James Berg
10 April 2005
TURBIDITY METER
PROGRESS REPORT II
The last month has been spent on two major developments: finding an appropriate micro controller and other associated circuitry and determining if a smaller sample container can be used.
My partner Jesse Adland, who is an electrical engineer, has decided that, at least the for the purpose of development, the Atmel AVR MEGA32 micro controller will work best. This controller will do the conversion from analogue to digital and will talk directly with the LCD screen. The MEGA32 has far more features then is needed for the turbidity meter but because it is what Jesse is most familiar with it will be easier to develop the turbidity meter using the MEGA32. We would switch to a cheaper micro controller for a production model. The basic design of the electronics will be two photodiodes feeding into the micro controller, the micro controller will then feed directly into a small LCD screen that will output the turbidity and instructions during calibration. A 9V battery will provide power for the micro controller and the laser via a voltage controller.
Tests to determine if the 1” cuvettes from the Hach 2100 could be used in this turbidity meter have been quite successful. Though there are several problems with the graph below it shows clear distinctions can be seen at low turbidities. Since the 1” cuvette is a preferable to 2L bottle the turbidity meter was modified. These modifications came in the form of a square box in the middle of the turbidity meter that was just big enough for the cuvette to slide down into. Three holes were drilled in the sides of the box, one for the laser, one for the 180° sensor and one for the 90° sensor. A similar design would be applicable for the final design. Possibly reducing the whole unite to a 3x3x4 inch box.
As I have said there are several problems with the above graph but despite these it does show that the photodiodes are capable of being very sensitive at low turbidities. At higher turbidities the readouts from both sensors become rather horizontal. This is a bit worry some but I am not satisfied with the validity of the numbers after 435 NTU. As you can see between 435 and 495 there is a dramatic change in the sensor readings. There was a problem with the Hach at this point, the sample after 435 should have been around 500 NTU but the Hach read the sample as around 225 NTU. Several attempts to correct this problem were unsuccessful. I mixed continually dirtier samples until finally the Hach gave a value above 250 NTU, the Hach read this sample as 495 NTU, but I would estimate, based on the NTU values the Hach gave before the 435 NTU sample, that the 495 NTU sample should have been somewhere around 650 NTU. I am still unsure what caused this problem. Another problem with this data is that up to 74.9 NTU the light intensity for the 180° sensor was to much and the sensor produced a voltage above what the data acquisition system was set to read. This caused the computer to read the voltage from the 180° sensor as 0.4998 V, when in fact it was higher. This is an easily fixable problem; just looking at the data I would guess that the voltage input on the data acquisition system should be set at ± 0.6 V.
I have also looked into getting a different laser, one that is not a disassembled pen laser. Newark Electronics has several diode lasers, they present a minor cost savings but they would be much smaller. The down side is that they all run on 5V as opposed to the 3V of the pen laser, and would likely produce a more intense beam of light. This is not so much a problem it would just mean that we would have to gather more data. This has just been a cursory search, there could be many more complications with using a diode laser, but it is certainly worth looking into.
Right now the final system is looking very promising and very cheep. As you can see in the table to the right the electronics should cost between $29 and $39. This does not include the cost of case or the cost of the sample containers. The cost of the case is very hard to estimate, for me it would probably be in the range of $5 for the black nylon that I would then machine as needed. In a production design the case would be a few pieces of black injection molded plastic and I have no way of figuring out how much that would cost, if I had to guess I would say $2-3.
While I am now using the cuvettes provided by Hach for use with their turbidity meter, it would be undesirable to use them in a production design as they cost $20 for a pack of six. There is also no need to store samples so at most two sample cells would be needed (one spare in case the first one breaks). Cuvettes are available at a much lower price but two decisions must be made. Cuvettes come square or round (mostly it seems square) and in glass or plastic. Glass has the advantage of not being as soft as plastic and thus being less likely to scratch and distort the light but is also more fragile. Another concern is whether or not the turbidity meter is water tight at the very least from splashing. It would be nice to put all the electronic parts in a sealed portion of the container but the laser, photodiodes and LED screen all need to be visible to some unsealed portion of the container. This makes it hard but certainly not impossible to make the turbidity meter water tight.