PCM4222 Modulator Output

Hi, Charles,

Figure 22 of the d/s shows linearity from input to output across the range.

If you want to verify operation in your system, we have an EVM available for this device.

-d2

Hello Don,

Thanks for the message. Figure 22 shows the linearity of the PCM output. We are using the modulator output.

In the old days of flash or successive approximation ADC's, a full-scale input signal would result in a full scale output signal (eg, 0x00 0000 to 0xFF FFFF on a 24-bit device). But with these modern delta-sigma devices, the high order feedback loops create instabilities if the output is too high or too low. The output is more or less a standard output overlaid with a lot of noise, so that if one looks at the modulator output pins, it is more of a probability distribution function than a fixed output.

We have a unit up and running. When we go to full scale minimum DC input signal, the modulator pins go to zero for the four MSBs and then the 2nd LSB has about a 10% duty cycle and the LSB has about a 50% duty cycle. So it seems that putting in a full-scale input results in an output that is roughly -7 dB below the expected full-scale signal of 000000b or 111111b.

I am looking for confirmation as to what the exact full scale range of the modulator output is. Thank you for your kind assistance.

Charles Hansen

Hi, Charles,

Wow, that was quick! Soemthing bothered me, so I was looking into this a little more and realized I missed your question!

Let me look into this for you. It's probably going to take several days to get an answer to you.

-d2

Hello Don,

Yes, the notification system is very efficient!

If you can get an answer in a few days, that will be great. I'm sure that you'll have to contact the designers at Burr-Brown Japan. It's too bad that there has been a trend to leave more and more and more information out of the data sheets. Go back and look at the datasheet for the PCM1804 to get an idea of what I am talking about. Ones older than that have even more information. Remember when they used to include a simplified schematic?

I'm not sure who makes the decisions about how much information to include. Are they afraid of divulging secrets? Whatever the reason, it makes designing things much more difficult.

Thanks, Charles Hansen

Hello Don,

While you are at it, maybe you could collect another piece of information that would be extremely valuable to us. Specifically we are looking for the spectrum of the noise floor from the output of the modulator. These data are given on the datasheet for the PCM1804 (Figure 31 on page 17), but not for the newer parts. Any help with this would be appreciated.

Thanks,

Charles Hansen

Hi, Charles, 

Due to some special sauce in this part, you can run the modulation as high as 0.9.

I couldn't find a plot of the noise floor for the modulator.

You hit the nail on the head, some of the guys here like to hide information for competitive reasons. It makes supporting customers a lot of fun!!!

-d2

Hello Don,

Thanks for the info. I must admit I am growing a bit more confused. While we were waiting for the info on the range for the modulator, we ran our own crude test. We put in a full scale DC signal and looked at the outputs of the six bits. The LSB had a 50% duty cycle, the second-most LSB had about a 10% duty cycle, while the other four were steady. This implies about a 7 dB loss from a theoretical full scale if the modulator could run all six bits from 000000b to 111111b, which corresponds to a different measurement we had made earlier (and opened this can of worms to begin with). So now we have two questions:

a) Why did your guys tell you that the modulator could be run up to 0.9? I'm not sure I'm doing my math right, since your guys gave you the answer in different units than the modulator outputs. But if 100000b is mid-scale, then by my calculations a full scale of 0.9 would either be a loss of around -1 dB or a 10% duty cycle of the LSB, or else a loss of 000011b which would be -12 dB. Neither case makes sense to me so clearly I am doing something wrong.

b) Howzabout we sign an NDA so that we can take a look at the noise floor of the modulator? The whole idea is to make our own PCM conversion. But without knowing where we are starting, it is very difficult to figure out how to end up in the desired location...

Thanks, Charles Hansen

Hello Charles, 

I found myself pretty much in your exact situation and I have the exact same questions and I was wondering if you managed to get your questions answered. 

I will be very thankful for your answer. 

Thanks, 

Vasil Savov