Expanded scale voltmeter

[GTBecker]

Let me pick your brains, once again, please. Here's the simplest no-calibration expanded scale voltmeter I can think of. Is this likely to work?

[twesthoff]

It can be done a little simpler. All in series, 10V zener, resistor, meter. Three parts. Make the resistor a pot to calibrate.

Not as accurate as what you had, but for battery levels it should be good enough...
Tom W

[GTBecker]

I don't want to calibrate. Isn't the Zener temperature sensitive and, running at 1mA, out-of-spec? A JFET could be used as a constant current load, but it still would need a pot for calibration, no?

[twesthoff]

[GTBecker]

You're right about tolerances, of course. Every time I've looked at one, the 7805-regulated rail of a project has been something like 4.98v or 5.03, within a percent or so - warm or cool. They seem to be generally well within the +/- 5% error quoted in specs, and they have internal temperature compensation. Still, you're right that the theoretical worst-case error in my circuit would be 10% if the meter resistance (200 ohms, as it happens) is considered in its 5k series resistance.

A single calibration pot in the meter resistance wouldn't be sufficient, I think, since that will only correct full-scale - and not the offset. A proper solution would also need to set the total subtracted voltage to 10v.

Yes, this is for the boat so it isn't terribly critical. I've installed four shunts to allow monitoring charge currents and voltages from the primary shorepower charger and the alternator, and load currents and voltages from two independent battery banks.

I was looking for a simple but still reasonably accurate method with parts on hand. Most accurate ones use two opamps and two pots to set offset and gain, more than I care to build.

[GTBecker]

How's this? It should satisfy us both, I think. I don't have a 317L, but I've got 317s. Still, I'll build the 78L05 version first to see how it does without calibration.

[twesthoff]

[GTBecker]

No, then the dropout voltage becomes a problem; below Vin=11.5v or so it wouldn't work. Using two regulators eliminates that.

Yup, uncalibrated, simple and accurate were the goals.

[GTBecker]

First feedback:

The simple version, using a 78L05 and a 79L05, with two stock resistors, 2k2 and 2k7 (4k9 sum) and the measured meter resistance of 130 ohms, yields very good accuracy, maybe 2% error, between 10.0v and 14.2v.

Something happens at 14.2v, though; above 14.2v the meter response flattens out some, so 15.0v reads as 14.5, and 16.1v reads as 15.0. I haven't yet determined if this is an electronic error or mechanical.

Tom

[twesthoff]

If you have time, I'd like to see how the simplest circuit works in comparison.  The one with a 10v zener, resistor and meter in series.  Don't mean extra work, just curious...

On 8/12/2010 3:27 PM, General wrote:

[GTBecker]

If I can find a 10v Zener here, sure; otherwise it'll need to wait for my next parts order. I think I have a pair of 5.1v Zeners, and I think that's a stocked part in the Radio Shack "force-feed" drawers.

Tom

[GTBecker]

Most odd.

The two regulators behave as expected with the meter disconnected. When the meter movement or a 120-ohm dummy load is connected, though, the inverted negative regulator fails to subtract 5v from its input voltage when the input exceeds ~-14.2v. I see no oscillation and no bypass cap placement helps.

Maybe it needs more than a 1mA load? If that's so, why does it appear to track with no load?

[edit] Indeed, adding 1k5 to the 79L05, Com to Out, solves that problem. That adds a ~5mA load.

[twesthoff]

Have you tried larger values for the 1.5K resistor? How about adding a resistor to the 7805 as well. Maybe 10K on each of them perhaps?

[twesthoff]

Looking at the 78L05 data sheet closer it says:
"Note 4: Recommended minimum load capacitance of 0.01μF to limit high frequency noise."

Also both regulators are not working as they were intended and that is to source current. These are both being asked to "sink" a little less than 1 ma.

So in "theory" (as far as I can tell) for this circuit to work correctly you should have a 5K resistor from output to gnd on both regulators, and at least a .01uf cap too.

It also says in the data sheet that if the input leads are greater than 3 inches long there should be a minimum of 0.33uf from input to gnd as well. In your application, my guess is that the wires are a lot longer than 3" from a filter capacitor. You did say you didn't see any oscillations, but it is better to be safe...
Tom W

[GTBecker]

OK, here are Zener results.