Custom Syscon fan profile?

[sandungas]

Btw @db260179 take a read at what i wrote here https://www.psx-place.com/threads/ps3-temperature-vs-ambient-temperatures.30855/page-3#post-257320
I never thought in it before, i just realized about it while looking at the results of the tests made by @leandronb
And is related with the "optimal" fan settings you mentioned you are trying to prepare

The point is... is needed to match the fan settings with that theoretical "efficient" range (if posible)
Anyway, is important to keep that concept in mind to have a better idea when we are configuring the fan speeds "out of the efficient range"

 

[db260179]

Hi Sandungas,

Yes, you are correct in your assesment.

From all of my fan tweaking tables, for a default optimal setting for the SEM-001 and COK-001-002 boards as they both have the 90nm CPU and GPU, fan speed of 30% to 45% seems to be the sweet spot of cooling and low noise.

One thing i have noticed with these old models of PS3, the rotation speed of the fans really affects the cooling process - there has been many a discussion over the years about this, and it is true, a good fan will maintain a good cooling threshold.

I will give an snapshot of the optimal fantbl range for the CELL and RSX now:

SEM-001 example

Default fan table from sony

CELL

fantbl get 0
fancon No:00
P0: TempD:0.0(0x0000) - TempU:76.0(0x4c00) duty:20%(0x33)
P1: TempD:56.0(0x3800) - TempU:77.0(0x4d00) duty:29%(0x4a)
P2: TempD:56.50(0x3880) - TempU:82.0(0x5200) duty:30%(0x4d)
P3: TempD:57.0(0x3900) - TempU:83.0(0x5300) duty:31%(0x50)
P4: TempD:57.50(0x3980) - TempU:84.0(0x5400) duty:32%(0x52)
P5: TempD:58.0(0x3a00) - TempU:85.0(0x5500) duty:34%(0x57)
P6: TempD:58.50(0x3a80) - TempU:86.0(0x5600) duty:36%(0x5c)
P7: TempD:59.0(0x3b00) - TempU:87.0(0x5700) duty:55%(0x8d)
P8: TempD:79.50(0x4f80) - TempU:88.0(0x5800) duty:65%(0xa6)
P9: TempD:80.0(0x5000) - TempU:91.0(0x5b00) duty:100%(0xff)
P10: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)
P11: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)
P12: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)
P13: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)
P14: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)
P15: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)
P16: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)
P17: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)
P18: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)
P19: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)

RSX

fantbl get 1
fancon No:01
P0: TempD:0.0(0x0000) - TempU:84.0(0x5400) duty:20%(0x33)
P1: TempD:56.0(0x3800) - TempU:85.0(0x5500) duty:29%(0x4a)
P2: TempD:56.50(0x3880) - TempU:86.0(0x5600) duty:30%(0x4d)
P3: TempD:57.0(0x3900) - TempU:87.0(0x5700) duty:31%(0x50)
P4: TempD:57.50(0x3980) - TempU:88.0(0x5800) duty:32%(0x52)
P5: TempD:58.0(0x3a00) - TempU:89.0(0x5900) duty:34%(0x57)
P6: TempD:58.50(0x3a80) - TempU:90.0(0x5a00) duty:36%(0x5c)
P7: TempD:59.0(0x3b00) - TempU:91.0(0x5b00) duty:55%(0x8d)
P8: TempD:77.0(0x4d00) - TempU:92.0(0x5c00) duty:65%(0xa6)
P9: TempD:78.0(0x4e00) - TempU:95.0(0x5f00) duty:100%(0xff)
P10: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)
P11: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)
P12: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)
P13: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)
P14: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)
P15: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)
P16: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)
P17: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)
P18: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)
P19: TempD:255.75(0xffff) - TempU:255.75(0xffff) duty:100%(0xff)

Tweaked settings from my testing:

0x4d = 30%
0x50 = 31%
0x52 = 32%
0x57 = 34%
0x60 = 37%
0x73 = 40%
0x80 = 50%
0x8d = 55%
0xa6 = 65%
0xff = 100%

Tweaked shutdown
CELL
tshutdown set 0 91.0 - originally 95

RSX
tshutdown set 1 91.0 - same 95

CELL 0 hex set
fantbl set 0 p0 0.0 70.0 0x4d (30%)
fantbl set 0 p1 71.0 73.0 0x50 (31%)
fantbl set 0 p2 72.50 74.50 0x52 (32%)
fantbl set 0 p3 73.0 75.50 0x57 (34%)
fantbl set 0 p4 74.50 77.0 0x60 (37%)
fantbl set 0 p5 75.0 80.0 0x73 (40%)
fantbl set 0 p6 76.50 82.0 0x80 (50%)
fantbl set 0 p7 77.0 85.0 0x8d (55%)
fantbl set 0 p8 79.50 88.0 0xa6 (65%)
fantbl set 0 p9 80.0 90.0 0xff (100%)

fantbl setini 0 p0 0.0 70.0 0x4d
fantbl setini 0 p1 71.0 73.0 0x50
fantbl setini 0 p2 72.50 74.50 0x52
fantbl setini 0 p3 73.0 75.50 0x57
fantbl setini 0 p4 74.50 77.0 0x60
fantbl setini 0 p5 75.0 80.0 0x73
fantbl setini 0 p6 76.50 82.0 0x80
fantbl setini 0 p7 77.0 85.0 0x8d
fantbl setini 0 p8 79.50 88.0 0xa6
fantbl setini 0 p9 80.0 90.0 0xff

RSX 1 hex set
fantbl set 1 p0 0.0 70.0 0x4d
fantbl set 1 p1 71.0 73.0 0x50
fantbl set 1 p2 72.50 74.50 0x52
fantbl set 1 p3 73.0 75.50 0x57
fantbl set 1 p4 74.50 77.0 0x60
fantbl set 1 p5 75.0 80.0 0x73
fantbl set 1 p6 76.50 82.0 0x80
fantbl set 1 p7 77.0 85.0 0x8d
fantbl set 1 p8 79.50 88.0 0xa6
fantbl set 1 p9 80.0 90.0 0xff

fantbl setini 1 p0 0.0 70.0 0x4d
fantbl setini 1 p1 71.0 73.0 0x50
fantbl setini 1 p2 72.50 74.50 0x52
fantbl setini 1 p3 73.0 75.50 0x57
fantbl setini 1 p4 74.50 77.0 0x60
fantbl setini 1 p5 75.0 80.0 0x73
fantbl setini 1 p6 76.50 82.0 0x80
fantbl setini 1 p7 77.0 85.0 0x8d
fantbl setini 1 p8 79.50 88.0 0xa6
fantbl setini 1 p9 80.0 90.0 0xff
On normal operation - running games, then idling will hover around these settings

p1 (31%)
p2 (32%)
p3 (34%)

This might seem high, but the noise level on cleaned fan is not even noticeable, plus its keeping the air circulation around the console.

Its a good balance.

Btw @db260179 take a read at what i wrote here https://www.psx-place.com/threads/ps3-temperature-vs-ambient-temperatures.30855/page-3#post-257320
I never thought in it before, i just realized about it while looking at the results of the tests made by @leandronb
And is related with the "optimal" fan settings you mentioned you are trying to prepare

The point is... is needed to match the fan settings with that theoretical "efficient" range (if posible)
Anyway, is important to keep that concept in mind to have a better idea when we are configuring the fan speeds "out of the efficient range"

 

[sandungas]

Wait, are you sure "fancon No:00" is CELL and "fancon No:01" is RSX ?, im reviewing this talk and never was mentioned before
https://www.psx-place.com/threads/s...o-what-does-it-mean.26148/page-11#post-227184

But now it makes sense why the COK-001/002 have a total of 6 fantbl (but 3 of them doesnt contains valid data), and the other motherboard models have a total of 3 fantable (but 1 of them doesnt contains valid data)
So...the 2 fantabl with valid data are for CELL and RSX individually ?... i guess this explains this screenshot and why the values of fantbl 1 are a bit bigger than fantbl 0

Btw at that days i made a layout to represent this values in a graphic, i think we should use it to talk about them, to discuss them and to compare them, is a lot more noob friendly because we do it visually

 

[sandungas]

I been improving a bit the graphic and made a version for SEM-001, please tell me if is right, i just made but not reviewed it, so maybe there is some mistake
Also tell me what do you think about representing the values this way, with straight lines in colors and arrows

 

[sandungas]

New version, i added some more info and i just realized about something interesting

-We can show fantbl0 and fantbl1 together because the duty values are identical (in some way we are lucky about this, otherway the visual look could be too much chaotic)
-The lines for fantbl1 are in black color by now, i guess i could use a better color code or reassing colors, but by now im not sure how
-The tempU and tempD for fantbl1 differs a bit, but the width of the horizontal lines for fantbl 1 seems to be always bigger (im guessing is going to happen the same in other PS3 models, i dont understand why but doesnt seems to be a coincidence)
-I added the tshutdown values (and marked it with a nice star for the big prize), if someone have suggestions about adding more info just tell

 

[sandungas]

Another version of the graphic, i added the info and lines/arrows from the third fantable (in pink), and the other info that is common for all fantables, this image includes pretty much everything (even a couple of unknown values that seems to be interesting)

If someone is really interested to see in detail how the PS3 thermal control works i suggest you to download the SEM-001 dump uploaded by @M4j0r here
https://www.psx-place.com/threads/s...o-what-does-it-mean.26148/page-11#post-225349
https://workupload.com/file/rHcbgYhk

You can use this code structures as reference, also published by @M4j0r here
https://www.psx-place.com/threads/s...o-what-does-it-mean.26148/page-10#post-223893
https://pastebin.com/Wtc7NcJ4

And you can use this list i just made where i "cropped" the hex chunks from the dump in a hexeditor (is more noob friendly)
https://pastebin.com/Wb94NuLe

Open the SEM-001 dump in a hexeditor and use the image and the references in pastebin to investigate his contents, there are still some "unknown" fields


----------------------------
I dont hink im going to make any more versions with the 3 fantables together... all that arrows are too much confusing... and i guess using different color codes is not going to help much to clarify it
Also, i think our PS3's only loads 1 fantable... so we just need to care about configuring 1 of them

 

[sandungas]

Also, i think our PS3's only loads 1 fantable... so we just need to care about configuring 1 of them

OMFG, thats wrong, by looking at the graphic i just realized there is a fantable for every thermal sensor and the PS3 is using/loading all the fantables at all times to monitor them. The reason why im saying is because:
max temperature of the first fantbl is 91.0 (exactly the same value than shutdown_temp_cell)
max temperature of the second fantbl is 95.0 (exactly the same value than shutdown_temp_rsx)
max temperature of the third fantbl is 84.0 (exactly the same value than shutdown_temp_bevr)

So...
fantbl 0 = CELL
fantbl 1 = RSX
fantbl 2 = BEVR (the voltage regulators next to CELL/RSX)

This is a pita to configure it (and for the graphic i was doing because there are too much lines in it), but being positive... it means the PS3 thermal control is a lot more precise than what i though
Also, this explain why the COK motherboards have room for 6 fantables (more info about it here), the first PS3 syscon models seems to be able to monitor up to 6 thermal sensors using a fantable dedicated for everyone of them


Edit:
*The "ini_hyst_cell", "ini_hyst_rsx", "ini_hyst_bevr" means hysteresis, the value 2 seems to represent a temperature treshold https://www.linuxquestions.org/ques...-explain-'hysteresis'-in-this-context-552271/

*The "initial_fan_time" seems to be in decisecons (where 20 deciseconds = 2 seconds)



New version, minor changes

Another one

And one more. The way how are displayed the texts seems completed (this version includes almost everything), next thing i should try to do is to change the color code :/

Edit:
The values of BEVR TempD (represented with pink down arrows) seems to be so much displaced to left because the smallest of them (BEVR P1 TempD:35.0) matches with a sentence in the PS3 user manual that tells:

Ambient Operating temperature 5°C - 35°C (41°F - 95°F)

In other words, it seems the engineers adjusted the BEVR P1 TempD for a theorethical max ambient temperature of 35ºC
So... there is no reason to modify them, i bet are identical for all PS3 models (at least the group at left, that starts with 35ºC, and increases 0.5ºC for every step, this start point and the increments looks something generic)

 

[sandungas]

Final version... i guess
Btw, the image have a size of 1920x1080 (and the little cells of the grid have a size of 9x9 pixels), is intended to download it and open it at full screen in a PC monitor

 

[sandungas]

fantbl setini 0 p0 0.0 70.0 0x4d - same as above, if the ps3 reaches 0C then goes to this setting
fantbl setini 0 p1 71.0 73.0 0x50 - if its min C get to say, 71C then selects this profile

Im reading your posts again and this worths to be clarifyed
The lowest value of TempD (always 0ºC for the profile "p0" in all the fantables of all PS3 models) is a bit conceptual, because it would mean the PS3 is frozen

In some way it means... if the temperature is reduced to 0ºC then the fan control should change to the profile previous to "p0"
But there is no profile previous to "p0" because the PS3 is not going to have 0ºC ever while working

The only way to return to "p0" fan speed (20% in all PS3 models) is when you are in "p1" and the temperature decreases to the value that appears in "p1 TempD" (in your SEM-001 motherboard this happens when CELL = 56ºC or lower)

When i was making the graphics i painted the arrows using this concept, i think is right, but if you dont agree feel free convince me that im wrong, im open minded

 

[db260179]

You are correct, i think p0 is just a guage profile, so from a cold start.

If you monitor closely on a live ps3 you see the float between p4 down to p1
p0 is just starting point profile - so in theory you could do p0 as - 20c to 70c

In this case the fans wont even spin until it hits 20c, i´ve tested this at 40c to 70c and no fan came on on P0

Im reading your posts again and this worths to be clarifyed
The lowest value of TempD (always 0ºC for the profile "p0" in all the fantables of all PS3 models) is a bit conceptual, because it would mean the PS3 is frozen

In some way it means... if the temperature is reduced to 0ºC then the fan control should change to the profile previous to "p0"
But there is no profile previous to "p0" because the PS3 is not going to have 0ºC ever while working

The only way to return to "p0" fan speed (20% in all PS3 models) is when you are in "p1" and the temperature decreases to the value that appears in "p1 TempD" (in your SEM-001 motherboard this happens when CELL = 56ºC or lower)

When i was making the graphics i painted the arrows using this concept, i think is right, but if you dont agree feel free convince me that im wrong, im open minded

 

[sandungas]

You are correct, i think p0 is just a guage profile, so from a cold start.

Nice, some days ago i was not so sure, but after looking at the syscon dumps from different motherboards i was more convinced about it

p0 is just starting point profile - so in theory you could do p0 as - 20c to 70c

In this case the fans wont even spin until it hits 20c, i´ve tested this at 40c to 70c and no fan came on on P0

This test was risky but is the definitive confirmation

If you monitor closely on a live ps3 you see the float between p4 down to p1

But you mean with the custom settings you suggested in this post ?
https://www.psx-place.com/threads/custom-syscon-fan-profile.30873/page-2#post-257336

I been looking at them and there is something i would like to suggest you, the difference of temperatures in between "TempD" and "TempU" (in all profiles p1 up to p9) is too low
In the profiles with slow fan speed the difference is only 2ºC or 3ºC
And in the profiles with high fan speed is 6ºC or 8ºC

The results is your fan speeds are going to change frequently
If you are playing "ingame" usually the RSX is changing his temperatures up/down a bit (because the game have random workload peaks in RSX that causes temperatures peaks in it)... lets say... 4ºC up and down
When this variations of temperatures does a "TempU" is fine... but you should make the "TempD" less sensistive to the changes
Sony abused a lot of this effect in some PS3 models because is more confortable if the fan speeds doesnt changes "ingame"
We should do it too, but we should not abuse of it, is needed to find the sweetspot

I think to achieve confortable noise levels (in other words, to prevent the frequent fan speed changes) we should use a difference of 10ºC or 15ºC in between "TempD" and "TempU" (as minimal, but could be bigger)

In the factory settings of the DIA-001 and DIA-002 motherboards is made this way, the settings of the DIA motherboards are the best from the syscon dumps publically availables... is just sony made other mistakes in it (are too much permisive with the overheatings, and there are some "garbage bytes" here and there that are pointless because are not even used, are the reminiscenses of previous experiments never cleaned up, lol)

 

[db260179]

Only recent months ive been discovered alot about the ps3 workings.

wait until you see my pile of RSX i've fried, on my first adventures with a IR machine. lolz

so reason for my settings, is noise level and operating temps. 30% to 37% fan speed is barely audible. If u look at default sony settings its pretty much similar but at high temps. Because its running most of the time at 32% when jumping from idle to gaming that micro increments makes sense because the Temps dont jump that high, pretty much all of the sem-001 and cok-002 boards have this behaviour. Cell stays stable from 69C to 73C never changes, unless there is something wrong with it then it goes 80C, but thats when the the other profiles protect this behaviour or fault. RSX swings wildly from 59C to 75C. Newer models like Dia-001,002 only allow p0 to p4 to be set, sony decided to not allow further changes.probably due to reduced silicone dies and really they dont need to.

so allowing too large (fan steps) gaps can cause negative effects on other components and increase build up of heat. the whole old ps3 heat structure is to push out as much heat as it can, reason for big ass fans.

But for me, if more people experiment and let me know or share then there is a good medium.

Thanks for the detailed response though.

if only someone could advise on my new thread mode_auth_data contents on the syscon.

Nice, some days ago i was not so sure


This test was risky but is the definitive confirmation


But you mean with the custom settings you suggested in this post ?
https://www.psx-place.com/threads/custom-syscon-fan-profile.30873/page-2#post-257336

I been looking at them and there is something i would like to suggest you, the difference of temperatures in between "TempD" and "TempU" (in all profiles p1 up to p9) is too low
In the profiles with slow fan speed the difference is only 2ºC or 3ºC
And in the profiles with high fan speed is 6ºC or 8ºC

The results is your fan speeds are going to change frequently
If you are playing "ingame" usually the RSX is changing his temperatures up/down a bit (because the game have random workload peaks in RSX that causes temperatures peaks in it)... lets say... 4ºC up and down
When this variations of temperatures does a "TempU" is fine... but you should make the "TempD" less sensistive to the changes
Sony abused a lot of this effect in some PS3 models because is more confortable if the fan speeds doesnt changes "ingame"
We should do it too, but we should not abuse of it, is needed to find the sweetspot

I think to achieve confortable noise levels (in other words, to prevent the frequent fan speed changes) we should use a difference of 10ºC or 15ºC in between "TempD" and "TempU" (as minimal, but could be bigger)

In the factory settings of the DIA-001 and DIA-002 motherboards is made this way, the settings of the DIA motherboards are the best from the syscon dumps publically availables... is just sony made other mistakes in it (are too much permisive with the overheatings, and there are some "garbage bytes" here and there that are pointless because are not even used, are the reminiscenses of previous experiments never cleaned up, lol)

 

[sandungas]

Let me show you an example of what i meant and why i said they abused of this effect
The values of TempD are always smaller than TempU, so we can generate the values of TempD by substracting "X" from TempU
In DIA motherboards they followed this rule:

p0  TempD = 0

p1  TempD = TempU - 12.5
p2  TempD = TempU - 13
p3  TempD = TempU - 13.5
p4  TempD = TempU - 14
p5  TempD = TempU - 14.5
p6  TempD = TempU - 15
p7  TempD = TempU - 15.5
p8  TempD = TempU - 16
p9  TempD = TempU - 16.5
p10 TempD = TempU - 17
p11 TempD = TempU - 17.5
p12 TempD = TempU - 18

p13 TempD = TempU - 15
p14 TempD = TempU - 8.5

I separated them in 3 groups, the first group is only p0 because is always 0ºC
In the group of the center it can be seen they increased the difference of temperatures in between tempD and tempU following a strict rule (half a degree bigger in every step), and they started with a difference of 12.5 in p1
In the third group for p14 and p15 they stopped incrasing the difference. Because this fan speeds for p14 and p15 are like the "def con" warnings anyway, the fan should not stay in them for much longer time

Im not trying to convince you to change your custom setings or try what im going to say, but i want to show how different would be your settings if we apply "the sony style" to calculate the vaues of tempD
The values you used for RSX was:

RSX 1 hex set

fantbl set 1 p0 0.0 70.0 0x4d
fantbl set 1 p1 71.0 73.0 0x50
fantbl set 1 p2 72.50 74.50 0x52
fantbl set 1 p3 73.0 75.50 0x57
fantbl set 1 p4 74.50 77.0 0x60
fantbl set 1 p5 75.0 80.0 0x73
fantbl set 1 p6 76.50 82.0 0x80
fantbl set 1 p7 77.0 85.0 0x8d
fantbl set 1 p8 79.50 88.0 0xa6
fantbl set 1 p9 80.0 90.0 0xff

Im going to change your values of tempD using "the sony style" starting with 10ºC of difference in p1 (and increasing the difference half a degree in every step)

fantbl set 1 p0 0.0 70.0 0x4d
fantbl set 1 p1 63.0 73.0 0x50 // TempD = TempU - 10
fantbl set 1 p2 64.0 74.50 0x52 // TempD = TempU - 10.5
fantbl set 1 p3 64.50 75.50 0x57 // TempD = TempU - 11
fantbl set 1 p4 65.50 77.0 0x60 // TempD = TempU - 11.5
fantbl set 1 p5 68.0 80.0 0x73 // TempD = TempU - 12
fantbl set 1 p6 69.50 82.0 0x80 // TempD = TempU - 12.5
fantbl set 1 p7 72.0 85.0 0x8d // TempD = TempU - 13
fantbl set 1 p8 76.0 88.0 0xa6 // TempD = TempU - 12
fantbl set 1 p9 80.0 90.0 0xff // TempD = TempU - 10

In the practise, what is going to happen with "the sony style" is the fan is going to need more time to step back to the previous speed
Using my example... imagine we are in p5 (it have a tempU or 80ºC) and the temperature increases 1ºC... this pushes us into p6 (with a tempD of 69.5)... so the fan is not going to reduce speed unless the temperature decreases down to 69.5 (and this could take several minutes)

With your settings for p6 (with a tempD of 76.5) you are allowing the fan to step back to the previous speed in less time than "the sony style"

The effect sony was trying to achive is nice if it happens in-game because it makes the fan less "sensitive" to the temperature changes, and this is achieved by keeping the fan at a speed a bit higher than the actual requirements (of this specific second), so if RSX is moving his temperatures up and down the fan will stay at the speed required by the highest temperature peaks of the last... dunno 5 minutes or so

At the lower temperatures (when we are not in-game) dunno, playing a video, or inside an app, or ftp, whatever... is fine to make this difference of temperatures in between tempD and tempU smaller... this way the fan is a bit more "sensible" and is going to be adjusted better to the real requirements, sony decided to not doing it though, but i think there is a good reason for it

Is a bit psicological, when we are exposed to some repetitive noise for several minutes there is a point where our brain starts "filtering" it and you dont realize about the noise (is like if the noise doesnt exists because you dont think in it). I like this effect, it really works in me
But suddenly the fan speed changes and you are ike "ops, the fan is doing something"

 

[db260179]

Ok, i see what you mean. Yes will give a go and experiment with this.

Thanks

Let me show you an example of what i meant and why i said they abused of this effect
The values of TempD are always smaller than TempU, so we can generate the values of TempD by substracting "X" from TempU
In DIA motherboards they followed this rule:

p0  TempD = 0

p1  TempD = TempU - 12.5
p2  TempD = TempU - 13
p3  TempD = TempU - 13.5
p4  TempD = TempU - 14
p5  TempD = TempU - 14.5
p6  TempD = TempU - 15
p7  TempD = TempU - 15.5
p8  TempD = TempU - 16
p9  TempD = TempU - 16.5
p10 TempD = TempU - 17
p11 TempD = TempU - 17.5
p12 TempD = TempU - 18

p13 TempD = TempU - 15
p14 TempD = TempU - 8.5

I separated them in 3 groups, the first group is only p0 because is always 0ºC
In the group of the center it can be seen they increased the difference of temperatures in between tempD and tempU following a strict rule (half a degree bigger in every step), and they started with a difference of 12.5 in p1
In the third group for p14 and p15 they stopped incrasing the difference. Because this fan speeds for p14 and p15 are like the "def con" warnings anyway, the fan should not stay in them for much longer time

Im not trying to convince you to change your custom setings or try what im going to say, but i want to show how different would be your settings if we apply "the sony style" to calculate the vaues of tempD
The values you used for RSX was:


Im going to change your values of tempD using "the sony style" starting with 10ºC of difference in p1 (and increasing the difference half a degree in every step)


In the practise, what is going to happen with "the sony style" is the fan is going to need more time to step back to the previous speed
Using my example... imagine we are in p5 (it have a tempU or 80ºC) and the temperature increases 1ºC... this pushes us into p6 (with a tempD of 69.5)... so the fan is not going to reduce speed unless the temperature decreases down to 69.5 (and this could take several minutes)

With your settings for p6 (with a tempD of 76.5) you are allowing the fan to step back to the previous speed in less time than "the sony style"

The effect sony was trying to achive is nice if it happens in-game because it makes the fan less "sensitive" to the temperature changes, and this is achieved by keeping the fan at a speed a bit higher than the actual requirements (of this specific second), so if RSX is moving his temperatures up and down the fan will stay at the speed required by the highest temperature peaks of the last... dunno 5 minutes or so

At the lower temperatures (when we are not in-game) dunno, playing a video, or inside an app, or ftp, whatever... is fine to make this difference of temperatures in between tempD and tempU smaller... this way the fan is a bit more "sensible" and is going to be adjusted better to the real requirements, sony decided to not doing it though, but i think there is a good reason for it

Is a bit psicological, when we are exposed to some repetitive noise for several minutes there is a point where our brain starts "filtering" it and you dont realize about the noise (is like if the noise doesnt exists because you dont think in it). I like this effect, it really works in me
But suddenly the fan speed changes and you are ike "ops, the fan is doing something"