ΣSigmaStudio Ask AI Русский Log in

SigmaStudio Parameter Value Output bug

SigmaDSP & SigmaStudio Natan Jan 16, 2013 00:00 7 replies Original on forum ↗
Subscribe
Natan Question
Jan 16, 2013 00:00

When SigmaStudio creates its output .h configuration files it outputs values in its parameter list greater than 28bits; for example "0xFF, 0x81, 0xFE, 0xB0". This causes a mismatch if you attempt to verify these values via readback. For the time being I created a quick program that takes the SigmaStudio output and parses in the correct values for example:

"0xFF, 0x81, 0xFE, 0xB0"

becomes

"0x0F, 0x81, 0xFE, 0xB0"

Sojourney Reply
Jan 16, 2013 00:00

Because it's negtive, notice the highest Bit.

Natan Reply
Jan 16, 2013 00:00

No, the 4 most significant bits sent out of the 32 bits are "no cares" because they don't exist in hardware. Therefore I see it as a bug when the output data makes any bits in this most significant nibble high; because they just don't exist and can not be high. See this thread for a similar complaint although it doesn't mention the output files specifically: https://ez.analog.com/message/33053#33053

ADIApproved Reply
Jan 16, 2013 00:00

Natan, you're correct, the top nibble is "don't care" because it doesn't exist in the hardware.

In fact, the most accurate way to display the data is neither

"0xFF, 0x81, 0xFE, 0xB0"

nor

"0x0F, 0x81, 0xFE, 0xB0"

but rather

"0xF81FEB0"

I'll check with the software team to see if they can modify the hex conversion method to leave the first nibble as zeros.

ADIApproved Reply
Jan 16, 2013 00:00

Actually, this is a bit of a tricky situation. CoolJavelin requested sign-extended data. Natan is requesting zero-padded data. Is there some kind of standard for this?

Natan Reply
Jan 16, 2013 00:00

My concern is only related to the output data and how it is inconsistent with data read back from the chip. For example if you write the following parameter:

"0xFF, 0x81, 0xFE, 0xB0"

You will read back the following:

"0x0F, 0x81, 0xFE, 0xB0"

In my mind you should only write data to the chip what you expect to read back, excluding read only registers. Since with I2C you can not read or write in nibbles but in full bytes you can not have "0xF81FEB0" (although I do understand why this is more correct to the actual hardware).

I do not understand what CoolJavelin is saying exactly because I am only talking about output data and there is no Sign relevant data in the most significant nibble.

Sojourney Reply
Jan 17, 2013 00:00

So the "0F....." , is correct to send to DSP.

When I using 5to23() function ,the results is "0F....".

But the sigma studio capture window is output "FF....".

So I modify 5to23() output as "FF...." when input is "-x.xx" to follow

The sigma studio output.

Is this way is right?

ADIApproved Reply
Jan 17, 2013 00:00

Sojourney,

If you are sending data to the SigmaDSP via I2C or SPI, you can send either 0Fxxxxxx or FFxxxxxx. The first 4 bits will be ignored, since they don't exist in hardware.