I am planning to use an instrument that gives a 4-20 mA analog output. I am not sure whether to use a SMPS based power source (which is cheaper) or a regulated power source for the DC power supply? Any ideas?

Regards, Abhishek

  • \$\begingroup\$ Your terminology is wrong. A SMPS means switch-mode power supply. A SMPS does "regulate" power. SMPS are often not cheaper, they are more complex than a linear (non switching) regulator. How are you even generating your 4-20mA loop? Op-amps? Why don't you read about what the op-amps need in terms of power. \$\endgroup\$ – KyranF Oct 29 '15 at 15:28
  • \$\begingroup\$ I searched on the internet, a DC regulated power supply is generally more costly than an SMPS based power supply. I am not designing the transducer that generates the 4-20 mA signal, rather I have an instrument that gives its output as 4-20 mA and an ADC card. I am just debating which type of power supply to use. \$\endgroup\$ – Abhishek Oct 29 '15 at 15:32
  • \$\begingroup\$ They would be more costly because for the same amount of power needed to supply, the efficiency is extremely low for a linear regulator. Low efficiency means high heat \$\endgroup\$ – KyranF Oct 29 '15 at 15:49
  • \$\begingroup\$ High heat means components must be larger, higher ratings, more material, heatsinks and fans everywhere! \$\endgroup\$ – KyranF Oct 29 '15 at 17:26
  • \$\begingroup\$ Are you planning to use a passive 4-20 mA device (you only have 2 wires which are supply voltage and current readback at the same time) or an active 4-20 mA device (you have a separate supply and an active output drive, so it uses a 3 or 4 wire configuration)? The answer is depending on that, mine was assuming a 2-wire device. \$\endgroup\$ – Arsenal Oct 30 '15 at 15:45

For best measurement accuracy you want to use a linear regulator as the last step to power your sensor.

So to get from line voltage to a usable level (say 26V or something) you'd use a switchmode power supply because of the efficiency. After that you place a linear regulator which handles the dropping to a clean 24V.

You should also protect against all thinkable errors and EMI problems. So you make sure that an external short will not damage any of your components (current limiting). As some operator will be handling all the wires and stuff, you place some protection against ESD at the out/input. For real heavy duty you might even need protection against lightning strikes.

Another thing in industrial environments that your device might be used in explosive environments, which puts a whole new level of requirements on your circuit. This is getting much to broad to answer all the "industry standard".

What you are trying to do reads more like a small job which doesn't require all the heavy duty industry stuff and probably not sub 0.5% accuracy. 4-20 mA sensors are usually built to withstand quite a lot EMI, so powering them with some cheap switchmode wall-plug adapter actually works reasonably well (at least those we build do).

The problem with switchmode power supplies is the voltage ripple, which prevents the current control stage from regulating to a clean current value as it is likely slower (in the order of a few kHz) than the switching frequency of the power supply (could be anywhere between a few kHz up to a few MHz).

Or you just buy an industrial power supply for 4...20 mA sensors. But actually no one sells you just that, typically they come with some circuit to measure the current already (that's what the industry wants), either in a PLC or a standalone device.

  • \$\begingroup\$ "Or you just buy an industrial power supply for 4...20 mA sensors." I think that is what the OP is trying to do, but doesn't know which type to buy. \$\endgroup\$ – KyranF Oct 29 '15 at 17:27
  • \$\begingroup\$ @KyranF well the problem is, that no one sells you just a power supply, it's integrated in the device for measuring the current typically. I've updated my answer to be less misleading in that regard. \$\endgroup\$ – Arsenal Oct 29 '15 at 17:40
  • \$\begingroup\$ If I was designing a product which had to output 4-20mA, I would design my on-board power circuit well enough to deal with industry standard power supplies (Perhaps based off the ATX PSU standard as a benchmark of what to expect on the input) and provide all the necessary filtering and linear converters myself - to basically guarantee a good quality 4-20mA output accuracy of the sensor. Industrial devices are designed to be idiot-proof, as long as you follow whatever simple guidelines are listed in the datasheets. \$\endgroup\$ – KyranF Oct 29 '15 at 17:55
  • \$\begingroup\$ @KyranF maybe we think of two different things. A 4..20 mA loop implies for me a passive sensor, so it takes all the energy it needs just from the loop. It's a 2-wire device where the supply is the output at the same time, so you have not control over all parameters and a large ripple is a cause for reduced current accuracy. Don't get me wrong, the input stage and current regulation of the sensors are very carefully designed to be as robust and precise as possible, but there are limitations. If you are looking for 0.05% accuracy (<8 µA) you better put some effort in the supply as well. \$\endgroup\$ – Arsenal Oct 29 '15 at 21:57
  • \$\begingroup\$ The ATX PSU standard provides way too little disturbances for industrial designs. The sensors must be able to handle at least 0.7 V "ripple" for 16 to 400 Hz (let's say ours are able to do that). \$\endgroup\$ – Arsenal Oct 29 '15 at 22:07

Either will do. If you use a switching supply, make sure to put some capacitance on the supply pin of your instrument, to get rid of leftover switching noise from the supply. You'll want both a largish electrolytic (tantalum for choice) and a smaller (~0.1 uF) ceramic to handle all frequencies. A small resistor in series will also be a good idea. Depending on the details of your instrument, filtering may already be incorporated, but a little more never hurts.

  • \$\begingroup\$ hmm... so does it mean a regulated DC power source is worth the extra cost ... rather than fiddle around with loose capacitors? \$\endgroup\$ – Abhishek Oct 29 '15 at 15:35
  • \$\begingroup\$ @Abhishek - There is no way to tell without knowing how much noise the smps produces, how much noise filtering is built into the instrument, and how sensitive the receiver is to what noise may be carried on the signal. In theory, a current loop ignores noise pickup, but there's a limit to real-world noise rejection. Sorry, but there's no way to tell at this point. \$\endgroup\$ – WhatRoughBeast Oct 29 '15 at 15:38
  • \$\begingroup\$ thanks! ... but I would like to know what is the "industry standard" practice for powering a 4-20 mA loop in the feild? \$\endgroup\$ – Abhishek Oct 29 '15 at 15:41
  • 1
    \$\begingroup\$ To the OP, please stop using incorrect terminology. Both SMPS and linear supplies are normally regulated. The noise from a SMPS can be reduced arbitrarily low. Most likely, you will be fine with an SMPS. Another option is to use an SMPS slightly higher than the voltage you actually need, and then use a linear regulator to generate the final voltage. The linear regulator in this case would be acting both as a regulator and a noise filter. \$\endgroup\$ – mkeith Oct 29 '15 at 16:24
  • \$\begingroup\$ @Abhishek the industry standard in the field would be to buy a device which is was specifically built to do that. \$\endgroup\$ – Arsenal Oct 29 '15 at 16:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.