Different behaviour on ADAU1450 and ADAU1452 with external SPI delay
Hallo guys,
we're developing an application based upon the ADAU1450 including the external SPI memory for additional delay (23LC1024-I).
Although the same project works well on the ADAU1452, it doesn't work when we use it on the 1450.
After some tests we found the following issues:
1) Bitrate inconsistency
On the 1450 the measured bitrate on the SPI bus is half respect the one set in the dialog window (i.e. setting 16000kHz on the bus we get about 8000kHz). This problem is not present in the 1452 where setting 16000kHz gets about 16000kHz on the SPI bus.
2) Sample rate inconsistency
The project is running at 96ksps; in order to accomodate a complete SPI data read/write transaction, the SPI bitrate must be higher than 30000kHz (15MHz on the SPI bus). If we set a lower bitrate, 26000kHz for example, we get the circuit working but the read/write cycles on the SPI bus happens at a reduced sample rate (48ksps as you can see in the following picture).
Again, in this condition two extra byte are added at the end of the data write cycle: they are always 0x00 or 0xFF, and we suppose they are overwritten during the subsequent bus transaction because the memory pointers are correct (see figure below: 8bytes read cycle and 8 + 2 bytes write cycle).
On the other hand, when a bitrate greater than 30000kHz is set, the SPI transaction seems correct but the audio signal is halved in amplitude and its distortion reaches 1%.
Did someone deal with this problem?
I attach the two projects in order to check the correctness of the DSP settings.
Thank you very much for any replies,
Marco.
test_delay1450_1452.zip
Audio Signal Processors
adau1452
spi
external delay
adau1450
sigmastudio
SigmaDSP Audio Processors
SigmaDSP
Show More
No updates about this topic?
Marco.
Hello Marco,
Sorry, but things have been rather busy. I will look into this soon.
Dave T
Thank you Dave,
I'm waiting to hear from you soon.
Marco.
Hello Marco,
Sorry for the delay. I was able to verify this with the lead designer for the part.
The SPI rate is tied to the core rate for the calculations. SigmaStudio does not compensate for the difference in the two parts. So you will have to put in a rate that is double in the SPI configuration page.
I do need to mention that the max rate for the SPI Master port is 24 MHz. This specification is over the entire IOVDD range and the low IOVDD voltage of 1.8V is where it has difficulty with the high speed clocks. So if you are running at 3.3V then you should be able to operate the SPI port at 30 MHz and it should work for most parts. However, we cannot guarantee that it will work for all silicon skews and all temperatures.
Dave T
Hello Dave,
thank you for the reply.
Our design is running at 3.3V, even for IOVDD, so we should be able to reach the full SPI speed of 24MHz. We set 30000kHz in the configuration page measuring 15MHz on the SPI bus, well below the 24MHz maximum speed stated on the datasheet, but we noticed halved and distorted signal.
By the way, please confirm that the maximum SPI speed is related to the port MCLK pin and not to the SigmaStudio configuration page.
Thank you,
Regards.
Marco.
Hello Marco,
The configuration page in SigmaStudio is only for developing the correct parameters for the serial port. The speed limitation is with the actual pin drive circuit and internal circuits in the silicon. So set the SPI speed for 48MHz and you should see around 24MHz out of the actual port.
Once this is operating at the speed you require then look at the audio. You might have been skipping samples or something similar.
Dave T
Thank you Dave.
We tesed our board mounting first the ADAU1452, setting the SPI pin frequency to 17000kHz, and everything worked well.
Then we substituted the 1452 chip with the 1450, setting the SPI frequency to get 17000kHz on the pin (34000kHz on the SigmaStudio interface), but the delay block in the same project did not work (half amplitude and more distortion).
The point is that we expected the same correct behaviour (like the 1452) between the two chip with the same SPI settings and project.
Marco.
Please log in to reply.