperature is raised 2c.

Sixth Firmware: 3.41
0.64v = 25c-49c – The starting temperature is the same as 3.15.
0.80v = 50c
0.91v = 51c
0.93v = 52c
1.01v = 53c
1.03v = 54c
1.07v = 56c
1.01v = 57c
1.14v = 58c
1.16v = 60c
1.94v = 62c – The maximum temperature is 2c above 3.15, or 6c above 2.53.

Seventh Firmware: 3.50
0.64V = 25c-48c – The starting temperature is dropped, although the amount of drop is right on the edge of the margin of error.
0.80v = 49c
0.84v = 50c
0.90v = 51c
0.93v = 52c
1.01v = 53c
1.03v = 54c
1.07v = 56c
1.10v = 57c
1.14v = 58c
1.16v = 59c
1.94v = 60c – The maximum temperature clearly dropped here…it is still 4c higher than it was on 2.53, but it is lower than 3.41

Eighth firmware: 3.55
0.64v = 25c-48c – The same as 3.50
0.80v = 50c
0.90v = 51c
0.93v = 52c
1.01v = 53c
1.03v = 54c
1.07v = 55c
1.10v = 56c
1.14v = 57c
1.16v = 58c
1.94v = 59c – I cannot confirm if this was actually a drop; the maximum readings from firmwares 7 & 8 were closer than the margin of error with the equipment I used.

Things used for test (other than things needed for downgrading):
Fluke 73III Multimeter
Ardunio Pro Mini 5.0v/16mhz (for temperature collection)
100K@25c/3% thermistors
Arduino Duemilanove (Arduinoscope)
Win7x64 laptop (for reading from the Arduinos)
CECHH01 (40GB Fat) PS3

This is as close to proof that Sony is trying to kill older PS3s as I can offer with my limited equipment…but the results are compelling; even assuming that 3.55 has a maximum temp of 59c, this is still 3c higher than it was on 2.53…and the PS3 fat is notorious for having heat releated failures; if anything, sony should be moving the maximum temperature down.  I did not continue on to 3.56, 3.60, or 3.61; as there is no way to go back; so I cannot confirm or deny any changes in fan speed for those firmwares.  Several people have contated me stating that 3.61 dropped the fan speeds in the launch models, but I cannot confirm this for myself.

I am sure that there are people out there with high quality ‘scopes and super-accurate thermal sensors, as well as a PS3 fat with firmware 3.55 or earlier installed…If any of you would like to take some more accurate readings, I would very much like to see them.