TMS320VC5505: Estimating the maximum value of the CVDD current

Hideaki Nambu

Part Number: TMS320VC5505
Other Parts Discussed in Thread: TMS320C5504

Hello,

My customer needs to estimate the maximum value of the CVDD current. A sample value would be accepted. Could you please advise?

Condition:
* 1.3V
* Processing at Full load.
* 100MHz

From the DS p-56, we found 0.31mW/MHz(typ), that is 31mW@100MHz(typ). Then could you please add a sufficient margins to estimate a sample max value?

over 5 years ago

0 Hideaki Nambu over 5 years ago

Mastermind 9125 points

Hello,
Please try to reply. Thank you very much.

0 Mark M over 5 years ago

TI__Mastermind 29440 points

Hi Nambu-san,

The C5505 Power Estimation Spreadsheet provides CVDD power consumption based on device utilization. These power estimations are extrapolated from extensive power measurements.

There are some macros to mimic typical applications too.

See the TMS320C5504/05/14/15/32/33/34/35 Power Consumption Summary wiki page:
http://processors.wiki.ti.com/index.php/TMS320C5504/05/14/15/32/33/34/35_Power_Consumption_Summary

And download and use the spreadsheet:
http://processors.wiki.ti.com/images/2/24/C55xx_PowerSpreadsheet_RevB.zip

Regards,
Mark

0 Hideaki Nambu over 5 years ago in reply to Mark M

Mastermind 9125 points

Mark,
Thank you for your response.

Let me try to explain the reason why my customer sticks to have maximum current consumption rather the typical value. Their application is a portable equipment with a limited battery capability. The power consumption of the titled part is relatively large in their application and gives significant impact to their product specification.
Thus, to estimate the worst-case scenario, the maximum power number is required.

I know it's not specified in the datasheet, so no exact value nor guaranteed value are required. You may have some distribution data from your production test line mean and maximum under different conditions, that gives some sort of indications.

Thank you again for sharing the C55xx_PowerSpreadsheet.
I filled a full load case based on the conditions, heard from my customer.

/cfs-file/__key/communityserver-discussions-components-files/791/SPRABN0A_5F00_C55xx_5F00_PowerSpreadsheet_5F00_100MHzFullLoad.zip

The result shows the CVDD power 55.930mW @ cell U141.
Could you please guess the maximum case value?

0 Hideaki Nambu over 5 years ago in reply to Hideaki Nambu

Mastermind 9125 points

Mark,
Your response is highly appreciated.
A tough question it should be, but could you please give a sample value?

0 Hideaki Nambu over 5 years ago in reply to Mark M

Mastermind 9125 points

Please see our posts after this. Thank you very much.

0 Mark M over 5 years ago in reply to Hideaki Nambu

TI__Mastermind 29440 points

Hi Nambu-san,

You make a good point. The wiki for the Power Estimation Spreadsheet states that the power numbers for it are from nominal units.
"The data presented in the accompanying spreadsheet was measured from nominal units."

Sometimes we measure the power with hot units to provide a maximum power instead of nominal power.

Let me find out if there is a kind of scaling factor that we can use to make the power spreadsheet numbers look like maximum power numbers instead.

Regards,
Mark

0 Hideaki Nambu over 5 years ago in reply to Mark M

Mastermind 9125 points

Mark,
Thank you.
We would wait for the the maximum case value estimation.

0 Mark M over 5 years ago in reply to Hideaki Nambu

TI__Mastermind 29440 points

Hi Nambu-san,

Thank you for your patience.

I think I have found a solution to this issue. Instead of using the C5515/05 power estimation spreadsheet, which supposedly measured nominal devices, we can instead use the C5517 power estimation spreadsheet, which used devices from the high end of the power consumption spectrum.

The two cores are almost identical, but C5517 is expected to consume slightly more power than the C5515/05 under the same circumstances. This is due to its support for faster SYSCLK frequencies than C5515/05.

The C5517 power estimation can assertively be used as an upper bound for the C5515/05 devices. I plugged in the same scenario into the C5517 power estimation spreadsheet and observed a higher current than with the C5515/05 power estimation spreadsheet.

Refer to the wiki page:

http://processors.wiki.ti.com/index.php/TMS320C5517_Power_Consumption_Summary

"The data presented in the accompanying spreadsheet is an estimate that is based on measuring units from the high end of the expected power consumption spectrum."

And download the spreadsheet here:

http://processors.wiki.ti.com/images/e/e3/C5517_PowerSpreadsheet.zip

While digging around for this issue, I also found some interesting recommendations regarding power consumption that might be helpful for your customer.

1) Static or Leakage power is a function of voltage, temperature and process only. The static power does not change with frequency.

2) Active or dynamic power is a function of frequency, voltage and capacitance. The active power does not vary significantly over process (nom/cold/hot) or over temperature in most cases.

3) The IO current consumption does not usually vary over temperature and can be assumed to stay the same.

Hope this helps.

Regards,
Mark

Processors

Processors forum

TMS320VC5505: Estimating the maximum value of the CVDD current