[SOLVED] J-Link target power

jev · Feb 21st 2019, 3:55pm

I'm feeding my target hardware from the J-Link Pro when debugging. The 5V output is regulated down to ca. 3V using an LM317T. The hardware connected to it is known to peek at ca. 10 mA. However... the J-Link Control Panel however says:

Yes, it really says "18161 mA". That's 18A and a continuous current of 6.2A. Whoa! I guess it is using smoke signals inside the processor core!

(hint: there's a couple of decimal points missing).

On a more serious note: how precise are these current measurements? My software checks Vbat at 2.866V which is pretty close to the 2.85V shown by the device driver's dialog so I guess the presented voltage is pretty good. Current however is slightly less accurate (checked against a Keithley DMM7510 - about 5 mA (including the LM317 regulator) versus the 6.2 mA shown by J-Trace).

SEGGER - Nino · Mar 4th 2019, 2:04pm

Hello,

Thank you for your inquiry.
First of all good cooling of your J-Link and target CPU there

Jokes aside this is most likely a calculation error.
The J-Link software version you are using is quite old, could you try the latest one and see if the values improve?

jev wrote:

On a more serious note: how precise are these current measurements?

The current measurement has a resolution of 50 µA for J-Link PRO/Ultra.
For J-Trace PRO it is possible that it is yet a bit inaccurate, this will be fixed with a future FW update and should be in line with J-Link Pro.

Best regards,
Nino

jev · Mar 4th 2019, 5:13pm

Hi Nino,

The latest and greatest v6.44 "suffers" from the same issue. It's no biggie since it's clear that either there's just a decimal point missing or there's a mismatch in unit of measure of course, it's just silly.

SEGGER - Nino wrote:

The J-Link software version you are using is quite old, could you try the latest one and see if the values improve?

Anyway, 50 uA accuracy would be good enough for daily testing and power optimisation.

SEGGER - Nino · Mar 5th 2019, 2:39pm

Hello,

Thanks for verifying. We will investigate this further and notify you here once we have more information.

Best regards,
Nino

SEGGER - Nino · Mar 5th 2019, 4:07pm

Hello,

Quick update on this matter. We found the source of the issue and it should be fixed with the next J-Link software version.
However the Windows based J-Link Control panel is going to be discontinued and replaced by the webserver version:
wiki.segger.com/J-Link_control_panel

For proper power profiling especially with your J-Link Pro probe we recommend using Ozone and the new power profiling features:
segger.com/products/debug-prob…chnology/power-profiling/

Best regards,

Nino

jev · Mar 5th 2019, 11:06pm

Ah, thanks for the heads-up. Will I still be able to see power consumption and voltage levels through this webserver?

Also, thanks for the pointer to the power profiling features from oZone. I'll sure play around with it when I find some spare time!

It is a pity the probes provide 5V Vcc only. I think it's safe to assume most (if not all) modern low-power designs require lower voltages (my hardware for example will run from a single CR2032 battery once it goes into production - the nRF52 and the sensors I use all accept anything from 1.7V up to 3.6V). "Proper power profiling" thus won't really be possible since in development an extra voltage regulator between J-Link and target is required. May I suggest to add programmable Vsupply output to J-Link (or J-Trace) in future models?

SEGGER - Nino · Mar 6th 2019, 9:49am

Hello,

jev wrote:

Will I still be able to see power consumption and voltage levels through this webserver?

Thats the Idea, the webserver version should replace the "Windows" versions 1:1. It is currently a transition phase so it could be that some features are not yet ported completely. But once the "old" Control panel is gone you should see all functionality in the new control panel.

jev wrote:

May I suggest to add programmable Vsupply output to J-Link (or J-Trace) in future models?

We had already considered this in the past, however the feature is not so often used so the extra cost to have a modular voltage regulator on the J-Link itself could not be justified as it would increase the general cost of J-Links which all customers would have to carry.
This is why we offer an adapter for this setup case as a workaround for customers that need it:
segger.com/products/debug-prob…rs/target-supply-adapter/

Best regards,
Nino

jev · Mar 7th 2019, 11:56am

Fair enough. The workaround basically is what I built myself. The disadvantage of using such a solution is that the extra hardware spoils the power profiling, so I'll just use the good'old 1 Ohm resistor and oscilloscope.

SEGGER - Nino · Mar 7th 2019, 2:15pm

Hi,

jev wrote:

he disadvantage of using such a solution is that the extra hardware spoils the power profiling, so I'll just use the good'old 1 Ohm resistor and oscilloscope.

It is the best trade off currently available for that setup. Also the relative current changes will still be accurate enough to take conclusions from the readings. You can also calculate it absolute again as the dropout voltage of the used voltage regulator on our adapter rises linearly with the current so it is a linear offset over the load. Quiescent currents are in the single digit µA area which is below the displayed resolution so it should not have any effect.

As the initial inquiry has been answered this thread will be closed now.

Best regards,
Nino

Share