TMS320F28379D: Concurrent data acquisition via multiple channels on the F28379D ...

Hi Saleh,

There are several examples too in C2000Ware where you can see how concurrency for ADC conversions work.  One example is suggest for you to look at is "adc_soc_epwm".  In this example, it shows you how to configure the epwm to trigger ADC conversions.  Each EPWM has 2 start of conversion (SOC) triggers.  You can use these SOC signals from the epwm to trigger different channels from different ADC modules of the TMS320F28379D for true concurrent conversion where 2 channels are converting at the same time.  In the "adc_soc_epwm" example, only one ADC module is shown.  You can just duplicate the ADC setup to configure the other ADC module you want to use for concurrent conversion (note that the TMS320F28379D has 4 ADC modules).  In this context, 2 ADC modules have to be running exactly at the same time (same ADCCLK rate and ACQPS for the ADCs that are active) and conversion is triggered by the same epwm SOC.

There is also another mode where you can convert on different channels on the same ADC but conversion has to follow the SOC order (e.g. - SOC0 converts first before SOC1 and SOC2 has to convert before SOC1....and so forth), but this is not concurrent and rather sequential.  A good example to look at that illustrates how this is carried out is "adc_soc_continuous".

Best regards,

Joseph

Hi Joseph,

Many many thanks for the great suggestions and information. I really appreciate it.
Will take a look at your recommendations shortly. Hope you don't mind me asking further
questions on this. A colleague and I are trying to figure out the operations of those
channels. So far we have identified one channel and looks like 3 more to go. As mentioned,
we were successful to use one of those code examples to capture the sinusoidal from the
TMS320F28379D board to a 100MHz oscilloscope but that is only through one channel. We would
like to achieve such acquisitions with other channels at the same time and be able to display all
of it on the mentioned oscilloscope with direct connection to the board. So I guess the key point
here is the very immediate goal of ours to pinpoint/precisely identify those 4 channels
(one down and three to go) in order to accomplish the concurrency I have in mind.

Thanks again and looking forward to more of your excellent thoughts on this.

All the best,
Saleh

Hi Joseph,

Apologies to bug you again but couldn't locate the "adc_soc_epwm" example you
mentioned. Perhaps was listed somewhere under a different name? I noticed one file
by the name of EPwm.c but I guess that one is a bit different? Would be really great
if you could point me to where those examples that hopefully would resolve those
problems are I'll certainly try to download and check them out.

Thanks again,
Saleh

Hi Saleh,

No trouble at all. I hope you have the latest copy of C2000Ware, but previous versions have the examples as well. Anyway, the version I have is 1_00_05_00. The examples can be found in device_support/f2837xd/examples/cpu1. You should see "adc_soc_epwm" and "adc_soc_continuous" along with other examples.

Best regards,
Joseph

Hi Joseph,

Again many thanks for your help. Looks like I do have the latest version of 1_00_05_00
as shown in C2000Ware - v: 1.00.05.00 under the software folder of the code composer
studio (Version 8.1.0.00011). So I gather those examples will be listed in a sub-folder down
the hierarchical software tree? Will have a look at it very shortly.
I'm sure will be asking more questions on this so I very much appreciate it. Looks like it
is something of an urgent matter to us and we really like to completely solve it before we
go any further. Our intention is really to fully understand the architecture as well as the
full functionality of F28379D board we have in order to be of some use to us and then quite
likely order more of them from TI.

All the best,
Saleh

Hi Saleh,

You are welcome.  I hope you will find the examples useful.  There are several applications that the TMS320F2837D (with dual C28x processors and with 4 independent embedded ADC modules and various analog subsystem modules) would be able to support.  If you have further questions on how to use the ADC (or any of the analog peripherals), please do not hesitate to post them in this forum.

Regards,

Joseph

Hi Joseph,

I'm working on it for most of this eve so I'm sure will have more
questions to ask about those ADC modules, channels etc.
A couple of my colleagues and I plan to make a heavy use of it
for a number of applications we have around the mechanical, electrical
and biological domains. It would be in conjunction or coupled with
our own hardware systems and devices.
Therefore if all goes nicely and I understand it quite well, I'll definitely
stick with it. In the mean time, I really do need to make sure I fully
understand it. I see there is quite a bit of materials there to try out and
review.
Back to work and will chat more soon. Thanks again.

All the best,
Saleh

Hi Saleh,

I'll monitor the forum for some time tonight and may be able to answer any immediate questions you have. I'll be out of the office for a couple of days but i already notified another ADC expert about this forum post in case you have more questions while i am away. Either way, we will do our best to answer your questions.

Best regards,
Joseph

Hi Joseph (as well as colleagues there),

A couple questions or so around the two files you mentioned. I've noticed
few there so I'd like to make sure that I'm looking at the correct files. I gather
the "adc_soc_epwm" would be the file listed under adc_soc_epwm_cpu01
which will be "adc_soc_epwm_cpu01.c" ?
Same goes for file "adc_soc_continuous" again listed under
adc_soc_continuous_cpu01 as "adc_soc_continuous_cpu01.c" ?
Those are the names I've noticed there. Let me know if I have been probably
looking at the wrong folder and such files would be located somewhere else.

I have not examined them closely yet but would like for example, just as an example,
to modify Lab2_cpu01 (one of the multi-day workshop from few months ago) so that
the ADC (sinusoidal data in this case) would be captured or acquired not only by
a single channel as shown there in that example but also by multi-channels
(say 4 channels). Having an outcome of multiple sinusoidal graphs that could be
shown on the CCS and then through multiple(?) cable connection from F28379D
board to the oscilloscope for an external display of those sinusoidal multi-graphs.

Would something like this or along those lines be possible? I thought perhaps to
just grab one of those already given lab examples, say Lab2_cpu01.c, and then
modify it in such a way according to the above mentioned files "adc_soc_epwm"
and "adc_soc_continuous", as you pointed to, in order to involve multiple channels
concurrently and then display the converted data in a multi graph form ...
Furthermore, within this ADC process we need to pinpoint or precisely identify
those involved channels on the board. This latter point is a requirement for us
since we plan to utilize them in subsequent work.
Hope these rough questions/thoughts are not too vague or unclear.

I really thank you for your thoughts on this as well as any suggestions on the
best possible way to approach or handle all this. Any information in this regard
is quite welcome and very much appreciated.

All the best,
Saleh

I gather Joseph is out of the office today and tomorrow as he
mentioned in his previous reply. Would appreciate any thoughts
from any of his colleagues there on some of our questions.

Thanks,
Saleh

Hi guys,

Any expert out there on the C2000 Launchpad XL TMS320F28379D ?? If so
would love to chat with you via this forum regarding issues of concurrency
by the system... as well as on the use of multi-channels during the ADC phase,
etc. If you happen to be an expert or quite experienced using this system,
I really would love to chat.

Many thanks,
Saleh

Hi Saleh,

If you want to sample 4 channels in parallel start with the adc_soc_epwm project:

\ti\c2000\C2000Ware_1_00_05_00\device_support\f2837xd\examples\cpu1\adc_soc_epwm

What you want to do is have 1 trigger source (ePWM in this case) trigger SOC0 on all 4 ADCs and then only 1 ADC triggers an ISR where you collect the data for all 4 ADCs.

So for the epwm example just duplicate

Uint16 AdcaResults[RESULTS_BUFFER_SIZE];

And

    AdcaRegs.ADCCTL2.bit.PRESCALE = 6; //set ADCCLK divider to /4
    AdcSetMode(ADC_ADCA, ADC_RESOLUTION_12BIT, ADC_SIGNALMODE_SINGLE);
    AdcaRegs.ADCCTL1.bit.INTPULSEPOS = 1;
    AdcaRegs.ADCCTL1.bit.ADCPWDNZ = 1;

And

    AdcaRegs.ADCSOC0CTL.bit.CHSEL = channel;  //SOC0 will convert pin A0
    AdcaRegs.ADCSOC0CTL.bit.ACQPS = acqps; //sample window is 100 SYSCLK cycles
    AdcaRegs.ADCSOC0CTL.bit.TRIGSEL = 5; //trigger on ePWM1 SOCA/C

And

AdcaResults[resultsIndex++] = AdcaResultRegs.ADCRESULT0;

For ADCs B through D.

Probably you can do something similar for the workshop example.  

Hi Devin,

I really appreciate your message and apologies for the late reply.
It has been a very hectic schedule so have not had much chance to
take a look at it till just a minute ago.

I'll take a look at that and will definitely be chatting more with you
today. One example I had in mind when I start chatting with Joseph
the other was the example in Lab2 of the workshop. The sinusoidal data
there go through the ADC then get displayed in the graph module of CCS.
We also have managed to get the same graph shown in an oscilloscope. All
that done through one channel that we captured. If there is a way perhaps
to modify that source a little so that ADC is done through multiple channels,
hence, multiple displays simultaneously that would be wonderful.
I gather there are in total 4 channels there and we already figured out one
of them so now we need to also pinpoint the other three. Our objective in all
this is to couple/interface the F28379D board with one of our own devices
which we have built in-house.
Such an interface would need to be done through those 4 channels.
As mentioned to Joseph, if we are able to do that we could harness the
power/functionality of the micro-controller in a much more efficient and
effective way fashion. Once this becomes successful we intend to order
more units from TI.
That is our current plan therefore we now need to understand it quite
thoroughly. I very much appreciate your help in this regard.

Cheers,
Saleh

I have now a project that seems to build and run fine on the CCS (workspace_v8)
as adc_soc_epwm_cpu01.
The changes you suggested will only involve the source in adc_soc_epwm_cpu01.c or
also I need to handle other files under the project?

Could you also suggest another simpler example? Perhaps something along the lines
of lab2_cpu01 of the workshop to handle a sinusoidal data set or something similar?

I need to hand all of this over later this week to my colleague who is a hardware engineer
but not much into the software side ...
Got a long list of paper deadlines that I really need to take care of.

Thanks for your help.

-Saleh

Hi Devin,

The replication of those lines of code would be in the same file? Would they need
to be moved into certain function(s) within it or anywhere in the file... ?

Thanks again,
Saleh

Hi Saleh,

To be blunt, we aren't going to write your application for you.  The next step you need to take is to thoroughly read the ADC chapter of the TRM () and then run the workshop lab.  Look at each line in the code and reference against the bit and functional descriptions in the TRM until you understand why the lab works.  Repeat for the epwm example. 

Once you understand both examples, use the hints from my previous post to extend one example or the other to sample multiple ADCs in parallel.  Note that there is also a mini code-snippet example in the TRM about sampling multiple channels simultaneously that you can refer to.  

We are more than happy to help with understanding the HW and SW and happy to give some guidance as to the best route to implementation, but I can't just give you the code you want.  This is necessary because you need to be able to support the code you write and extend it further as your application grows.   

Hi Devin,

We are more than capable to code our own in several PLs as well as various

algorithmic approaches etc. so we are not asking for others to do our work .

Sorry if you got the impression that I was asking you to do that.

We acquired a hardware system from you guys that we're trying our best to fully

understand so we could better utilize it in conjunction with other hardware systems

we have in-house for a number of planned applications.

If we don't understand it we just won't be able to use it regardless of how new or

innovative it is. In any case, thanks for the back and forth chats and really

very sorry to waste your valuable time as well as for any misunderstanding on this.

Have a good weekend.

-Saleh

Hi Saleh,

Sorry, I was just trying to be very direct that I think the next action in getting your application successfully working is on you to understand the documentation and examples a little better.

We're happy to spend time helping you. Feel free to come back if you run into specific questions about the documentation or about how the SW examples work, or if you get stuck once you move onto implementing the simultaneous sampling.

Thanks Devin. I do understand so no problem. A colleague who has a very long experience in
hardware design as well building circuit boards was supposed to handle all this but unfortunately
does not have much of strong software background. In any case, he is heading out out of town
so looks like the project is back to me. The schedule on my side is quite swamped with paper
deadlines and a couple talks to come but I'll get back at those documents and your suggestions.
Coding is not the key issue here as far as I understand it, but it is the interplay between software and
the hardware of the given systems. Therefore I really need to understand the latter quite well.

In any case, I hope you don't mind me asking more of those questions. If any of those
questions are not too clear or perhaps I'm asking too much or you have no time to
reply/answer them, that would be fine as well and no worries.
Many thanks again and sincere apologies for any misunderstanding.

Cheers,
Saleh

Hi Devin/Joseph,

I hope you don't mind if ask about those lines of code replications you suggested.
Just got the chance to look at adc_soc_epwm_cpu01 project on CCS. It builds nicely
with no problems. Those lines of code replications are in adc_soc_epwm_cpu01.c
file so any of those replication would need to take place there?
I gather some changes would need to be done with F2837xD_Gpio as well as
F2837xD_DefaultISR, no? I guess would need to dig a little deeper into the code set.

Thanks,
Saleh

Hi Devin,

Wishing you an enjoyable Labor Day holiday. Back again to the project to
hopefully wrap it up soon. Plan to work on it over the weekend so will
try to keep you posted. Thanks for your very useful pointers and help
in this regard. I appreciate it.

All the best,
Saleh

Hi everyone,

Back to CCS to wrap up our project but noticed a Java-type error that keeps
occurring at every function:

In a small window: java.lang.NullPointerException

Regardless of the function/step taken, it keeps popping up... Any suggestion
on how to fix this? I really need CCS back in operation to finish our work. We
don't have much time to start all over again from scratch on this.
I appreciate any help.

-Saleh

Looks like related to Eclipse I believe but so far unable to resolve. Any help on
this would be great.

Hi Saleh,

I'd recommend creating a new thread with a detailed description of the CSS issue you are seeing. If you tag this as CCS and not with a specific C2000 part it'll end up in the CCS/Tools forums which will be much better equipped to debug this type of issue.