[SOLVED] Comparator/bidirectional zener problem

[marfirefly]

This is a simplified version of a circuit I'm using to determine whether a signal is positive or negative at any given instance. However, no matter what the input is to the left of the current limiting resistor, the voltage at the input of the comparator is never negative. It bottoms out at 0.47 volts. I've applied various negative voltages so it to the same result. My guess is that the clamping diodes in the comparator are causing this somehow but I'm not sure why that would be. Also of note, this only happens when VCC is applied to the MAX964. When VCC is off and a negative voltage is applied left of R(lim) then I get a nominally negative voltage at the input pin of the comparator. It's not anywhere near the zener voltage or -9v, however. Any ideas/help is appreciated.

 

[Audioguru]

Your schematic has no Vcc supply voltage and does not say whether Vcc is positive or negative.
It also has no zener diode voltages or part numbers.

The datasheet for the MAX964 shows that its maximum allowed input negative voltage is -0.3V so why are you trying to destroy it by feeding much more negative input voltage? Your zener diodes will not limit the negative input voltage to anywhere near -0.3V. The datasheet does not mention input clamp diodes but maybe you can use a 1N5817 Schottky diode.

The input should normally go to ground or to the supply voltage. Since your input clamps itself to no less than +0.47V then I think you have destroyed the IC by feeding the negative input voltage.

 

[marfirefly]

VCC is +3.3V. Thank you for pointing that out about the negative voltage of the input. I'm sure you're right and the IC is fried. I'll have to redesign it so that VCC and VDD are biased downward by a volt each, I suppose, so that zero crossing will remain at GND. Thanks for your help.

 

[Warpspeed]

The circuit should certainly work, but I suggest you replace both zeners with a single Schottky diode.

You are NOT supposed to see a negative voltage into the comparator, it is bad news, potentially destructive with most chips. But apparently not with the MAX964 as it must have some sort of internal protection.

Just monitor the output of the MAX964, it should be working exactly as you expect.

 

[FvM]

The circuit seems to completely ignore MAX964 maximum ratings. The zener diodes are effectively useless, just adding some (probably unwanted) input capacitance. To protect the inputs, you would want to place schottky diodes, preferably two in antiparallel circuit to also clamp positive input.

It's not uncommon that the internal substrate diodes of fast analog devices are rated for a certain current level like 5 or 10 mA, in this case external clamping diodes might be omitted, avoiding additional capacitive load. But no such specification is given for MAX964, so the maximum input voltage range of -0.3V to (VCC + 0.3V) should be kept strictly.

 

[marfirefly]

Funny thing is I went to another forum before edaboard and asked about that specific issue, that is, the limitations of the inputs of a linear comparator, and everyone seemed to think that all that mattered was whether one input was higher than the other regardless of their relationship between VCC and VDD. Now that you all have pointed out that it does in fact matter and that it's not as if I'm dealing with two sides of a galvanic isolator, the problem in my circuit makes perfect sense. My only remaining question is whether anyone sees a problem with me biasing VDD to -1v and VCC to 2.3 volts so that I can reference my polarity to ground? Or, is there another/better/quicker way of doing this?

 

[FvM]

If you don't mind that your input source is loaded a bit by clamping the input, you can operate MAX964 well with single supply and ground referenced input because its common mode range includes the supply rails.

 

[Warpspeed]

It will work perfectly well right down to ground as FvM says, but its advisable to not let it go much beyond ground in the negative direction.
Hence my earlier suggestion of using a Shottky diode.
While there will be some clamping action within the MAX964, its better not to rely completely on that, and add an external diode.