Low cost digital readout of velocity and flow

[Tony]

Yup, Motorola, and also Honeywell make a virtually identical product range.

The first problem is to decide on a suitable pressure range. These sensors all have a very small electrical output signal, so it is always best to use the most sensitive pressure range you can. But if too low a pressure range is chosen, the sensor may then be permanently damaged by an overpressure accident.

Second problem is that these sensors all change sensitivity with temperature, which is a definite problem if good consistent accuracy is required. Some type of effective temperature compensation is definitely required. But there are many different ways to go about solving that.

You can buy just a bare very low cost sensor, and build your own external temperature compensation. These types of sensors are very cheap, but compensating them properly can be very difficult to get right.

Internally temperature compensated sensors have some simple electronics buried inside, much better, but more costly too.

The ultimate is an internally temperature compensated sensor with an inbuilt amplifier. Automotive MAP sensors are like that. Beautiful ! but also too expensive to consider for a project like this. And anyway, a MAP sensor measures absolute pressure with only one input port, not differential pressure, which s what we require.

But a similar differential, fully calibrated, dual port sensor, would be perfect, but way out of our price range unfortunately.

So the choice of sensor type is pretty wide open. There are definite tradeoffs between cost and convenience. I rather like the Measurement Specialists Model 1220 transducer suggested by Slracer. It is simple and uses a very ingenious temperature compensation method that seems to be unique to them.

I have e-mailed Measurement Specialists for a quotation, see what they say.

[106-1194218389]

Tony,
I never saw any response to my post in this thread showing the transducer I found at

Is that something like we are looking for only way too expensive?

John

[Tony]

Sorry John.

That is certainly a very good quality transducer, but it has a 4-20mA current loop output which is a lot less convenient to use than a straight voltage output.

I am hoping to keep the total parts cost below $100 and still have very good performance.

[115-1172523331]

Hi Tony, One of their distributors is Digikey. They quote a $36.58 for 1 of the 1220 transducers with either 5 or 15 psi differential. They don't show a 2psi in stock, but the cost should be similar. (I would hope!) You can see what they have here;



This could get to be fun! -- Doug (aka slracer)

[Tony]

Once again a great find Doug. They only have one in stock, and it is 0-5psi but that will certainly do for starters. The price is right too.

Before I try to grab that, I will see if Measurement Specialists have an agent here in Australia.

The problem with ordering stuff from the US is the cost of currency conversion, electronic funds transfer by my bank, and freight will triple the price.

[86rocco1]

Tony,
Ideally, what voltage range would you like to see for your sensor?

[115-1172523331]

Ed, I couldn't find our previous correspondence. Is the pressure transducer you already have going to work for this?

Doug

[86rocco1]

No, I don't think so, it a bidirectional + 5" w.c sensor with a linear 5.0V output range.

[R Thompson]

I have been searching through this forum for a picture or some evaluation data of the PTS plans flow bench and have not found anything. Is there something I could look at, a prototype or something somewhere?

[106-1194218389]

In the Readers Flow Bench Projects there should be some pictures of what some readers are doing. Although you probably will not see the "plans" as most of us here have bought the plans from Bruce. This does 2 things, gives you an excellent set of plans to go on and helps Bruce pay for this forum. We really appreciate the expense and time Bruce has put into this. I have never found another site like this. I have built my bench off of the plans and it is very repeatable and super accurate. Buying a set of plans also gives you access to the plans discussion area which is closed to anybody who has not bought a set of plans. I will guarntee you will not be disappointed. These plans work! You will not find a commercial bench better.

John

[Tony]

[106-1194218389]

Here are some more pressure transducers. I do not know if they fit the bill. They have an international phone number. They have a millivolt output. ALso have temperature compensation.

[Tony]

I still rather like the Measurement Specialties 1220 transducer. The 2psi range is ideal, and the quoted 0-50C temperature error is only 0.3% which is outstandingly good. And it is lower cost. It also has an unusually high output signal voltage of 50mV at full scale pressure, which is also very desirable.

The Omega comes in 1psi and 5psi ranges, and the quoted 0-50C temperature error is 1.5% , and it costs more.

[115-1172523331]

Tony, When I was looking at the transducers on Measurement Specialties, I filled out their info form. Today, a rep contacted me and asked if I had found what I was looking for. I told him a bit about the forum and what you had proposed. I also told him we (you) wanted a 1220 @ 0-2psi, but the Digi-Key folks had only 0-5 and 0-15 psi models. He will check the factory tomorrow and email me, but is going to try to get "a couple of samples". You might want to wait a couple of days before ordering the 0-5 model! ####, I love the internet!

Doug

[Tony]

Unfortunately guys, I HAVE DISCOVERED A HUGE FUNDAMENTAL PROBLEM.

I was initially rather carried away by the whole idea. I tested my prototype display and it certainly appeared to work, in that it gave a smoothly rising output display that followed the rising input pressure. At first I only checked it at three points in the range, where I found the numbers to be exact. Later I did some more checking at many other pressure points, and discovered some very significant departures from a correct inverse square law function.

Now, I have seen these displays used in this way in the past, to display in non linear functions, including square law and logarithmic law, so I know it is possible to do. But I could not remember exactly how it was done.

I have spent a whole day researching this, and have found nothing either in my library or on the internet about generating specific non linear transfer functions with these chips. But slowly the recollection is coming back to me after a night time of dreaming about the problem.

While the main input, and reference input can be used together to generate some non linear functions, it is not the only method. External current can be sourced or sunk into the internal integrating capacitor, introducing deliberate distortion into the timing ramp. This is highly complex, but that is how I now remember these highly non linear systems worked, although the exact details still escape me.

This method can only really be used to develop an original digital display module design from scratch, where you have full control over all the internal component values used.

This goes way beyond being a simple to build open public project, where a low cost commercial module (of uncertain origin) can just be hooked up and expected to work properly.

Sorry guys, but this unfortunate discovery has really killed the whole idea.

It all becomes just too difficult and complicated. Much better to build a software based system, where real numbers can be crunched in software, and the result will be reliable.

This has all been rather embarrassing.

It is still entirely possible to build a direct reading digital manometer with a liner readout of pressure, but converting this to velocity or flow directly in analog is not as simple as I had at first hoped.

So back to plan "A" of building a vertical water manometer with a scale calibrated in feet per second velocity.