Bias current, leakage, DC offset? Remote trouble shooting

What's with the ground connection to the toggle switch?

...Jim Thompson

Singapore is pretty humid pretty much of the time, I think, so the flux residue might be fully hydrated already. Did you recently change to water soluble flux? 1 megohm times 1000 is 1 gigohm effectively, so all you need is 7 nA of input bias current. I've seen worse than that with water soluble flux.

Could also be a bad socket, e.g. one made from nylon-66. That also misbehaves in high humidity, but wouldn't respond to being breathed on.

Alternatively, they could be next to a big radio transmitting tower.

Or maybe the 49.9 ohm resistor on the x1000 scale is wrong, or shorted. Do you get 0.7V offset on the x100 scale?

Cheers

Phil Hobbs

If the other 4 units work ok then swap input cables and whatever else is easily swapable to see what the problem "follows".

Does sound like an RFI or moisture issue.

Mark

What do you see when both inputs are grounded?

Have you changed the capacitor?

The other thing that'll do this is cleaning off the flux using poor quality solvents, _especially_ drug store rubbing alcohol.

Cheers

Phil Hobbs

Almost surely leakage due to the wrong type of flux such as so-called 'no clean' flux.

The current increasing with time **is** an effect I have observed. As them to breath on the PCB- you should see a rapid increase in leakage.

Carefully and thoroughly scrubbing the area around the op-amp input with a toothbrush and strong solvent such as lacquer thinner should cure it (care is needed to avoid getting that kind of solvent on certain other parts).

Or you could buy me a plane ticket and I'll go over and do it for free. ;-) I could use some genuine Singapore hawker food.

--sp

There are two inputs to the instrument amp, so if they are just using one they can ground the other.

George H.

You know I just sent off an email asking about the humidity.

Hmm OK I could check if we've gotten a new shipment of sockets.

Well the other four units that they have don't show the effect.

Yup and ~70 mV on x10... I forgot to mention that we checked that.

George h.

A very small offset. Also if both inputs are AC coupled then the output offset goes away... well is much reduced.

George H.

Ahh well there is drug store rubbing alcohol in the production area. (Why is it bad?.. I'll google.) The boards are washed in an ultrasonic bath with whatever "gunk" Kester recommends. But there is a process step where the the switches are soldered and then cleaned. I'll ask what they use for the cleaning.

George H.

I've seen something like that before. The capacitance was acting with the device input becoming a somewhat detector of noise. The output showed as a DC offset that moves around depending on the external conditions.

But our findings also showed the output had some noise on it and what we did was put an R inseries with the cap, that seem to remove the problem.

Hiz inputs are a bitch to work with, we normally suspend them and enclose that section.

Jamie

OK your ticket is in the mail. :^)

I asked him to breathe on it and see what happened.. he said he saw no change. But this is by email, and well, he is a typical tech... we had a few days of discussion about bias current and why the offset is bigger on AC than DC. So what do you recommend for cleaning? I asked him to try some cleaning, and looking back at the email I see that he used IPA. (Perhaps not the best according to Phil.)

The one problem with cleaning is that the pcb is held to the front panel by the switches all of which have to be removed to get to the under side.

I'll try breathing on some of the boards here and see if I can get them to respond.

George H.

(Have I ever mentioned that I have a love/hate relationship with flux.)

You can get 99%+ IPA over the counter but the store will most likely need to order it. Takes a day or two. If you really want to go crazy, get the ultra pure HPLC grade 99.5%+ for somewhere north of $200 for 4L glass bottle plus shipping.

Just don't leave the top off less it becomes 91% soon enough.

Don't forget to clean inside the switch and under the chip. Might even use ultrasonics.

What type capacitors are you using?

Telepresence would be nice.

Not sure it's useful for you, but I used this 'hardware store' stuff- it's not specifically recommended for electronics, but it works for me. YMMV. It can attack some plastics and take off non-epoxy paints, but so far have not seen any harm to ordinary components.

Toluene. 60 - 80% Methyl ethyl ketone 10 - 20% Methanol 5 - 9% Acetone 1 - 5%

Not too bad to work with- it should be used in a well ventillated area.

There's purpose-made spray stuff we get for prototypes but I've never tried it in this particular situation.

Isopropyl Alcohol 15 - 20% Trichloroethylene 80 - 90%

My recollection is that it took a lot of scrubbing to get the contaminated boards to >5G leakage. Unfortunately you can't *see* the contamination.

Have you perchance found a 'stronger' liquid flux- I've got some expensive through-hole parts that have lost a bit of solderability. I'd hate to have to discard ~700 pieces just because of that.

--sp

Rubbing alcohol has oil in it. ...Jim Thompson

Is he in a high RF environment? The AD620 is really good, but if there is a lot of RF, it can rectify it and look just like there is a DC bias. I always put in some RC filtering of inputs that come from the outside world.

Jon

Hi Spehro and others, So I got a few of these off the shelf, and tried the breath test. On the back side, nothing. But on the other side I spiked the meter! So I can reproduce it. (which is nice.)

The front side (between the pcb and the front panel) is where all the components live.

I bathed the area in 91% IPA, dried it and that stopped it!

I'm now going to try removing the socket and reinstalling the I-amp. (on a different pcb that also fails the breath test.) Thinking that maybe it is the plastic socket that is the problem... I don't know.

More later. Tanks, George H.

Hah, it's the socket! Phil wins the kewpie doll prize.. and my heartfelt thanks.

George H.

Watch the ultrasonic bath - you could trash bond wires if the frequency happens to be exactly wrong. I've also seen plastic packages (TO-92) develop free floating pins after an ultrasonic adventure.

RMA flux really needs a water wash with an appropriate detergent very soon after soldering to get at ionic residue. If you wait a week, the flux becomes harder to remove (oxidized/polymerized). If you overheat the solder you get a very difficult to remove residue.

We've used this with good results:

Alconox also has a neutral detergent called Luminox - not quite as aggressive as D-8 but will not corrode aluminum.

The detergent wash should be followed by a deionized water rinse and an overnight 45-50C bake.

I reserve the IPA wash for cleaning after the installation of unsealed, rotating or lubricated parts (trimmer caps, switches, sensors etc.). IPA won't hurt o-rings and generally won't strip lubricants (very much).