AMC1333M10: SDFM output spread is too much

Hi Jitendra,

Thanks for your question. Some questions for you:

1. Could you please clarify what you mean by "Data samples spread is too wide to be used for feedback control loop of DC-DC converter"? Can you provide any reference pictures or similar so I can better understand the issue?
2. Which filter capacitors are you referring to regarding the 100pF and 4.7nF caps?

Best,

Samiha

Hi Samiha,

Thanks.

1. Yes, input is 740V (ripple peak to peak < 4%) while the SDFM data output has variation of 12%. I can't measure low voltage analog signal at the input of the device as we don't have low voltage isolated probe. Picture of high voltage DC signal is attached (note that channel coupling is AC).

2. Section 8.2.2.1 Input Filter Design in the datasheet refers to the filter design at the input. It is C30 in our schematic.

Regards,

Jitendra

Dear Samiha,

Thanks.

We're using OSR - 256 which is maximum. Do you mean software based filtering? Or do you mean using 20 MHz sigma delta modulator.

We also want to use this output for protection. Each SDFM channel has two filter units: data filter unit and comparator filter unit. Max. OSR possible for comparator filter unit is 32. Comparator filter unit is being used to interrupt the PWMs.

Also more oversampling results in more delay.

Below is the data of SDFM filter of 1000 consecutive samples. With OSR 256, data rate is 39.1 ksps (~25 uS per sample).

Looking forward for your help.

Regards,

Jitendra

Dear Samiha,

Thanks.

As I mentioned in the earlier post, we are using this input for protection using comparator filter in TI MCU. This is hardware based and has maximum OSR value of 32. Software has no role to play in it.

We are following most of the operating conditions mentioned in the datasheet of the device. Can you suggest how to use AMC1333M10 to get good results?

Regards,

Jitendra

Hi Jitendra,

Jitendra Jain said:

Below is the data of SDFM filter of 1000 consecutive samples. With OSR 256, data rate is 39.1 ksps (~25 uS per sample).

Is the graphed data you're showing the output of the SDFM data filter (and scaled), or something else? At this point in time are you trying to improve the SDFM data filter readings / performance? The C2000 SDFM's data and comparator filter are equivalent, other than the OSR max difference.

Can you verify what input qualification you are using on the SDFM GPIO pins of F2837x device? We recommend 3-sample qual be used, see info below.

If the issue is not related to noise on the SDFM signal inputs to the C2000 then this may be a system noise issue on the sensing side, like Samitha mentioned.

Best,

Kevin

Dear Kevin,

Thanks.

Yes, the graph is output of the SDFM filter. We're using DOSR 256 with 9 bit right shift which gives data output of 15-bit (signed). This output is scaled in according to resistor divider on the input of sensing device (R1 - 800k, R2 - 667R).

In the current application, we are using both data filter (OSR - 256) and comparator filter (OSR - 16). Comparator filter is used for protection to trip PWMs in hardware. But the problem is that it is always trips in current scenario.

We were using aync settings on the GPIO which we took from the C2000ware example. Today tried with 3-sample qual settings but the data filter result is same as earlier.

How to improve the system noise issue with AMC1333M10? Any suggestions.

Regards,

Jitendra