The Zinc-air O2 sensor debate

This was a real firestarter. Rip Quinby swears by zinc-air hearing aid batteries as O2 sensors, and Will Smithers took him to task. For brevity and clarity, I have edited the messages for length. But it's still a long thread. If you want the final analysis, skip to the chase

Date: Wed, 28 Jan 1998 00:05:53 -0500
From: Ripley Quinby III <rquinby@snet.net>
Subject: Re: o2 sensors

I have successfully used zinc-air (read zinc-oxygen) hearing aid batteries as O2 sensors. They work very well for me and of course are rugged, reliable, cheap, and readily available.

At this point, everyones ears perked up. "Hearing aid batteries???" several queries followed the following pattern:

Date: Wed, 28 Jan 1998 00:12:16 -0500
From: "Jay K. Jeffries" <coolwater.design@mci2000.com>
Subject: RE: o2 sensors

Ripley,
It would be interesting to learn more about your work with the zinc-air batteries and the supporting circuitry

Date: Tue, 27 Jan 1998 23:47:58 -0600
From: David Drake <diving-dd@mindspring.com>
Subject: Re: o2 sensors

Do explain how you've managed to apparently made a better mouse trap than the commerical guys.

Do you have details on the batteries and circuit. How expensive was it to make your circuits? For the rest of us?--DD

Date: Wed, 28 Jan 1998 01:24:20 -0500
From: Ripley Quinby III <rquinby@snet.net>
Subject: Re: o2 sensors

I'm game as always. ACAD 12 and Word 7.0 files for my design are available. Would you like them; can you view them? 18 files total, many megabytes!

Date: Wed, 28 Jan 1998 01:32:20 -0500
From: Ripley Quinby III <rquinby@snet.net>
Subject: Re: o2 sensors

Hi Dave;
I guess I just keep my eyes open for the unusual. Yes, I have details. Please, please, somebody post my files for ALL to view! Dave, if you can view Autocad 12 and Word 7.0 files, I can send you copies of the data for my unit.

Date: Wed, 28 Jan 1998 01:40:03 -0500
From: Ripley Quinby III <rquinby@snet.net>
Subject: Re: o2 sensors and electronics

Actually they do "put out the amps". Frankly, gentlemen , I'm getting a tad fatigued reading no-sayers who aren't willing to shell out a couple of bucks at Radio Shack to CHECK OUT THE FACTS! Buy a pack of zinc-air cells and try 'em! They are galvanic sensors just like the ones everyone else uses, but the manufacturers haven't cottoned to the fact that they serve another function other than powering hearing aids! Me, I've spent a few hours UW with these cells, so go your own way.
Regards,
Rip

David Drake wrote:

> Uh, yea, okay here's the crap load...
> Sensors have VERY little output which requires some consideration to avoid
. . .
> underwater/pressure situation. What have other home builts come up with?
> Remember the water problem. Do those zinc batteries deal well with 100%
> humidity and condensation?
>
> Second, water proofing the electronics and the the pressure problem when
> you seal them is not easy. A couple of the early homebuilt guys in Canada
> were working on it. Everything I've heard of envolves putting the meter in
> an computer housing that is sealed against water and withstands pressure.
> Adjustment and calibration must be a brease all sealed up in the housing.
>
> So how bout them zinc boys? They would put out the amps, wouldn't they?
> --DD "Tennessee"

At this point Rip offered the following:

Date: Wed, 28 Jan 1998 02:32:22 -0500
From: Ripley Quinby III <rquinby@snet.net>
Subject: Re: o2 sensors

Bill, talk to me if you can do this. Let's diddle with these files (if they are of general interest) until those equipped to view them can contact me with specific questions, comments, criticism.

I sent off e-mail to Rip and put a directory up on the rebreather users site for him. It's still empty.
At this point, Will's interest was piqued:

Date: Wed, 28 Jan 1998 09:05:49 -0500 (EST)
From: "William M. Smithers" <will@tradeware.com>
Subject: Re: o2 sensors and electronics

This is interesting - I don't need to see huge files - how about a simple explanation of the behavior of the zinc-air cells under increased PO2 - response times, etc And what about the humidity issue?

-Will

There followed some support for Rip:

From: rrl@kmd.dk
Subject: RE: o2 sensors and electronics
Date: Wed, 28 Jan 1998 15:27:23 +0100

Take a look at Duracell's web site:

http://www.duracell.com *link broken*

There are some very detailed technical documents (.pdf) available for download. The chemistry of the cells are explained => how many "free" electrons (read: current) you get, each time a O2 molecule hits the interface.

About pO2 response - it's basic statical mechanics - the number of O2 molecule hist per area (the interface of the ZinkAir cell) is proportional to the pressure.

René

Date: Wed, 28 Jan 1998 20:35:52 +0100
From: "Atle Aamodt" <aaamodt@online.no>
Subject: Zinc Air battery

Here's some links to more info.

The Duracell site is very good.

http://www.duracellnpt.com/primary.d/zinc.d/znkind.html *link broken*

http://www.advancedhearing.com/battery.htm

http://www.aclhearingcenters.com/aclwww/batteries.html

http://www.aern.com/primer12.htm

http://www.siemens-hearing.com/pprods/batteries.html *link broken*

http://www.mse.berkeley.edu/faculty/Evans/Projects/zincair.html

http://www.lbl.gov/Tech-Transfer/techs/lbnl768.html

Date: Wed, 28 Jan 1998 20:43:44 -0600
From: David Drake <diving-dd@mindspring.com>
Subject: Re: o2 sensors and electronics

Slow down Ripper! When I first read your post, I though someone else was bothering you. Then I realized you were squawking at me. ALL of the cells I've delt with when I wrote the post below are the tiny amperage Pequod/Miniox cells that are touchy and DO require calibration to check for drift and DO change over their useful life. ANY battery/galvanic cell will. I doubt you'll need to adjust the meter often, but if it's in a case, that has to be opened and done, that is a design consideration our new fellow should be aware of, he most likely not being a whiz bang inventor/engineer.

The major advantage I see for your batteries is they are designed to "put out the amps"! Therefore my post to you on the better mouse trap. If you made them work, I want to use 'em. No sarcasm. No bull. I'm on your side. I think its hilarious you've found a solution for you that has escaped all the commercial O2 meter makers out there. This is a major issue for accurate rebreather diving. I won't jump in the water based on your meter until I see the results and evaluate the design for myself. Sort of a healthy respect for O2 toxing. How do the batteries handle condensation?

We should all be friends here right?--Tennessee

P.S. I'm an engineer, you're preaching to the choir on the design steps. Have a beer an enjoy life instead of being fatigued.

Rip starts to back up his claims:

Date: Wed, 28 Jan 1998 22:38:31 -0500
From: Ripley Quinby III <rquinby@snet.net>
Subject: Re: o2 sensors and electronics

Hi Will!
Buy 'em and try 'em! Response time is load dependant. The more current they supply, the faster they respond obviously. I lod my size 13 cells with 560 ohms, giving about a 2.4 mA drain. Response time is about 2 seconds with my circuit, and lifetime is about 110 hours.
These cells contain a Teflon gas exchange amembrane to keep the internal electrolyte water level relatively constant. I have completely flooded my unit, includinf submersion of the sensor cells, and recovered within 30 seconds. Since they are at ambient pressure, they don't seem to mind a bit of extra water. Open circuit, these cells deliver about 1.4 volts. Under a 560 ohm load, this drops to about 1.2 volts at atmospheric conditions.The voltage ranges from 0.8 volts with 0% oxygen (a destructive condition), to 1.65 volts at 100% oxygen.
These are Nernst reactions, and are totally partial pressure dependant.

Date: Wed, 28 Jan 1998 23:21:06 -0500
From: Ripley Quinby III <rquinby@snet.net>
Subject: Re: o2 sensors and electronics

Greetings David;
No response from me will ever reach this site unless I have personally verified it. And that applies to recycled CO2 absorbents, as well as a few other select topics.

Date: Wed, 28 Jan 1998 23:27:12 -0500
From: Ripley Quinby III <rquinby@snet.net>
Subject: Re: o2 sensors and electronics

Gotcha Dave. Thanks!
Yeah, I get a bit stuffy at times. But its so easy and cheap to verify that zinc-air is as good (or better) than other available sensor technologies. Seems that folks want all the results with none of the effort, know what I mean? A bit of creativity and about $100 will allow anyone to verify my results. I can only report on my own experiences and efforts.

OK, so now people need to independently verify the results. After the rebreather party, some people tried

From: "William Sewell" <customworks@hotmail.com>
Subject: zinc air sensor test
Date: Sat, 31 Jan 1998 01:07:33 PST

Hi Ripley,
OK so I shelled out five bucks for four DA13 duracell zinc air batteries and under the cloak of Alaska darkness me and my henchman Daniel do some tests in his secret lab. Hooked up the voltmeter and looked at 1.20 volts..then we put it in a stream of pure 02 and watched..seamed like forever but was more like twenty seconds then the volts climbed very slowly to a peak of 1.41 volts..cut off the 02 to see the volts fall and this is where things got stuck. Over a minute later it still showed 1.40 volts so we put a telledyne 17 med on the meter and it showed 9mv in air and climbed to 22mv in the 02 stream in a few seconds and dropped back down to 9mv when we shut off the gas. After that we put the zinc air back on and it still showd 1.40 volts ( we blew air on it as well) So i'm a little nervous but not giving up.
Bill

Date: Sat, 31 Jan 1998 15:45:33 -0500
From: Ripley Quinby III <rquinby@snet.net>
Subject: Re: zinc air sensor test

Hi Bill;
The cell has to be under load! It will consume no oxygen when in an open circuit condition, and the response time will therefor be effectively infinitely long. I load my cells with a 560 ohm resistor. This draws about 2.4 mA from the cell. Lifetime of the cell under these conditions is about 100 hours. My circuit presents a voltage comparator with a very low frequency low pass response to the cell. With this input and the 560 ohm load on a size 13 zinc-air, response time is about 2 seconds breathing on the cell.

Once again, numerous people asked for schematics. I don't personally know of anybody who received these, but I imagine somebody did.

From: "William Sewell" <customworks@hotmail.com>
Subject: part of a schematic
Date: Tue, 03 Feb 1998 18:11:18 PST

so this is an attempt at sending a schematic along with a question. The question being; am I on the right track. This is a start for a summing amplifier for an air cell based system..

/neg__(cell)______+_____/1 meg 0hm resistor/______________/
/                                              /     /      /
\__neg__(cell)______+_____/1 meg ohm resistor/____/      /
/                                             /    / "741" op-amp
\__neg__(cell)______+_____/1 meg ohm resistor/__/
\
\common negative

the cells would have a 560 ohm resistor to pre-load and the 1 meg ohm
jobs blind each cell to the other allowing the op amp to do its summing
then off to the display. Any comments?

Bill

At this point, we start getting some negative feedback

Date: Wed, 04 Feb 1998 19:04:52 +0100
From: calypso <calypso@grn.es>
Subject: Zinc-air batteries and He

Has anyone checked what happens when you calibrate a zinc-air battery on nitrox and then change to heliox?

The reason i ask is 'cause i had some o2 sensors once that had a problem with this, due i think to what was called the Gardner effect- the reading depended on the square root of the diluent/nitrogen molecular weights, i.e. the diffusion rate of the gas. It was why i abandoned these freebie cells- no gas switch without changing re-calibration. I am not sure why these are different, because they are the same type of cell i think, ie galvanic, but i seem to remember that the gas inlet looked to be little more than a pinhole. They looked very much like 1/2 AA drycells, if you know what i mean. Made in the UK i think. I will try to trawl through the tekdiver list when i get time, as i posted it to there some 2-3 years ago.

Julian

Date: Wed, 04 Feb 1998 22:10:37 +0100
From: calypso <calypso@grn.es>
Subject: Re: Zinc-air battery linearity

Just occurred to me:

are they temperature compensated, or do they need to be?????????????

just a BIT important if they are like normal cells!

Julian

Date: Wed, 04 Feb 1998 16:47:42 -0500
From: Bill Elliott <bill@nwdesigns.com>
Subject: Zinc-air battery linearity

Hey you guys messin' with the zinc-air...

Has anyone checked the linearity of the response as p02's and pressures and temperatures change on these batteries? It's one thing to see the voltage increase, but I am just wondering if it does so in a completely linear way. I guess what I am saying is has anyone checked the response under various conditions and compared the response to a known linearly reacting sensor?

I missed most of the early posts on this subject, so if this has already been gone over, ignore me.

Regards,

Bill

Date: Wed, 4 Feb 1998 18:41:46 -0500 (EST)
From: "William M. Smithers" <will@tradeware.com>
Subject: Re: Zinc-air battery linearity

I don't have my pressure pot yet, so I can't test linearity, but I was down at BMI today talking with Dick King and mentioned the air cells. FWIW, he says he'd heard that air cells are non-linear, and (more interestingly), somebody has a patent on their use in rebreathers, not that would matter to the homebuilders.

As soon as I get my pot working, I'll plot a graph.

-Will

Date: Wed, 04 Feb 1998 21:31:42 -0500
From: Bill Elliott <bill@nwdesigns.com>
Subject: Re: Zinc-air battery linearity

Will,

I don't know about a patent on their use, but P.Readey also says they are not linear, and they do need temperature compensation, this is what made me ask about the linearity in the first place. I'm looking forward to your tests and graph, should be interesting.

Bill

And here comes an actual experiment

Date: Wed, 04 Feb 1998 22:53:03 -0500
From: Tim Taylor <ttaylor@mich.com>
Subject: Re: Zinc-air battery linearity

Bill Elliott wrote:
> Hey you guys messin' with the zinc-air...
>
> Has anyone checked the linearity of the response as p02's and pressures and temperatures change on these batteries?

I tried a few points and basically decided that the battery would probably work if you were able to calibrate it at your preferred diving PO2 setpoint. The cells appear to me to be too nonlinear to extrapolate a 1ATA calibration beyond 1ATA. Feel free to disagree, but I'm not comfortable with the results I saw.

This is far from a rigorous experiment, but.....
Saturday- Hooked a Radio Shack 675 cell up in parallel with a pot, dialed the pot to give a 1.100v output in air on an old uncalibrated RS multimeter. Put the sensor in an enclosure full of 40% nitrox. Read 1.141volts. Purged the chamber with O2, read about 1.180v. Ran back and forth through the different gasses, and my O2 point drifted down to 1.176v then 1.164v. This could be the result of aging of a brand new cell, or maybe I was running too close to the capacity of the cell (1.1volts is about 90% of current capacity if you trust the duracell page for radio shak batteries), or maybe the cell was cooling off from the blast of cold gas from my tanks. BTW, Argon pulled the cell down to about .4 volts in the couple of minutes I was willing to wait.

Sunday, dialed the pot to a starting point of 1.200 volts in air. let it sit for 2 hours, read 1.198v, dialed back to 1.200. Put the cell in O2, let it sit for an hour, read 1.228v. put the cell in 40%, let it sit, read 1.214v.

Haven't had a chance to rig up a pressure pot to try 1.2, 1.4, 1.6 ATA,

Interpretation (based on the minimal observations above and reading the duracell web page):
1. They are not very linear with O2 concentration (slope from 0.21 ATA to 0.40 ATA is roughly 3x the slope from 0.40-1.00) - So, are you going to calibrate at 1ATA and extrapolate the curve higher during the dive? Are you going to dive 1ATA or less in the range you can easily measure? Are you going to bring along a pressure chamber to recalibrate on the dive site?
2. They have a pretty strong temperature dependency- I let the cells sit the second day to let them warm back up to room temp after being hit with a flow of cold air from my inflator hose and got better repeatability (or maybe there is a time dependency on the battery output as the cell reaches equilibrium, or maybe the issue was too much current draw the first day, or maybe......) The duracell web page has a graph that shows a strong dependency on temp below room temperature. Probably not an issue in your canister during the dive, but will the canister be up to temp when you calibrate?

So, can anyone suggest a good brand & model of oxygen sensor for use in a rebreather? How about solenoids and deep submersible electrical connectors and switches?

--
Tim Taylor, ttaylor@mich.com
Great Lakes Shipwreck Festival, Ford Seahorses Underwater Photo Contest
Michigan Underwater Preserves Council Web Pages
http://www.mich.com/~ttaylor

Date: Wed, 04 Feb 1998 23:02:45 -0500
From: Bill Elliott <bill@nwdesigns.com>
Subject: Re: Zinc-air battery linearity

Tim,
Thanks for the notes on your tests... I wouldn't dive em! As for other sensors, if you have "acceptable" training, Steam Machines has some VERY nice sensors which react fast and ARE linear and were designed around the operating conditions of rebreathers (high pressures, and p02's). They cost about $110.00 each and so far seem to last something over a year with no special care.

Many of the list members have used Teledynes, but I am not sure which ones. BMI sells sensors as well.
Bill

Date: Wed, 4 Feb 1998 23:07:48 -0500 (EST)
From: "William M. Smithers" <will@tradeware.com>
Subject: Re: Zinc-air battery linearity

Tim,

Interesting to hear of your tests - from your post, it wasn't clear what the load you were applying was - did you use something like the 560 Ohm load Ripley was mentioning?

-Will

Date: Thu, 05 Feb 1998 22:06:19 -0500
From: Tim Taylor <ttaylor@mich.com>
Subject: Re: Zinc-air battery linearity

Yeah, sure, give or take a lot. I used a different size battery (same voltage, different limiting current) These cells aren't ideal batteries with a constant voltage regardless of current draw. The voltage rolls off as current draw increases. I believe it relates to oxygen starvation. Big cells can put out a lot more current before they roll off. Don't ask me for a clear explanation. My field is engines, not electrons.

Ripley's note said (#13 cells, 560 ohm, 1.2 volt, 2.4mA drain) I used a #675 cell, 152 ohm, 1.2volts. that works out to 7.9mA. Sunday I just dialed the pot to get 1.200 volts. I went back to the basement tonight to see where my resistance ended up.

If you can apply duracell specs to radio shack batteries, (see http://www.duracellnpt.com/explor.d/zsumsh.html *link broken*) then my test was running at 31% of my cells limiting current and Ripley was running at 20-24% of his cells limit.

When I reread Ripley's note to get his test conditions I noticed that his results were very different from mine. Maybe the different cells behave differently. Maybe my technique left something to be desired.

Date: Fri, 06 Feb 1998 21:12:58 -0500
From: Ripley Quinby III <rquinby@snet.net>
To: Bill Elliott <bill@nwdesigns.com>
Subject: Re: Zinc-air battery linearity

I checked 'em for linearity over pressure, temperature, diluent gas pressure, moisture level, and contaminant gas level. It's my life after all!

Date: Fri, 06 Feb 1998 21:23:55 -0500
From: Ripley Quinby III <rquinby@snet.net>
Subject: Re: Zinc-air battery linearity

Yeah Guys:
The cells are very different. Different sizes have different gas diffusion rates, different internal impedance, etc., etc. etc...
If you're sucking almost 8 mA from one of these things, it is current limited AND oxygen starved. Read the specs first or you'll get nowhere in a big hurry! Duracell has the best technical info section.
Regards,
Rip

Now, Will jumps on the bandwagon, and does some testing

Date: Sun, 15 Feb 1998 23:50:32 -0500 (EST)
From: "William M. Smithers" <will@tradeware.com>
Subject: Zinc Air Cell Tests.

I ran some preliminary tests with my new homebrew pressure pot, and while detailed testing will have to wait until I can find a 0-300 PSI guage, it does look like the cells perform OK, not great.

Response time is several seconds with a 650 ohm load, linearity is OK, not great, and they appear at first glance to be subject to a degree of thermal drift, which may or may not be a problem, as these suckers warm up under high PO2's.

The key is that you can't look at voltage response, which is all over the place, you have to look at current response, which reacts much more quickly and shows less drift.

From my quick two hours of experimentation, I'm leaving my mind open until I have the time (and LP guage) to do proper graphing of results vs. other sensors.

One thing I *did* note is that the cells I used, Duracell DA 630, seemed to peak at a current output of about 175mA, which is a) stunning for a button-type cell, and b) limits the measureable PO2 to about 1.7, which isn't a big problem is my book.

This does raise concerns about electrolyte burning, and cell life under duress (higher PO2's).

As you might expect, the reaction time was good, but the *settling* time (time it takes to arrive at +-5% of the actual vs. measured PO2 value), was better at high PO2's than values below about .7

What you typically see is a quick drop to within 10-15% of the actual PO2, then a several-to-thirty second slide to the exact value, with the slide time getting longer as the PO2 is lessened.

-Will

Date: Mon, 16 Feb 1998 14:04:01 -0500
From: Ripley Quinby III <rquinby@snet.net>
Subject: Re: Zinc Air Cell Tests.

Hi Bill;
Keep in mind that the reponse of any cell of this type is hardly linear, it's logarithmic. If you are getting temp drift problems, you may be over loading the cell (too much current draw). Yes, they do have a phenomenal amp-hour rating for their size, but that's because they don't have to carry the oxidizer internally. I believe the type 630 cells you are using have a 650 mA hour spec. I went to size 13 cells on my breather because of the faster reponse time of the smaller cell, but I still get over 100 hours life out of them (into a 560 ohm load).
Regards,
Rip

Will says not to write these off just yet. . .

Date: Tue, 17 Feb 1998 06:25:37 -0500 (EST)
From: "William M. Smithers" <will@tradeware.com>
Subject: Re: Zinc Air Cell Tests.

Rip,

Yes, I thought that most of the issues I noticed would probably be resolved by a much smaller cell - the response time vs. current load of the 630 was indicative of oxygen starvation within the inner workings of the cell, particularly for drops in PO2.

Folks, don't write these cells off just yet, from what Rip has said, everything makes sense for the smaller cells - it's just possible that the $5 O2 sensor is a realistic possibility, although one would have to divide 100 hours of useful life vs. 6-12 months of useful life from a $75 O2 sensor.

The economics obviously don't pan out, and the sensors probably require digital compensation (read software) to be at all worthwhile, but the reason I'm interested in looking further is the incredibly small size, which is very conducive to use in bailout rebreathers...

-Will

Date: Fri, 20 Feb 1998 20:38:59 -0500
From: Ripley Quinby III <rquinby@snet.net>
Subject: Re: Zinc Air Cell Tests.

Uh...
That's 100 hours of load time. They last about three months unsealed, but unused. Better recalculate the economics. You sound as if this is a nice, but farfetched, idea. I repeat, I've built my rebreather based on these cells, and am extremely satisfied with its performance, reliability, servicability, comfort, stability, etc. Is there anything I would change? Of course! But the zinc-air cells are not on that short list.
Regards,
Rip

Date: Fri, 20 Feb 1998 20:41:37 -0500
From: Ripley Quinby III <rquinby@snet.net>
Subject: Re: Zinc Air Cell Tests.

Oh, and PS. In earlier posts I've mentioned that I don't use microprocessors of any kinf. I don't trust code when a life is involved, especially my own! My circuit may be deceptively simple and unorthodox, but it's self calibrating, stable, and SIMPLE, SIMPLE, SIMPLE!
Regards,
Rip

Here is the final straw to the whole theory

Will goes and does it.

Date: Thu, 19 Mar 1998 00:00:47 -0500 (EST)
From: "William M. Smithers" <will@tradeware.com>
Subject: Zinc-Air Test Results.

After the noise on the list about zinc-air cells, and some initial quickie trials I did (which were indecisive, tending toward negative), I decided to give the cells a more thourough evaluation. I did this because if they turned out to work, the small size makes them extremely appealing for use in a bailout rebreather.

For the test, I did a couple of dozen runs in my test pressure pot (totalling over 14 hours of tests, using 5 differnt cells), using compressed air to create O2 partial pressures of between 0 and 2.6. [Will adds later: Actually, air was used for >.21 PO2. I used an Argox bath for 0.0-.21.] I used compressed air because it offers more accuracy due to the wider pressure range, and because it also better simulates actual ambient pressure at depth (for a given PO2). Plus, since many batteries are directly affected by pressure, I figured this was a good thing to do.

The depth sensor I used is calibrated for 0-1500fsw, with an accuracy of +-0.25% of full scale range.

These tests are far from what I'd consider exhaustive, but I think they do offer a true feel for the dynamics of the cells.

For the tests, I used Duracell DA13 cells, loaded with a 1k Ohm resistor, and used about a 100fpm descent rate, and a 60fpm ascent rate. The raw millivolt values you see are relative to a constant voltage offset that I applied (something like 1.1v)to get finer-grained results, and reflect (more-or-less) the baseline voltage of a new cell in air. No amplification of the raw output was performed. So the reading is a "differential" voltage between the baseline and the current sensor reading.

I also did a series of tests using current as the correlative, but found that the results flattened out so much over a PO2 of about 1.0 that it wasn't worth going further with.

I'm not going to list all of the test results here, because I took down the data in a notebook, and I don't feel like transcribing the whole thing. Instead, I've selected representative single test results, which more-or-less represent an average case for the individual test series that was performed.

The first test was a general linearity and repeatability test, which I performed with various PO2's about a dozen times. For this test, the depth was moved as noted above. After each adjustment, the sensor was given about 10 seconds to settle before the sample was taken. Total time at each depth was typically 20-30 seconds, as this is the time it took to write down the data and start adjusting the pressure. I haven't listed the correlation and error rates here, but feel free to calculate them if you're in the mood.

0 fsw, 16.8mV, .21 PO2
66 fsw, 30.4mV, .62 PO2
130, 37.5, 1.03
214, 58.0, 1.56
365, 67.0, 2.53
217, 34.7, 1.59
131, 32.2, 1.04
66, 26.1, .62
0, 12.9, .21

As you can see, things are looking pretty good on the way down, but something happens on the way back up to screw it up. I ran a bunch of identical tests, and was confused by this, until I figured out what was going on.

Notice that when the sensor gets back to a .21 PO2, it is 3.9 mV below the voltage it was at the start of the "dive". If you add that voltage to the ascending results, your linearity is more-or-less restored. Also, the sensor baseline voltage stays at 12.9 and does not bounce back to the original starting voltage, so the next test run starts at 12.9 at 0fsw. At the end of the next run, you've got an even lower voltage.

After a bunch of runs, I suspected that what was happening is that the cells are tremendously stressed by operating under elevated PO2's, and consequently burn off a large amount of their "vital capacity" under these conditions. Burning off vital capacity translates into a lower output voltage. [Will adds later: Permenantly.]

I figured that the best way to counter this would be to raise the load resistance, so as to lighten the peak load under high PO2. I used a 5.5k Ohm value.

I did another series of runs, this time diving to a fixed depth, and remaining at that depth for a typical bounce dive period. The results were pretty much the same, with the voltage quickly dropping when a depth/PO2 of 1.6 was reached, only the drop rate wasn't as fast.

I raised the resistor value to 10k Ohms, and it improved things a bit, but not much. Here's a typical run (the pressure adjustments are because my test post seems to lose about 1fsw every few minutes due to leakage):

0fsw, 3.8mV, 8:41 PM
33, 13.2, 8:42
65, 19.1, 8:42
99, 24.5, 8:43
133, 24.3, 8:44
220, 42.6, 8:45
219, 42.6, 8:47
217, 39.5, 8:51
(restored pressure to 220)
220, 39.2, 8:52
219, 37.4, 8:56
218, 36.1, 9:00
217, 35.0, 9:03
216, 33.5, 9:07
215, 33.1, 9:09
212, 31.2, 9:16
211, 29.4, 9:21
209, 29.1, 9:26
(restored pressure to 220fsw)
219, 29.8, 9:27
219, 30.8, 9:29
218, 28.7,, 9:31
217, 27.8, 9:35
216, 27.1, 9:38
214, 26.5, 9:43
155, 18.2, 9:45
155, 18.0, 9:47
132, 16.1, 9:48
132, 15.9, 9:49
99, 14.3, 9:50
66, 6.7, 9:51
66, -7.5, 9:53
33, -15.0, 9:54
33, -17.1, 9:55
15, -21.6, 9:55
0, -27.3, 9:57
0, -28.0, 9:58

As you can see, more of the same. After an hour at an actual PO2 of 1.6, the sensor is reading the equivalent of a 1.1 PO2. If this was running a setpoint controller, your PO2 would have been elevated to well over 2.0, as a result of the drift.

Finally, I raised the cell load to 15kOhms, which produced downright nutty results, which I won't bother to reproduce here.

Specifically, it appears that the higher the load value, the lower (and less accurate) the differential voltage response, and the slower the response time. However, the cell does not burn down as quickly at higher loads. [Will later adds: That should be "higher resistance", and hence, lower load.]

So, here are my conclusions on small zinc air cells:

[1] They burn off too much of their life under high-oxygen partial pressures to be useful. This burn-rate occurs so swiftly that it creates a high degree of continuously increasing drift. The drift increases so quickly (minutes) that it quickly brings the voltage readings outside of the acceptable error margin.

[2] I suppose it would be possible, with a huge expenditure of effort, to create a drift compensation table, then digitally (or otherwise) apply this table, but I'm not going to bother.

[3] Use of these cells without drift compensation, or for any prolonged period (measured in minutes or hours, depending on PO2 levels), to control or monitor O2 is extremely dangerous, as the cells will begin to show lower-and-lower PO2 readings during the course of a single dive.

The results weren't encouraging enough to continue with tests for temperature and humidity, as I had planned.

My final note is to Ripley: for your own sake, don't take my word for it. If you are really diving these sensors, don't do so again until you've verified these results for yourself, especially with a 1k load, as your life could be measured in minutes if you run a high PO2.

-Will

Date: Fri, 20 Mar 1998 16:39:57 -0500
From: Ripley Quinby III <rquinby@snet.net>
Subject: Re: Zinc-Air Test Results.

Thanks Will;

I ran very similar tests, and more besides. Still diving successfully, but as you mention, there's more to an electronic RB than the sensors! An extreme low pass filter function on my sensor comparators deals with the drift problem. BTW, a great deal of what you are seeing is a physical crushing effect of the pressure on the cell. There is a Teflon membrane just under the cell gas ports which keeps the internal electrolyte (potassium chloride) water level more or less optimised. Pressurizing the cell too fast results in fairly severe changes in the internal structure and impedance. Different manufacturer's cells respond differently in this respect due to differences in port size, membrane thickness, case strength, etc.

As I've mentioned before, I run my cells into a 560 ohm load. A type 13 cell has a life of about 100 hours under these conditions. The lifetime is NOT oxygen concentration dependent, it is load dependent (i.e, it is zinc exhaustion which ultimately depletes the cell). Under open circuit conditions, a zinc-air cell will last just as long stored in pure O2 as it will stored under helium (barring secondary effects like moisture loss, etc).

I'm really glad to see you've made the effort to double check my work. The interesting part is a judicious application of circuit design to the specific requirements of the zinc-air cells (as is also true with all other galvanic sensors)!

Regards,
Rip

Date: Sat, 21 Mar 1998 01:24:07 -0500 (EST)
From: "William M. Smithers" <will@tradeware.com>
Subject: Re: Zinc-Air Test Results.

On Fri, 20 Mar 1998, Ripley Quinby III wrote:
> Thanks Will;
>
> I ran very similar tests, and more besides. Still diving successfully, but as
> you mention, there's more to an electronic RB than the sensors! An extreme low
> pass filter function on my sensor comparators deals with the drift problem. BTW,
> a great deal of what you are seeing is a physical crushing effect of the pressure
> on the cell.

Three of the test runs used a descent rate of 15fpm, because I suspected physical damage might be playing a part. This made no difference in the rate of drift.

Furthermore, an O2 sensor that fries at a descent rate of 60fpm is pretty much useless, as many people typically do 75-100fpm on descent.

> As I've mentioned before, I run my cells into a 560 ohm load. A type 13 cell
> has a life of about 100 hours under these conditions. The lifetime is NOT oxygen
> concentration dependent, it is load dependent (i.e, it is zinc exhaustion which
> ultimately depletes the cell). Under open circuit conditions, a zinc-air cell
> will last just as long stored in pure O2 as it will stored under helium (barring
> secondary effects like moisture loss, etc).

We are not talking about open-circuit conditions. You are supplying a constant resistive load. That does not mean that the current through the resistor is constant - a 560 Ohm load will pass almost as much current as the cell can produce, which is a bad idea, because there is a specific drain-rate that the manufacturer states as peak. In the case of the DA13 cell, Duracell lists 0.85mA as typical for a 1.5k Ohm load. Exceeding the peak current drain will result in cell damage, as was illustrated in the tests, since I measured peak drain rates in excess of 60mA - about *80 times* the suggested load.

The DA13 is rated for 260mA hours, or about 4.2 hours under that type of load, which correlates to a loss of vital capacity of about 25% per hour, which is roughly consistent with the measured drift results. This doesn't factor in the cell stress encountered by exceeding the load limits. It's quite amazing, in fact, that such a tiny cell managed to crank that kind of current for a prolonged period.

That PO2 affects cell "burn rate", and that it wasn't just physical cell damage, is supported by the fact that lowering the load resulted in improved drift characteristics, while the descent rate was kept constant.

Clearly, while the Zinc is getting consumed, the amount of available O2 determines how *fast* the zinc gets consumed, otherwise the cell wouldn't be much use as a PO2 sensor, would it? The amount of O2 determines the energy produced. The amount of total energy produced is limited by the milliamp hour rating of the cell. The more energy produced, the more raw materials get consumed. The more the raw materials get consumed, the less the output. The less the output, the bigger the "drift" as measured against the original cell energy output. Physics 101.

>
> I'm really glad to see you've made the effort to double check my work. The
> interesting part is a judicious application of circuit design to the specific
> requirements of the zinc-air cells (as is also true with all other galvanic
> sensors)!
>

Ripley, I am doing my best not to be snide here, but your enthusiasm and emphaticism has wasted over 20 hours of my time at this point, and I'm not wasting much more on this topic, which has now been exhaustively covered, as far as I'm concerned.

But, because I am willing to grant you some small margin of doubt, I will waste up to three more hours on this subject. Please fax me your drift-compensating circuit, and I will run a series of chamber tests using your own design.

I will then publish the raw results to this list.

My fax # is 212-809-7049

The casual reader will notice that while Ripley originally crowed about the fantastic "linearity and repeatability" of these cells across temperature, humidity, and O2 variances, it suddenly appears that these "linear" cells have a "drift problem", that requires "judicious" compensatory circuitry to correct.

All I have to say is that it must be judicious circuitry indeed, since the degree of measured drift is directly dependant on PO2 exposure level per second - the higher the PO2, the faster the degradation.

Curiously, the rate of sensor decay is roughly equivalent to the NOAA PO2 exposure tables...a 1.6 fries a cell beyond useful limits in 45 minutes or less.

-Will

And Will has now fairly well refuted the theory and backed up his results. And Rip replies:

Date: Sun, 22 Mar 1998 01:09:59 -0500
From: Ripley Quinby III <rquinby@snet.net>
Subject: Re: Zinc-Air Test Results.

Sincerest Apologies Bill,
I spent over 600 hours of my time developing a reliable RB which works quite well for me. You've got a lot of hours to go. Keep experimenting, you'll get there. Anyone who believes that verification of research is a waste of time is himself a waste of time. Sorry to have hampered your social life.

Regards,
Rip

At this point, Lee sums up Rip with the following:

Date: Mon, 23 Mar 1998 13:13:10 -0500
Subject: Re: Zinc-Air Test Results.
From: nssdiver@juno.com (Lee C Kresge)

Arrogant sucker, isn't he?. . .

Lee

Date: Sun, 5 Apr 1998 12:18:24 -0400
Subject: Who's the Dumb Fuck?
From: vdervooten@juno.com (Christopher Van der Vooten)

Who's the dumb fuck that recommended the zinc air batteries? This is the single stupidest thing I have seen on here, and that is saying something!!!

So, do you still want to use Zinc-air batteries? I've yet to hear anybody but Ripley doing it. From the results Will obtained, I think it's safe to say this is a good thing.