Connect with us

On/Off Voltmeter

Discussion in 'General Electronics Discussion' started by AvrDude2012, Aug 1, 2012.

Scroll to continue with content
  1. AvrDude2012

    AvrDude2012

    6
    0
    Aug 1, 2012
    Hi, I'm new to mosfets and up until a few weeks ago, I didn't know know that p-channel meant high side switching and n-channel meant low side switching.

    What I'm trying to do is build a voltmeter for a micro controller to measure up to 500 volts. I'm using a voltage divider to bring the voltage down to 5v for the ADC. However I need the voltage divider disconnected until I read the voltage. To do this I'm using a p-channel mosfet with an n-channel mosfet for the saturation of the p-channel for high side switching of the voltage divider. I also used another n-channel to disconnect the ground object being measured. I want a complete disconnect when not measuring voltage. I also need the speed of mosfets, relays won't cut it. Here is the circuit I'm using:

    [​IMG]

    The problem is when I measure around 100 volts or more, the p-channel locks always on (assuming I fried it) and I have to replace it. It does not lock on until I read the voltage (turn on the p-channel) when it is above 100v or so. Any help would be appreciated.

    Thanks,
    avrdude

    edit to add: The 180k resistor is actually a 10k resistor. Sorry
     
    Last edited: Aug 1, 2012
  2. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,418
    2,788
    Jan 21, 2010
    I imagine that you're exceeding the absolute maximum values for Vgs for Q1 when Q3 turns on.

    One option is an opto-coupled mosfet driver. Redesigning around Q1 would be the alternative.

    Something like this would allow you to switch the mosfet on and off. Note that you really need a pair of mosfets in series so that the body diode does not provide a current path even when the mosfet is switched off.
     
  3. KrisBlueNZ

    KrisBlueNZ Sadly passed away in 2015

    8,393
    1,270
    Nov 28, 2011
    Could you go into more detail about what you're doing? You say you're measuring a "voltage of up to 500V" and that you need both sides of that voltage disconnected from your measuring circuit, but that doesn't really give us any idea of what you want to do.
     
  4. Harald Kapp

    Harald Kapp Moderator Moderator

    10,585
    2,360
    Nov 17, 2011
    Do you need to disconnect the high side? If so, why? Maybe another solution can be found for the issue.
    For isolating the µC you could disconnect the low side, avoiding the problem of high side switching.
    Do you measure DC or AC?
    At what rate do you have to operate the switches (aka MOSFETs)?
    Do you realize that there are fast electromechanical relays, too (e.g. reed type)?
     
  5. john monks

    john monks

    693
    1
    Mar 9, 2012
    I frequently protect Mosfets with a Zener diode between the gate and the source.
    A 6.2 or 7.5 volt will usually do.
     
  6. Harald Kapp

    Harald Kapp Moderator Moderator

    10,585
    2,360
    Nov 17, 2011
    You will then need a series transistor in the gate control signal to limit the currrent through the zener diode. This resistor must be capable to withstand the 500V. Beter use 2 or 3 resistors in series, that should work.

    Harald
     
  7. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,418
    2,788
    Jan 21, 2010
    Resistor? I was thinking along those lines, but there are difficulties when the input voltage is close to zero (i.e. not being able to turn the mosfet on without a negative supply).
     
  8. Harald Kapp

    Harald Kapp Moderator Moderator

    10,585
    2,360
    Nov 17, 2011
    Right, I guess we were all thinking along the 500 V line.
     
  9. AvrDude2012

    AvrDude2012

    6
    0
    Aug 1, 2012
    Wow, thank all of you for your quick responses. There is a few things I want to use the 500v switch for, not just a volt meter. For the volt meter I'll be measuring small farad capacitors up to about 400v, I'm just an overkill kinda guy. I want to switch the high side because of this.

    For future use I would basically like a switch that I can turn on and off that will handle up to 500v low amps, with a speed of at most 5khz. I'm more into the logic of electronics then the ohms law stuff and from what I read, mosfets are fast and can handle alot of power if needed. The circuit I posted, I just threw together from what I could find on the net.

    I was kinda using this page to build the high side switch, but I must be misunderstanding what the gate needs to be turned off and on. I thought off meant give the gate the source voltage to turn it off, and ground it to turn it on. Know of a good tutorial on mosfets?

    Can opto's handle 5khz? Thanks for the part number :)

    It will be for various parts of a larger circuit. I need to monitor certain capacitors. Some of those capacitors are being charged by high side switching, and some are being charged by low side switching. Each voltmeter can be wired for for low side or high side to turn off the voltage divider. Since this circuit will more then likely be used for future projects I figured why not just switch both sides now and not worry about which voltmeter I need to use in the future. I could go into more detail, but I don't think much more is needed to understand what I'm trying to do.

    See above. DC measurements only. Relays would be perfect if there are any that can handle 5khz.

    Does the gate of the p-channel only need 5 volts to turn it off? This says something about 5 volts not being enough voltage to turn off the mosfet or something. I guess I need to study more on mosfets.

    Thanks all very much. :)
    avrdude
     
  10. KrisBlueNZ

    KrisBlueNZ Sadly passed away in 2015

    8,393
    1,270
    Nov 28, 2011
    Thanks for the description of your project. It still leaves a lot of questions though. Are these capacitors fully isolated from the voltmeter circuit, is that the issue? So you want to be able to measure the voltage across each one using the same DAC. And you want to avoid loading the capacitors when you're not measuring them? What accuracy do you need? You really should go into a lot more detail if you want to improve your chances of getting suggestions that are feasible.

    Here are two ideas that might help.

    First, you might be able to put a MOSFET at the bottom end of the top resistor of the voltage divider. That is, connect the source terminal to the ADC input (with the bottom end resistor connected from the ADC input to ground), connect the drain to the bottom end of the top voltage divider resistor, and connect the top end of that resistor to the voltage being measured. The MOSFET's source voltage will always be in the range 0~3V so if you apply, say, 10V to the gate when you want the MOSFET to be ON, and 0V when you want it OFF, it will switch regardless of what actual voltage it is measuring. The big problem here is the gate-source capacitance, which will disturb the voltage at the ADC input when the gate changes state. You might be able to clamp that input to 0V (for example) during the switching periods but that capacitance will also slow down the stabilising of the ADC input voltage. Aim for the smallest MOSFET you can get that is rated for the maximum input voltage and remember to consider the drain-source diode inside it. This is just a suggestion and it may not be workable.

    Second, you could measure the capacitor voltages continuously using a small circuit that is permanently connected across them. You can convey the measurement to the AVR in various ways; a simple way would be a linear optocoupler such as the Vishay IL300 (HP make them too). This would require a power supply on the capacitor side; probably a few AA cells would last a while. Accuracy is reasonable but not stunning. You could also perform the analogue-to-digital conversion on the floating side and pass the information across using a fast digital optocoupler. If your micro has lots of spare time, you can decode serial data from multiple streams simultaneously without needing a UART for each. You would have to monitor the capacitor voltage constantly, and replace the batteries on the isolated side periodically.

    If you're thinking "that's totally unworkable" about either of those suggestions, then you may be starting to see why you should explain your project in much more detail.
     
  11. AvrDude2012

    AvrDude2012

    6
    0
    Aug 1, 2012
    Some caps are isolated on the other side of an induction coil, some are not isolated. I have different ADC's for each voltmeter. As far as accuracy on the 500v scale, I don't need much, to the 1volt would be accurate enough. The 10 bit ADC on the AVR is enough for this when using a voltage divider.

    I guess the main question is, and I should have made this the topic, How do you switch on and off a 500v P-Channel Mosfet using an MCU? The FQP3P50 from Fairchild says in the data sheet that it is a "500V P-Channel MOSFET". To use a mosfet do you have to know what voltage you are going to be switching in order to pick the correct resistance? Can it not just switch 0-500v? If so, is there a circuit that can switch it with 5v from an MCU? I don't mind reversing the logic level to always high to keep the mosfet turned off. Thats just programming.

    I guess I just don't understand p-channels. All of the examples I have seen switch p-channels using an MCU at 5v to switch 12v to 50v. I haven't seen an example of an MCU switching say 450v using a 500v p-channel. Thats really what I need.

    Your second example is not feasible for me. I think I tried the first example, but I had the drain and source backwards from what you are saying. I'll try that though, thanks :) (I will still need to switch 450v at 5khz eventually).
     
  12. KrisBlueNZ

    KrisBlueNZ Sadly passed away in 2015

    8,393
    1,270
    Nov 28, 2011
    Accuracy of 1V in 500V is 0.2% which can actually be difficult to achieve.
    A P-channel MOSFET, just like an N-channel MOSFET, responds to the voltage between gate and source. In the case of a P-channel device, polarities are reversed, so the gate must be driven negative with respect to the source to turn it ON, and the drain is negative with respect to the source. The internal diode is reversed too. So a P-channel MOSFET behaves just like an N-channel MOSFET with the polarities reversed.
    The principle is the same as for the low-voltage case. A device such as an NPN transistor or an N-channel MOSFET provides a voltage that pulls to ground (from its collector or drain), and this is coupled into the gate of the P-channel MOSFET using a circuit consisting of one or more resistors, in the simplest case. A common design uses a two-resistor voltage divider, with the values chosen so that when the bottom end is pulled to ground, the voltage across the top resistor will meet the gate-source voltage requirement of the P-channel MOSFET and turn it ON. This arrangement has problems if the voltage on the source of the P-channel MOSFET could vary over a wide range, because that voltage determines how voltages are distributed on the voltage divider, so if the incoming voltage is low, the gate-source voltage seen by the P-channel MOSFET will be too low to turn it on. You can replace the gate-source resistor with a zener in parallel with a high-value resistor but this slows the turn-off of the MOSFET. It's all a lot of compromises. Other problems with this approach are that significant current flows through the voltage divider when the bottom end is pulled to ground, which loads down the voltage supplying the P-channel MOSFET, and that the device that drives the voltage divider needs to be rated for the maximum voltage on the P-channel MOSFET. These problems are all eliminated by my suggestion, but it introduces problems of its own.
    That suggestion uses an N-channel MOSFET. Sorry I didn't make that clear. So the gate voltage is positive with respect to the source (the source is connected to the ADC input and will be in the range 0~3V). When the gate voltage is driven to 0V the MOSFET turns OFF and its drain is pulled up to the voltage being measured. Only the MOSFET's leakage current will flow. You need a MOSFET rated for that voltage, but choose the smallest one you can find, to minimise the gate-source capacitance, because that capacitance will inject a disturbance into the ADC input when the gate voltage changes, i.e. when the MOSFET is turned ON and OFF.
     
  13. AvrDude2012

    AvrDude2012

    6
    0
    Aug 1, 2012
    A 10 bit resolution ADC gives me a number from 0-1024. If I set my voltage divider from 0-500 down to 0-5, won't that give me an accuracy of around .5 volts or am I missing something? I'm probably missing something as I always do, lol.

    Thank you, a lot of light bulbs went off in my head that time. (I had to read it a few times) I will learn these p-channels eventually. I have a couple questions though.
    When you say "will meet the gate-source voltage requirement of the P-channel MOSFET and turn it ON.", In my data sheet it says:
    VGSS Gate-Source Voltage ±30
    Does that mean I need to make a voltage divider that will take the 500v input down to 30 volts for the gate?
    And when you say ON, I thought that when you brought the gate to +V it would turn it off (oppisite of an npn). I'm kind of confused on that.
    The companion NPN or n-channel when used with a P-channel, is that to get -V for the gate of the p-channel?

    Ah, that makes more sense, And a good idea, I will try that, would just a BJT work better then? Or should I make it a mosfet?

    Thank you, so much :) :)
     
  14. KrisBlueNZ

    KrisBlueNZ Sadly passed away in 2015

    8,393
    1,270
    Nov 28, 2011
    Your ADC has a RESOLUTION of about 0.1% but that's not the same as ACCURACY. There are many sources of error. Check the data sheet for your voltage reference for the initial accuracy and the temperature coefficient. The ADC has a parameter called differential non-linearity. You can get resistors with 0.1% accuracy but they're expensive, and may not be available in the high-resistance and high-voltage types that you need. Even these have a temperature coefficient of resistance. The reference and the resistors also have long-term drift.
    No, that number is in the "maximums" or "limits" section of the data sheet and is the gate-source voltage that should not be exceeded to avoid damaging the MOSFET. To turn the MOSFET ON, you need to exceed its gate-source THRESHOLD voltage. The voltage you need also depends on the drain current you need the MOSFET to pass; for a higher drain current, you need a higher gate-source voltage. (I mean a higher MAGNITUDE of gate-source voltage; for a P-channel MOSFET the gate-source voltage is negative.)

    Look in your data sheet for a "mutual" graph. It shows the gate-source voltage on the horizontal axis, and the drain current on the vertical axis. This graph is TYPICAL, not a specification, but it shows that the MOSFET conducts harder as more gate-source bias is applied.
    I was talking about the MAGNITUDE of the gate-source voltage, sorry this was unclear. For a P-channel MOSFET, the gate-source voltage needed to turn it ON is negative, i.e. the gate is negative with respect to source. I was talking about the magnitude of that negative voltage. So you're right, when you bring the gate of a P-channel MOSFET to be as positive as its source voltage, it turns OFF.
    Yes. When the micro drives the base or gate positive, the NPN or N-channel MOSFET turns ON and pulls its collector or drain down to 0V. Since 0V is negative with respect to +V (on the P-channel MOSFET's source), the P-channel MOSFET sees its gate going negative with respect to its source, and conducts.

    The details of the drive circuit are important because you need to ensure that the magnitude of the gate-source voltage is higher than the P-channel MOSFET's threshold voltage, but not so high that its limit (the +/- 30V you mentioned) would be exceeded. Also, MOSFETs have significant gate-source capacitance, which requires current to charge and discharge; this argues for low-value resistors in the divider, but that means that the divider draws more current from the positive supply, which is a no-no in your application.
    It has to be an N-channel MOSFET because no DC gate current flows in a MOSFET; the gate is insulated. With a transistor, current flows into the base to turn it ON, and this current flows out the emitter, and would affect the ADC reading.
    You're welcome :)
     
    Last edited: Aug 2, 2012
  15. KrisBlueNZ

    KrisBlueNZ Sadly passed away in 2015

    8,393
    1,270
    Nov 28, 2011
    BTW I suspect you won't be able to achieve 5 kHz switching rate without losing a lot of accuracy. The high voltages and high resistances you want to use mean that any capacitance slows things down, and capacitance is hard to avoid. The faster you go, the more effect the capacitance will have.

    Several people have suggested solid state relays. Have a look at these options and see what you think.

    http://www.digikey.com/product-detail/en/CPC1393G/CLA276-ND/1277131
    http://www.digikey.com/product-detail/en/TLP170J(F)/TLP170J(F)-ND/3056586
    http://www.digikey.com/product-detail/en/CPC1393GRTR/CLA270CT-ND/1277144
    http://www.digikey.com/product-detail/en/PLA170STR/CLA344CT-ND/2417230
     
  16. AvrDude2012

    AvrDude2012

    6
    0
    Aug 1, 2012
    I think I see what you are saying. I used a scope to get the exact voltage (as close as my scope will get) then read the voltmeter with my avr, divided the 0-1024 number it gave by what my scope said the voltage was and got (A). I use that constant (A) on that particular meter whenever its read. I think that is the accuracy you are talking about.

    So, I need the gate voltage to be around 10 volts LESS then the source voltage (but no more then 30 volts less) in order to turn it on?

    Those are a little too slow for me, according to the charts, their turn on + turn off times are just a tad bit slower then I'm aiming for. However I will probably order some as I love stock, thanks :)
     
  17. KrisBlueNZ

    KrisBlueNZ Sadly passed away in 2015

    8,393
    1,270
    Nov 28, 2011
    That doesn't take into account nonlinearity in the ADC, variations due to temperature and variations over time. I don't think it will even give an initial accuracy of 0.2% anyway. But there aren't many options for improving the accuracy, so you will get what you get, I suppose.
    I prefer to say that the gate voltage needs to be around 10 volts NEGATIVE WITH RESPECT TO the source. Assuming you're only talking about absolute positive voltages, then yes.
    I doubt you'll find any discrete solution that's faster.
     
  18. AvrDude2012

    AvrDude2012

    6
    0
    Aug 1, 2012
    Thats where I was getting confused, in the wording. This whole negative voltage thing confused me, but when you said with respect to the source, I understood. THANK YOU! Everything I have learned about electronics came through experimenting and info from the internet.

    Again, thank you KrisBlueNZ. :) I'll have to chew on this for a while.
     
  19. CDRIVE

    CDRIVE Hauling 10' pipe on a Trek Shift3

    4,960
    648
    May 8, 2012
    The Operation of a PCHFET when used for high side switching is nearly identical to a PNP BJT. In this circuit R2 & R3 form a voltage divider. As R2 is reduced in value Vgs increases until Vgs(Thld) is reached and the Drain-Source junction closes.

    If the FET is used as a switch, R2 would be replaced by a switch or solid state junction that would pull the gate to ground.

    Heading out to the watering hole.

    Later!
     

    Attached Files:

Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day

-