Your Electronics Open Source

The following article is an example of how even an electronics project with a requirement rather simple, needs system considerations and not just circuit ones.

In practice, it is often necessary to measure signal voltage. Each of us can imagine many ways to archive that. The methods various from analog systems to digital ones, various in complexities and performances.

But if we add more requirements to the initial one ( the voltages should be measured at more than one taps and the isolation is mandatory) then the possible solutions number is decreasing. Of course it can be still trivial, I can put a transformer and that's it. But if the voltage is continuous?

So it is still a simple one, but we have to start to look for some particular components to help us.

So we have to sort out some problems hereafter:

• the first, quite trivial, is that of the analog/digital conversion (today it's not worth to think only to the analog architectures, is always better to convert to digital and then work)

• The second one is isolation.

The A/D conversion is not a worrying problem now-a-days, you just need to choose the most suitable converter (sometimes it involves more logical considerations such as costs, supply times, the component knowledge, more than technical issues).

For isolation we have to try something more valuable or particular.

The first architecture that come into mind is:

1. Take the signal, convert it and through, for example, a digital serial line, transmit it to a processor and do here the case elaborations.

2. Take the signal, match it through an opto-isolator and then convert the voltage so obtained in a digital one following then the previous "step".

3. Take the signal, convert it into frequency and transmit the obtained signal on a line galvanically separated from the others (it depends on transformer or digital opto-isolators capacity).

4. Or any other architecture that doesn't come to mind right now

But now let's see:

1. The important points of the first presumption are the facility to isolate some simple digital lines. We must realize a small acquisition function for each voltage to be measured and than a simple serial line. But, perhaps it is absorbed too much, or it is required that the voltage is stable or the resultant cost is too high both economically and in terms of space and complexity of the circuit.

2. The second solution is the easiest one, it doesn't cost much and it doesn't have technical problems. But we know that the opto aren't very stable (their parameters vary with the temperature, with the oldness or vary one from each other and then down with the calibration potentiometers, the procedures¡­), in few words there is a problem if we have insight.

3. The F/V conversion and therefore V/F was done a few years ago (at the dawn of electronics our blessed fathers projected complex circuits to resolve a problem like this). Today not a few companies continue to produce this type of converters (Analog Devices, maybe National but then...).

And now, let's take another look to the solution N.2, this has a critical point but maybe is just one (perhaps the technology comes into help) because once we manage to take the signal the problem becomes very easy to be solved.

Maybe you can use the isolated amplifiers rather than an opto (there are excellent from TI coming from ex Burr-Brown), but these components request two carrying tensions and therefore aren't easily applicable, yet it would be much more simple to use an opto

Hewelet-Packard today Agilent opto section and there it is our linear optoisolator HCNR201, see fig.1 (Check datasheet for details), but it is strange as it has two outputs, that is the input signal applied on the LED and transformed into light, it is sent on two PINs. A look at some applications notes and you understand that Agilent suggests a circuit where those who “transmit” rereads the converted voltage (through one of the two PIN) and reactivate the input so that the proportionality between input and output (on the other branch, the isolated one) is constant. The circuit proposed is quite complex (but not too much) however requires stability and precision typical for an analog circuit measurement perhaps we're not there yet.

Fig. 1 HCNR201

The first sensation that we got is that the criticality basically remained.

Let's summarize, different architectures, different solutions each one with its own problem and its advantages, in these conditions we are sure that "Aren't we ignoring a facet?"

Let's make these simple considerations, the variations of the parameters of the presumed opto are very slow because are tied to some phenomenas as temperature, oldness, or tied to the dispersion of characteristics.

A solution could be to select the best components, choosing those whose parameters return in a range sufficient for our purposes, but probably our boss will object because this solution is expensive and we don't have well-known results.

The suggestion that will be illustrated (it has been realized and is working) solves these problems easily with a system architecture reliable and repetitive.

The solution is remade to that what we are doing in the daily practice, to know if the circuit on which we are working is functioning correctly or not, we send known signals and we control the output. But we can do more if the output moved away from the values that we expect, we can correct them by compensating the introduced drift (the perfectionists might say that is necessary to know what relation is between the variations of parameters and the output, but in the first approximation we can assume a linear dependence, if you want, you can change the correction function until you reach the desired result).

Well we are arrived at the end of the test, our circuit will be formed (see fig. 2) by:

• reception block of the signal to be measured (composed from filters, protection of disturbs, possible amplification or reduction of the dynamic)

• propagation block of the reference signal (normally a voltage reference)

• Switch between the two signals (analogical switch with low impedance)

• Adjustment of the signal to our opto

• Signal reception and conversion of the same in numerical

Fig. 2 Insulated Voltages Acquisition Block Diagram

The HW is finished and now that the signal is converted, any processor can apply all the corrections that we deem to do.

The advantages of this type of solution are the absence of criticisms circuits parts (tolerances, stability) the price and the space on the PCB content.

The disadvantages may be a certain quantity of SW to design (but remember that you have to pay for the SW only once while the HW is a repetitive cost).

Conclusions

All this highlights how requisites even simple can hide some dangers and how is necessary to confront the problem from more points of view and therefore to acquire a technical awareness not necessarily hyper-specialized but as "round" as possible.

Finally I have to say that the circuit was developed and is operating reasonable.

On a Eurocard board (100x160 mm) it has been put a circuit which measured the voltage, the current and the frequency of 16 signals, took control decisions as shutting down or modifying the sensing of power supplies (which produced the 16 tensions). Provided also controls and reports for all these voltages. The circuit was also not even too "full".

As a last note, the fig. 3 shows the schematic of one of the "n" sections implemented, it should be noted that the A/D converter was inside the micro (a MCS9 Freescale).

Fig. 3 Voltage measurement Schematic Drawing