12F675 based voltage monitor?

[badcockr]

Hi chaps,

While I'm waiting for my instrumentation amps for my other project I'm thinking about a simple on-board voltage checker.

I've been thinking that the common volt checkers mean nothing to the average user unless they are familiar with battery discharge curves. So knowing that a battery is at 5.3 Volts implies nothing about safe to fly!

Using the PicKit1 I've been playing with using a simple two LED indicator to show 4 states of charge for common RX packs; i.e. 0-25%, 25-50%, 50-75% and 75-100%.

In order to make the indicator as realistic as possible I've produced graphs of capacity and voltage for a 700 mA 4 cell NiCd under discharge rates of 100, 200, 500 and 750 mA. (see attached gif file)

Therefore, using the 100 mA file as being the most representative of a plane at 'rest', I have calculated the following LED conditions and voltages can be used to indicate state:

Colours (4Cell)
GREEN (flash) 5.614-5.057 (100-75%)
GREEN 5.057-4.947 (75-50%)
RED 4.947-4.806 (50-25%)
RED(flash) 4.806-3.598 (25-0%)

All-in-all this should lead to a much more easy to understand voltage monitor that can be implemented using only two LED's, some very simple code and should offer some weight savings over conventional checkers.

Any thoughts?

[badcockr]

Attached is the proposed schematic - simple indeed!

GP1 is the Vref, generated from a 2.5 V Zener diode
GP2 is the Analog input from the supply line across a potential divider to bring it in range of the reference zener.

The two LED's indicate the charge state.

HTH

Rod

[Mr.RC-CAM]

Looks like fun.

FWIW, the curve and rest voltage is highly dependant on the temperature, load current, battery chemistry (NiCD vs. NiMH), and other little gotchas. So, your accuracy will ultimately vary too much to be reliable for others to use. Your scaling will be a fair match for the battery packs you tested, assuming that the test protocol, temperature, and load currents are controlled.

For reliability and accuracy, a battery fuel gauging method is best. Here is one such design: http://www.digitalproductsco.com/batmon/

Good luck. PIC projects are always a blast to create.

RC-CAM

[badcockr]

RC-CAM:

Thanks for the reply - you open up an interesting discussion! I think that this is a discussion that *should* be brought out in the open for other modellers to realise the implications.

1. All of the battery monitors available in model shops (that I have seen) such as the voltspy and similar rely purely on a bar graph display of voltage.
2. Voltage is not a *true* indication of charge, and it does vary with load, capacity, chemistry and temperature. The attached voltage capacity graphs at different loads prove that.
3. Most modellers who have on-board checkers use a voltage checker without understanding what it means!

The design you have linked to is far superior (although I will need to check the data sheets more closely as some method of identifying individual packs is required).

What I have proposed, I believe, should be more reliable than the common voltcheckers for *NiCd* packs ONLY.

The issue of change in voltage / capacity behaviour at different environmental temperatures is very interesting. Does anyone have links, or actual data, for information on NiCd behaviour with temperature? If this is not apparent then I do have access to an environmental chamber where I can do some controlled discharge studies.

With the temperature issue aside I had envisioned the capacity checker to be of use under close to no-load conditions i.e. RX switched on but plane static on the ground. The discharge curves show that below 1/5 C the critical relationship between voltage and capacity is negligible and therefore the non-linear scale should be more meaningful than a voltage display.

I have, of course, made a number of assumptions which may not be correct:
1. Almost all NiCd packs (different manufacturers) will have similar voltage capacity curves under zero-load conditions.
2. Under no-load conditions the relationship between voltage and capacity is similar for packs of different cell capacity.
3. Most people use NiCd packs for receivers.
4. Most people would check their on-board monitor *before* take-off in close to no-load conditions in order to assess capacity.

It would be nice to expand on this thread as I have witnessed that most modellers use an on-board voltage checker without understanding what it *really* means - this could have consequences for some. I would be very grateful for additional comments here.

Rod

[steve lewin]

I haven't done very much really scientific testing but from what I've seen the no-load voltage of NiCd or NiMH cells is a very poor indicator of charge, particularly when the real use is with relatively high loads. They rebound a lot when the load is removed. Even the the voltage under a fairly realistic static load seems a bit more useful which is why a lot of the retail ESVs impose a simple resistive load, often around 250mA.

But for a device to be left in a model the voltage I'd find most interesting is the lowest voltage reached in use. If you are continually sampling it wouldn't be too hard to freeze the display at the lowest value seen and that would give you some indication of the state of the battery and possibly help to trap things like stalling servos too.

Steve

[DanC]

rather than rely on just voltage of the pack why not measure accross a resistor and get the current as well there for giving the power taken from the pack.

There are specialised current sensing resistors about or if thats not acceptable get a hexsense fet and an opamp arrangement.

But I think are there "gas gauge" chips readily available for measuring capacity left in the battery .

I mean the micro you are using has a couple of A2D'd on it so you can measure current as well

[badcockr]

DanC and Steve,

I can understand the issue with monitoring the lowest voltage and highest current seen in use - but the question is what would you do with the information?

Dan is right - there are many 'gas gauge' ICs by nearly all the manufacturers - RC-CAM gives a link to a fully working circuit further up the thread. I don't suggest that such a simple circuit as this can replicate these IC's, but should offer some advantages over the conventional voltchecker circuits without the additional complexity of the advanced circuits. After all if there is such a large variation why use 10 leds to indicate a linear voltage range when the information from two LED's may be as precise and more usefull?

The discussion so far has got me intrigued - I know of many (IC) aeromodellers who do rely on a voltchecker. But what is the variation in relationship between voltage and charge under no-load conditions for temperatures ranging from 0 - 40 centigrade? If wide then there can be sever implications.

Rod

[Mr.RC-CAM]

There haven't been any successful designs, at least for R/C'ers, that use voltage to determine capacity. It can work with pre-characterized packs under controlled conditions (old military & space apps did this). But that is not possible in our app.

The hobby battery checkers rely on some assumptions. They do not offer capacity (MAH) data, Instead, they only offer a pass/fail sort of result. That is about as bold as they can get.

Basically, they apply a load (most model checkers use a fixed 300mA, but C/5 is industry standard), and if the voltage is over 4.75V then there is *probably* at least 30% remaining capacity (Green). This means that many modelers go home with plenty of reserve capacity (better safe than sorry). Some apply a warning scale (yellow) at the voltage knee, which is just above the fail (Red) condition.

It is all a finger-in-the-wind sort of thing. But it works, especially if the limitations are understood. It can be worthless info in the wrong hands.

Your proposed design (check voltage without load) may work just fine for you. However, it goes against the common methods that the industry uses. Just about every method has been tried, most without any real accuracy (except fuel gauging). I have a feeling yours will work for you, once you characterize if for your packs, but may be less accurate with widescale use.

The battery makers offer generalize data on the temperature issues, so just google around. However, this is just one of the gotcha's that are in the way, and perhaps is not a show stopper on its own.

Just test a lot of packs and compare your color coded meter vs. a cycling system's reported mAH. With enough data you will be able to determine if your method will work for R/C'ers.

[badcockr]

RC-CAM:

I agree!

What I find especially interesting is that there are *so many* true 'gas-gauge' IC's out there that will communicate with a microcontroller (something I had not appreciated until I followed your link). I don't know why there is not a commercial solution based on this technology available to modellers through the model shops in the UK!

What started as a few thoughts about could something simpler than a voltage bar-graph display be used to impart pertinent information has opened up a whole can of worms. I have looked through the manufacturers data sheets (Sanyo, Panasonic and others) and searched on the internet for information about the relationship between voltage, capacity and temperature but find it does not exist.

For me this has created an academic interest: what is the relationship and what are the (quantified) implications for modellers who rely on voltchecking? I do know of many modellers who do this withut understanding the limits of the technique.

As I said if I find the information does not exist I can certainly do some controlled studies and report back here.

Many thanks,

Rod

[badcockr]

ALL:

OK I have found the capacity / temperature / voltage data I was looking for on the general Cadnica engineering specs: http://www.sanyo.com/batteries/pdfs/cadnicaT_E.pdf.

If you look at Figure 3.9 on page 13 the variation in going from -20 to +60 centigrade is as great as going from 0.2C to 8C discharge!

What are the implications: using purely a measurement of voltage to indicate state of charge is only valid *for a known temperature*. Hmmm ponder this one and look at the graph and consider the implications for those modellers who do rely on voltage.

In summary: we should be using a full 'gas-gauge' circuit like the one RC-CAM posted the link to. The use of 'conventional' voltage display is not practical where the temperature of the pack might vary from freezing to warm summer (like in the UK).

This has certainly been an education for me and will mean I look closely at the new 'gas-gauge' offerings.

I hope everyone finds this useful.

Rod

[Ironsides]

Rod:

Good work. Did not know how complex this was - I have flown glow for years using an Expanded Scale Voltmeter (ESV) to check the receiver pack. I have flown at temps below zero Celsius. I guess I got away with it because at low temps one flight is about all the body can stand, let alone the receiver pack ;-)

David