Multiple adc measuring same input?

Discussion in 'General Electronics Chat' started by geekoftheweek, Jul 17, 2015.

  1. geekoftheweek

    Thread Starter Member

    Oct 6, 2013
    42
    2
    What I'm looking to do is measure a single input with two separate ADCs. The first one being a PIC18 something or another and the second one most likely will be a mcp3425 (http://www.microchip.com/mymicrochip/filehandler.aspx?ddocname=en533572). They will be both fed from the output of one section of a mcp6002 op-amp. The reason for measuring with two is due to timing. The 10 bits of the PIC ADC will be precise enough and fast enough for a rough idea to use to control the relay controlling the circiut. The 16 bits of the other adc will be what I'm really after for calculations for the user, but will be too slow to control the circuit before disaster.

    Are there any considerations I should be aware of while doing this, or just simply branch the output of the op-amp into the adc's and measure.

    As always... thanks for the help. You all have taught me a lot!!
     
  2. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    5,809
    1,105
    Apart from ensuring that the opamp output doesn't go beyond the supply rails of either ADC I see no problem.
     
  3. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,546
    1,252
    PIC A/D converters are famous for needing a low source impedance driving them. To drive two in parallel, you might need to buffer each input with a voltage follower; but 1 mV of offset error equals 4 LSB of ambiguity at 16 bits. This effectively turns a 16-bit converter into an 12-bit converter.

    You don't say how fast the fast conversion needs to be, but if the result is used to control a relay, that relay will have several milliseconds of delay that probably swamp out any conversion latency.

    Also, there are megahertz-speed 16-bit A/D converters out there. AD, LTC, TI...

    ak
     
  4. crutschow

    Expert

    Mar 14, 2008
    13,053
    3,244
    AK has a good point. Even a slow 16-bit A/D is much faster than a mechanical relay.
    Of course if it's a solid-state relay, that's a different story.
     
  5. geekoftheweek

    Thread Starter Member

    Oct 6, 2013
    42
    2
    Haven't been back to check for a few days... My original design has circuit breakers instead of the relays, but after a little studying I've learned that they just aren't going to work. I was hoping to use relays and fuses in hopes I could shut off the relay before the fuse blows and save money and time by not having to swap out fuses. The project is going to be used to detect short circuits in the end. Ideally I don't need any circuit protection as the source is fused already, but there's always the chance that something won't go right. Maybe I'm being overly cautious...

    I'll check into other ADC's. I mainly chose the one based on cost, lack of time to look deeper, and it would be fast enough for what I had envisioned for it. I've learned the hard way about the PIC liking the low impedance... that simple problem caused me to learn a bit about op amps in general. It's also what prompted me to ask this question in the first place. I wasn't too sure if running two off the same input would work or would cause problems.

    My ultimate goal is to use the ADC's to calculate the difference across a shunt resistor. It's a pretty ambitious project for me as I'm still learning in the analog world, and starting to wonder if there's just too many variables and things I haven't yet considered that are going to make it difficult. I may end up just sticking with my voltmeter and box that I have so far. I just thought it would be cool to build an all in one unit.
     
  6. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,546
    1,252
    What is the peak current to be measured and the desired resolution? How many shunts / monitored power runs? The shunt is a current sensor for what? AC or DC, DC high side or DC ground leg? AC isolated from the mains or live? Single or 3 phase?

    Lotsa things combine to reduce the large number of circuit techniques available to the few that really make sense for your particular project. If you are measuring DC current, AND IF the shunts are in the ground leg, AND IF that ground is the same ground as the signal ground in the measurement system, THEN what is called a single-ended opamp (one that does not need two power supplies to operate, and has an input common mode ranthe than includes its negative rail) can deliver an analot voltage that is directly proprtional to the current, as in 1 V = 1 A or some other tidy little conversion factor.

    ak
     
  7. geekoftheweek

    Thread Starter Member

    Oct 6, 2013
    42
    2
    I guess a little background would help...

    Ideally 14 volts DC... it may be slightly more or less depending on the source at the time. Shunts are rated at 50A = 50mV... unfortunately I haven't been able to locate a datasheet for the full specs. The max current is kind of one of the variables since I don't have control of the power source. I want to use a relay to break the circuit on one side of the shunt along with a fuse to provide my own protection to the project.

    The shunt will have to be on the high side since another variable is the actual ground system. I've drawn up some voltage dividers with zenners (planning for the worst)... to feed the op-amps that will then feed the ADCs. Minumum resolution of .0001V since the normal overall current draw on the circuit is normally between 1A and 5A and with the voltage dividers I'm taking away a bit of resolution.

    There are six shunts to be fitted so cost is a potential issue. All six shunts are separate circuits with potentially different input voltages, but the same ground.

    I also need to be able to detect current direction to check for back feeding (which is why I was looking into differential ADCs). I have a few other ideas to use the same circuits with additions to measure overall voltage also... so the ADC's maximum speed will be divided in half if I were to use differential.

    If I were to use single ended ADCs it would effectively eliminate the extra circuitry I have planned and I could run continuous sample mode on the ADC (if it has it... which will eliminate some communication time).

    I liked the I2C part of the MCP3425, so that I could build replaceable modules (a just in case sort of thing).

    I would although be just as happy with something with a parallel interface. I've never been a big fan of SPI so I more or less avoid it when I can.

    I could just use the ADCs on the PICs to detect if there is a high current draw and a rough overall voltage, but I thought it would be nicer to get a little more resolution and make a more informative display.

    What would be a good ADC to look into? I was looking at the sigma-delta ones because of the price. The current draw of the circuit will pretty much stay the same so speed isn't a real big issue other than trying to open the relay before the fuse or the shunt goes.
     
    Last edited: Jul 24, 2015
  8. pwdixon

    Member

    Oct 11, 2012
    488
    56
    Maybe it would have been better to start by saying what you are trying to achieve.
    It looks like you are trying to control an output that if an over-current condition exists can be shut down but I could be completely wrong. If this is true then there are chips for that.
     
  9. adwina2

    New Member

    Jul 14, 2015
    18
    2
    For better resolution, there are plenty of shunts out there that are rated closer to the current you are operating at. You could use a 10A=50mV or 10A=100mV. If you want to display or record the current,you can separately use the slower adc, but use a short circuit protection design to turn on a FET and re-direct the current for actual protection. Also, there are slow-blow fuses that will buy you a little extra time, although I normally stay on the safe side and use a fast blow to better protect the system.
     
  10. geekoftheweek

    Thread Starter Member

    Oct 6, 2013
    42
    2
    Thanks all for the suggestions... I gave the idea a little more thought.

    I think you're on the right track... It will be a device that when placed between the source and the load it will be able to break the connection between the source and the load. Sorry for leaving that detail out. I started to rethink things and decided although it would be a neat feature it's really not necessary. It would be simpler just to use fuses to do it since there are only two possible problems that would occur with what I'm going to test that would cause a fuse to blow. The source fuse is usually already blown so I've learned how to recognize them... it won't be that much trouble to carry a little box of spares around. If I find I end up going through fuses quickly I'll put some thought into it again.
     
  11. geekoftheweek

    Thread Starter Member

    Oct 6, 2013
    42
    2
    Unfortunately I'm stuck with what I have at the moment in the way of shunts... the price was right.
     
  12. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,546
    1,252
    Linear Tech has some nice electronic circuit breaker control chips that can be set up to be faster than the fastest magnetic breaker.

    ak
     
  13. pwdixon

    Member

    Oct 11, 2012
    488
    56
    I've used LT4363HMS-2 (Farnell 2356037) to do exactly this so you might have a look at that device.
     
Loading...