Multiplexing PCM data from 2 channels A/D CHANNEL (SOLVED)

[juanodiy]

Good afternoon, I'm from Bogota Colombia and i need help, because i don't find information about my topic. I need to digitize 2(two) analog signals, and send them for the same channel. It means multiplex the signal, after that for can be listened, but the signals don't have to mix, the signals must be able to be heard after a conversion D/A.
I decide to do 2 programs, and use 2 pic's of course. The pic is 18f452. In the first pic, I take the sampling of the two signals using A/D conversion, and send it every so often. In the second pic, I take the data (8bits in the portc) of the conversion (PCM word), and send it through the ports, (portc and portd) every so often. The problem is that i think the samples are not sufficient. But I'm using a delay of less than (50 us), it means sampling the signals at 20000 Hz, but isn't true because the time between the instructions take a time that i don't know.

Please help me i don't know what to do, for the signals can be multiplexed at the same time, the codes are below, for any suggestion I'll be very grateful.

PIC1, sampling

[PCM programmer]

[juanodiy]

Hi, thanks to PCM programmer, I have been checking the code, and I make some changes, here are the codes:

PIC1

[juanodiy]

Hi, i solved the problem and the pic 2 is multiplexing in the same time as the data arrives, but now i'm trying to increase the sampling time, and when I do that, the multiplexer does not work because the data arrives more faster, and i can' t find the just time for the program works, please HELP,

PIC1

[PCM programmer]

You still have all the incorrect #byte statements. Just delete them all.
To keep bad code in your program is a mistake. To use it is also a mistake.
---------------

Your project consists of PIC #1 that collects data and then it sends the
data to PIC #2. The problem is PIC #1 puts the data onto PortB, but
PIC #2 doesn't know if new data has arrived, because there is no
synchronization method used.

You need to think of a synchronization method and do it.

The easiest way would be to use hardware SPI to send the data from PIC
#1. Then in PIC #2, call the spi_data_is_in() function in a loop.
If PIC #2 receives a byte, then call spi_read() to read it. Look at this CCS
example file:

[Ttelmah]

As a comment, you are not going to sample the signals at 50000Hz, with this PIC.

Sampling an ADC channel, takes:
Tacq (the time needed for the signal to acquire, and charge the internal capacitor - 13uSec on this PIC).
Plus 12*Tad ADC clock cycles
The fastest legal ADC clock on this PIC is 625KHz.
So an ADC conversion, doing nothing else, not even storing the numbers, takes 32.2uSec. An absolutely fastest conversion rate of just over 31KHz. Reading two channels, each will get half this rate.

You are not going to get even close, without switching to a PIC with a faster ADC. Most of the later chips have ADC's supporting faster Tad clocks, and also faster Tacq times.. The 4520 for example, which is pin compatible with the 452 has a Tacq, of only 2.4uSec, only needs 11 cycles of the ADC clock to perform a conversion, and this is allowed to be over twice the speed, giving a best case conversion time just on 11uSec. Basically 3* as fast....

You need to rethink your design.

Best Wishes

[juanodiy]

First of all, i really appreciate that you have been reading my post. In my country there is not information about this topic, your answers show your knowledge. Thanks very much.

[juanodiy]

[juanodiy]

I appreciate any information about psp (Parallel Port). I'm trying to recieve data like this way, but it doesn't work.

The code is below:

[Ttelmah]

The obvious question is 'how are your sending this'?.
With the PSP, you need a select line to be strobed by the sending device to trigger the transfer.
Effectively your master needs to hold the WR input on the slave device low, and the RD input high, put the byte onto it's output port, and then strobe the CS input low then high. The PSP interrupt occurs when the CS line goes high at the end of the transfer.

A series of comments:
This is going to use a lot of pins. Are you sure you want to do this?. As PCM programmer says, SPI is an alternative. Since this can be clocked at FOSC/4, the byte can be transferred in just 1.6uSec with a 20MHz master clock, and use just three wires.
You are still ignoring the need to _identify_ the bytes. Problem is that unless the two chips wake up exactly synchronised, and data is never missed, you have the potential for the bytes to get swapped. Using SPI, you could use a single other line as a 'A/B' identifier, and change this according to which channel you are sending, allowing the slave device to always 'know' which channel it is receiving. You need to think about this.
You still have the great unwanted 'swathe' of port identifier defines. These are not wanted/needed.

Best Wishes

[benoitstjean]

If you want to make your life easier, just use an external CODEC like the PCM1803A from Texas Instruments. You can get the device mounted on a breakout board for 16$USD from Sparkfun electronics. Simply power it, configure the format via two pins and that's it. Plug-in an analog audio signal that is within the chips's specs and the digital output will be your audio, digitized, in whichever format you selected through the configuration pins. Audio quality is outstanding.

The only problem I am running into (I'm upgrading my stuff to circumvent the problem) is that I am using a PIC18F4620 and with a clock of 24.576MHz, I can sample the digitized audio without a problem and store it internally in an array on the PIC but trying to write it to external memory or an external device, even at 8 bits 8 kHz, is too big of a task for the PIC since the PIC must rely on a few signals from the CODEC to be synchronized with the audio. Don't forget that 8 bits @ 8kHz, it's 64 kbps.

The PCM1803 will give you a word clock indicating if the audio is from the left or right channel. And to be synchronized, you need to sample audio by also reading the bitclock which goes from low-to-high everytime a data bit is valid. If you sample the data from one channel only, then you can *maybe* (and that's what I'm trying to figure-out) use the right-channel's time slice (62.5us) to manipulate the data you just sampled from the left channel.... But in order to maintain your audio quality and not have any errors, you need to be perfectly synchronized... therefore you only have 62.5us left to do something with the data before having to read the left channel's data again.

I tried using the A to D on the PIC for audio before and it was going to be quite a difficult task, if not impossible. I even contacted Microchip and they said it would be very difficult and complicated. So instead, I used an external CODEC. It does all the conversion work for you and simply gives you the digitized audio for you to manipulate along with the BCLK and WCLK signals for you to synchronize yourself with the digitized data.

Or, you could simplify your life even more and purchase the audio development kit from Microchip which comes fully functional with a Wolfson audio CODEC interfaced to a PIC24 (I think... or PIC32), an LCD display and all the stuff you need to get your development going.

Good luck.

[bkamen]

[juanodiy]

[juanodiy]

Hi I have changed the code using SPI but it does not work, when I'm trying to simulate the proteus show me a large number of messages, the messages says:

"Stack overflow is forcing device reset" , I suppose is because the data arrives so fast, but what can I do? , use other method?, what is wrong?, I have only 2 pics 18f452, and I'm sure it is possible with this hardware.

i dont know much about SPI, the codes are below, please i need some help with this project, thanks

PIC#1

[juanodiy]