Audible for exact voltage reading

Thread Starter

Knowledge seeker

Joined Dec 10, 2022
6
I’m testing batteries at a battery storage construction site with a Fluke 87V looking for 50.5 VDC (~0.01%). I’m looking for a solution that produces audible indication so I don’t have to look a the DMM each time I probe for verification. TIA
 

Alec_t

Joined Sep 17, 2013
14,280
Welcome to AAC!
That's a tall order. 0.01% is an extremely tight specification. Are you sure you really need it that accurate? Even the Fluke87's rated accuracy is 'only' 0.05%. on its DC setting.
I think all components in any designed measuring circuit would need individual tolerances even tighter than 0.01%, or the calibration of the overall system would need a similar accuracy, and the whole measuring set-up would need to be accurately temperature-controlled and electromagnetically screened. So the sourcing of suitable components is going to be a challenge and the system cost could be high. What is your budget for this?
 

Ian0

Joined Aug 7, 2020
9,679
Are these batteries in a temperature controlled environment? Batteries, like any chemical process, are temperature dependent. The terminal voltage will depend on the temperature of the battery.
 

crutschow

Joined Mar 14, 2008
34,285
Look up "talking voltmeter for the blind".

But to get 0.01% accuracy, you would likely need a precision bench multimeter, not a hand-held one.
However I think that degree of accuracy is much higher than you need to measure battery voltages.
 

ronsimpson

Joined Oct 7, 2019
2,989
We are measuring batteries. I don't care the absolute value, but we want to know if all the batteries in a pack are the same.
We do not care about accuracy but resolution.
Your meter will measure 50.51V for resolution, which is not enough for us but the 0.05% is much better than what we need.
In our case, there is a difference between what the boss wants and what we need.
 

Thread Starter

Knowledge seeker

Joined Dec 10, 2022
6
Welcome to AAC!
That's a tall order. 0.01% is an extremely tight specification. Are you sure you really need it that accurate? Even the Fluke87's rated accuracy is 'only' 0.05%. on its DC setting.
I think all components in any designed measuring circuit would need individual tolerances even tighter than 0.01%, or the calibration of the overall system would need a similar accuracy, and the whole measuring set-up would need to be accurately temperature-controlled and electromagnetically screened. So the sourcing of suitable components is going to be a challenge and the system cost could be high. What is your budget for this?
Thanks to everyone for their thoughts on this and I will read all your replies just as soon as I correct my typo that should have read 0.1% tolerance and for all practical matters it could go be 0.2% and still work for what I’m doing. I apologize folks :(
 

Thread Starter

Knowledge seeker

Joined Dec 10, 2022
6
Look up "talking voltmeter for the blind".

But to get 0.01% accuracy, you would likely need a precision bench multimeter, not a hand-held one.
However I think that degree of accuracy is much higher than you need to measure battery voltages.
Now that’s funny. Not just because it’s funny but because it’s got the additional Gary Larson sting that it would work for me. Aaaarrrgghh. I knew this was coming when I got in here.
 

MrChips

Joined Oct 2, 2009
30,720
Even 0.1% is rather tight tolerance to me.
±0.1% at 50V is ±0.05V. Thus you are seeking a window between 50.45V and 50.55V

±1% at 50V is ±0.5V, with a range between 50.0V and 51.0V. Now that is more realistic.
 

Thread Starter

Knowledge seeker

Joined Dec 10, 2022
6
I’ll post pictures Monday. All I’m trying to do is make a monotonous, repetitive task more efficient. Out of 179,256 total batteries there are roughly 150k left to test. Batteries go in buildings, tested for voltage, 22 battery tall racks are linked together in series, we go back and megger the stacks at 2Kv, then racks parallel to a cabinet then go to smaller buildings alongside housing inverters then 1Kv to 21Kv transformers and out to (or in from) the grid. Out of 30k batteries 3 have been bad so far (and not by much). But at $10k per they get replaced by the manufacturer.

If I could just get my hands on one of those talking voltmeters for the blind... I was thinking something really simple like stacked banana plugs at the meter with a potentiometer and buzzer field adjustable to match the meter. the buzzer circuit doesn’t need actual precision just not to beep at <50.3 and >50.7
 

ronsimpson

Joined Oct 7, 2019
2,989
Even with 10 turns post it will be hard to set to within 0.01%.

The comp cannot see 50V directly so I built it for 5.05V. That is an easy fix.
Reduced the post to 100 ohm, ten turn posts.
Put only 10mV across the post. (now that I rethink it maybe 1mV on the post)
1670780633594.png
There are still some problems, but it is a start.
The CMPs have internal errors that need to be addressed.
 

AnalogKid

Joined Aug 1, 2013
10,987
Even with 10 turns post it will be hard to set to within 0.01%.
True, but the spec has changed ...

Oops not to beep at <50.4 or >50.6
That is a range of +/-1 in 500, or +/-0.2%. I know that's not the same as the setpoint accuracy he mentions, but it does give a starting point for the two trip points. Because he accepts the idea of using the voltmeter to calibrate the comparators (as opposed to the comparator circuits being inherently accurate), the only real issue is stability. If the temperature coefficients of the trip points voltage divider string are tightly matched, then the problem reduces to

1. the output voltage stability of whatever the voltage source is for the setpoint divider string. This could be two LM317's (or something newer and tighter) in series, with the first one stabilizing the input to the second one.

2. the input offset voltage stability of the comparators. If a high quality (not an LM393) dual comparator device is used, these should temperature-track pretty well.

Bourns used to make trimpots with a series resistor on the same substrate for very tight temperature tracking.

ak
 

ronsimpson

Joined Oct 7, 2019
2,989
but the spec has changed ...
Oh, I see.

Get a really good reference. $0.50 to $1 gets you a really good known and stable voltage.
Limit the range of the pots. Keep calibration simple.
By setting the pot range from 50.2 to 50.8 the bad effects of the pot are limited. Cal is faster. Simple pot.
(If you are not going to have a good reference then you will need more voltage across the pot)
Make a "0.2V" with a voltage divider so CMP2 is always 0.2V less than CMP1. No Cal is needed.
Use good resistors.
CMP1 and CMP2 should be a dual IC. Not two singles. (helps match the two parts)

1671045924587.png
I would never make this circuit as shown. 50V is too high. Make a 5V or 12V version and divide the 50V down to something reasonable.
 

AnalogKid

Joined Aug 1, 2013
10,987
Schematic almost done.

1. What output do you need? Beeper? LED? Other?

2. What power sources are available for the circuit. The circuit plus indicator could require over 20 mA. If powered by the 50 V battery it is measuring, then 1 W of heat has to go somewhere. Not a big problem, but it will affect the sizes of some of the components.

ak
 
Top