Ability of transducers to produce sufficiently amplified signal

Question

Transducers are responsible for conversion from one a form of natural signal into electrical signal. Example - heat variations into electric voltage variations.

In the book Microelectronic Circuits by Sedra and Smith, it is mentioned that transducers produce weak signals that are susceptible to noise and can get corrupted easily and that is the reason we need to amplify the signal obtained from a transducer before we can do further processing on the signal.

Why can't we build transducers that do not produce weak signals?

Answer 1

While there are certainly transducers (a better definition of which would be something that converts one form of energy to another form of energy) that can produce large signals (like a thermistor or force sensitive resistor), for others the physics involved just doesn't permit it (for example, a wire strain gage, which uses a teeny tiny change in the geometry of a wire to generate a signal).

The real answer to your question, though, is that in most cases, amplification is cheaper and easier than creating a transducer with higher sensitivity.

Answer 2

The energy in the electrical signals initially comes from the ‘natural’ signal. It would in many cases be possible to capture a larger amount of energy e.g. with a large horn on a microphone or a massive thermopile, but for most purposes we want a small and inexpensive device.

Answer 3

Why can't we build transducers that do not produce weak signals?

Well, we could; by adding a signal amplifier (aka another circuit that increases errors). So, does a transducer manufacturer, on reading this question do any of the following: -

• Add a high performance amplifier that might suit 10% of the market
• Add a low performance amplifier that might suit 40% of the market
• Manufacture a range of transducers with different amplifiers/specifications and cost
• Do nothing and rely on the purchaser to know what to do to suit their target circuit

Other things to consider are, if the transducer has a built-in amplifier and, that amplifier's tech gets superseded 6 months later, do they bring out a newer model. What happens if the amplifier suffers from lead-time issues; what should the transducer manufacturer do now? What if the added amplifier becomes obsolete?

Answer 4

In some cases you can do that, but often better technology just leads to weaker signals. For example, the microphone capsule that is in a 1960s "black" phone is sensitive and requires no amplifier, but is large and expensive to manufacture. An electret microphone has a signal that is so weak it needs the amplifier built into the capsule but it's cheap, very small and quite good. Ribbon microphones are very, very good but have extremely low output voltage (fortunately also with a low source impedance).

If you want to get a temperature signal that is mV/degree rather than uV/degree you could build a thermopile with 1,000 junctions but it would suck 1000x the heat away from whatever it is that you are trying to measure (all other things being equal), so it would be a terrible thermal sensor. Even before amplifiers were practical, the output of a single junction was typically used for measurement and control purposes (using finely balanced galvanometers).

A photoconductive cell has large output but it contains cadmium and is undesirable for that reason (and others). A photodiode has a weak output, but is cheap to manufacture, small, easily amplified and is faster and far more stable.

At some point you may run into the limits of physics such as Johnson-Nyquist noise (more generally the fundamental fluctuation-dissipation theorem) - the signal is so weak that you cannot achieve the required signal to noise ratio even with the finest amplifier practical or even possible, and in that case you may have to make the sensor larger and heavier to achieve the required performance.. for example with accelerometers and gyros. The tiny ones tend to be rather noisy and drifty.

So sometimes you have to make the transducer big and heavy and expensive (see LIGO), for example, but if you can, economics tends to lead to making the transducer small and cheap and less sensitive and adding an amplifier. More so as amplifiers tend to get better and cheaper over time and material costs tend to rise over time. An engineer that uses a $100 more expensive transducer to avoid a 5 cent amplifier for no good reason won't likely be long employed. As one data point, a low-drift DC amplifier from 1972 can be outperformed by couple orders of magnitude in a device that today costs a couple orders of magnitude less in real terms (~10,000:1 improvement in price/performance).

Answer 5

Whatever you're wanting to measure, and whatever your measurement system is, a larger signal requires a larger force. But we never want our measurement process to affect what we're measuring, so this is generally the opposite of what we want.

In many situations too, producing a larger signal would need a physically larger (and possibly more expensive) transducer. This could become large enough to be impractical within the system, or could be too expensive to be used. As an extra complication too, a larger transducer will typically be slower to respond (due to inertia, thermal lag, or other causes), and this will affect the quality of our measurements.

Electronic amplifiers are cheap and reliable, and enable us to use transducers which have minimal impact on what we're measuring and give us fast, accurate responses. Of course dealing with weak transducer signals is a complication, but methods of processing weak signals are very well understood and highly mature, having barely changed for 40-50 years. In general, this makes it a better solution to the problem.

Answer 6

The other answers are all good. Another factor to consider is the bandwidth of the transducer. As an example, take a hydrophone which converts acoustic energy into electrical energy. Most hydrophones are relatively small because most applications require significant bandwidth and omnidirectional response. This, however, results in a low conversion efficiency. If you are willing to give up a wide bandwidth and/or a near omnidirectional response and if size and weight are not an issue, then you can build a larger hydrophone to increase the sensitivity. This is probably true of other sensors also where sensitivity and size must be traded off.

Answer 7

Suppose you want to count visible photons using a microcontroller. The energy in a logic pulse is about 100 million times the energy in a photon. So, you'll likely to want to use something like an avalanche photodiode as a sensor. That has built-in amplification, maybe 1000x in energy. You're still a factor of 100,000 short of a logic pulse. So, an APD measurement chain will have several more stages of amplification.