Questions about EnergyTrace

It is possible to use the energytrace-enabled MSP-EXPFR5969 eZ-FET lite with all MSP430 devices. The process is described in the MSP430™ Advanced Power Optimizations appnote. You do need to have the new revision of the board that has a production MSP430FR5969, not the old one which has an XMS430FR5969.

The appnote specifies the sample rate for that configuration as being 1kHz. Due to the nature of the energytrace method spikes shorter than 1ms aren't lost, but will be averaged out over the 1ms period.

You can monitor external loads as long as they are powered from the eZ-FET lite's V+ connection to the target MCU. You'll have to be careful not to overload the DC-DC regulator, however.

Edit: Regarding continuous capture, CCS6 does not support that at present. The maximum capture duration is 5 minutes. IAR Workbench might be able to perform continuous capture (surprisingly their EnergyTrace UI appears to be more mature than TI's).

Hi Robert,

Thanks for answering and pointing me to the sources that I have missed.

With a limit on sample rate, I am wondering why TI pick that method over the standard current shunt monitor approach, with which much higher (often desirable) sampling rate is possible. Is there anything there that I don't understand?

I have tried IAR Workbench, their continuous representation contains bars, not chart, but I guess it is due to low sampling rate.

Regards

Jan

The 1kHz limit is only applicable to the Ez-FET Lite on the FR5969 launchpad. I don't know what the limit is for the MSP-FET, but it's higher than 1kHz AFAIK.

The EnergyTrace patent has some information on the claimed benefits of the technique compared to current shunt measurement.

Thanks Robert,

Can anyone answer if MSP-FET has the same sample limit (1kHz) as Ez-FET Lite, please?

Jan

OK,

Have some answers here http://www.ti.com/lit/ug/slau157ad/slau157ad.pdf

Q: What is the sampling frequency of EnergyTrace++ technology?
A: The sampling frequency depends on the debugger and the selected debug protocol and its speed
setting. It typically ranges from 1 kHz (for example, when using the Spy-Bi-Wire interface set to SLOW) up
to 5 kHz (for example, when using the JTAG interface set to FAST).

and here http://www.ti.com/lit/an/slaa603/slaa603.pdf

Specifically, when using the eZ-FET on-board emulator via
Spy-bi-Wire the sampling frequency is 1 kHz, and the MSP-FET via JTAG runs at 10 kHz. This means
that some information may be lost due to rapid code instruction execution and associated device state
changes.

The numbers given are different.

Jan

“surprisingly their EnergyTrace UI appears to be more mature than TI's”

jan2809 said:

With a limit on sample rate, I am wondering why TI pick that method over the standard current shunt monitor approach, with which much higher (often desirable) sampling rate is possible. Is there anything there that I don't understand?

The problem with shunts is that the measurable current range only spans one or two decades. You need current-dependent range (shunt) switching and you’ll lose energy during the switch.
Also, you have a voltage drop on the shunt. For small currents, you need a large shunt, which will generate a large voltage drop for larger currents.
IN addition, digital currents, especially with MCUs, are not uniform, but multi-MHz peaks. A simple voltage-over-shunt measurement is next to impossible.

The EnergyTrace measures energy put into the supply rail. Knowing the output voltage and the energy consumed over times gives the averaged current over this time. And since the energy is counted by counting the switching cycles of the supply (with a fixed amount of energy supplied per switch cycle), precision can be kept across several decades, from nA to mA. Without much additional circuitry.

Hi J-M,

My main concern now is EnergyTrace (not EnergyTrace++).

Do you know if when running in "FreeRun" mode (no communication between JTAG and target MCU) the sample rate (of current measurement) is the same for MSP-FET and eZ-FET?

As well my feedback is to bring a continuous (in time domain) display of current. voltage and power - something in the style of attached picture. (with adjustable time axis). My guess is that It need an implementation of a circular buffer on PC, which should be possible. As it is at the moment taking the snapshots is very inconvenient IMHO.

Can you bring it to the attention of whoever is in charge of it?

Jan

J-M,

For for that explanation, appreciate it.

However I am not fully convinced. For most of my application the current range to be measured is 0.1uA - 50 mA and my main concern is the lowest range of it. The simple, relatively accurate, practical and reasonably priced tool would be nice.

With a shunt resistor let's say 4.7R you have 4.7R * 100nA = 0.47uV and 4.7R * 50 mA = 235mV. Both acceptable voltage drop for 3.3V (or similar) supply and measurable, if switching the gain of the amplifier. TI has a number of reference designs for a similar solution, so most likely they know what they doing when making the choice. However everyone is different and have different needs, so there is no universal solution to make everyone happy.

Making the EnergyTrace user interface more friendly will most likely gain more acceptance from me.

Regards

Jan

Since eZ-FET and FET430UIF use the same hardware (same processor and clock) and the plain EnergyTrace doesn’t access the MSP, I would expect the sample frequency being the same.

 Regarding your image, you want sort of a scope display. Well, shouldn’t be too difficult.
Perhaps the source for the standalone UI is available soon, and you can implement it on your own?

Thanks again.

And again I have to disagree with you. I know at least 2 commercial solutions doing it the way, that you describe as very challenging and there might be more in that class. Only recently I discovered that TI's partner IAR has I-jet and I-scope, but it works only with ARM, not with MSP430 as I need. I still need to study and experience it deeper to see more clearly.

I'm not going to follow the " if you not happy do it yourself" suggestion.

Providing my feedback is solely for for a purpose of improvement, (even if others might disagree with it) and I am sure that the EnergyTrace solution will keep improving.

Regards

Jan

Thanks, appreciate it.

Will keep sending feedback both appreciation and a constructive criticism.

Jan

William Goh said:

I took a look at the I-scope and this requires you to provide an external shunt resistor for the tool to profile the current. They do so in such that you can switch to different resistor values based on the current load of your application or how much resolution you are looking for.

It looks like the I-jet has a built-in shunt resistor that allows it to measure the current supplied to the target board as a whole. The user guide specifies the I-jet's current measurement resolution as ~160uA*.

* The I-scope manual lists the I-jet's current resolution as ~160mA, which is hopefully a typo :)

jan2809 said:

I'm not going to follow the " if you not happy do it yourself" suggestion.

Well, that’s your choice. It's amazing, how often people use the words "it's so simple" and "I won't do that" in the same sentence. Few (including me) ever say "It's so simple, see, I have done it".
Of course you can “live unhappily ever after” or hope for improvement. Which is rather unlikely due to practical and financial considerations. Once more, the current EnergyTrace approach is barely more than a (very clever) software solution of counting energy pulses injected into the supply. It barely requires additional hardware. So it is dirt cheap in production and can be easily built into LaunchPads or FETs without noticeable additional cost. In fact, you get it for free when buying a FET or LaunchPad.
A switching shunt resolution is way more complex.