I was just about to start naively storing all the bits sequentially, but thought of a better method (I'm probably not the first to think of this).
Instead there is a pattern of 5 bytes, the first 4 contain exactly what was in the least significant bits of the first 4 analog to digital conversions. The 5th byte is packed with four sets of the extra 2 bits from each conversion.
I'm pretty sure this is much faster than doing it the other way, both packing and unpacking.
I haven't included the code for setting the FSR to the start point of the store, the loop or the readAD subroutine because they are more application specific.
Code:

    call readAD                ;First byte
    movf ADRESLtemp, W
    movwf INDF
    movlw 0x04
    addwf FSR, F            ;FSR + 4
    clrf INDF
    btfsc ADRESHtemp, 1
    bsf INDF, 7
    btfsc ADRESHtemp, 0
    bsf INDF, 6
    movlw 0x03            ;FSR - 3
    subwf FSR, F    

    call readAD                ;Second byte
    movf ADRESLtemp, W
    movwf INDF
    movlw 0x03
    addwf FSR, F            ;FSR + 3
    btfsc ADRESHtemp, 1
    bsf INDF, 5
    btfsc ADRESHtemp, 0
    bsf INDF, 4
    movlw 0x02            ;FSR - 2
    subwf FSR, F    

    call readAD                ;Third byte
    movf ADRESLtemp, W
    movwf INDF
    movlw 0x02
    addwf FSR, F            ;FSR + 2
    btfsc ADRESHtemp, 1
    bsf INDF, 3
    btfsc ADRESHtemp, 0
    bsf INDF, 2
    movlw 0x01            ;FSR - 1
    subwf FSR, F    

    call readAD                ;Fourth byte
    movf ADRESLtemp, W
    movwf INDF
    movlw 0x01
    addwf FSR, F            ;FSR + 1
    btfsc ADRESHtemp, 1
    bsf INDF, 1
    btfsc ADRESHtemp, 0
    bsf INDF, 0
    movlw 0x1            ;FSR + 1
    addwf FSR, F

 

Careful....if you don't clrf the 5th byte prior to your packing routine, your data is going to be corrupted for every pass following the first.

 

Good point, I've fixed it now.

 

May I ask why you need to store four (4) ten bit ADC results?

Regards, Mike

 

Actually, l'll be storing a few hundred for later processing. Four 10bit samples fit into five bytes.

 

Actually, l'll be storing a few hundred for later processing. Four 10bit samples fit into five bytes.

If you're measuring a signal that changes relatively slowly over time, you might be able to store the first 10-bit reading and then store a signed 8-bit delta value for subsequent readings.

Good luck on your project...

Regards, Mike