60HZ pick-up?

Are you sure it's the cap? You can seriously lower the value of a 10M resistor just by fondling it with grubby fingers.

Try cleaning the resistors and the board thoroughly -- I'd use 99% isopropyl alcohol, because that's what I have, but any volatile board cleaner should work. Get the resistors, get any expanse of board that bridges the resistors, get anything that might be forming a path around those resistors. Make sure that you're putting on an excess of solvent and mopping it off -- if you just rub on some alcohol and let it dry all you're doing is moving the dirt around, you're not cleaning it off.

Let us know if that makes a difference.

--
http://www.wescottdesign.com

s

is

Tim,

The 10Mohm is the input impedance of the Data logger. One of the test's I did was to disconnect the wires to the data logger. However on measuring I still had ~ 0.4V at the output (This was without adding the Cap). I guess it's still a good idea to give a clean though.

Cheers,

Ozzy

to

).

ays

a

ng

is is

ny

M t

nt

l

If you used the "no clean" flux you are screwed. You have to use very aggressive solvents to get the stuff off

If you use alcohol, put the board in the hot box for a day or so before you re-apply the high voltage.

Can you try a different meter to make the measurement? It would be a shame if the whole mystery was the meter was oing bad on you.

The data logger may not like being driven from a 30 megohm impedance signal source. A small amount of bias current or mux charge injection could make a big error. If you can't scale down the divider resistors, you could add a picoamp-bias-current follower opamp on the divider board, suitably guarded/coated to keep leakage down.

John

PS- if the problem is charge injection, that can mess up the opamp, too.

John

A PCB is not the right approach for this. Dedicated HV probes are very high value resistors coated in oil to keep airborne humidity out, cased in a plastic tube. Touch the resistor and you just shorted it out. It is also quite long to accomodate breakdown voltage. Please specify a part number for the resistor. You might be better off with a longer chain of smaller resistors.

I have a probe like this:

Note the shielded cable. This is much like a record player pickup, all very high impedance circuitry.

(Also if you just picked up your probe at the post office and happen to take the Metro through the gay village, refrain from opening the box and smiling when holding the large red cylinder with a metal spike. I seriously did this because I was happy to get a good eBay deal on the probe and I needed the probe. Some guy almost hit on me because of this. Open the box at home.)

As for the AC DC thing, any stray capacitance will form a heck of a low pass filter with the multi gigohm resistor so it's likely you're probing DC.

What kind of cap?

n to

-

R3).

elays

g a

he

ding

This is

G

of

any

10M

hat

nd

vent

all

Hi,

I actually did handle these 10G resistors myself - I took them off the board one time, so they could be dirty. I could actually stand the 3 resistors off from the board to tell me whether it's leakage from the board or resistor. On hindsight guard tracks would have been an option, but I was tight for space.

Cheers,

Ozzy

Firstly, perhaps you forgot to mention the input current which may be a contributing factor. Just put the 30Meg resistor on the input, shield it if necessary and see what kind of DC offset you get.

That cap could retain a charge and give that reading, or charge up due to input current.

Use of a DVM (most have 10Meg input) would be a good alternate / substitute.

I missed that -- as well as just sort of skating over the "giga-ohm".

It might be a good idea to generally lower your impedance all the way around. Just as a diagnostic I'd be tempted to remove that G-ohm resistor entirely and see what sort of results I got -- I'm not sure that I'd trust _anything_ capable of absorbing moisture (like FR-4) to have more than a 1G-ohm resistance without specifically verifying it.

You're basically building a high voltage probe -- have you investigated what their construction and materials are?

--
http://www.wescottdesign.com

There's two problems here. First, high value resistors can fail, you want to clamp your datalogger inputs. Second, your datalogger has a capacitance as well as input resistance. Clamp diodes won't help this.

I'd estimate the input capacitance, use an inverting op amp configuration (so two clamp diodes to GND don't get forward biased), maybe add a few pF to the HV resistor, then parallel the feedback resistor of the inverting amplifier with a trimmer capacitor. The benefit here, is that the

10G meter resistor feeds a LOW impedance point, with good sparks-a-flying immunity. Drift is best if the (+) input to the op amp is ground-connected through a resistor that matches the feedback resistor. It could be bias current in your datalogger input that is causing you a DC drift/error, this will fix it.