PDA

View Full Version : Sensor team needed


bstreib
03/11/2002, 07:19 PM
Hi All,
We will putting the first project DIY's up on the new web site in the next couple of days. We will start by building a couple of sensors for our project. One will be a level sensor and the other a current transformer. A current transformer will tell the controller that a critical device [main return pump] has failed.
I need a E-tech lead, an associate with PCB design experience and a another E-tech that retest and verify the results. This three associate team will be responsible for documentation and submitting reports and DIY instructions to the web master.
This team will report directly to the project board. There are many proto-projects that are based on sound physics and electronics that need to be investigated. While the board will supply ideas and fundamental designs, this team will also some autonomy to investigate and explore other options.
This project is real and will develop into some really great things for all of us. Now it's your turn, any volenteers?

DJ88©
03/12/2002, 01:03 AM
I don't mind sticking my fingers in here. I am an E-tech and am currently at school again for another engineering diploma. It is pretty good timing as I am working on another research project of my own and will be designing some amplification circuits for data logging purposes(also to do with reefkeeping).

If I can get a bit more of an explanation of what will be required as for equipment, testing and such I'd appreciate it. Jsut so I can get an idea of how much time I can expect to put aside.

DJ88©
03/15/2002, 01:28 AM
BIll,

Send the info you have to my email address listed on here. Now that mid-terms are done I can focus on projects again.

I'll take a look at when you have and then see how it compares to what I have been looking into.

One area I am wondering about is what range do you want readings to be taken from the tank. ie temp. say 74-88? greater? What temp sensors did you think would be a good one to use? If the outputted voltage from the sensor is low it won't be anything to set up an intrumentation amplifier to boost the signal level to a more readable level for the 8bit card. If we were to set it up to read say a 20 degree range we can give it a resolution of ~0.01 degree per step of the ADC in the card. Depending on the impedance of the probe I can build the amp for the signal for a couple of bucks. If it is a high impedance we can use an instrumentation amp and get a gain up around 100-200 quite easily. more than enough to boost the signal to where we want it.

Where we are looking at whole or half degrees for precision the 8 bit card should work great. 256 steps is plenty. I need 12 bit for the work I am doing. But once I have the card I can do some testing of prototypes.

I will do some reading into some other simple op am setups to ensure we don't start oscillating with heating and cooling cycles. Some form of integrator.

what sensors/probes are you looking at using? I'll start digging at school and look into temp sensors we can get that are already used in industry. I have a meeting tomorrow with the head of computer control systems and will pick his brain about this. I will also get in touch with a few other instrumentation tech's I know.


I see that my email address is not actually displayed for you to send me mesages on your own.. my address is dj88@shaw.ca

DJ88©
03/20/2002, 12:31 AM
Anyone else interested with this part of it? Personally I think it will be fun. But maybe I am a bit wierd. :D :D

I'd like to start swapping ideas with anyone interested.

schwaggs
04/09/2002, 11:48 AM
Originally posted by DJ88
Anyone else interested with this part of it? Personally I think it will be fun. But maybe I am a bit wierd. :D :D

I'd like to start swapping ideas with anyone interested.

I will put my hat in the ring for the sensor part. I'm not a circuit or ADC guru, but do have an entry level understanding of what is to be measured and what needs to be done to the data. I've done some PIC programming in the past trying to build my own aquarium controller but pitched the project (UI too hard for me).

Here is my list of sensors (inputs):

Temp- we can go two ways here, either use one of the existing temp probes on the market or build our own. In any event, I think we should make our conditioning circuit compatible with the off-the-shelf probes just in case we can't make our own reliable. I believe that the off the shelf probes are nothing but a themistor embeded in plastic.

Ph - not much choice here, must buy probe. The challenge here is calibration. Do we want to feed the "brain" raw data, or calibrate it in the sensor.

Orp - again, not much choice here, must buy probe. I don't think there is any calibration involved.

Conductivity - again not much choice here, must buy probe. Do we want this?

Disolved Oxygen - not much choice here, must buy probe. Do we want this? Probes cost over $400.

Electrical Current sensing - used to sense the current going to a device, like a main pump. Not sure if we need this, I did see a post in here somewhere that someone wanted this...

0-5V input - Not really a sensor, but an input. This could be valuable for several reasons. Should be optically isolated to prevent blowing other components up!

Digital input - Not really a sensor, but an input. This could be valuable for level sensors, water flow sensors. Should be optically isolated to prevent blowing other components up!

Any others?

Jon

BSide
04/09/2002, 01:14 PM
Temp sensors: PLEASE use something like DalSemi 1820s. They are pretty cheap, and amazing devices. Way better than thermisters (temp calibrated, good range, digital, have serial #s, can put 64 on 2 wires, etc.)

Current flow: is this really the best way to handle knowing if the main pump is working? I would think something like WATER flow is more ideal. There are definitely flow-rate sensors that can be purchased, possibly expensive though for OTS.

I'm definitely interested in getting the circuits for pH probe amplification, etc. That kind of analog stuff is way beyond me. Looking forward to it!

schwaggs
04/09/2002, 01:39 PM
Originally posted by BSide
Temp sensors: PLEASE use something like DalSemi 1820s. They are pretty cheap, and amazing devices. Way better than thermisters (temp calibrated, good range, digital, have serial #s, can put 64 on 2 wires, etc.)
I agree, but if we use an integrated temp sensor, we MUST build our own probe. I'm not sure how hard this is...

Current flow: is this really the best way to handle knowing if the main pump is working? I would think something like WATER flow is more ideal. There are definitely flow-rate sensors that can be purchased, possibly expensive though for OTS.
I agree, I only put the Electrical Current sensor in here as a placeholder. I don't think we need it...

I'm definitely interested in getting the circuits for pH probe amplification, etc. That kind of analog stuff is way beyond me. Looking forward to it!
Me too! I researched this a while ago but couldn't find any good circuits. I will start researching this tonight.

bstreib
04/09/2002, 08:39 PM
Hi all,
Just to start off no we don't have to build all the sensors but we can build some better and cheaper than the OTS versions. I look at sensors as belonging to one of 2 classes for simpilcity: discrete sensors; simple on/off and analog sensors; a dynamic value over a specified range.
I have sent plans to our website for several of these as starter projects. Everything you have posted so far cound be a discrete or an alog sensor. A temp sensor for example could be both types for different applications. An alog sensor for measuring water temp and datalogging it. But a simple thermocouple made very a few twists of thermocouple wire and couple transistor and relay, we can control when fans in our hoods, electrcal cabinets and stand turn on and off.
Level sensors are another example. Discrete sensors are cheap and dependable to signal when levels in reservoirs, mixing tanks, dosind and additive reach certain levels. On the other hand the level in the main display and/or the sump would probably be better controlled with an analog sensor based on capicitance sensing. We can build this around a basic stamp and divide the level into, I think 4964 increments.
I have been playing with a magnetic stack flowmeter. It works in principle but circuits still need some tweeking and work. Hopefully these can be build for less than $20.00 a pop if we can work the bugs out.
I think this part of the project will be the most fun and interesting part. I would welcome you both to jump, if your interested. I would suggest you go to the omega website and order their free bound reference library, the URL is www.omega.com/. They have a wealth of info in their manuals and it is easier to have it sent to a business address, than a home one.
I have a couple of pH controller schematics. The one I like is based on all Radio Shack parts. I made 2 and sold them on the west coast and they worked great. The last 1 is still in the basement, like a lame duck, no stability or dependability. I think the power is the power supply. Trying to derive +-15 VDC for a single wall wort, but haven't had time to work it out yet.
For the last 3-4 weeks, I haven't had time to dedicate as much time as I wanted to, to this project because of the big "R" at our plant [reorganization]. I have a new job and the hours are hectic, 16 yesterday and 15 today. This is a good project with some very talented people involved and I think you all will enjoy working with it. I hope to have you all working the sensor team. I just got back my zip drive and disks as well as my 5 removeable hard drives. They were all collected and locked up when the plant shut down for reorg. So if your interested I can send you some of my stuff. Owe the webmaster a ton and then DJ88, the first sensor team member. I will get it out to you but mat take a few days. Thanks again

DJ88©
04/10/2002, 08:51 AM
HI Bill and everyone.

I have some good news for input to the PC

Dataq has 10 bit a/d's on sale. $24.95. I am buying two shortly.

I also have been quite busy with school, midterms right now, but do have some ideas for this. I have been doing a fair bit of work with PICs lately with this and some other ideas I have.

I"ll try to get back here in the next day with those ideas and such.

I'll be able to dedicate more time shortly.

schwaggs
04/10/2002, 11:36 AM
Bill,

I you mention putting things on the project web site. Is there one and what is the URL?

DJ88,

You mention using ADC on cards. What type of system are you using? $25 for a 10 bit ADC seems kinda high for an individual ADC chip. Just a quick browse of digikey I found some for $5 and less. Although if you are using some protyping system, that would be cheap. Just curious what you are using.

I think whatever we choose for an ADC, should be used for all the sensors to keep it simple. SO, what range do we need for each sensor? I will take the first stab at it.

Ph - Assuming we want to display Ph to the X.XX level. Ph varies from 0-14

8 - 2.56 not enough
10 - 10.24 would allow us to display a good range
12 - 40.96 would allow us to display the whole range

Athough 12 bit will allow us to display 0-14 Ph, i don't think that is required. We are not making a lab tester. So either a 10 or 12

ORP - I'm not sure what the actual range is for an ORP probe, but we need to at least be able to display 0-700 micro siemens per centimeter so a 10 bit ADC would work

Temp - assuming a X.X degree reading.
8 - 25.6 Probably enough
10 - 102.4 a good range
12 - 409.6 the whole range

Conductivity I need help here, I can't find the range we need to measure.

Seems like 10 bit would be good. I might be all wet here but thought I would take the first stab.

Jon

schwaggs
04/10/2002, 11:54 AM
DJ88,

I took some inititive and searched for DataQ and found their site. That $25 4 channel 10 bit DA package looks cool! I am going to look into it, it may be a great starting point and look like an easy way to integrate DAC. I found the following Ph probe circuit the other day. It is actually a Ph controller but we can use the amp part to feed the DAC.

Check this out this scematic (http://www.thekrib.com/Plants/CO2/hurley-schemat.gif) and documentation (http://www.thekrib.com/Plants/CO2/hurley-co2.html) .

I ordered a Neptune Aquacontroller from Advanced Aquatics yesterday so I will have a Ph probe to play with in a week or so. For anyone that wants an Aquacontoller, talk to Jason at Advanced Aquatics for the best price. Mention you saw a message on ReefCentral about it.

I will spend some time this afternoon surfing DataQ's site, it looks awesome, thanks for the lead!

Jon

DJ88©
04/10/2002, 05:00 PM
Jon,

Sorry for not giving you the link to DATAQ. lol. was in a rush today.

I just ordered the 10BIT package as I have some other testing that it will work great for.


For the price you can't beat it right now. I am glad I waited. A few weeks ago all they had on sale was 8 bit. :) :D

Snailman
04/15/2002, 09:49 PM
Digital pots that the processor can control would be nicer and more stable than screwdriver pots R2 & R3.

schwaggs
04/15/2002, 10:57 PM
Originally posted by Snailman
Digital pots that the processor can control would be nicer and more stable than screwdriver pots R2 & R3.

I was thinking about that very same problem the other day. As I see it, there are 3 ways to handle the calibrating issue.

1) Little trimmer pots

2) Digital pots

3) Feed the processor raw data and apply a mathmatical adjustment

I'm not sure how to do it in software but I would think it is easier and more accurate than 1 and 2.

Anyone have an idea how to do this in software? Lets say for Ph you have the typical 7 and 10 cal solutions. At 7, the probe read X but should read 7. At 10 the probe reads Y but should read 10. What formula do you apply to measured value Z to get the calibrated real value?

Jon

bstreib
04/17/2002, 02:58 AM
Hi all the software that comes with the DataQ allows use to calibrate and scale data in the program. You enter a low and high value and dimensions and software scales the inputs accordingly. I'm sure it asssume a linear function between the 2 points which is ok with most probes if the scaling is kept small. I.E. for pH 7 and 10 or for conductivity 300 to 500 etc.

schwaggs
04/17/2002, 08:46 AM
Assuming our own program would be written for the user interface on the PC, we need to know the formula.

bstreib
04/20/2002, 06:44 AM
I don't think there is any real formula involved here. The datalogger is measuring a voltages and then scales between them. For example, the pH probe front end outputs voltage X with ph calibration fluid 7.0 and voltage Y with pH calibration fluid 10.0. The software then scales the values in between them.
A lot of the values and acurracy will be determined by the front end we design for the probes. The greater the voltage range between values X and Y the greater the accuracy. Assuming a 0 to 10VDC output: if we have pH 7.0 equal to output 7VDC and pH 10.0 output 10VDC, we only have a range of 3VDC. We can increase the accuracy by increasing the range. While this true only to a point, the limits of the circuitry to maintain linearity, it is quite feasible to do.
This is true for most of the probes we want to use. I'm not sure about DO, I have never work with a DO probe before.

banger
05/17/2003, 04:03 PM
Originally posted by schwaggs
Bill,

Ph - Assuming we want to display Ph to the X.XX level. Ph varies from 0-14

8 - 2.56 not enough
10 - 10.24 would allow us to display a good range
12 - 40.96 would allow us to display the whole range


Jon

You might want to look at the math. 8 Bit would be enough for PH. If PH is 0-14, then take 14 and divide by 256. This gives you a resolutiong or reading ph in increments of .05 Likewise, 10 bit would give you a resolution of .013

schwaggs
05/18/2003, 07:52 AM
Exactly why I said 8 bit would not be enough. you need higher resolution than .05 for pH. With .05 resolution, pH of 8.00 could bounce between 7.95, 8.00 and 8.05. To make an accurate meter, you should not display the highest measurable resolution. Thats why I reccomended a 12 bit ADC since it's resolution would be .0034 over the entire pH range. The other option is to measure only a range of values, say pH 7-9. Then an 8 bit ADC would work since the resolution would be in the area of .0078.

Gudwyn
09/25/2003, 11:33 PM
For the in-tank probe, you need the pH range to cover from about 6-11. The calibration fluids are at 7 and 10. Using a fixed analog circuit, you can roughly center that range on the DA input. Then the slop at the ends gives you room to use software adjustment on the raw data for probe specific calibration.

To bad my wife isn't working in a lab anymore. Would be pretty easy to have her do a titration of kalk into vinegar and measure the pH vs. voltage curve on a probe. Then we would know if the software correction over the 6-11 range was a linear adjustment.

It may be wise to cover the entire range from 3-11 to allow freshwater and Ca reactors to use ph4.0 calibration fluid. That would make units: 8bit 0.03, 10bit 0.01, 12bit 0.003. So, yeah you need 12bits.

BTW. My preference is to use commercially available probes where possible. This includes temperature. Added capability like multiple pH probes, 4-5 float switch inputs, water flow sensors, etc. is what interests me more than saving money on probes.