# Steady Voltage Supply for extended runtime: Batteries in Parallel?

I have a device to measure tilt which is powered by 2 batteries (9 Alkaline Battery, Industrial by Duracell). So essentially, the voltage supply is 18V. I am using it for a project that came with a "black box". This black box has essential has a switch that changes the supply from battery to a supply from mains and a converter to step it down to 18V.

I need to log the tilt measurements for large amounts of time (a few weeks or months even). The mains supply can run forever but it is just too noisy and distorts the data. Digital filtering and smoothing doesn't help as the recovered value is wrong and accuracy is very important. The battery supply is fas less noisy and I am able to get accurate readings, however, the batteries only last about 3 days.

My question is, to provide a stable supply for a long time; would connecting multiple batteries in parallel solve the issue a will there be some shorting if there is a mismatch in voltages.

Is there any other solution (to make it work with batteries)? Smoothing the mains even further?

• If I were you, I'd talk to the suppliers of the "black box". Tell them it's not working like it should, ask them to fix it or send you the necessary specs so you can make a workable solution yourself. Ask them about input voltage tolerance, for instance. And current consumption. – Dampmaskin Mar 10 '17 at 9:55
• Running batteries in parallel is generally not a recommended scheme. – Michael Karas Mar 10 '17 at 10:53
• You (& we) need to know the allowable voltage range of the 18V supply. It could be 18 +/- 0.1V or it could be eg 14-20V or ... . If the spec is tight then almost any battery by itself is unsuitable as the voltage will fall as it discharges. A possible solution is a 5S laptop battery of nominally 18V . These are actually 5 x 4.2V = 21V when fully charged, fall to 18V at about 1/2 charge and are at about 15V fully discharged. Even a small 18V laptop battery will have VASTLY more capacity than 9V PP3 batteries. You may need a voltage regulator BUT 1st tell us the acceptable Voperating. – Russell McMahon Mar 10 '17 at 11:13
• @Dampmaskin the black box was made by a lab technician a few years ago for a spirit level model that is not being produced anymore so getting such data might be out of the question – Baba Mar 10 '17 at 11:30
• Ok, some more suggestions then: You could test it with different voltage inputs. Start at 18V and reduce the input voltage until the box stops functioning. You now know the minimum input voltage. The maximum input voltage is riskier to figure out, but you could open the box and try to reverse-engineer the circuit. – Dampmaskin Mar 10 '17 at 11:35

If you use 12 cells of alkaline batteries with 1.5 V each instead of 2 9 V blocks you get also 18 V but with much more capacity in amphours. The 9V alkaline blocks have only about 0.5 Ah, AA cells 2 Ah, size C 8 Ah and size D 18 Ah. If the 9 V blocks last about 3 days, C cells will last about 16 times longer or 48 days. D size cells will last about 100 days. Using the small AA cells, a run time of only 12 days will result.

• A step in the right direction - BUT Alkaline cells have qute a wide range of Voltage during discharge. Brand new you typically get 1.65V, dropping t slightly above 1.5V very quickly. The nrogressively to about 0.9V fully discharged. So for eg 1.0V endpoint and an LDO with about zero dropout he'd want 18 cells for Vbat >= 18V – Russell McMahon Mar 11 '17 at 5:29

You may want to consider a solution with a different type of "black box" altogether. Use a re-chargable battery technology to provide the operating voltages for the device. The alternate black box would be designed as a mains powered charger for the re-chargable battery. This charger would have a means to shut it off during times when the tilt measurements are actually being made.

You will have to also run some experiments on the setup to determine the type of shutoff control to use for this charger. You may find it suitable to electronically simply shut off the output of the charger. On the other hand if the mere connection to the mains is what results in the instability of the tilt readings then it may become necessary to disconnect the charger through true galvanic isolation such as a two pole relay. This could either switch the AC mains connection or the charger output connection to the battery system.

• Oh thanks hadn't thought of that. It's a useful solution. The only issue here I think is that, the tilt measurements need to be made continuously over like a month without interruption and stopping it to recharge will interrupt. Might combine your idea with the one @Uwe suggested, as that would probably be the best long term solution – Baba Mar 10 '17 at 11:04
• @Baba - When you say continuous measurement I rather doubt that is the real case. Much more likely is that you are digitizing the measurements and doing that on a periodic basis. This would allow charging to be a process that is averaged between each reading that you take. Tilt measurement indicates that you are sampling some system that may have a certain amount of mass and inertia. By careful evaluation you can determine just how slow you can perform the sample measurements and still track tilt. This in turn can provide guidance on the timing characteristics of the switched charging. – Michael Karas Mar 10 '17 at 11:13
• You're right by that. But that adds a level of complexity, I felt would be overkill as it is not the main aim of the project. I'm tracking what should be a 3 second oscillation in tilt but because I'm running from a Raspberry Pi and using an ADC some noise gets added. So I need to sample at faster rate and average out the noise, currently it samples at ~16Hz. The issue will be getting the Raspberry Pi to be able to communicate with the new black box without slowing down the effective "sampling rate" – Baba Mar 10 '17 at 11:26
• @Baba - As the rule always states - "Nothing comes for free". Added complexity and / or cost is the norm of increased performance. The real engineering job is learning how to balance off the alternatives and finding the solution that is adequate without going too much one way or the other. – Michael Karas Mar 10 '17 at 11:31
• Yeah thanks! Think I'll go with using rechargeable size D batteries and recharge them after 30 days rather than in between reading. Seems like the easiest compromise for my use case. Thanks! – Baba Mar 10 '17 at 11:35