1. Home
  2. Embedded Programming
  3. Cirrus Codec CS4270

Cirrus Codec CS4270

I am considering using the Cirrus CS4270 codec in a new design and wanted to ask if anyone had used this before and if it had any gotchas that weren't described in the errata sheet.

I was going to use an older AK4552 because of the tiny TSSOP16 package (without going to a leadless part). But I realized that it isn't ROHS compliant, so its out with yesterday's fish.

The only other codecs I have used are TI parts and they proved to be a relative nightmare from the software side. I think there are something like 100 registers to setup. The CS4270 can be used in a stand alone mode and still have control over the sample rate using a couple of pins.

One thing that I am not clear about on both of these parts is the single/double/quad speed "modes". The AKM part seems to not actually sample above 50 kHz since it looks like it is saying that in double or quad speed mode the input/output data is decimated so the higher sample rate doesn't buy you much. The CS4270 seems to actually provide the full sample rate in double and quad speed modes and so should provide a wider frequency response.

I don't need the wider response for this application, but I have been asked before about ultrasonic audio and this looks like a nice part for that too.

Anyone used the CS4270 before and can tell me about your experience?

Reply to
rickman
Loading thread data ...

I didn't use 4270 however I have used many other similar codecs from Currus. Here are my impressions:

  1. Cirrus codecs can be put to work however it may take a week or two of effort.
  2. Cirrus codecs have hardware bugs, especially if you are using the features like gain control, etc. So I would advise not using the features.
  3. Currus datasheets have bad mistakes and inconsistensies, like wrong registers and/or incorrect I2C addresses.
  4. Wrong programming makes Cirrus codecs to do really weird things like the input only pins become the outputs. There is absolutely no information about that.
  5. The performance is not outstanding by any of the parameters.
  6. Cirrus support is good and accurate, but their typical response time is
2-3 weeks.

Well, not as many as 100, but something like 32 registers for sure, and the TI documentation is not an easy reading. However the TI codecs work exactly as described.

You may be unpleasantly surprised. I advise you to use the jumpers to select the different config options as well as connect I2C or SPI to it.

They all have the response to 85%...90% of Nyquist of the selected sample rate, however the higher sample rate doesn't buy any improvement in the SNR.

What is the application? What are the performance parameters?

Vladimir Vassilevsky DSP and Mixed Signal Consultant

formatting link

Reply to
Vladimir Vassilevsky

This is all interesting and scary to hear. Maybe I should avoid the Cirrus parts. Mainly I picked them because they came up as in stock at Digikey and the package fits my needs. I prefer a leaded part and don't have a lot of board space. Originally I was going to use the AKM part as it comes in a 16 pin TSSOP which fits nicely, but it is old enough that it is not ROHS. A lot of the TI parts are in larger QFPs or leadless packages.

The requirement for this effort is just standard audio freq, 8 or 48 kHz SPS (two modes). It would also be nice to have 16 and 32 kSPS just in case 8 kSPS in case a third mode is needed. Extreme audio fidelity is not required, 80+ dB of SNR is good enough. A lot of converters seem to give you 24 bits, but I'm not clear on what the extra bits buy you in parts that are worse than 96 dB of SNR or dynamic range.

The part I used had close to 100 registers! The software guy was pretty ticked at me about it so I had to help him set it up. Turns out you can ignore most of them as they default to typical settings. But you have to read them all to figure out which to ignore and which to set.

From what you have said, I think I want to avoid the Cirrus parts. That was why I asked about it here. Thanks...

The control signals all go to an FPGA, so they can be used either way. But it is much simpler to use the stand alone mode over I2C and I don't want to push things into the software. That just spreads the "fun" around too much.

But that depends on what Nyquist is compared to what they are calling the "sample" rate. If they internally interpolate before sending to the DAC, you are still bandwidth limited to the data rate you can pump into the part. The AKM data sheet is very terse and it looks to me like in double speed mode you can only use half the "samples" and in quad speed mode you can only use a quarter of them. So no extra bandwidth.

Strictly audio 20 Hz to 20 kHz with nothing special in SNR or dynamic range. I figure 80+ dB is ok. But I would like to get familiar with a part that can sample at 96 or 192 kSPS with corresponding increases in bandwidth.

Reply to
rickman

features.

the

about

is

Some more notes:

  1. The EMC is terrible. Needs really good layout and heavy decoupling. Do not separate the planes.
  2. The ESD can hang up the codec. The hardware reset is required to get out of that.
  3. The power dissipation is quite high especially in 96 and 192kHz modes.

Well. Cirrus codecs can be put to work, they are available and not too expensive.

Looks like fairly modest requirements. A cheap codec from NXP would probably do.

I agree. This is mainly for the compatibility with I2S modes.

It is always nice to hear the voice of experience :)

Vladimir Vassilevsky DSP and Mixed Signal Consultant

formatting link

Yes, this is the correct assertion.

Reply to
Vladimir Vassilevsky

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.