isolation & level shift

Discussion in 'General Electronics Chat' started by wmgehl, Aug 17, 2009.

  1. wmgehl

    Thread Starter Member

    Jul 21, 2009
    12
    0
    Hello.
    I am new to analog design and need to monitor a 1-10V signal using the ADC in a microcontroller. So I need to level shift the 1-10V down to 0 - 5V and I have a dual op-amp circuit I intend to breadboard to accomplish this.

    I gather that using an op-amp circuit should impedance match better into the ADC input (2.5Kohm max)as well as provide serious isolation (high z) between the line and the MCU. To protect this circuitry further from signal line overvoltage and misuse I assume that a 12V zener on the op-amp input would help.

    However, given my very limited design experience, there may well be a better way. So I would appreciate any comments and or suggestions anyone has on my thoughts. Any suggestions on the best op-amp (or op-amp circuitry) to use for this application, etc?

    Thanks in advance.
     
  2. nanovate

    Distinguished Member

    May 7, 2007
    665
    1
    Can you post a schematic of your proposed input circuit?

    Are the opamps rail-rail input (if single-supplied) or are you using a dual power-supply?

    Often a Schottky diode from the input to the VCC on the ADC works very well for clamping over voltages.

    As far as isolation goes... you likely have a resistive path from the opamp inputs to the ADC input so it is not "serious" but it is much better than a direct connection.
     
  3. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    This shifts the input from 1-10V in to 0-5V out. If you do it this way, the output is automatically limited to 0 to +5V.
     
    Last edited: Aug 17, 2009
  4. wmgehl

    Thread Starter Member

    Jul 21, 2009
    12
    0
    Thanks for your reply Nanovate.

    I'm at the block diagram level so no digitized schematic just yet - my CAD lady is on the BD right now. I will post schematic asap and, yes, I was intending to source single supply rail to rail op-amps (12V supply). Thanks for the Schottky to Vcc tip.

    That said....if I may be so bold as to shift the topic a little and impose on you further....the isolation is a larger issue than I thought originally and management wants "real" isolation (not just serious)...so it appears that an opto-coupler is the way to go.

    Can I also level shift 1-10V to 0-5V through an analog opto-coupler?...... ...from what I've read it seems that I can....which would theoretically cover both requirements quite well. I see there are a few Fairchild CNY17-3 opto-couplers here that may work for this?

    Do you agree? Please advise/comment. Thanks.
     
  5. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    I forgot about the isolation requirement. I added a voltage follower to the schematic in my previous post. The LM324 is a quad op amp, so there is still a section left over.

    As I said, it automatically limits the output to 0-5V, even if the input exceeds the 1-10V range (e.g., 0-12V).

    Edit: This obviously does not provide true isolation - just high input impedance and low output impedance.
     
  6. nanovate

    Distinguished Member

    May 7, 2007
    665
    1
    Is this an analog signal or a digital signal? How much isolation do you need i.e. is there a regulatory requirement? Optocouplers provide good galvanic isolation but if you are using it as an amplifier then it gets tricky. TI and Analog Devices have "isolation" amplifiers that are designed for this but they are costly. Maybe this can be split into two small micros: one that does the A/D then transmits the data over an isolated link to the second micro that does everything else. Or use a dedicated A/D device and communicate with it over an isolated link.
     
  7. wmgehl

    Thread Starter Member

    Jul 21, 2009
    12
    0
    Thanks so much Nanovate and Ron H.

    No regulatory requirement on isolation, but this design is part of a system using 230VAC in a very harsh and noisy environment. Lots of machines, etc. The input 1-10VDC is an analog control signal coming in on a very long line. Can't use two micros here.

    I had the same op-amp level shifter circuit in mind as Ron H provided without the voltage follower (or two of them) on the front end.

    But can I not achieve the best isolation as well as level shifting by using an opto-coupler (or isolator) assuming that I can level shift as desired with the analog opto device? maybe put a voltage followerr stage ahead of the opto device?

    Are these devices (such as CNY17-3) not linear such that I can make a 2:1 voltage conversion (level shift 1-10V down to 0-5V)? Has anyone created this functionality with such a device? Or is true linear conversion unlikely?

    Assuming the opto device can be used as the desired level shifter, the only other consideration would be Z to the micro ADC input which is limited to less than 2.5Kohms. I assume this could be controlled by the resistor in the collector of the opto-transistor or, if not, a voltage follower stage could be implemented between the opto device and the MCU ADC input.

    My apologies for my ignorance on these matters....and I really do appreciate your advice and comments.
     
  8. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    AFAIK, there are no linear optoisolators with decent transfer gain (K3 in the datasheet) predictability. LOC110 might work, but its transfer gain appears to be highly variable from unit to unit. Here is an LOC110 application note.
    What is the frequency range of the input signal? If it is low enough, a digital optocoupler such as the CNY17-3 could give you better results, using pulse width modulation (PWM), if you can tolerate the lag (time delay)of PWM.
     
  9. wmgehl

    Thread Starter Member

    Jul 21, 2009
    12
    0
    Thanks Ron H.
    This is a straight DC control signal, no frequency, 1 - 10V.
    Just need to isolate and level shift it to 0 - 5V into an ADC input.

    I'll try the CNY17. Thanks again.
     
  10. wmgehl

    Thread Starter Member

    Jul 21, 2009
    12
    0
    Guys, I just found the Fairchild HCPL3700 AC/DC to Logic Optocoupler, which looks like it might just be what I'm looking for. What do you think?
     
  11. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    No, that has a digital output. If you don't want to use an isolation amp ($$$), IMHO you will need to convert the input to PWM, pass it through a digital optocoupler (optoisolator), and then convert it back to analog. As I said, there will be a lag, caused by the low pass filter in the output stage, and it can be on the order of tens to hundreds of milliseconds.
     
  12. wmgehl

    Thread Starter Member

    Jul 21, 2009
    12
    0
    Thanks Ron H.
    You're right about the 3700.....I don't know what I was thinking. But it looks like I can use it or the MID400 to sense AC Mains presence to the system, as that is also a requirement.

    I see that the TI ISO124 isolation amp is around $14 in low quantity. And I'd sure like to avoid the hassle of messing with PWM, etc. I'm hoping there is an easier way.

    If I use a >10V Vcc, single supply, rail to rail, dual op-amp (like TI972 ?) to amplify the signal (input on a long noisy line) and level shift it from 1-10V down to 0-5V (as previously discussed) and then run that through a CNY-17 optocoupler for true isolation, it appears thats about as good as it gets $-wise and hassle-wise. The vishay CNY-17 datasheet says "signal information, including a DC level, can be transmitted....".

    What do you think? Is there any reason this should not work?
     
  13. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    You're not gonna give up on thay CNY17, are you? :eek:
    It's just a a standard optocoupler - a light-controlled transistor. There is no way you can use this to stably transmit defined DC levels through it. It is meant to be used as an isolated digital switch: No current into the LED, the transistor is off. Sufficient current into the LED, the transistor is on. (The definition of "sufficient" depends on the load current of the transistor, i.e., the power supply voltage divided by the load resistance.)
    There is no transfer function of these devices which is linear, stable with temperature, or repeatable from device to device. Not Iout/Iin, not Vout/Vin, not Iout/Vin, not Vout/In. The only parameters you could count on are things like weight, dimensions, color, etc. These are not too useful in creating an isolation amp.
    I have no idea what Vishay means by "signal information, including a DC level, can be transmitted....". Perhaps the datasheet was written by an intern. Maybe he was interpreting notes from an engineer, who had said something like "signal information can be transmitted between systems between which there exists a large DC potential", or something of that ilk.
    Think about this: If this would work, there would be no market for isolation amplifiers.
     
  14. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    Have you looked into 4-20mA current loops? Do a search, if you aren't familiar with them. They are generally used to transmit DC info over long distances, but they do not provide isolation. Why do you think you need isolation?
     
  15. wmgehl

    Thread Starter Member

    Jul 21, 2009
    12
    0
    RonH, sorry for the lengthy silence on this. Was away then pulled off the project and now I'm back but on a different section of the design for now. Meanwhile I talked to an engineer in another company office that claims you can indeed use the CNY17 to isolate and level shift on its own and its quite linear??? I am going to try it as soon as I can get back to it and will let you know how it goes (I certainly have my doubts about that too - as you have expressed above). And, no, I have not looked into 4-20mA loops yet.
     
Loading...