Let's build a laptop

[kwankiu]

• plugged in :
  5V DC input => TP4056 Lithium Battery Charger Module => Lion battery => Boost converter => 12V PD output
• unplugged :
  Lion battery => Boost converter => 12V PD output

I don’t think there is any PD involved in input,

Well i mean,

The rock 5B usb-c should input PD 20V when plugged in, for example if the device is plugged to a USB-C monitor with DP and PD 60W. In that case, the ups should be in charging mode to the battery, and the ups will not output any power to the rock 5B. However, the ups charging input seems wants no more than 5V input, which means when a usb-c monitor is plugged in, 20V is supplied to rock 5B which the rock 5B will work normally, however, the ups board’s charging ic may be broke when 20V is supplied.

If you are using the board i suggested on this post

I found a module that output the battery voltage instead of using a buck or boost module, it can output upto 50W, it runs 3S battery. However, it seems only support input upto 15V

It will be possible to have pd input upto 15V (which means u have to aware of whether your power supply support 20V, if it support 20V, it might not be a good idea using it, or you might have to modify the u-boot / kernel to disable 20V negotiation.

[kwankiu]

I am actually thinking of supplying the ups battery output using the PoE power 5V pin as one of the solution to this 20V PD issue.

However, in this case, we need a module that converts our battery to something around 5V 8A output.

And I have a question on the 5V pin, when the Rock 5B is powered by 5V using GPIO or the PoE, does the usb-c port output 5V power?

Since I see that the usb-c vin goes to the mp8759 and negotiate with the pd controller to allow dynamic input from 5-24V, if the rock 5B is powered using gpio or poe, we are powering everything upto the 5V net, but i am not sure what will the USB-VIN behave in this case when a usb device is plugged into the usb-c port.

[Etienne]

The rock 5B usb-c should input PD 20V when plugged in, for example if the device is plugged to a USB-C monitor with DP and PD 60W. In that case, the ups should be in charging mode to the battery, and the ups will not output any power to the rock 5B. However, the ups charging input seems wants no more than 5V input,

ah yes I didn’t consider power coming from screen itself, this will be an issue indeed

And I have a question on the 5V pin, when the Rock 5B is powered by 5V using GPIO or the PoE, does the usb-c port output 5V power?

Didn’t dig much into rock5b powering topic, so I’ll be of little help here but would be interested if you get more info on that.

It will be possible to have pd input upto 15V (which means u have to aware of whether your power supply support 20V, if it support 20V, it might not be a good idea using it, or you might have to modify the u-boot / kernel to disable 20V negotiation.

Have you considered PD trigger board? I’ve got two of them, I know at least one is able to passthrough video signal while forcing a predefined PD setting (like 15V).

Otherwise yesterday was experimenting/hacking things out of curiosity but now realizing it won’t be appropriate here (due to external screen PD power).

PD power supply => 20V PD trigger => 3S charger => 3S Li-ion + BMS => PD car adapter => USB-C breakout board

with both screen and rock5b connected to usb-c breakout board

The PD car adapter accepts 12-24V DC input and is able to deliver PD output with enough power to rock5B. While working well with only rock5b connected this probably wont be doable with another PD device in the loop…

Finally this UPS topic requires more reflection that I though…

[kwankiu]

ah yes I didn’t consider power coming from screen itself, this will be an issue indeed

However this seems like what the first post aims to achieve.

Didn’t dig much into rock5b powering topic, so I’ll be of little help here but would be interested if you get more info on that.

It would be perfect if somebody could do a test (simply if somebody is using a poe or a 5V power supply from the gpio, he/she may be able to plug a normal usb drive to the usb-c port to see whether it works or not), I don’t currently own a usb to pin header power cable, so i couldnt test that out at the moment.

[kwankiu]

Have you considered PD trigger board?

I got two of them (one is lost, another one is used to trigger 12V direct input to the rock 5B’s usb-c port as a solution to boot loop / armbian kernel that does not support pd.

Tho u can use a pd trigger board on the usb-c port to limit a max 12V power input, but the rock 5B has a pd trigger chip already, so i would still prefer using it instead of needing an additional pd chip, on the other hand, if you choose to use the direct soldering method, the pd trigger board will be plugged after the rock 5B’s pd trigger chip, which means the additional pd trigger board will be ignored as the power supply first got response from rock 5B’s chip to negotiate 20V.

[Etienne]

the pd trigger board will be plugged after the rock 5B’s pd trigger chip

I guess this only makes sense in front of rock 5B, as this was an alternative to hacking kernel PD negotiation.

Upon further consideration, seems like PD is not wanted on the UPS at the end?

• while being connected to display, this would make 3 PD devices trying to talk to each others, not sure what the outcome would be
• considering negotiation is done through data pins (TBC)… UPS plugged on power pins only, would make PD useless

Seems like requirements for UPS would be:

• bidirectional input/output
• able to take 5V-20V as input
• and at least 20W output

[kwankiu]

Upon further consideration, seems like PD is not wanted on the UPS at the end?

We definitely don’t need PD, the reason to negotiate 20V PD is to allow more power going into the rock 5B which is then converted to 5V in the rock 5B.

Same purpose for UPS, allows more current to go in instead of only 5V 3A for the whole Rock 5B and its other connected devices/HAT/UPS

hacking kernel PD negotiation.

If kernel limiting at 15V is a solution, the board i mentioned earlier should just works.

[ThomFranken]

Just found out that 5V 6A via GPIO seems to be possible:

Rock5 does not work on most PD power supplies

[IMG_20221129_003645] Tried a few PD chargers and they all resulted in endless restarts. 30w of power through the GPIO using a 5v 6a power brick. I used 2 in-line 3A fuse to prevent potential overcurrent, but it may not be necessary. This is just proof of concept at the moment, I’m now planning a breakout board eventually.

[ThomFranken]

I do not fully understand your Ideas here - so maybe it’s pointless - but does that ZC828 mentioned here could be helpful in anyway:

Rock5 does not work on most PD power supplies

For me, RADXA POWER PD 30W is sold out right now. Since I still want to use my PD chargers, I found out that using YZXSTUDIO’s ZC828 (which uses the Zhirong SW3518S chip) as a relay (PD in, PD out, it also supports DC input) can make it work on every PD charger that I have. It can negotiate 20V very quickly and is reliable. I will use this as a temporary solution until a stable U-Boot firmware solution is available. [image]

And do I understand it right that “Let’s build a laptop” would be much easier if one would drop the feature of an external USB Type-C (data) port?

Like just using a powerbank with passthrough and one of these usb-c to dp/hdmi/vga adapters that has an aditional usb-c port where that powerbank could be plugged into.

[Etienne]

Yeah that was overall the idea, passthrough data from Display to rock5B, then regarding power pins, I see two options available

• passthrough everything, UPS acting as man in the middle, meaning it won’t interfere with R5B and Display powering but just take current to charge itself (that was my first idea with the male to female breakout board)
• only passthrough data, and redirect power to UPS (I think that corresponds to your previous design)
  In that case I would presume, the R5B will still be in charge of PD negotiation with Display but won’t receive its power directly from it. Instead negotiated voltage will go to charge UPS, and UPS will power rock5b independantly (as if it was unplugged from screen). This has to be confirmed, as I’m not sure this usb-c hack is possible

In your design, you used PowerBank instead of UPS. Although it seems to have 2 bidirectional usb-c connection, usually PB are not able to charge/discharge at same time. Which means it may cut power to rock5B while charging.
That’s why I spoke about UPS (Uninterruptible Power Supply) instead of PB

Based on that, overall design could be:
Display —> External UPS —> R5B

with two USB-C connection exposed outside external UPS

• USB-C input connected to display
• USB-C output connected to R5B
• breakout power pins internally connected or redirected to UPS charging/discharging circuit

An example of custom UPS implementation then could be (supposing 12V and 3S Li-ION is chosen) :
(PD trigger board) => USB-C Breakout board => 3S charging board => BMS => 3S Li-ion batteries

PD trigger is optional, just in case UPS would only accepts narrow input voltage range, but I think it is possible to find 5V-20V input range

Please note these are only assumptions, that would be good if someone could confirm it would work