Discussion "Bestiolino" Nano + Full New Power Management System (open source)

[Luigid]

Hi guys, while working on my project lately I developed two options, one "custom" and the other "complete" for portable applications. It has currently been developed and tested on the PlayStation 2, but obviously if it reflects the settings for other consoles it can also be used for them. At the moment the Wii should be cut out since it only has one power stage for the CPU/GPU, but I could think about remodeling it and adding an additional stage, perhaps for my XII Boy, since the PMS v1 has the 3.3V problem.

Bestiolino "Nano":

What is it and how does it work?

"Nano" is, like the word, a highly small and efficient micro regulation module. It uses a step-down model: TPS62816 from Texas Instruments. It is designed to be a "modular" module and to be "patched" directly onto the console's motherboard in order to apply the power directly without disturbances or voltage drops. It is easily modifiable and these are the technical specifications:

1. AEC-Q100 qualified for automotive applications – Device temperature –40°C to +125°C
2. Input voltage range: 2.7 V to 6 V --> therefore it is essential to use it with a 1S configuration (li ion - Lipo - LiHv etc.)
3. Output voltage from 0.6 V to 5.5 V
4. Maximum current delivered 6A

The output is quite simple to set just change the value of R1 - R2 - CFF , here is an example table:

Obviously, being a step down and if powered directly from the battery, I do NOT recommend output voltages below 2.5v.
Here it is in action on PS2:

Making a quick comparison with the tlv62095 testing outrun 2006 in demo mode I got 3h and 5min, with "Bestiolino" I went up to 3h and 27min only by replacing this regulator, about 12% more efficient.

Link for this first release:
GitHub

A quick tip:
The PCB has an F1 output for testing (I added a 3A self-resetting one), but it's best to remove it if it's working properly.
The inductor I chose was a 1uH inductor, but if you have height issues, you can use a 0.22uH like: AWLU0420F-R22MPA after a few quick tests, it seems to work properly.

 

[Luigid]

Bestiolino "Full":
currently in development / debugging ready for release
"Full" beast as the word suggests represents a real all-in-one PMS, let's study it more closely:

1. TPS62816 --> the heart of the "Nano" 6A maximum for cpu/gpu core
2. TPS628438 --> the replacement for the ldo there are 3 solderable models but all with the same package / pinout variable voltage from 0.4v up to 3.6v easily via a single resistor - in this setup 1.8V - 600mA Max
3. TPS628501 --> There are also various variables available (see datasheet) in this setup it will be used for 2.5v - 1A Max
4. TPS63802 --> Here is the "champion" IC , already extensively covered on my 79001, a hybrid converter capable of changing its internal operation to constantly deliver 3.5v 2A Max
5. TPS61022 --> "small" step up 5V - 8A available
6. Sw6106 --> last but not least, the heart of it all is a power bank IC that in addition to "giving us" a second 5V rail so as to separate for example PS2 / audio / LCD as we prefer, manages the cell charging and much more, currently it will be limited to 5V 2.5A input charging but can handle over 18W with fastcharge (to be tested), it is also easily customizable thanks to a single resistor to program the type of battery to use

Some comparison photos between "nano" and u-amp:

 

[Legend]

These are awesome! Happy to see some more power solutions out there for PS2 portables! Great job!

 

[Luigid]

Hi guys, some small updates on the trimmed 90004, with dualshock, 1 memory card and the mx4sio:

• Idle: 4,42V x 0,993A --> 4,39W
• 
  
  
• Outrun demo menù : 4,42V x 1,043A --> 4,61W
• 
• Outrun 2006 in Game : 4,42V x 1,082A --> 4,78W
• 

I'd say it currently looks like it's "broken" the 5W limit!

obviously all that glitters is not gold, I made some mistakes on the PCB on the 5V side, nothing serious, the main problem is on the SW6106 side, it seems like a small disaster, I will have to understand better how to "fix" it, in the meantime I ordered another board so that I can understand pin to pin how to fix it

 

[Stitches]

I wonder what the undervolt potential silicon lottery is like for the PS2

 

[Legend]

I wonder what the undervolt potential silicon lottery is like for the PS2

Its nice because we know that the ps2 can run off of more standard voltages by combining the 1.75v and 2.5v rails and running it at 1.8v, and running everything else respectively at 1v and 3v3 instead of 1.25v and 3.5v (which only gives tiny gains but its gains none the less), so I'd almost like to try my hand at making a different iteration of the PMS to support undervolting with an EE app for adjustments similar to what we have with thundervolt!

 

[Luigid]

Hi guys, let's say the table should be more or less like this:

• 1,25v --> max 0.95v The little "bestiolino" is set to around 0.975v which drops by 0.15v-0.20v under load. Both this and my 79001 work without problems.
• 1.75v --> max 1.5v (I'm testing it)
• 2.5v --> as said by legend at 1.8v NB: my 79001 wouldn't boot at all and this 90004 model shows flickering lines on the AV output (I haven't tried the VGA) so I'd say the video encoder doesn't like it
• 3.5v --> 3-3.3v? at 2.5v the ps2 won't boot, probably the mechacon remains in a permanent reset state
• 5v -->The console boots and runs at 3.3v but the image has very "clipped" colors and shades

so technically by eliminating the video encoder and applying a RVL-DD (I think this is exactly the work Mister X is doing) you can apply as above
Returning to the project, I fiddled with the SW6106 and managed to get the 5V output, I also connected the charging status/diagnostic LEDs and they are ok. I was using MOSFETs that were not digested by the IC, currently they are temporary but I manage to get stable above 5V and the LCD on so I'd say at least for now it's ok.
but currently it seems that I can't recognize the 5v input to charge the batteries, an error that I had "discovered" I forgot to connect a vout line of the USB C socket (which in this configuration is precisely our input and not "hybrid") the other one discovered yesterday and that to simulate the cc1 and cc2 I inserted 51k resistors and not 5.1k. but even with these precautions it completely disgusted me lol , but I'll figure it out somehow.
taking away the last tidbits I'm just "undecided" about these two ic:
-TPS628438
-TPS61022

The TPS628438 is truly microscopic, it's as big as a 603 resistor but it has 6 pins, so I realize that it's not exactly easy to solder, so in the meantime, since the console seemed to boot at 1.5V as written, I would like to try directly with an LDO. Eventually, the loss of using an LDO coupled with the undervolt could bring us to the same level of using this IC but with fewer components and soldering complications.


The TPS61022 is not convincing me, it seems too fragile, it goes into internal short circuit as soon as you make a "false step" while the other ICs respond very well, another example just by removing it you gain another 30ma vs sw6106, so another option to be able to align the board is to use a second TPS63802 like on my 79001 which works very well without too many "stories"

 

[Luigid]

Hi all small update:

1. I forgot that for the cc1 and cc2 lines you need a resistor of 5 or 10k but towards vbus (so 5v) and not towards ground otherwise the integrated "sleeps" also confirmed that you need mosfets with vgs 0.9v like the ones unsoldered from the original module that I currently purchased, even with those collected from old motherboards it works and supplies 5v, the battery recharges both at 5v and 9v input, between 7 and 12w (to be verified better with the new mosfets) so it is optically finished.
2. TPS628438 --> for him I did a flying test to understand if as written it could be abolished by simplifying the circuit the answer is yes and the RAM can be stably brought to 1.5v without problems I used an ams1117 waiting for the rt9193. in the end this undervolt saves about 25ma on the line but increases it by 6ma on the battery I would say that it is a current that can be safely negligible, moreover therefore anyone who has problems can simply change the rt9193 from the 1.5v variant to 1.8v without modifying anything to the circuit simply by changing the IC.
3. TPS61022 --> TPS61023 for him it seems to have found a feasible replacement we will not have that very high ceiling of 8A (which was also useless) but it should also be more efficient for the current range that we need around 1-1.5A (assuming ps2 - usb and possibly as I did a linux board for SMB) ,It has almost the pinout of the old IC so I can adapt it to the tracks and try it on this board already , same components inductor etc, great.
4. I also did tests with 3.5v, as I suspected the limit is 3.1v but there is a risk that the console will freeze like 1.8v if it goes too low, in the end we are talking about very few mA saved which honestly is not worth "shaving off" to maybe find a paperweight during the games.

• 1,25v --> max 0.95v - 0.93v (unnecessary risk of consumption not relevant to doing this stay between 1v and 0.95v)
• 1.75v --> max 1.3v (unnecessary risk of consumption not relevant to doing this)
• 2.5v --> with the original video encoder it is better to leave it stable
• 3.5v --> 3.1v (unnecessary risk of consumption not relevant to doing this)
• 5v --> 5v same like 2.5v

Another modification will be to optimize the board so that the outputs are parallel to the motherboard so that we can have the shortest cables possible

some photos of the two rails tested 1.8v and 3.5v of how much they can go down and how much they consume, the other two total consumption console + lcd test with TPS628438 (lower consumption) and without with ams1117 at 1.5v for 1.8v (consumption 6ma higher)
It will take a while to get the new components and do further testing but at the moment I feel quite "calm"