Back Up Battery failure inspires new test tools.

RE: Back Up Battery failure inspires new test tools.

mcb1 — Mon, 03 Dec 2018 02:56:21 GMT

[mention:f80b53cee57c44bc9d7c577d07d7c791:e9ed411860ed4f2ba0265705b8793d05]

I'm curious to know the float voltage you had for the battery.

When we were maintaining remote phone systems way back when, they used a normal car battery and float charger.

The initial setting from the electrical people meant they always needed water, and finally we settled on 13.2v float to keep them charged but not requiring constant watering.

As [mention:ef601b47037147bfb9b0b4a3ab63026a:e9ed411860ed4f2ba0265705b8793d05] as said, I suspect your battery is now scrap, and given that you want it to work when required, not something you need to keep.

Why does it fluctuate.

Most of these small UPS are designed to shut down the computer, either by a connection to it, or that beeping noise.

I suspect that 75w is at the lower end of it's design range and the unit is struggling to regulate because the load is too small.

You also have a battery that has a high internal impedance, and this may be impacting on the inverter as well.

But essentially these things are simply and inverter and a AC generator and not a precison voltage source.

For the 650w models we used, they were a SPS and the incoming mains was fed via an auto transformer and then provided 'inverter' mains when the voltage was outside the limits.

Mark

RE: Back Up Battery failure inspires new test tools.

jw0752 — Sat, 01 Dec 2018 06:01:52 GMT

I should probably start a new blog for this but you are all probably getting sick of my Process Duration Timer so I will add it to this stream. After I finished with the interface I saw another simple improvement that I could make to the PDT unit. Up until now the PDT only worked on situations where the voltage from the Device Under Test (DUT) was falling towards the Target Voltage. By adding a switch that the Arduino could read as High or Low I was able to add another Mode whereby the PDT can also process DUT voltages that are rising towards a Target Voltage. The software of the Arduino also had to have a simple upgrade to accommodate the reversed process. Here are some pictures of the upgrade:

[View:/resized-image/__size/620x465/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-d761e501-95d6-483b-a9c2-9b54dc0abdc4/4745.contentimage_5F00_189217.jpg:620:465]

A little hard to see but the blue SPST switch in the center of the picture will pull the 5 volts supplied by a 10K resistor on Pin 11 of the Arduino down to 0 volts when it is in the closed position. This directs the computer to process the input voltage data to the PDT as Rising or as Falling.

[View:/resized-image/__size/620x465/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-d761e501-95d6-483b-a9c2-9b54dc0abdc4/3240.contentimage_5F00_189218.jpg:620:465]

A view of the end control panel of the PDT

After I finished with the upgrade I performed a test of the unit. I set up an experiment with a 10,000 uF capacitor which the PDT monitored. I set the bench power supply to 10 volts and the Target voltage on the PDT to 6.32 volts which is one Time Constant. I selected 2 random 10K resistors from the bin. I then charged the capacitor through each resistor while the PDT monitored and timed the process as it went to the Target Voltage. Resistor "A" took 142 seconds and Resistor "B" took 145 seconds. The Ratio of Resistor "A" to Resistor "B" was 142 / 145. I next took an ohmmeter reading of Resistor "A" which turned out to be 9,869 Ohms. Using the ratio predicted by the PDT I calculated the value of Resistor "B" to be 10,077 Ohms. Finally I measured Resistor "B" and found that it was actually 10,038 Ohms. The reading of the PDT was therefore off by 0.4% which is acceptable for any experiment that I would likely do.

I have begun to gather the materials and my thoughts as I hope to build another PDT and put it into one of the more professional looking cases. This is the usual way that I develop an idea into a tool. If you remember, my first iteration of the PDT was a device with a electro-mechanical clock and a comparator circuit that had to have an external target voltage supplied to it from a bench supply. Little by little as the short comings were recognized and the ideas came improvements were made. In case you want to look at the first PDT here is the blog from April 2015.

https://www.element14.com/community/people/jw0752/blog/2015/08/16/battle-of-the-batteries--e-vs-d-vs-bargain

The current model is still a hodgepodge of commercial clock, power supply, and microprocessor modules interfaced together to do a job but it is much more stable and accurate than the original build.

Eventually I will try to work towards the following goals:

A timer that will cover more than one 24 hour period before cycling to 00:00:00

Accuracy to 0.1 or even 0.01 second.

Auto start that better coordinates initiation of the process with zero seconds. Currently there is a slight delay in the start sequence (about a second) as the processor resets the clock.

Slightly higher input voltage.

It is all a process and it takes time. There was about a year between the latest upgrades to the unit and the previous one in which I installed a switch, a relay, and a software upgrade where I could choose to have the unit disconnect the external load or not when the Target voltage was achieved. If I finally get my mind in the right place to proceed with the building of a second unit in a more commercial case this might provide the opportunity to make some other as yet un-thought-of upgrades.

John

RE: Back Up Battery failure inspires new test tools.

three-phase — Wed, 28 Nov 2018 20:58:51 GMT

Nice project, I do like the process duration timer build, I shall go and find those blogs in due course and have a read through.

Loss of water in a flooded cells causes sulfation of the plates, seen as a crystalline deposit across them. This reduces the active area of the plate and hence the capacity of the cell. If sulfation is not present, then you can have measured success with replacement of the acid and equalising charging, but it can be very difficult and time consuming and is often easier and more cost effective to get a new battery.

The electrolyte also allows for even heat dissipation throughout the cell, without this the plates can overheat and deform / buckle altering the distance between adjacent plates and again affecting the performance of the cell. Once plates are deformed, then there is no chance of recovery.

Sealed lead acid, is obviously 'maintenance free', however sealed cells are much more vulnerable to heat than flooded cells. So the 8 year nominal life specified for the battery will be for a certain temperature, usually 20 Deg C. For every 5 Deg C increase in temperature, the battery life will halve. Looking at the setup you have for the UPS, it looks quite enclosed, so you may want to verify the ambient temperature conditions for the battery.

Impedance testing is a well practiced on sealed cells, the nominal impedance is given by the manufacturer and a doubling of that impedance is an indicator for a discharge test to verify the capacity.

You can pretty much run whatever you like of a UPS, you just need to size it for the capacity you require, we run whole control rooms with all their associated computers and controlled systems from either centralised UPS or multiple UPS units feeding individual aspects - as usual, there is a solution if your budget allows. UPS autonomy can be increased by using parallel battery banks, up to what the charger can handle, remembering that it has to charge the battery and supply the load under mains operation.

Kind regards.

RE: Back Up Battery failure inspires new test tools.

Jan Cumps — Wed, 28 Nov 2018 19:16:18 GMT

John, a little recycle tip: once your computer isn't able to run at a decent pace and you buy a new one, you can give it a second life (edit: in your case third life) by installing a linux OS on it.

It's more fun than dumping it in a recycle container.

The requirements for running linux as a desktop system are a little lower than windows - don't expect the computer to fly though [View:/resized-image/__size/16x16/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-d761e501-95d6-483b-a9c2-9b54dc0abdc4/contentimage_5F00_1.png:16:16].

It's a chance to learn a new OS and try to make it work together with your new purchase in a home network.

RE: Back Up Battery failure inspires new test tools.

DAB — Tue, 27 Nov 2018 21:41:21 GMT

Good post John.

DAB

RE: Back Up Battery failure inspires new test tools.

Gough Lui — Tue, 27 Nov 2018 05:46:07 GMT

Flooded lead acid cells are fairly bulletproof, but indeed, every three or so months, topping up is highly recommended. Unfortunately, once the plates are exposed to air, I believe they oxidise and the capacity can't reliably be recovered in any way. Or if you do manage to get the layer off, it settles in the bottom where it eventually shorts out the cell, manifesting itself as high self discharge. Generally speaking, deep-cycle batteries are best suited, with most automotive "starting lighting and ignition" batteries not suitable for cycle use as their plates disintegrate rapidly from just a few deep cycles.

AGM cells or even gel cells can have a fairly decent float life but do watch the voltage. If the float voltage gets too high, the recombination catalysts in the cap may not handle the load and the cell may eventually dry up. Charge rates are important for gel as well, as sometimes I've heard of the batteries swelling due to trapped gas in the gel itself. But this is what most telecommunication facilities use nowadays, and it's actually much safer for use "in the home" as all that evaporated water would have meant hydrogen evolution into your workspace. Also of note is that the vented nature of flooded cells may also lead to corrosive vapours which can attack metal nearby - terminals, etc.

Ironically, where I live, most of the time the consumer grade UPS fails before the power does - in a very similar way as you mentioned. During a self test, the battery would not be able to handle the high rate load, the voltage drops, the UPS complains and attempts a switch back to mains ... but not quickly enough (or a relay fails in an inopportune time, welds contacts, etc). Once that happens, I have to disconnect the load and replace or service the UPS. I got sick of this as I did run a number of experiments and servers where it would be nice to have no downtime operation, and without a bypass transfer switch to cut the UPS out of the loop for servicing, I basically had to shut everything down. Aside from that, I find that over longer periods, many of them also have a habit of overheating, many produce "modified" sine wave (or square wave with "dead" periods) which results in noisy power and those with AVR (automatic voltage regulation) basically cut in boost/buck coils at their whim when the mains drifts about 10% and these changes cause power line noise during the transfer time (i.e. 4-8ms with nothing on the line) which I find more stressful as some precision devices will not ride these out without consequences (e.g. multimeters which are "mains line synchronized" to reduce the influence of mains hum on measurements).

And that's the story of how I learned to stop worrying about power outages, live with a simple but effective MOV based surge suppressor, and saved myself the hassle of maintaining a bank of batteries that would save my bacon less often than I would have to test/service it!

- Gough