Important changes to forums and questions
All forums and questions are now archived. To start a new conversation or read the latest updates go to forums.mbed.com.
Need help with SPI com
Kenneth Pinpin.
25
replies
My current Code
#include "mbed.h"
#define OUT_MAX 2104
#define OUT_MIN 1632
#define PRES_MAX 5
#define PRES_MIN 0
SPI spi(p5, p6, p7); // mosi - no connection, miso - connected to pin 3 of sensor, sclk connected to pin 4 of sensor
DigitalOut cs(p8); // chip Select line connected to pin 5 of sensor
Serial pc(p13,p14); // tx, rx
int main() {
// Setup the spi for 8 bit data with a 800KHz clock rate
spi.format(8,0);
spi.frequency(800000);
while(1)
{
// Select the device by seting chip select low
cs = 0;
wait(0.1);
// Send a dummy byte to receive the contents
int byte_1 = spi.write(0x00);
int byte_2 = spi.write(0x00);
int byte_3 = spi.write(0x00);
int byte_4 = spi.write(0x00);
float temp = byte_3<<3;
temp = ((temp/2047)*200)-50;
float psi = byte_1<<8|byte_2;
//psi = byte_1|byte_2;
psi = ((psi-OUT_MIN)*5)/(OUT_MAX-OUT_MIN);
pc.printf("Byte 1 = %X; Byte 2 = %X; Byte 3 = %X; Byte 4 = %X; PSI = %.4f; TEMP = %.2f\r",byte_1,byte_2,byte_3,byte_4,psi,temp);
// Deselect the device
cs = 1;
wait(0.5);
}
}
And my results;
The TEMP is fairly close. it is of by 2 degrees (I'm only using byte 3), but I assume the inside of the chip is warmer than the outside air.
The PSI is somewhat close, but not right. (This is a 0 to 5 psi sensor BTW, http://sensing.honeywell.com/index.cfm/ci_id/156218/la_id/1/document/1/re_id/0 )
I'm fairly sure I don't have the math correct, but I don't know what the OUTPUTmin and OUTPUTmax are supposed to be. I tried using the 2^14 but thats way off. If I use the value at 0 psi for OUTPUTmin and then use the value at 5 psi for the OUTPUTmax I can get it close but still varies wildly (as in the current code) - so I don't think that is right as it is not very accurate to even call it a sensor.
So I am still missing something here.
Thanks for your help, much appreciated.
Andy K wrote:
Quote:
the data should be captured on the falling edge (chapter 2.0)
This is incorrect. Chapter 2 Figure 2 shows the data is sampled on the rising clock edge.
This was actually what I meant to write - when you read further you see that I came to the same conclusions regarding CPOL/CPHA.
Andy K wrote:
So you might like to try spi.format(8,0); and spi.format(8,3);. I susspect you will get the same results as the important factor is the bit data is latched on the rising sclk edge.
I think they will differ. As wikipedia states, these formats not only differ on which edge they sample (rising / falling), but also on which edge they start. So Glenn should really try all formats, especially since the data sheet is not really clear here.
Andy K wrote:
Quote:
Also, it might be better to read all values within one SPI read. The data sheet specifies a minimum SPI clock rate (50 kHz), so you cannot have too long delays between clock cycles. Doing 2 reads might cause that, and therefore lead to your wrong reading. You could try to read 16 bits for a start, I'm not sure how many bits the SPI read can handle.
This is also incorrect. The sclk frequency on an SPI bus is "burst frequency". The sclk on an SPI bus is always high or low between accesses (depends on CPOL and CPHA as to which). You should stick with 8 bit transfers as this is what the data sheets/tech note is saying.
Figure 3 in the data sheet show clearly continouos reads (and, btw., shows SCLK=idle high, so its consistent with figure 2 - or do you mean figure 3 on the last page?). Circuits with anlog parts in them tend to use the SCLK for their internal workings, so they might do something wrong when SCLK gets too slow (e.g. most ADCs with SPI to this, and they get much more incorrect when SCLK is too slow). For example, in the small delay between 2 8bit-reads internal capacitors in the sensor might get discharged, resulting in reading which are too low. That's why I was suggesting reading all data in a single call.
Btw: I just saw the wait(0.1) after the chip select - I think this should be more like a wait_us(10). The data sheet says a minimum of 2.5 microseconds, waiting too long might also cause some misbehaviour of the analog circuits (though the data sheets says nothing here).
With purely digital devices the SPI timing is normally not critical, but with analog devices one should really try to stick to their specifications (esp. clock frequency and delays), becaise when they use the clock for their inner workings they might just misbehave otherwise.
Also when interfacing the mbed with your device, how do you do the voltage level translation for the SCLK signal? Reading the data sheet the sensor seems to expect at least 4 volts for the high signal...
I really thought you were on to something there as I keep forgetting that the mBed is a 3.3v device, I have it connected directly to the sensor.
So, I took a second sensor and plugged it into my Arduino Test board which is 5volts, I'll let the results speak for them self. Note the Arduino data is in Decimal and the mBed is in HEX
mBed results, 0 to 5 psi
Arduino results, 0 to 5 psi
The pressure isn't quite exact, but the Temp is.
since you are getting a count of 0x0681 (1665) with no pressure applied, which is quite close to the specified zero-pressure-value (1632) for this sensor, i would assume that your SPI reading is working well and is correct. So IMHO either your math is wrong somewhere (but i don't see an error there) or more probably your applied pressure is not as high as you expect it to be. I don't know your setup of course, but how are you setting up and measuring the pressure you apply? One thing to pay attention to is that with a gage sensor you have to apply x psi + atmospheric pressure to the port to actually measure x psi.
I use a mini pump with a mechanical guage to apply 5 psi to the sensors, the counts only increase by about 400 to 500 counts. Not nearly enough. I can verify the psi by applying the same to my analogs. I can also verify the psi using water to pressurize the tube (no water touches the sensor), 27.6807 inches of water (vertical) is equal to 1 PSI
AtmosP isn't going to effect it that much.
Hello,
I need to communicate with a Honeywell pressure sensor over the SPI bus, but I am not familiar with the SPI bus at all.
Can anyone tell me how to read the pressure, or at least explain what I need. This is a link the SPI information but it seems incomplete to me.
http://sensing.honeywell.com/index.cfm/ci_id/156989/la_id/1/document/1/re_id/0
Thanks