OK, I got some numbers today. The voltage tends to drift downward slowly but I don't understand exactly why at this point. I took the reading when it looked like the voltage drop had slowed significantly. I switched between resistors as quickly as possible to try and get similar conditions throughout the range of resistor measurements.

infinite = 0.88 v
10 M = 0.88 v
1 M = 0.88 v
100 K = 0.88 v
10 K = 0.62 v
1 K = 0.20 v

At the end of the last reading with the 1K resistor I went back to infinity to see if I had the same conditions. My reading started out low and drifted UP to 0.84 v. Also, I did a kind of trial run in which the 10K and 1K readings were within 5 to 10 percent of the numbers above.

Hopefully that gives you something to work your equations with. I can redo the test anytime if you think it necessary.

Thanks.

 

Daniel:

Thanks. That's all we really need is an order of magnitude. A real number would be to put a 100K potentiometer across your probes and adjust the voltage to 1/2 the initial value. Remove the pot and measure the resistance, but who cares. So a resistor greater than 100k||10M and <10 Mohms won't disturb the measurement. The lower the resistor the less unacceptable to noise.

The actual impedance of the "battery" is somewhere between 10K||10M and 100K||10M. The || denotes "in parallel with" . The current is likely <100 uA. So, we have some "ballpark" numbers. They would vary based on the soup mix too.

So far, I have an idea that will do positive and negative voltages, with a filter and good 60 Hz rejection, so you can put it in all in one place with one 9V battery. I just have to figure out a way to draw it. Digikey's Scheamit (sp?) browser based package sucks. You can't add a pin properly to the top or bottom of a triangle AND you can't copy an existing symbol to a custom symbol and modify. I confirmed it with DK's support yesterday. I didn't get to try this https://easyeda.com/ one.

"Battery test" should be easy. Probably put a load on the battery and check the voltage with a comparator"

Reverse polarity can be done the easy way of the "hard way". I've successfully done the "hard way" with a 9 V battery before. The easy way has higher leakage or lower voltage - pick one, but you have a power switch. It's easy to temporarily put a 9V battery wrong on the clips.

I haven't spent much time of the 30s thing and you may not need it.

Plan to protect the inputs with a 200 ohm resistor and amplify x1 first and shift the output to 1.5 V with no input with an Instrumentaion amp. You can "pick" whatever resistor you would like at the front end. An IC will create a 3V supply and 1.5V reference at 20 mA from 9V. The output of the IA will have a filter following it. Then use two comparators. One set for 1.5+thresh and one set for 1.5-thresh, so > than the first one or less than the second would generate an alarm. Will have some level shifting issues to deal with and the 20 mA limitation.

So, that's how it sits in my mind.

 

Thanks, Keep it. I understand that something needs to be done in the event the 9v battery is inserted backwards. I'm not sure I understand needing to alarm with a negative voltage, though. With an aluminum pot on one lead and a stainless probe on the other, and only one way to attach the leads to the "box," then why do we need to plan for negative voltages? Is it a matter of protecting the components? If it is an option that doesn't cost any more than alarming on < +400mv then I guess it would be fine.

I also take it, then, that we will be able to power everything with a 9v battery, components and alarm inclusive?

 

Basically, you can reverse the leads on the pot with no ill effects. For now, I'll assume clips, but that might change.

The only real way to implement a filter is to buffer the signal first. I chose an Instrumentation Amp, so common signals get rejected.
The filter so the wires can be long.

I also take it, then, that we will be able to power everything with a 9v battery, components and alarm inclusive?

That's what I see so far.

Do you see multiple pots being on the stove at the same time?

Corollary: Do you envision, say a 4 channel unit?

Take a peak at allied, mouser, digikey and do some searching for buzzers.

Is there anything wrong with allowing a wall-wart power supply? i.e. one that disconnects the battery when plugged in?

One 9V battery holder I've seen is sort of a drawer.

 

Oops. can't do the 9V only battery trick and the part may be too small.

 

I think the wal-wart (odd name) power supply would be a nice feature. I do not think multiple channels for more than one pot at a time is necessary. Maybe a good thing for version 2.0.

I'll get back to you on a buzzer. Edit: Just to be sure, what is the acceptable voltage range for a buzzer?

Thanks!

 

what is the acceptable voltage range for a buzzer?

Various ones available from 3V up.

 

I don;t think the sound from a 3V buzzer will be loud enough, but you can look into that too. If I'm making soup, I'm far from it.
So, let's look at a typical 9V battery: http://data.energizer.com/pdfs/522.pdf So, 9 to 4.8 V or so. You can look at discharge current and MA.

Not thinking about exploring other battery options right now, like 2 AA batteries or any other battery technology. Neither is it worth exploring regulators such as this http://www.linear.com/product/LTC3388 to make a lower voltage at this time. It's just "low power" usually means "low voltage". There are "watch buzzers", but I really doubt you want one of those.

So, I'm basing a buzzer on stuff that's around the: wireless doorbell (4-D), Kitchen timer (9V) and a freezer alarm (2-AA).

 

OK, here's my big parts selection - no package variations. The general problem is not wanting a +-supply, ow voltage (where 5V is common), Battery operation (where low power is common) and a fair amount of buzzer power.

For now, the design would mostly operate on a CR2032 Lithium battery which generally gives a relatively flat discharge curve. A 9V battery would operate the buzzer. Battery test might be an option a well as eliminating the CR2032. This would be regulated down to a 2.5V and 1.25 V supply with a REF2025 which is capable of delivering 20 mA. http://www.ti.com/lit/ds/symlink/ref2030.pdf

Long leads and a high input Z (Impeadance) will cause issues, so a load of 100K to 10M for the "soup" would be selected. Your DVM was liklely 10 M and you pretty much shown that 100K to 10M is the range you need to use.

By using an IA (Instrumentaion Amplifier) AD8231http://www.analog.com/en/products/amplifiers/instrumentation-amplifiers/ad8231.html#product-overview , the output can easily be level; shifted to 1.25 V for 0 V input further allowing common mode rejection. The rejection of a 60 Hz signal common to each input. Hopefully, it will not be necessary to put part of the circuitry next to the pot. The IA is a relatively expensive part. A breakout board is available for $75.00 or so. The uncommitted amplifier would be biased uncommited (x1, input resistor to ground).
two 10M resistors would provide a bias return path. 200 ohm or so input series "protection" resistors would be used as well

The IA would be followed by a filter (say 0 to 10 Hz or less) would allow longer leads and reject 60 Hz interference.. Twisted pair would be preferred, but probably not essential.

Now a dual comparator (dual polarity with 400 mV reference). The LT6700-1 or 6700HV-1 http://www.linear.com/product/LT6700
The thresholds would be set to 1.25+800 mA and 1.25-800 mV using high value resistors to conserve power. Their outputs would be combined (wire OR) such that the output is on when less than the minimum voltage and greater than the max voltage. Now, probe polarity would not matter either.

At this point, you really have to select the buzzer (9v, x mA). So, this
https://www.digikey.com/product-detail/en/pui-audio-inc/AI-3135-TF-LW100-R/668-1350- ND/1745459 could work.

To turn it on, I'll co unconventional and use an optocoupler or optomos relay (CPC1006N)
http://www.ixysic.com/home/pdfs.nsf/www/CPC1006N.pdf/$file/CPC1006N.pdf It uses up 1.2V @ 0.5 mA for the LED (which is very, very low) and has an on resistance of 10 ohms and good for 75 mA. Digikey stocks it.

So bypass caps and filter values are also required and a whole bunch of misc stuff. Seasoned guys know that the case/power supply/connectors can sometimes break the bank. The other rule is "real estate" (the PCB) is expensive.

It's not the simplest, nor the cheapest, but it's a start. Questions it has to answer are:
1) Will long leads be a problem - I suspect not.
2) Is it necessary to add "condition must exist for 30 seconds" before sounding alarm?
3) How long will the batteries, "buzzer time" last?

It may also have some room to grow too.
1) CR2032 battery elimination
2) Multi-channel
3) Wall-wart power supply
4) Battery(s) test

Comments:
The 30 s delay would compromise the ersatz battery test (probes out of water)

Warning -- parts are tiny.

Now, have to figure out how to draw it.

 

Thanks, KeepIt. I had a couple of questions and comments (actually I have about 1000 questions, but I don't know enough to ask them all yet, so these are just things that come to mind right now.)

Regarding the amplifier: It looks like it costs $4.67. Can you give me an idea why the amplifier is needed? I thought the comparator could look at the input from the pot directly and alarm at the 400 mV trip point. I know I'm missing something. Does the amplifier smooth out the choppy signal when the water level gets down to the tip of the probe?

The $75 breakout board is just something I need for putting the components together, but is not required for the final product, correct?

Regarding the comparator, you said, "The thresholds would be set to 1.25+800 mA and 1.25-800 mV..." Is this a typo? Should it be +800 mV? I'm not sure I understand how the amplifier and the comparator work together. I don't have to understand every detail at this point, but this isn't really a detail I suppose.

I like the idea of the on/off switch, but I also think it is important to know when either battery is too low on power. The unit will be left on for hours at at time, during which time it could run low on power. I saw that the CR2032 was rated at 220 mah, but I don't have a feel for how many ma this device will consume during operation. I think the 9V battery is likely to last a long time as it will only be used for a short time while the buzzer is sounding.

Speaking of the buzzer, I think something slightly louder than a microwave alarm is all I need. Based on looking into dB levels online, I would propose we shoot for something above 75 db, but not greater than 80 db.

Also, I don't think we need the 30 second delay. Allowing the unit to alarm when turned on is actually a good check that the battery is OK to start. I might even suggest someone turn it on just before inserting the probe in the liquid so that the alarm is heard. Does the filter quiet down the signal just before the water level goes below the tip, as mentioned above, or does the amplifier do that? That would be the only reason I could see wanting a 30 second or 60 second delay.

 

Have you seen this?

Pet Water Disk Alarm Sensor

 

yes you can do it with a battery, but you need to drop the current down to microamps in the probes, so a mosfet transistor sensor would be best

3uAnot low enough? the design works as is.

 

I need to leave soon. The Amplifier is a "buffer" or it has a x1 gain. The real purpose is that it's required if you want to implement a filter, which I think is absolutely necessary.

An IA is a special amplifier that amplifies the difference between two leads. It specifically rejects 60 Hz input present on BOTH Leads. They also allow easy level or offsetting.

A lower-cost version would just include a single ended buffer and the connection would be one-way.

Remember when I mentioned www.proto-advantage.com. They make breakout boards. Breakoits type boards from analog devices are really pricey, but are specifically designed to explore a given part.

The delay was a possibility if boiling exposed the probes momentarily, but it creates more problems. Yes, I envision the same thing. Basically power on right before insertion to check the battery. e.g. hand on clip and hand on switch. Attach clip shortly after turning on and inserting.
The elimination of the CR2032 is possible. It's also an "expensive battery" at the consumer level. I get them routinely for $.89 each with a $15.00 minimum order and free shipping. I can get them cheaper. Not $5.00 to $6.00 at the retail level.

If you shift the level so zero is 1.25V, then 800 gets added and subtracted from the new zero. The subtraction is almost free and it allows probe reversal.

The filter would hopefully remove 60 Hz pickup. So, it's like really short leads or filter.

 

I guess the buzzer you posted would be fine. It is rated at 105 db but I guess you can't really tell until you hear it. Louder is better of course if it doesn't make a big difference in battery life. This sounds like something that would be easy to switch out without changing the other components.

Wendy - Yes, I saw that when I originally posted here, but my application is a bit different. Boiling water, food grade requirements, etc..

 

I guess the buzzer you posted would be fine. It is rated at 105 db but I guess you can't really tell until you hear it. Louder is better of course if it doesn't make a big difference in battery life. This sounds like something that would be easy to switch out without changing the other components.

There's a plot of intensity vs current on the manufacturer's website, I was just looking for anything that would work.
For fun today, I went to www.easyeda.com and: 1) Cloned a package (the comparator) and 2) Modified a cloned package (the reference).
The next step is harder: Create the IA.

It doesn't do a great job of listing "Your devices", so adding IC in front allows you to list "your" IC's. It's always wise to work with a copy of a footprint/symbol, so everything stays the same when you do it the "next time". So, I'm learning.

I'm just trying to come up with a schematic and not necessarily do everything to get the complete project done. So far, so good. So, I might pick "a buzzer" and a "battery holder", but not the "right ones". Example stuff that would not be included, but would be required would be "connectors". Everything starts getting interrelated.

Here's https://www.sparkfun.com/products/retired/11368 a RETIRED chassis mount 9V battery holder. There are all sorts

 

Interesting! There's a lot more to consider about designing these circuits than I figured. I'm trying to learn more but I don't want to get in the way of your design. I understand the approach you are taking though. At some point I'd like to understand exactly how each component interacts with the other but that's for another day.

 

Can't do the CR2032 battery and the 9V battery version in the way I was planning. V1 will likely be more of a USB phone charger version (testing only) and not battery powered. It's an easy and cheap way to get 5V at 100 mA. That was basically yesterday's progress. The IA has a minimum voltage of 3V. The battery won't operate at 3V for long.

The LTC3388-3 would make the single 9V battery work.

Battery or DC adapter would be more practical, but a mini-USB type of connection just doesn't seem right? A ( Mini-USB cell phone "charger" or 9V battery) type of product may be more difficult to do than say a (12 V wall-wart or 9V battery) type of product.

 

If it can be done with one battery instead of two, even if it requires some more components, that would be better from a user point of view.

Is the problem that we have different components that have opposing voltage requirements, making battery selection difficult?

 

If it can be done with one battery instead of two, even if it requires some more components, that would be better from a user point of view.

Is the problem that we have different components that have opposing voltage requirements, making battery selection difficult?

Primarily "low power" means "low voltage'. Think a 1.5V watch battery that lasts a year. 5V used to be a common voltage for logic. Now we are seeing 5V, 3.3, 2.x, 1.8V etc. 3.3. is sometimes written 3V3.

Totally agree, I just don't think the "super-simple" version is viable. Where we started. I have a reasonable confidence that I'll come up with will work well.

So far the voltages will be 3.0 and 1.5 which have to generated from a 5V source.

The LTC3388 is a "Buck" converter, so it reduces th input voltage.It looks like you need about 200 mV more than the regulated voltage.
The 3 V1.5 V can work with 10 mV of headroom, so you need 3.010 V for it to regulate.

That in itself means a voltage from say 3.3V to 20V in. Then there's the buzzer, so that ups the voltage. Then there is the 5.5 V absolute max,so that sets a practical low limit. Then there how much current the regulators can source,

So, when your all done, you can take a simple approach with a 9V battery and a 12 V adapter with diodes that would select 9-Vd) and 12-Vd. Vd can be between (Vd ranges from 0.2 and 0.6V). Leakage could be an issue. The other choice is an ideal diode controller. So, without battery operation, 5V would be reasonable.

The "project" is public under KeepItSImpleStupid with the name of Play. Supposedly, you can look, but no modify.

 

I went to EasyEDA and found your project, but there wasn't anything there. Maybe you scrapped it?