Jetson TX1 supply votage range

Hello All,

Default power supply comes with JTX1 is 19V … Does it work properly with 12V too ?

Thanking in advance…

Hi KapilMehta,

I didn’t test it but it should (VDD_IN can be 5.5-19.6V).
If you are about to use the DC plug from Jetson TK1 for Jetson TX1 I warn you that it is not the same jack form factor…

Regards,
Ale

no not TK1 but any other adapter with 12V will work right ??
and is there any limitaion for current , in total any limitation for power watt?

The one provided by NVIDIA is rated for 4.74A/90W.
Even a smaller one would be safe

FYI, part of the reason for maximum power/current requirements are the provision of power to the PCIe x4 slot and to power USB ports. If you use externally powered USB or USB parts with low power requirements, and if you don’t use PCIe, then it would probably be safe to use far less than 90W rating. I don’t know how much less, but it would be “a lot less”.

@linuxdev : right and far less if you don’t use SATA :)

I managed to get the board to work with a 12V adapter, but it took two power button presses for it to turn on. I guess some capacitor had to charge first…

Be aware that it’s a 5.5mm outside/2.5mm inside plug, which is not the one most power supplies use (that one has 2.1mm inside).

However, see my other thread here about the problems I’m having with getting the board to boot with anything except the original power supply and that one 12V power supply. There’s something I’m missing there.

I got curious, and I hooked up my TX1 dev board to my HP bench power supply (an E3431A; yes, HP, not Agilent, it’s that old :) ). The 0-25V rail only goes up to 1A, so I started it at 24V. It did take a couple of presses of the on button for the board to power up. At 24V, it was drawing anything between 50 and 150mA; with all four arm cores running infinite loops (so, 400% CPU utilization), the current went up to 380mA or so. I don’t have anything handy to keep all maxwell cores busy too, and of course nothing was connected to either the usb or the pcie port. It kept powering up, with more current draw, down to 18V, again after multiple presses.

The funny thing is that although it would power up at 19V, if I lowered the voltage from 24 to 19 while it was running, it would turn off. Probably the power sense circuitry thought it was a transient and preemptively turned it off to avoid a potential subsequent spike.

It would not power up at 12V, but that may be because it was trying to draw more than 1A.

The maximum voltage rating is around 19.4V…24V might work, but it also might damage it. 12V should work, it’s well within the voltage range…needing more current might be the key there.

well, I guess I just publicly admitted voiding my warranty :)

I can confirm that it works, but there apparently is a huge current spike at bootup. I talked to a very good EE guy about the symptoms I’m seeing, and he said that he suspects that the capacitors on the board have to be charged up very very quickly when pressing the start button, otherwise the startup is aborted.

So, the board does only need <1A while it’s running, but for startup it probably needs much more. Maybe that’s why there’s a 90W adapter included.

Even when I used the official power supply but just added a 20cm extension chord (socket -> cable -> plug), it wouldn’t start up, maybe due to the added resistance. I can’t insert anything to the rail to measure the current while it’s powering up, so it’s hard to say what really happens there.

I’m thinking of adding a large capacitor at the power plug, but the EE guy I mentioned told me that this could lead to a current spike that exceeds the maximum amperage the board and the connector can bear.

If capacitor charge and discharge are separated via diode polarity (perhaps a Schottky with fast switching and low dropout), and a limiting resistance is inserted (perhaps a different resistor for charge versus discharge), then the capacitor would be practical to adapt lower max current supplies.