Protecting MUX from huge voltage

I'm using an Arduino Uno in my latest tube amplifier project to control startup, emergency shutdown, and other functions as well as monitoring the voltages and currents. My concern is measuring the 500V, 400V, and -80V rails.
Below is the voltage divider that I plan to use for the 500V rail which works fine. My concern is the problem that arises should R2 fail and the full 500V is presented to the opamp which is in front of the incredibly difficult to replace SMT 74HC4017 MUX. In the event of R2 failure, R1 has to dissipate about 1/4W and the zener about 2mW, and 5V is on the op amp input.
Is the simple zener diode sufficient? Is this the standard approach or is this a newbie error?
Thanks for your help.
Mike

I would divide the 1000k in 2 resistor of 500k, or 4 of 250k, just to stay safe, and add a diode from op input to 5V (anode to op, cathode to 5V).

Ciao, Ale.

Thanks, I think I'll divide R2 into 2 series resistors as well. What does the diode at the op amp input protect against?
BTW, its a 74HC4067 MUX. The above is a typo.
Mike

You can leave R2 like that, I suggest splitting R1 'cause of the high voltage that R2 doesn't see. The diode to V+ limit the input voltage to 5V + plus diode foward voltage, so around 5.5V, I see, you already put a zener for that, I like more the diode solution. You can even add another one from ground to input (anode to ground) to protect from negative voltage .

Ciao, Ale.

I like the diode solution, too. I'm going with that one. Thanks for your help.
Mike

The zener should do the trick.

[u]Protection diodes[/u]. The advantage to a zener over a regular diode to +5V is that the regular diode only works when the 5V works.

...the incredibly difficult to replace SMT 74HC4017 MUX.

Difficult to replace or not, you should protect your low-voltage circuits from the high voltages.

You shouldn't need the op-amp buffer since your MUX should have a high impedance input {relative to 5.6K) and the Arduino has a very high impedance input.

Are you sure you need the MUX? The Uno has 6 analog inputs.

For the -80V you can make a voltage divider between +400 and -80, or between +500 & -80, which gives you a positive voltage under normal conditions. In this case, you should protect the Arduino from high positive voltagesand any negative voltage. A "reversed" diode to ground (following the voltage divider) will protect against negative voltages.

In the event of R2 failure, R1 has to dissipate about 1/4W and the zener about 2mW, and 5V is on the op amp input.

The odds of a resistor failure are nearly zero unless it's over-powered and burns up. R1 has nearly 500V across it under normal conditions so I'd recommend a 1/2W resistor.

Use R2 = 100k, add 100nF in parallel to R2, and make R1 two 10M high-voltage resistors in series - that
will be much safer than the single low voltage resistor which could easily flash-over in damp conditions.
Also using high value resistors will reduce heating in R1.

High voltage resistors are a lot longer, and for instance won't fail if an ant crawls over it! Standard
resistors are definitely not rated for 500Vdc.

I'll definitely use the recommended higher-value, higher wattage resistors. I am monitoring 9 voltages + temperature, so I need the MUX. It's all programmed and works well and all the high voltages are well away from the arduino board and my tube amp shield. But I was unclear about how to protect the arduino and associated circuitry form the old-school tube amp voltages. The only reason I used op amps (DIP packages on IC sockets) was to sacrifice them instead of the MUX in case of a high voltage making its way onto the arduino board.

I feel comfortable with the solutions you guys provided. This is a really good forum.

The LM358 has problems with high value resistors.

An Arduino input can measure those voltages including the -80volt directly (no opamp).

Why the MUX. An arduino has more than three analogue inputs.
Leo..

Wawa:
The LM358 has problems with high value resistors.

Could you eludcidate? The input bias current is 50nA or so, I don't see the issue with
100k input impedance, that's 5mV or 1LSB.

An Arduino input can measure those voltages including the -80volt directly (no opamp).

indeed, so long as the voltage divider is present!
[/quote]

Something worth considering depending on the physical layout of stuff - consider putting a 2 stage divider in the system - put the first "stage" in with the amplifier and high voltages and drop the output to say 10 or 20v which you could then feed into the divider in the front of the Arduino. That would remove the high voltages from close proximity to the Arduino as well as not having those voltages wandering around in the cable you have from the tube section to the Arduino (I am assuming they are not physically on the same chassis?)

Input bias could also be 150nA, and I saw 2x10Megohm mentioned.
That opamp is not a rail2rail opamp either.
You might only have 700 A/D values with default Aref and powered from 5volt.

The "use schottky diodes to the rails" recipe won't work with high impedance voltage dividers.
Leo..

I agree with gpsmikey to split the divider and leave the dangerous part inside the valve amp.

High voltage is only dangerous (to humans and Arduinos) when there is sufficient current flow.
400volt through a 400k resistor and connected directly to an input pin (no resistor to ground) is not going to harm the Arduino. Arduino's internal input protection diodes will direct that 1mA safely to the 5volt rail.
No need to try that ofcourse.

You shouldn't have more than 100volt across a common small through-hole resistor.
Values above 100k are more likely to go open circuit (experience).
There are special (= expensive) high-voltage resistors.

This should work. Use "7 + 1" resistors for the 800volt section.
Leo..

DVDdoug:
The zener should do the trick.

The advantage to a zener over a regular diode to +5V is that the regular diode only works when the 5V works.

That does not make sense.
Max input voltage (if you ignore current) is VCC +0.5volt.

It should be:
A zener diode does not work when the 5volt supply is not there.
The advantage of a normal diode is that it also works when the Arduino is off.
Leo..

Wawa:
Input bias could also be 150nA, and I saw 2x10Megohm mentioned.
That opamp is not a rail2rail opamp either.
You might only have 700 A/D values with default Aref and powered from 5volt.

The "use schottky diodes to the rails" recipe won't work with high impedance voltage dividers.
Leo..

But the 20M is not the input impedance presented to the opamp by the divider, that's 99.5k
if you work it out, unsurprisingly the input impedance from a divider is less than the lower
resistor.

MarkT:
But the 20M is not the input impedance presented to the opamp by the divider, that's 99.5k
if you work it out, unsurprisingly the input impedance from a divider is less than the smaller
resistor