My envision of the the super simple version.

It's not likely that the "broth battery" can supply more than 10 mA, so we can call the device inherently protected. Let's assume a fixed 400 mV.

At the very minimum, you need a high value of resistance from the inputs to ground. This absorbs the input bias current. You might try values between 1 and 10 meg ohms. The one for the unused input could be lower.

Go for a LED rather than a buzzer for proof of concept.

A CR2032 battery and holder should work. That discharge cure is pretty flat. The 10 mA max sink current of the comparitor can drive it directly.

Get a version of the LT6700 that does > than and < than. You'll only use 1 comparitor. Get a few mounted on DIP boars at proto-advantage.

Instead of that resistance to ground, you might want to consider putting an IC socket to play around with and use a "header" so you can try out various input schemes.
One being a simple pull down to ground. e.g. 1 to 10 meg ohms.

For the time being make one main card that uses the < comparitor and one that uses the > comparitor, but the same header to change the input components.

The magic number for a filter is Fc=1/(2*π*R*C). This is the frequency (Fc is the corner frequency) where the output is down about 70%. RC is a time constant and 5*RC means the output is about 99% of the final value.

Find a LED. Color determines the forward voltage of the LED. So, use the discharge curve of the CR2032n to find a minimum voltage (<3V) and find a reasonable max Vf for the LED and calculate R. If you use the wrong input of the right selected comparitor, the light state would be wrong.

Thinking out loud, later you may be able to replace the LED with "something else", say an optocoupler and maybe use another power source for the buzzer, An A23 12 V battery comes to mind. 12 V buzzers are easy to find.

The 3V starting point paves the way of possibly using 5V type parts for a front end.

For now, we'll assume that the voltages you get will be positive. I think we are going with the idea of <400 mV turns on the LED or buzzer, so negative numbers would be interpreted as -0.3 V or so and turn on the buzzer.

What;s used for the output can potentially invert what you really want.

 

Keep the current and power down in the micro region and probe decay is not a problem.

Using low current induces other problems.

Some time back, I had tried such a simple circuit similar to this. But after a while it tends to give false alarms.

From what I have understood (or assumed ) is that the probes tend to act like electrodes and the water (acting as a mild electrolyte, i guess) tends to act more like a battery and gives a negative voltage. Although I have used same material for the probes, I suspected that that length of the probe exposed to water also contributes to buildup the unwanted voltage.

However if the bias current is kept at a sufficient level (got to find for yourself), then the reverse voltages will diminish; so does the probe

 

Although I have used same material for the probes, I suspected that that length of the probe exposed to water also contributes to buildup the unwanted voltage.

What was already determined is that the materials chosen were an aluminum pot and a SS electrode. Sign reversal showed up when using the same materials, I think, and different broths. The probes have to be of different materials. You want to make a battery which requires different electrode materials.

400 mV may or may not be too high of a voltage, but it's an easy one to get. Sensing if he probe is out of the water V> -40 mV and V< 40 mV is harder.

 

I used picture hanging wire

 

@keepit and wayneh: I'm pretty good with soldering, so that shouldn't be an issue for me. I believe I've got both silver and leaded solder sitting around. I've got plenty of experience with that, just not the electronics design part, breadboards, and whatnot.

 

@keepit and wayneh: I'm pretty good with soldering, so that shouldn't be an issue for me. I believe I've got both silver and leaded solder sitting around. I've got plenty of experience with that, just not the electronics design part, breadboards, and whatnot.

I know you have had lots of response to this and not sure where you are now. Seems like lots of differences of opinions. I have worked for years in the conductivity based electronics industry and still think that you could use conductivity with 2 probes. Even in your application the liquid is conductive. High sensitivity circuit will conduct through liquid ice cream etc.

 

My envision of the the super simple version.

It's not likely that the "broth battery" can supply more than 10 mA, so we can call the device inherently protected. Let's assume a fixed 400 mV.

At the very minimum, you need a high value of resistance from the inputs to ground. This absorbs the input bias current. You might try values between 1 and 10 meg ohms. The one for the unused input could be lower.

Go for a LED rather than a buzzer for proof of concept.

A CR2032 battery and holder should work. That discharge cure is pretty flat. The 10 mA max sink current of the comparitor can drive it directly.

Get a version of the LT6700 that does > than and < than. You'll only use 1 comparitor. Get a few mounted on DIP boars at proto-advantage.

Instead of that resistance to ground, you might want to consider putting an IC socket to play around with and use a "header" so you can try out various input schemes.
One being a simple pull down to ground. e.g. 1 to 10 meg ohms.

For the time being make one main card that uses the < comparitor and one that uses the > comparitor, but the same header to change the input components.

The magic number for a filter is Fc=1/(2*π*R*C). This is the frequency (Fc is the corner frequency) where the output is down about 70%. RC is a time constant and 5*RC means the output is about 99% of the final value.

Find a LED. Color determines the forward voltage of the LED. So, use the discharge curve of the CR2032n to find a minimum voltage (<3V) and find a reasonable max Vf for the LED and calculate R. If you use the wrong input of the right selected comparitor, the light state would be wrong.

Thinking out loud, later you may be able to replace the LED with "something else", say an optocoupler and maybe use another power source for the buzzer, An A23 12 V battery comes to mind. 12 V buzzers are easy to find.

The 3V starting point paves the way of possibly using 5V type parts for a front end.

For now, we'll assume that the voltages you get will be positive. I think we are going with the idea of <400 mV turns on the LED or buzzer, so negative numbers would be interpreted as -0.3 V or so and turn on the buzzer.

What;s used for the output can potentially invert what you really want.

@Keepit: Thanks so much for your interest in helping me along with this project. One problem is that you don't know what I know and what I don't, so when you talk about ground requirements, for example, you probably don't realize I'm pretty lost. However, I don't want you to think I'm not technically minded. I was an engineer for a good while and can pick this stuff up if I just get some good basic education. Along those lines is there any place online that teaches the basics of playing around with components and a breadboard like we are discussing? That way I can learn a few things without wasting everybody's time here. Having said that, here is where my head is at now:

I don't think I'll really need to measure negative voltages as I think the polarity will always be the same with an aluminum and stainless cell. Maybe I could just worry about alarming when the voltage drops below 400 mV for now.

Could you step back for a moment and give me the big picture of how a circuit gets designed? What I believe we are doing is starting with a comparator that "trips" at 400 mV. That comparator will take input directly from the leads coming from the pot and the probe. I will need to insert a resistance to ground so that the signal is seen by the comparator rather than just leaking to ground. Yes? Is this ground resistance attached to a ground connection on the comparator itself? Then I'll need a battery (3v suggested to start with) to make the comparator work. I assume this same battery will power the alarm?

When the voltage drops below 400 mV and trips the comparator, I assume it directs current from the battery directly to the alarm, or per your recommendation an LED. I understand the actual buzzer or alarm might need to work with a 12v battery instead depending on which one is chosen. So where does a filter fit in, and exactly what is the filter? I assume it is a physical component that is wired in to the comparator rather than a setting that is programmed in somewhere.

Is there anything else that needs to be figured into this circuit other than the comparator, a resistor to ground (does ground to the breadboard qualify as a ground?), a filter, battery and LED/alarm?

Hopefully that gives an idea of my "cluelessness" to date. Any suggestions, especially about learning more of the basics, are welcome.

 

I know you have had lots of response to this and not sure where you are now. Seems like lots of differences of opinions. I have worked for years in the conductivity based electronics industry and still think that you could use conductivity with 2 probes. Even in your application the liquid is conductive. High sensitivity circuit will conduct through liquid ice cream etc.

Hi Nick. I think I have a really good design for sensing the liquid level. To recap, I am interested in boiling a liquid down from a full pot to any predetermined level, say 3 inches from the bottom. As discussed earlier, I need the sensing mechanism to be insensitive to heat, both from the liquid, the pot itself, and the hot gas coming off the burner. It also can't be fragile and has to be food safe. Moving parts are less desirable because animal fats and moving parts leads to problems eventually. The solution I have now is to make use of the galvanic cell principle. Since the vast majority of stock pots are aluminum, I realized that a single stainless probe would meet all the requirements. I purchased a few thin stainless rods and rigged up a prototype gizmo to hold the rod in the boiling liquid. Using my ammeter and some alligator clips, I found that the signal was a good strong 1.0 to 1.2 V, even as the liquid got low enough to barely touch the rod. It seems the heat really keeps the voltage high. After a minute or two of "sputtering" readings, the voltage went to 0 with only about 1/8" gap between the rod and the boiled down liquid.

I'm pretty happy with that, and now we are on to designing a circuit that can make the alarm work. I don't know enough about circuits to understand whether this signal is lacking in some way, but the guys helping me to this point seem satisfied with using it as an input. What do you think?

 

The magic number for a filter is Fc=1/(2*π*R*C).

On the subject of filters it would be best to avoid electrolytic capacitors, since they won't like high temperatures. That places an upper limit on capacitor value if you want to keep physical size down, so implies a high value resistor. I'm thinking something like 220k and 1uF would do the job.

 

On the subject of filters it would be best to avoid electrolytic capacitors, since they won't like high temperatures. That places an upper limit on capacitor value if you want to keep physical size down, so implies a high value resistor. I'm thinking something like 220k and 1uF would do the job.

It isn't like that. The only thing subjected to heat is the stainless probe that hangs on the side of the pot. The electronics are in a box on the counter. Ever seen an oven meat thermometer? The probe sticks in the meat and the wire comes out of the oven and into a digital thermometer. In this case it will be a level alarm.

 

Could you step back for a moment and give me the big picture of how a circuit gets designed?

Tough one. One spec you don't have, is the amount of current you can get out of your cell. Why? >10mA can destroy the OP amp, with th OP amp powered < battery voltage. (simplified for now).

What I believe we are doing is starting with a comparator that "trips" at 400 mV. That comparator will take input directly from the leads coming from the pot and the probe.

Yes. Directly, maybe not. Example. Say we don;t know if you can generate 10 mA @ 2V, R=200 ohm. If we add a 200 ohm series resistance to your probe, then the OP amp/comparitor is protected. There could be issues with ESD (Electrostatic Discharge). This is a design issue that probably would not be found easily by the casual user.

I will need to insert a resistance to ground so that the signal is seen by the comparator rather than just leaking to ground. Yes?Is this ground resistance attached to a ground connection on the comparator itself?

The OP amp/Comparitor is not ideal. What that means it's input Z (Impedance) isn't infinate and there is an offset between the terminals.
I'm not going to get the numbers form the data sheet, but let;s say it's 10e-9 amps and <1 mV. If you don't have a place for that 1E-9 nA to go, then your in big trouble. It could go to ground, or it could go to +V; it just has to go somewhere. It would also provide the "default" or unconnected value.

Adding a resistor in series with the input of an OP amp/comparator really doesn't affect its function, because the comparator is drawing very little current; V=(1e-9) * 200 ohms. It does have other effects, but we won't worry about them.

So, simplifies, a resistor has to go across your probe. You have been using likely a 10 M input voltmeter, so let;s use at least that. Then the 200 ohm resistor to the input of the comparator.

You MIGHT need a filter on the input, but maybe not. I left that open. It could be because "bubbles" are lowering the potential. You don;t have an oscilloscope to know.

Then I'll need a battery (3v suggested to start with) to make the comparator work. I assume this same battery will power the alarm?

If you use a 3V lithium cell, you almost have a regulated power supply because the discharge curve is relatively flat. The 3V battery surely won't operate a buzzer, but can drive an LED. The comparator can drive about 10 mA, I believe, so it can directly drive an LED.
We're lucky that this comparator can operate > 5V.

Most of the single-supply, low power stuff operates <5V and if you wanted to go <400 mV, the 3V batter might be the way to go.

Here http://www.digikey.com/products/en/...t=0&page=1&quantity=0&ptm=0&fid=0&pageSize=25

I looked for buzzers between 3 and 12 V internally driven.

Questions, then are: Can the battery support the buzzer? Is the buzzer loud enough? What kind of sound do you want? How long will it sound, if continuous?

How long will it last in the environment? Make it plug-in replaceable? I put one in a car for a door open/headlights on alarm and I'd have to replace it periodically, like every 5 years. Moisture might kill it. That's where wireless can come into play.

I like to think ahead and anticipate possible problems. The more I think ahead, the better the design can be. You can always reduce later.
I don't like a lot of "start-over" points.

One reason, I don't like the 9V battery is that it's not sealed as well as the A23 http://data.energizer.com/PDFs/A23Z.pdf or the CR2032.
The 9V battery (non-retail version): http://data.energizer.com/PDFs/EN22.pdf
So, at 20 mA you might get an hour from the A23 and 20 hours from the 9V battery, so the 9V battery may be a better choice.

A word about the comparitors.

The dual comparitors come in different versions (how it responds). The version to light a LED will likely be different than the version that will drive the buzzer. Why? The output is Open Drain, so it can only sink current and we know that it has a 10 mA maximum.

If it works with 3 V and a 400 mV threshold, you can change a few components and make it work with 9. If it doesn;t work with 3 V, 400 mV threshold, you may have to use < 5V parts to lower the threshold.

The output circuit (buzzer drive)
It will have to be a FET. The gate drive will have to come from the positive supply and the comparitor will sink that current away from the gate.
The current is tiny, but the parameter Vgs(ON) has to be low and this may be called a LOGIC FET. So, driving the buzzer may require a different section of the comparitor or a different version of the comparitor because of the inversion. Low voltage logic-fets are a sort-of, new component.

IC's
Every IC requires power supply bypassing. This is typically no more than a capacitor close to the supply pins. It can get more complex than that.
The manufacturer usually has a recommendation. These are typically ceramic or metalized polyester capacitors.

Reverse Polarity Protection (Battery installed backwards)
Yep, you can do that with a 9V battery. Temporarily, anyway. That will have to be provided. There are pitfalls here too.

Low battery test/indicator
It's probably not necessary in your case because they can turn it on before insertion (it should buzz) which might be annoying.

So, let;s say as a value added option, you have a light that is immediate and a buzzer that sounds 30s later? Not saying it's possible or easy to, but it's a conceivable option and so is battery test, so is a switch that cuts out the probe. Just thinking ahead a bit.

All of these things get intertwined with each other. Low power generally means low voltage. Low voltage means low power. Loud audio means high power. Car keyfobs typically use the CR2032 batteries and so does the wireless DXS transmitter thingy. I do have an old technology IR remote that uses a 9V battery.

I was thinking of the "commercial kitchen", not a mom a pop kitchen.

My kitchen timer has a 9V battery and it's pretty loud. I have a timer that I can wear on my neck that I can barely hear. Forget me hearing a watch alarm. I also have tinnitus at approximately 2600 Hz, which is in the range of piezo buzzers. Kitchens can be noisy.

A secondary thing would be to look for an appropriate buzzer and see how well it holds up. Wireless might be the only way to go.

 

The electronics are in a box on the counter. Ever seen an oven meat thermometer? The probe sticks in the meat and the wire comes out of the oven and into a digital thermometer. In this case it will be a level alarm.

I was thinking of a self-contained unit. Long high impedance wires introduces other issues. Twisted-pair shieldled cables or a shielded cable would likely be necessary. BNC cables are not flexible, but RG179 cables with a BNC connector might work. For initial testing, keep the leads short.

With long unshielded leads, picking up 60 Hz from your body is entirely possible. So, now we might have to put a differential amplifier as a front end. Shielding reduces RFI and twisted pair reduces EMI. 60 Hz is more EMI. So, long leads has the potential of introducing other issues.

 

The electronics are in a box on the counter.

Ah. Like KISS, I assumed this was going to be a small self-contained sealed unit, which would simplify hygiene issues.

 

fter a minute or two of "sputtering" readings, the voltage went to 0 with only about 1/8" gap between the rod and the boiled down liquid.

This is where a filter MIGHT help some. It won't respond to some of the bouncing liquid. Another way might be to responds only ii the pprobe has been out of the water a certain amount of time or both. e.g. 30 sec to 1 minute Maybe have the lamp light nearly instantaneously and the buzzer only if the lamp has been on for a certain amount of time?

 

The LT6703-x is a single comparitor version of the LT6700.

I had a hard time figuring out what part you would need, but page 14, here http://cds.linear.com/docs/en/datasheet/670323fd.pdf#page=14 helped the most.

The pic (LT6703-2) with the LED, essentially says, turn on the LED when the input is > some value. You want the opposite, so you need a LT6700-3. This part can sink 40 mA, so substituting a buzzer should be easy.

The full part number has to include the type of package. The 5 lead TSOT-23 is going to be the easiest to work with.

The last page of the datasheet says to use a 47 ohm resistor in the power supply lead for "soft supples" like batteries and a 0.1 uf capacitor for decoupling.

For input voltages <0.3V, you have to limit the current to 10 mA.

Digikey doesn't have the LT6703, but they do have the LT6700 http://www.digikey.com/products/en?keywords=lt6700 ,so somewhat back to the drawing board. Note the +- signs on the input. You want the one in the package (whatever suffix you chose) that has a comparator +IN as an input. The unused comparator should probably have a 1-10 M resistor to ground on the input.

Note, that digikey has dig-reals, cut-tape and tape and reals. A real is the manufacturer's bulk quantity, a digi-real is a smaller quantity for pick and place machines with a fee to spool it. Cut tape is exactly that. The tape that holed the packages is cut to the desired # of packages.

 

So, the LT6700-1 has a +IN and -IN, the LT6700-2 has two -IN's and the LT6700-3 has two +IN's

http://cds.linear.com/docs/en/datasheet/6700123fh.pdf#page=10

So, for breadboarding, the -1 makes sense. If you needed a low battery detect as well and also needed the same or different comparitor sense, then you could use another package.

When driving the LED or a buzzer, you have to subtract 250 mV from the buzzer or LED supply.
e.g. A 10 mA LED @ 3 V battery ; R <= (3-2.1-.250)/10e-3. It has the potential not to work. V will vary and so will the 2.1 V which is the LED drop. There is a +- range specified for that too. The LED color changes the values.

 

@keepit: I'm trying to catch up to all this, but I can report one thing. I can't seem to get any reading for milliamps or even microamps. The most sensitive range is 0-20um, and I get a reading of 0.

 

@keepit: I'm trying to catch up to all this, but I can report one thing. I can't seem to get any reading for milliamps or even microamps. The most sensitive range is 0-20um, and I get a reading of 0.

I actually would have expected that. The usual cheap ammeter is noting more than a resistor that lets say is 400 mV full scale. You need to look at your meter, so it looks like a 40 ohm resistor.

One other caviot, is that there is usually a 2A fuse in most meters on the current scale and you usually have to use another post to measure current. That current fuse is almost always blown especially if you let someone else borrow your meter. FWIW: I made sure I had a complicated DVM at work, so no one would borrow it, just like no one borrows a calculator that uses RPN. They sometimes attempt to, but immediately give it back.

So, in theory, you could use say a 1000, 1000, 1M or 10 M resistor and try to measure the voltage across it, but when calculation the current, you have to assume the 10 M of the meter is in parallel, so for a 10 M and a 10 M meter, you have 10||10 M or 5 megohms. I = Vm/5e6; Vm is Vmeasured.

I know I gave you a lot to munch on. I sort of, led you astray, which may be a good thing when I found the single comparator version, but I didn't check stock first. Nonetheless, there is some valuable education there. Note, that I modified some PDF links to open at a specific page.

 

I picked a FET out of the blue to illustrate a few concepts: http://www.digikey.com/product-detail/en/diodes-incorporated/DMN3190LDW-7/DMN3190LDW-7DICT-ND/

But let's say you had to use FET to drive a higher current load. The first problem what voltage will turn it on which is related to Vgs. This one has a Vgs from 1.5 to 2.8 V and is known as a N-channel LOGIC FET. Unless, you did some selection, 2.8 might be too high to get good battery life from a CR2032.

Again, design say to put some resistance in series with the gate, 200 ohms is a good rule of thumb. Whenyour using it as a switch, you have 1 uA of leakage current, you have to provide a sink to, so you need some resistance from gate to ground. The Rds(on) parameter is just like Vce(sat) for transistors. It does provide a drop.

So, in the case of an open drain comparator, The positive gate voltage has to come from the positive supply, so you basically apply some small current above the leakage current to the gate via that 200 ohm resistor. So, to turn it off, you have to cause the comparator to sink all of that current away. This, in effect, changes the "sense" of the output.

The comparator data sheets for the comparators are extremely confusing and they sidestep what you really need. e.g. On the "-Y" part, a voltage >400 mV on the *IN terminal causes the open drain transistor to conduct.
where Y is replaced with -2 and -3 and * is replaced by + and -. That's what's missing.

The wording can be cleaned up a bit to include the hysteresis, but those words are missing.

 

I do appreciate your time with this, even if I don't understand much of it. Hopefully I will have more time this week to do a little research and come back to you with some meaningful questions. I think my goal at this point is to understand the issues that need to be addressed, such as draining off leakage, filtering a sputtering signal, and so on. Beyond that, I want to get a parts list together and just start building something to see what happens.

Regards!