Robovergne

Presentation

In this post i’ll describe a project which, despite being ridiculously titled, I am most proud.

I have had for several years a reef aquarium at home with living corals. These animals draw in seawater, and set in their skeleton certain chemical elements it is therefore necessary to replace calcium, carbonates, strontium, iodine and other trace elements. So we end up having to add in the tray a number of products to maintain the stability of the composition of the water.

A system for automatic distribution of these products in these conditions of great help. Commercial systems are all based on the peristaltic pumps. On the one hand, they are extremely expensive. The Grotech TEC III, reference device corresponding to the left figure, for example, sells (online) more than 350 euros in version 3 to 7 channels and 600 pumps!

I searched what I could find as peristaltic pump to make my own system, unfortunately, I could not find anything less than € 60 at the time the pump, which was far too expensive for me. Inspired by a DIY described “Alain34″ I sought a solution to drive the pump espresso machine.These are the pumps that are responsible for noise and vibration of the machine. The pump model most commonly found in most machine is the famous red pump produced by Ulka.

Result

The result is this configuration that integrates in one box, up to 9 pumps, and electronic control circuit. I even prepared a doc for details of the operation: Documentation nano dispenser

The pumps are installed perpendicular to the box below the entry and exits above. This organization is the weak point of the metering as this require to break the media on which it is asked to install the cans below. My first idea was to make a wooden chest, pierced to let the pipes. Thereafter, the dispenser simply moved on a shelf in the cabinet under the aquarium.

Adjustments are made with the help of two buttons and the LCD. You can choose the amount of product to be distributed each day, and the timeframe in which the distribution is made. The device is then responsible for calculating the optimal operating speed of the pump so that the dose of the product is distributed on an ongoing basis during the given period. The height of release, the differences in viscosity of the products and changes in the manufacture of a pump to another must be taken into account. For this, the device includes a procedure for an automatic re-calibration of each pump.

The system is perfect for the reef aquarium, but I guess it must be adaptable to other uses, dosage of fertilizers for the freshwater aquarium, for example, but probably hydroponics or why not maintain a swimming pool.

Some technical

Basically, a Ulka vibration pump is actually a metal coil with a plunger in the center. When a current through the coil, the piston moves. When the current stops, springs back to his original position. The movement of the piston displaces fluid and a valve system prevents it from going back.

In the 230V pumps (there are also versions 24V), a diode is placed in series with the coil. At each wave alternans sector, 50 times per second, the piston is a back and forth between the two valves. Very noisy vibration of the pump corresponds to the impact of piston slap, 50 times per second.

My idea was to find a solution to change the interval between two pulses of power to the pump, so you can precisely control its flow.

The operating principle of the system is inspired by the note of application Picdim Microchip. Basically, this document describes the use of a PIC microcontroller to perform a lighting controller based on the phase delay.Conceptually, what we trying to achieve is very similar to the problem of dimmer: one seeks to detect the zero crossing of the wave sector, in order to filter it and pass only certain vibrations. Thus, one can obtain an apparent frequency lower than that of the sector and will slow a flow rate of the pump. What we are interested in applying to the steering pump Ulka is the description of the detection of zero crossing of the wave area, and incidentally, the power supply without transformer

Assembly for the detection of zero passes in the sector from the application note from Microchip

The detection of zero crossing of AC is bringing the sector, via a resistor 20MΩ directly on an input pin of the peak.As the voltage is between 0 and 5V, the input is 0, and when 5V are exceeded, it switches to 1. The slope of the sine wave is very strong close to zero, in practice the 5V are located very close to zero. However, in doing so, we have an assembly that is not electrically isolated from the 230V, which requires to take all normal precautions when working with mains voltage.

The second very interesting thing in this application note is a description of the transformerless power supply. This is a montage that can power a 5V circuit, from the sector, with only a few inexpensive components, in particular, without a transformer, expensive and cumbersome. Obviously, with this type of power supply, no isolation of the lower voltage. By changing the values ​​of the note from Microchip, we can expect from 30mA without too much trouble, which is enough to power an LCD display, even backlit. If this type of power supply you are interested, Microchip has published a background paper on the subject with all the required equation for the size.

So the PIC will modulate the operating speed of the pump. After calibration, the software deducts the amount of product corresponding to a movement of the piston. Depending on the dose applied daily, it calculates the number of injections required and the interval for the dose is distributed so as linear as possible. To further simplify installation, the clock of the PIC is based on the frequency of the area that is a value extremely well controlled by EDF.

Case in point: we wish to distribute 1L per day. The pump has a flow rate of 250 ml / minute at 50Hz or 83μl by piston movement. Therefore 12,000 strokes for 1L or an injection every 7s. The PIC will then move to count the O sector to permit only rotation of the sine wave, all 7s.

It is worth noting that this circuit is perfectly capable of driving a pump much more hairy than small NME4 15W, a 50W large EP5 pump for example works perfectly without needing to change anything in the circuit. The only thing would probably be the appropriate range. Obviously, it will find a package suitable for each new type of pump.

Realization

Warning:
The assembly described in this article is not isolated from the mains voltage and is thus a mortal danger.Do you run this realization that if you master the concept of security associated with this type of circuit.Under no circumstances should you be able to enter into direct contact with an electrical conductor connected to this assembly when it is plugged. All must be enclosed in an insulated box, including the metal frame of the LCD.

You can download the project archive which contains the necessary file: the list of components with reference to purchase from Gotronic . The artwork, a drilling template and the program for the PIC.

Etching the printed circuit poses no particular problem. The artwork is available as a PDF in the archive project .It’s simple face, the tracks are wide. The welding is not a particular problem, except perhaps the assembly of two resistor and capacitor in CMS. I regret it now, but the code was developed on a proprietary platform Mikroelectronika pay with a compiler. I could always provide the code, but it’s a safe bet that you can not do much.

To prepare the cabinet must make holes on each side to pass the inputs and outputs of the pumps, and also cut a hole for the LCD. You will find a drilling template in the archive of the project.Another important point, the LCD must be covered by glass. On one hand it prevents any splash water entering the box, and secondly, it must be remembered that the assembly sector is not insulated, so it includes the metal frame the LCD. It takes two momentary switch. Be chosen only model with no metal parts accessible to the outside.

Finally, connect the pump to the circuit, be careful, because the control module allows only a half-wave impulse sector and there is a diode in series with the coil, the sense of connection its importance. If you reverse the connection, it simply does not work. I think the most painful is to connect the LCD to the circuit with the water. I would much prefer a solution or the display would have been welded directly to the circuit, but I’m unable to get something small enough for this.

I have in the past, made ​​for other devices. I no longer wish to do so now because it takes me a long time, as well as construction monitoring thereafter. Conversely, if you are motivated to make this assembly, I can provide the printed circuit board and the PIC programmed. I still have a little inventory of components, I have enough to prepare some kits for those who want to get. I have more cons by pumps in my boxes.

Feedback

Having done several of these for friends a few years ago, today I step back on it. Key points:

The circuit seems quite sensitive to electromagnetic disturbances. In particular, typically found in the aquarium ballast for lighting bac.La near the device may cause crashes dosing. Aside from that, the electronics are reliable and has never been found wanting. I had to repair some of the maps for welds that had dropped, but the design of the circuit is not in itself cause problems.

It seems that the pump flow varies over time. May be a phenomenon of running, or drift due to wear. As part of an aquarium use, we are not ready mL, and it does not matter much. For another use or the absolute value of the amount of product dispensed is important, I recommend to regularly check the calibration, using the test procedure in the software.

Over time, the PVC pipe that I use 6mm transparent on drying out and eventually cause leaks. In other words, a leak of corrosive liquid just above a 230V circuit is not a very good thing. To avoid the problem, I press a collar colson on the pipes the output of pumps.

The chlorine-based products are highly concentrated (balling method) used for the reef are very corrosive and over time they degrade the metal parts (spring / piston), however, stainless steel pump. These products must be diluted for use in the dispenser. Especially in thinner, it increases the volume, it is no longer working pump and injecting the product of more continuous. Given the low cost of pumps, having to replace the chlorine injection channel 1 once a year may not be so bad.