Load Cell ADC Parts Decision Help

Hello,

         I need to convert the analog signal of a load cell.  I want to use a 16-Bit Single channel ADC with SPI for that purpose.  The reason for the 16 bits is that I currently have other ADCs (single ended potentiometers) on the SPI bus.  After they are converted to digital, everything gets packaged together and sent off using rs-485.  Also, the existing code expects 16 bit values for everything.

The AD7715 looks like a fantastic part but its too slow.  Because I have to package all of the converted values I would end up waiting for the load cell or transmitting an old value with the rest of the ADCs.  Therefore it looks like I need to go to an SAR.  I am having trouble picking a good SAR + Differential ADC Amplifier combination.

Couple of things:

Load Cell is 2mV/V for excitaion.  Its currently wired up with 10V but if I could use 5V it would really simplify my design.  So max output is either 10mV or 20mV depending on whats practical for the input of an ADC amplifier.

The SPI bus is 3.3V.

The speed doesnt need to be incredibly fast 250-500K is plenty, over 1Mbps would be great.  10Mbps is overkill.

I was hoping someone could give me some suggestions on what they have used or what might fit my needs.  I have used parametric search but have been going in circles a bit.

Thanks,

Karl

Parents
  • Hello,

           As far as the required accuracy, that is slightly unknown.  Like I said I am trying to replace a parallel process that used a National Instruments DAQ.  The DAQ has an adverised 16 bit resolution but as far as its real world performance, I dont know.  I suspect due to the large amount of muxing on the part that you are probably not getting an error free 16 bits. 

    The more important aspect for me is speed.  This serial design must match the speed of the original design. 

    I'll give you a car analogy:

    I am taking the backroads now instead of the highway like I did in the past (3 wires in new design vs. 30 wires in the old). 

    Also now I am carpooling instead of driving by mself (Packaging all the sensors together).

    Cons

    Now I have a lot more red lights (Sensor->ADC->SPI->MCU->RS485 Transceiver->PCI RS485 Card->Software Parsing) to go through then before (Sensor->PCI DAQ).

    I have to drive a lot faster now to make sure I get home at the same time.

    Also, if I have a slow co-worker (Sigma-Delta conversion) I cant leave from work without them.  I have to wait for them to get in the car with everyone else before I take off for home (Due to packaging sensor data).

    Pros

    The advantage I get is the that the roads are much narrow(Less wires) and there are less cars polluting the air(No real analogy there. Ha!).

    The output rate of the sigma-delta is just too slow.  Therefore, I am going to concentrate on a SAR based designed (AD7687 or AD7688).   I will do the Sigma-Delta design aswell incase the effective precision bites me later on.  I doubt it will though.

    I was having trouble finding a nice Single Supply Rail to Rail True Differential ADC Driver Amplifier.

    I eventually found a really nice one.  Its just that you guys dont make it, but thems the breaks I suppose.

    Also, I realize Rail to Rail amps and single supply ADCs dont work so great at actual rail voltages.  I can keep my signals off the rails so that shouldnt be a problem.  I would rather not have a voltage inverter on the board.

    Consider my question answered.  I just thought I would write back in case someone else eventually looks at this thread.  Spread the knowledge is what I always say.

    Thanks,

    Karl

Reply
  • Hello,

           As far as the required accuracy, that is slightly unknown.  Like I said I am trying to replace a parallel process that used a National Instruments DAQ.  The DAQ has an adverised 16 bit resolution but as far as its real world performance, I dont know.  I suspect due to the large amount of muxing on the part that you are probably not getting an error free 16 bits. 

    The more important aspect for me is speed.  This serial design must match the speed of the original design. 

    I'll give you a car analogy:

    I am taking the backroads now instead of the highway like I did in the past (3 wires in new design vs. 30 wires in the old). 

    Also now I am carpooling instead of driving by mself (Packaging all the sensors together).

    Cons

    Now I have a lot more red lights (Sensor->ADC->SPI->MCU->RS485 Transceiver->PCI RS485 Card->Software Parsing) to go through then before (Sensor->PCI DAQ).

    I have to drive a lot faster now to make sure I get home at the same time.

    Also, if I have a slow co-worker (Sigma-Delta conversion) I cant leave from work without them.  I have to wait for them to get in the car with everyone else before I take off for home (Due to packaging sensor data).

    Pros

    The advantage I get is the that the roads are much narrow(Less wires) and there are less cars polluting the air(No real analogy there. Ha!).

    The output rate of the sigma-delta is just too slow.  Therefore, I am going to concentrate on a SAR based designed (AD7687 or AD7688).   I will do the Sigma-Delta design aswell incase the effective precision bites me later on.  I doubt it will though.

    I was having trouble finding a nice Single Supply Rail to Rail True Differential ADC Driver Amplifier.

    I eventually found a really nice one.  Its just that you guys dont make it, but thems the breaks I suppose.

    Also, I realize Rail to Rail amps and single supply ADCs dont work so great at actual rail voltages.  I can keep my signals off the rails so that shouldnt be a problem.  I would rather not have a voltage inverter on the board.

    Consider my question answered.  I just thought I would write back in case someone else eventually looks at this thread.  Spread the knowledge is what I always say.

    Thanks,

    Karl

Children
No Data