I'm connecting various transducers to ADC inputs, typically through a voltage divider or other matching circuitry. What is the accepted method to protect the input from overvoltage/overcurrent? The inputs seem to be resistant to minor overvoltage, but my environment has 24 volts available, which would surely let the magic smoke out, not to mention the possibility of an ESD event.
MOV's, Zener diodes, TVS, Bipolar clamp diodes, are all viable options, but I have experience in using none of them.
-Don
Protecting inputs
I believe all of the pins on the Mega32 and Mega644, when configured as inputs, have clamping diodes which protect against overvoltage, but you still need current limiting resistors in series with the pin.
For ADC inputs you might consider using an opamp configured as a buffer to isolate the input from the sensor and associated circuitry. An anti-aliasing RC filter between the opamp and ADC input is also desirable.
The lower right corner of the attached schemastic shows what I did with a PIC based wireless generic ADC module a couple of years back.
For ADC inputs you might consider using an opamp configured as a buffer to isolate the input from the sensor and associated circuitry. An anti-aliasing RC filter between the opamp and ADC input is also desirable.
The lower right corner of the attached schemastic shows what I did with a PIC based wireless generic ADC module a couple of years back.
- Attachments
-
- gen_adc.gif (4.61 KiB) Viewed 6935 times
These are great suggestions! I used something similar in one of my circuits, but used a comparator. However, the addition of an op amp adds to the circuit complexity.
The suggestion of an "antialiasing" filter is also good.
By this, I think he means a low-pass filter set to something lower than the ADC sampling frequency. I _think_ that his schematic will have a low pass filter fundamental of around 3MHz. With a 50 ohm resistor, maybe a 1uF cap would bring the fundamental down to around 30KHz, whouch should be a better low pass filter.
All rou REALLY need is a resistor sufficiently large to limit the current under your "worst case" scenario. The input is supposed to be able to tolerate 20mA, so a current limiting resistor for 24 volts would be
R=V/I,
R= 24V/.020A = 1200 ohms. So something larger that that ought to be OK. Maybe 2.2 - 10K.
You could still have a low pass filter by using a cap to ground after the current limiting resistor. You could also add your own schottky diodes to clamp the input voltage without relying on the CPU's protection (also after the current limiting resistor).
That should be pretty good for all "normal" overvoltages. ESD protection? MOV might do it. The Schottly's might be sufficient. I have used just a limit diode of 20-something K, a 0.1uF low pass filter cap, and clamp diodes, in one circuit with no fried CPU. And this was with fairly ample opportunity to have ESD damage: connector in and out a zillion times.
-Tony
The suggestion of an "antialiasing" filter is also good.
By this, I think he means a low-pass filter set to something lower than the ADC sampling frequency. I _think_ that his schematic will have a low pass filter fundamental of around 3MHz. With a 50 ohm resistor, maybe a 1uF cap would bring the fundamental down to around 30KHz, whouch should be a better low pass filter.
All rou REALLY need is a resistor sufficiently large to limit the current under your "worst case" scenario. The input is supposed to be able to tolerate 20mA, so a current limiting resistor for 24 volts would be
R=V/I,
R= 24V/.020A = 1200 ohms. So something larger that that ought to be OK. Maybe 2.2 - 10K.
You could still have a low pass filter by using a cap to ground after the current limiting resistor. You could also add your own schottky diodes to clamp the input voltage without relying on the CPU's protection (also after the current limiting resistor).
That should be pretty good for all "normal" overvoltages. ESD protection? MOV might do it. The Schottly's might be sufficient. I have used just a limit diode of 20-something K, a 0.1uF low pass filter cap, and clamp diodes, in one circuit with no fried CPU. And this was with fairly ample opportunity to have ESD damage: connector in and out a zillion times.
-Tony
Thanks for the input.
At first, I hadn't thought much about the ESD problem, but the more reading I do, the more common it seems to be. Call me incredibly lucky, but I have yet to lose a chip due to ESD. Now that I've said that of course....
I'm looking at MOV package sizes right now. If I can find one small enough, I think I'll go that route, as it seems the most robust protection I can add, hopefully without increasing the board size (just over 2 square inches, and there's already components on both sides). I like the idea of simply adding a resistor to limit current, but that might not be enough for a 10KV ESD strike.
-Don
At first, I hadn't thought much about the ESD problem, but the more reading I do, the more common it seems to be. Call me incredibly lucky, but I have yet to lose a chip due to ESD. Now that I've said that of course....
I'm looking at MOV package sizes right now. If I can find one small enough, I think I'll go that route, as it seems the most robust protection I can add, hopefully without increasing the board size (just over 2 square inches, and there's already components on both sides). I like the idea of simply adding a resistor to limit current, but that might not be enough for a 10KV ESD strike.
-Don
Well, they are pretty easily available in disc radial lead packages 5mm and 7mm diameters are easy. SMT is also available.Don_Kirby wrote:I'm looking at MOV package sizes right now. If I can find one small enough, I think I'll go that route, as it seems the most robust protection I can add, hopefully without increasing the board size (just over 2 square inches, and there's already components on both sides). I like the idea of simply adding a resistor to limit current, but that might not be enough for a 10KV ESD strike.
-Don
The schottky diodes are pretty good as well. I just am not sure of the typical turn-on time, nor how that compares to MOV, and how good this would be for ESD protection.
I have no special knowledge on this subject, but I did confer with an engineer about a power supply design, and he used a 7mm diameter MOV (about 33 volt DC max sustained voltage). For the power supply a polyswitch current limiter as well. The power supply design was not specifically intended to be ESD protected, but overvoltage in general, and ESD too. For the some of the external inputs, nothing special at all other than a low pass filter was used. On inputs to the CPU from internal sources (less ESD risk?), no protection at all was used.
-Tony