AMC1304M25 Delta Sigma Modulator and TMS320F28377S SDFM

As far as I got to know so far, i need to increase my Filter data rate somehow.

Filter Data Rate = Modulator Data Rate / DOSR

If I decrease the DOSR from 256 to 64 or 32, I get a nice sine wave again but of course, resolution of the filter will not be that high as before.

Clock 10 MHz, DOSR 32, Analog Signal : 10KHz 100mV Vpp

I would be happy if someone can confirm if this is was the right solution.

Best Regards,

Inam

Inam,

Why does your output waveform look amplitude modulated? Can you send a snapshot of SDFM settings in memory window?

Regards,

Manoj

Hi Manhoj,

Did you mean the waveform that I put under b) Analog Signal: 100mV Vpp, Input Frequency 10KHz, Clock 20MHz in my first Post?

That is the problem that I get when I feed an analog input signal of 10Khz with SDFM settings I mentioned before in my Post.

When I change my data filter OSR from 256 to 32 with 10MHz CLK, then I get the waveform that I uploaded in my last post.

My question is

a) Is it a right solution to decrase DOSR when input signal has a high frequency of 10KHz?

I tried out taking a snap of memory browser for SDFM1, as I am new at Code Composer Studio, I am not sure if I could get what you meant

SDFM settings snap in memory browser

Regards,

Inam

Inam,

My guess is we are looking into a aliasing problem. When you reduce your DOSR, you are increasing the data filter sample rate and hence you are able to reconstruct the signal properly. Why do you want to provide 10KHz signal to modulator? (is there a specific reason?)

What are your application requirements?

Pg 19 AMC1304 datasheet, shows that they have tested with 5KHz signal (with 20MHz CLKIN and OSR = 256). Did you already try that?

Regards,
Manoj

Hi Manhoj,

Do you think if input signal is 10KHz or 20KHz, Modulator Clock frequency 20MHz, SDFM >Mode 0 so Modulator Sample Rate 20MHz
can it cause aliasing problem with low DOSR such as 32? Data Rate of filter is hence 625KSPS.

I need to use AMC1304 for inverter current and voltage measurements for automotive applications. For current measurement I don't see any aliasing problem but I am afraid that voltage signals (phase-phase) must be having higher frequency around 10Khz and must be having alot of switching noise over them.

I tried an input signal of 5KHz as in AMC1304 datasheet but the filter output signal in time domain doesn't seem that good. Therefore, I had to reduce DOSR.

Regards,
Inam

Inam,

With 5KHz, you should get a clean sine wave in time domain. Something is wrong in your code setup / hardware setup. Please check.

Regards,

Manoj

Dear Manhoj,

I get this Sine with 5KHz input signal frequency 100mVpp with maximum resolution of filter (OSR_256, Sinc3).

When I increase input signal frequency to 10KHz, I have only 2-3 samples in one sine half with same filter settings and clock frequency of 20MHz

Regards,

Inam

Hi Manhoj,

You are right ! with 20MHz Clock, I am able to get clean sine wave with 5KHz input signal frequency.

But because my inverter switching frequency is 10KHz, I wanted to test it with 10KHz input signal. In that case, the output sine wave is not clean. A snapshot from input signal with 10KHz frequency:

Is there anyway to solve this problem without changing the output data rate of filter?

Inam

Hi Manhoj,

Thanks alot for drawing my attention to this.

I had some problem with calculating two's complement in my interrupt routine. I am able to get the similar signal now as you

When I sample at input frequerncy of 10Khz

Regards,

Inam

Inam,

Please close the post, if your query has been answered.

Regards,

Manoj