I want to sence a voltage signal (x(t)) which can vary from 0V to 5000V. The signal will be driven to a MCU unit whose maximum input voltage can be 5V. I want to divide this input voltage using voltage dividers if the voltage level is greater than 5V.
Only one of the switches will contact at a time.
Code:
if (0V < |x(t)| < 5V) ==> take it directly to the MCU
if (5V < |x(t)| < 50V) ==> divide it by 10, then take to MCU
if (50V < |x(t)| < 500V) ==> divide it by 100, then take to MCU
if (500V < |x(t)| < 5000V) ==> divide it by 1000, then take to MCU
But to this, I need four switching elements for switching on and off these four voltage dividers. To keep the sensitivity of the circuit high, these switching elements must have as low voltage drop as possible on them. In the schematic above, I placed four FET transistors for this purpose, but I'm not sure if I made the correct choice. As you guess, the swithcing element will work only in cut-off and saturation modes. FETs have lower internal impedence that BJTs, so they are more convenient for my project, isn't it? Is there any other semiconductor element you can suggest, which would have less voltage drop on it? I think old mechanical relays are best for this, but their response time is longer, they keep more space on the board, and I don't want any non-solid-state circuit elements on my circuit.
Re: I need a switching element which would have low voltage
You may want to check your basic idea against the properties of available semiconductor devices. You'll find MOSFET
with up to about 1000 V rating, but the circuit won't work due to their rather high leakage currents. A few special JFET types
can handle voltages above 100 V, if I remember right.
Apparently, you're trying to design a programmable voltage divider. Looking at existing instruments, you'll notice two realistic design options:
- Using a common high side resistor with switchable low side resistors. For 5 kV input voltage, it would have a considerably
higher resistance than 10 MOhm to achieve an acceptable input current and power dissipation.
- Using relays with sufficient voltage rating. They are available up to several 10 kV, but of course bulky - and expensive.
Overvoltage protection of the input amplifier and voltage divider parts should be considered as well.
There are some dubious design details in the resistor arrangement around the OP, but I like to ignore it as problem of minor importance.
Re: I need a switching element which would have low voltage
FvM said:
You may want to check your basic idea against the properties of available semiconductor devices. You'll find MOSFET
with up to about 1000 V rating, but the circuit won't work due to their rather high leakage currents. A few special JFET types
can handle voltages above 100 V, if I remember right.
Rather than handling high voltage, I want to know how to do this switching operation in general. Is it there any semiconductor meterial or any other solid-state technology which has zero voltage drop on forward biasing? Or should I pre-calculate this voltage drop and add its effect on the microcontroller side?
FvM said:
- Using a common high side resistor with switchable low side resistors. For 5 kV input voltage, it would have a considerably
higher resistance than 10 MOhm to achieve an acceptable input current and power dissipation.
If I were to place a resistor greater then 10MΩ, the output of the voltage divider would have had a very high output impedence. Since an ordinary opamp has about 1MΩ input impedance, wouldn't that considerably effect the opamp operation?
FvM said:
Overvoltage protection of the input amplifier and voltage divider parts should be considered as well.
There are some dubious design details in the resistor arrangement around the OP, but I like to ignore it as problem of minor importance.
This schematic above is not my circuit. In fact, I haven't decided how to design my circuit yet. I draw a simple example circuit to illustrate my question. I know that the opamp will invert the addition, and there is no high voltage protection, I am planning to add them in my real project.
In general, FET switches can be considered as controllable resistors. They have a finite rds,on, but no voltage drop at zero current.
At a MOhm resistor level, the switch resistance would be typically neglectable. But as said, there are problems than on-resistance.
As a general consideration, any real circuit has non-ideal properties. They should be calculated in relation to the specified
circuit accuracy. In so far, zero error (e.g. zero voltage drop) isn't a reasonable specification.
Since an ordinary opamp has about 1MΩ input impedance, wouldn't that considerably effect the opamp operation?
I would expect neither ordinary resistors nor ordinary OP-Amps in the circuit. But I only wanted to point out a
particular problem. Circuit calculation is your part.
Yes, as FvM wrote too;
You have FETs direct on HV, YOU MUST HANDLE this HV_you can not clamp it out...
The solution can be ONLY some "switch on the low side"....
Since an ordinary opamp has about 1MΩ input impedance, wouldn't that considerably effect the opamp operation?
This is NEW for me!
OP Amp can have 1Mohm, but its depend of type & circuit!
You must know, an OpAmp with very high Input Imp. is not inpossible to build_then all noninverting OPs are so....
Pls select another idea, with switches on the Low Voltage (at OP) side & you can change the outputs of an attenuator ladder to your Op for example....
K.
Re: I need a switching element which would have low voltage
Maybe you do without the switching elements. But this depends on the accuracy and conversion speed you need.
Consider a fixed 5000 on 1 voltage divider, and a 24bit sigma-delta ADC, with a full scale for 1Volt (as example).
This means that the lowest range (0 - 5V) becomes (0 - 1mV) and you still have a resolution of (2^24)*1e-3 = 16777 increments = ~14bits. All you need to do then is to select the wanted range in software.
I don't know if this is the best solution, bit it should be theoretically possible.