Problems associated with very large resistors (1M+)

The resistance of several megoms is not a very big problem. However the conformal coating is absolutely necessary. Also, when working with the high resistances, don't forget to take into the account the stray capacitance and the maximum voltage across the part. Consider the series connection of several resistors of 1M each.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

Leakage isn't an issue unless you expect unusual dust/dirt or condensation. If you're worried about that, use a 100M or 1G resistor in the top of your divider and coat the whole divider region with a polyurethane or something. In a benign environment, expect picoamp leakages on a clean board.

Touching a 100M resistor won't change things significantly, unless you're really iccky.

John

Paw prints won't do much. Board cleaning is a much bigger problem, especially with water soluable fluxes. I have had problems in the past where the assembly house didn't clean the boards well and messed up a circuit with a 10Mohm impedance and another with 350kohm impedance just recently.

He's a twit, and probably only has worked with digital stuff. You're just drawing power, not making a measurement, right? A couple of M will have a healthy 27.5uA flowing through it with 55V applied. That's like a 100K resistor at 3V.

With simple uncoated boards (no no-clean flux) in benign environments

Best regards, Spehro Pefhany

We had boards assembled by a plant that used mostly wave soldering (with a few hand soldered bits that were not compatible with wave), but used "fluxless solder". It was a small run (about 50) but substantial if you're handling assembly yourself.

Basically, it was fluxed solder, but it evaporated to leave nothing more than a mostly invisible film on the board.

After the 70-80% failure rate on our boards, we had to clean every one of them ourselves, and tell them if they didn't, we'd shop elsewhere. When it came time to make up a new batch of boards, we chose another outfit anyway - and pre-warned the new people. As it turned out, we needn't have, they cleaned all their boards regardless. They tried the fluxless solder themselves earlier and found you get a more reliable product using traditional methods with cleaning. Albeit at a higher cost.

Our boards were doing low level signal analysis, with about 60dB gain on some areas, and the "invisible" flux was causing grief on the higher gain stages. However, there were a couple of digital boards as well, (all at 5v logic), and those were causing grief too. So this wasn't ALL a case of special purpose boards that were not compatible with this fluxless rubbish either.

After cleaning it all worked as per spec.

Usually problems from (some kinds of) dirt, and touching (skin oils, acids, and bases) can be seen with 1,000Meg (and higher) resistors. One would be hard put to goof up a low value 1Meg resistor. Now if any power worth mentioning is needes at low voltages (the 3V mentioned comes to mind), then a DC-DC converter would be a lot more efficent (lower loss, less heat dissipation).

Check; even that divider you mentioned needs no particular coating if otherwise protected from crud.

Organic fluxes and some so-called "no-clean" fluxes are ionic in the right (for the unlucky, the wrong) environment. DI and PGA become rather useful....

It isn't a "very large resistor", it is a "very high resistance value resistor".

These parts can be affected by simply breathing on them. Not just immediate effects either. Long term effects. The epoxy most are dipped in is hygroscopic, for one thing. Very bad at high values. Conformal coating won't help if it isn't insured to be very clean before the coating process. Whatever is in or on the part gets trapped in with the coating. Must be VERY clean, and should use a vacuum of the assembly to remove any water before conformal coat step. Conformal coating should be oven cured as well, not merely air dried.

You need to get real with your requirements. It makes no sense to use a divider load that is 1% the self-discharge rate of battery, and it makes no sense to have any switchable sensing divider on all the time when you only need to sample it at low duty cycle. For example, if your divider loading is as much as 1mA and you only switch it on at 1% duty, then the average loading will be be 1mA/100=10uA, which is equivalent to a divider of 100x the resistor values, e.g. if the divider is 55K and switched on only 1% of the time, it draws no more than a 5.5M divider.

If using surface mount parts, to prevent trapped contaminants, underneath them where you can't clean it off I recommend mounting the parts upside down. That is the sealed/marked side to the board. This will allow thorough cleaning of flux and crud from your divider path and un-sealed parts backside. You don't have to worry too much about the gunk trapped below your parts messing up your divider ratio. al

As meter resistors go, that's not very high impedance, BUT are you doing this with teensy little surface mount parts, or are you using a half-inch-between-leads radial ceramic/metal film resistor?

If you want it accurate, use a big component. 55V and short spacing on a PCB is not a nice combination. Cleaning the little space under a surface mount resistor is ... an unwanted variable.