Help, Pi2 and MCP3008

Question

My analog board and Pi2 are inside a well ventilated metal enclosure. The MCP3008 is not getting warm to the touch. There is a fan pushing air though the enclosure.

All signals look clean on the scope.

Vcc, and Vref are 3.3v. All samples are within 0-3.3v.

I have tried all different SPI clocks form 500Khz to 3Mhz.

An identical board sitting on my desk has been working fine for 45 min. Three different PI and MCP3008 inside the metal enclosure will work for about 5 min after a power cycle then the readings go nuts, before all eight channels revert to zero.

I will leave the working board running all night. The AC cuts off here at about now. So the temp in this office will go up over night to about 30c (Singapore) so I guess I will know if the temperature affects it.

Here is my "transfer" code:

//==================================================================

//Private main read/write

//==================================================================

int CMCP3008::spiWriteRead( unsigned char *data, int length)

{

struct spi_ioc_transfer spi[length];

int i = 0;

int result=-1;

//One spi transfer for each byte

for(i=0;i<length;i++)

{

memset(&spi[i],0,sizeof(spi[i])); //Flush

spi[i].tx_buf = (unsigned long)(data+i); //Transmit from "data"

spi[i].rx_buf = (unsigned long)(data+i); //Receive into "data"

spi[i].len = sizeof(*(data+i));

spi[i].delay_usecs = 0;

spi[i].speed_hz = this->speed;

spi[i].bits_per_word = this->bitsPerWord;

spi[i].cs_change = 0;

}

result=ioctl (this->spifd,SPI_IOC_MESSAGE(length),&spi) ;

if(result<0) { printf("spiWR==[%i]\n",result); return -1; }

return 0; //Success

}

Please help

e14 Contributor · Accepted Answer

It appears to be dropping into low power mode and staying there. I do not understand it, because the readings are accurate to three decimal places. I will isolate the inputs. I am having the boards redone with low power op amps as voltage followers.

shabaz · Answer

Your code is a bit odd, you only need one of those structs, not an array surely? A single struct is sufficient to transfer multiple bytes, since the rx_buf and tx_buf are arrays. Anyway, although incorrect it looks benign, it shouldn't cause issues. I personally don't use the same locations for rx and tx, whereas you have them both pointing to the data array. If it doesn't cause an issue then great - it's just something I've never done. and I've never tested to see if it is possible. Apart from that I can't see an issue with the code but I've only examined it briefly. I think you're saying the MCP and Pi are in the same enclosure. Do you have anything like long cables from the MCP to the sensor? Perhaps you're getting noise on them?

shabaz · Answer

If the code works on your desk but not in the real environment, it is likely to be an electronic/electrical issue. The environment can have an effect. How long are the cables to the sensor, and what is the sensor? Probably some photo is needed showing in detail your arrangement, cables and sensor and the boards. Does the MCP3008 board have any input buffering? Most cheap ADC boards neglect this, and as a result can cause issues with incorrect readings. I've noticed the code eliminates two readings from six and then averages the remainder; was this because you're getting very noisy readings?