A digital panel meter is an instrument built to be cut into the front of a control cabinet. It reads a sensor, shows the result in engineering units, and — the part that matters most — does something about it: closes a relay, drives an analog output, answers a controller over a network. It is not a display with a screen. It is a piece of the machine.
Almost everything unusual on this specification sheet follows from where the instrument lives. A hand-held indicator is held by a person in whatever conditions that person is standing in. A panel meter is bolted permanently into a steel enclosure that also contains contactors, motor starters and quite possibly a variable frequency drive, wired to a load cell that may be a long way off on a machine whose ground is not at the same potential as the panel's ground. Nobody is watching it. It has to be right anyway.
So the noise rejection figures look excessive until you know that. Common mode rejection is specified at 130 dB from DC to 60 Hz. Expressed as a ratio rather than in decibels, that is better than three million to one — a worked textbook example of a differential amplifier rejecting interference typically comes out around a thousand to one. The instrument will withstand 250 volts AC of common mode voltage, and at 130 dB that arrives at the measuring circuit as roughly eighty microvolts. Published guidance is blunt about where such interference comes from: in North America most common-mode noise sits at the 60 Hz power line frequency, and in Europe and elsewhere at 50 Hz. It is the mains, and in a control cabinet the mains is everywhere.
The read rate is the other half of the same defence, and it is not a speed claim. The meter converts 60 times per second on 60 Hz supplies and 50 times per second on 50 Hz supplies. Those are the same number as the mains frequency, which means each conversion integrates over exactly one power line cycle. Published description of how integrating converters behave explains why that is done: if the integration lasts exactly time T, all frequencies at whole multiples of 1/T are theoretically rejected completely. Set T to one mains cycle and the mains cancels itself out. That is where the 90 dB normal mode rejection at 50/60 Hz comes from, and it is the reason a 50 Hz version exists at all rather than being an inconvenience of export paperwork.
And it is expected to act, not just inform. The relay boards carry 8 amps at 240 volts AC on changeover contacts, which is a contactor coil, a solenoid valve, a warning beacon or a motor starter — real switching rather than signalling. An analog output board can drive a controller directly. Communication boards will answer Modbus over a serial line, over Ethernet, or over WiFi. A meter that only displayed a number would need none of that.
The last thing worth understanding is that it is not really one product. The base meter is a chassis with slots in it. Power supply, relays, analog output and communications all arrive as plug-in boards that can be fitted in the field, which means the specification you buy is not the specification you are stuck with. Machines outlive the requirements they were built to, and this instrument is designed around that fact.
Excitation goes out to the sensor, millivolts come back, the meter converts them and applies a scale factor. What is worth understanding is how the choices you make at installation change the quality of what you get.
There are five input ranges and they are not all equal. The meter accepts 20, 50, 100, 250 or 500 millivolts full scale. The two smallest resolve to one microvolt; the three largest resolve to ten. So selecting the smallest range your signal actually fits inside buys you ten times the resolution, for nothing, and selecting a range far larger than you need throws it away.
Here the arithmetic is unusually tidy. A 2 mV/V load cell excited at 10 volts produces 20 millivolts at capacity — which is exactly the most sensitive range on the meter, resolved at a microvolt, giving twenty thousand counts across the sensor's working range. The commonest sensor in our catalogue and the most sensitive setting on this instrument are a precise fit for each other, and that is not a coincidence.
Excitation is jumper selectable, and the current limits differ far more than the voltages do. Five volts is rated to 100 mA, ten volts to 120 mA, twenty-four volts to only 40 mA. That last figure deserves attention, because a 350 ohm bridge across 24 volts asks for close to 70 mA and the rail will not give it. The 24 volt setting is for higher impedance sensors — 700 ohms and upward — or for powering a transmitter, not for a standard 350 ohm load cell. Ten volts is the setting most of our sensors are specified at and it is where the numbers work out best.
Higher excitation is also free resolution. Sensitivity is quoted per volt because output is a proportion of the supply, so the same load cell run at ten volts produces twice the signal it produces at five, which is twice as many counts for the same load. Where the sensor and the wiring permit it, ten volts is the better choice on both counts.
One supply can feed several sensors, within a budget you can calculate. Four 350 ohm bridges wired in parallel present about 87 ohms, which at ten volts draws roughly 114 milliamps — just inside the 120 milliamp rating, and precisely why that rating is the number it is. That is a platform, a hopper or a tank on four mounts, summed into one meter. Add a fifth and the arithmetic stops working, so count the bridges and check the current before designing the system rather than after.
Three different update rates appear on the specification, and all three are correct. The converter runs at the mains frequency. The relays and the analog output update fifty-six times a second. The display refreshes about three and a half times a second. Those are not inconsistent; they are three different consumers being served properly. A machine needs a setpoint to act within a few milliseconds. A person needs a number that sits still long enough to be read, and a display flickering sixty times a second would be useless. Control speed and reading speed are different problems and the instrument solves them separately.
The filter adapts rather than being set once. An adaptive digital filter selects its own time constant, holding heavy averaging while a signal is quiet and releasing it when the signal genuinely moves, so you are not forced to trade a steady display against a responsive one. That is the compromise most instruments make you choose in a menu.
Specifying one of these is mostly a matter of listing what has to happen when the reading reaches a number, and then choosing boards to suit. Our applications engineers will happily go through a panel schedule with you.
Start with what has to act, and how many thresholds there are. Two setpoints or four, and switching what. If the meter has to operate a contactor, a solenoid or a beacon, the changeover relay boards carry real current. If it is feeding a logic input on a PLC or a small indicator, the solid state boards are quieter, faster and have nothing to wear out. The choice is genuinely about the load on the other side, so find that out before ordering rather than guessing.
Then decide whether anything else needs the number. An analog output board turns the reading into 4–20 mA or a voltage for a controller, a chart recorder or a data acquisition system, at sixteen bit resolution. A communication board lets a supervisory system ask for the value instead: plain ASCII or Modbus, over RS232, RS485, USB, Ethernet or WiFi. Serial is the dependable choice inside a cabinet; Ethernet suits a plant network; WiFi is the answer when running a cable to the machine is the expensive part, and it comes with a built-in or an external antenna depending on how much steel is in the way.
Match the power supply board to the panel, not to habit. There is a mains version and a low voltage DC version. Machines with a 24 volt control bus are common and the DC variant saves bringing line voltage to the meter, which simplifies the panel and the paperwork.
Order the boards you need now, not the boards you might need. Because they are field installable, the usual argument for over-specifying does not apply here. If a supervisory system arrives in two years, the meter takes a board rather than being replaced — so buy what the job needs today and keep the option.
Specify TEDS on the sensors if there will be more than one of them. With a Cal-Teds plug and play sensor the meter reads the calibration off the transducer and configures itself, and a front panel indicator confirms it has done so. On a machine with a single permanently installed load cell that matters less. In a facility that swaps sensors between machines, or recalibrates and refits them, it removes the step where the right sensor gets the wrong scale factor.
Think about the front of the panel as well as the back. Sealed to NEMA-4X from the front once it is properly mounted, so wash-down and splashing are accounted for by design, and a splash proof cover is available where the environment is harsher still. The operating range runs from freezing to well above normal ambient, which covers most machine enclosures but not all of them — a cabinet in direct sun or beside an oven can run hotter than people expect, and that is worth measuring rather than assuming.
Send us the sensor you are using, the thresholds that have to act, what they are switching and whether anything upstream needs the data, and we will put together the board list. Tell us if the panel is a wet one.
DPM-3 Digital Panel Meter Applications.
The Transducer Techniques DPM-3 is a smart plug and play TEDS IEEE 1451.4 compliant digital panel mount load cell meter that scales to a full five digit display, from −99,999 to +99,999 counts, reading directly in engineering units such as grams, ounces and pounds.
- Machine Control: Setpoint relays act on force, load or torque thresholds directly at the machine, without waiting on a supervisory system.
- Batching and Filling: Multiple setpoints handle a fast fill, a dribble feed and a cut-off, with the meter driving the valves itself.
- Tank and Hopper Weighing: A single meter excites and sums up to four 350 ohm load cells in parallel, reading total contents from a vessel on multiple mounts.
- Press and Crimp Monitoring: Peak capture records the maximum force in each cycle and a setpoint flags any cycle that falls outside limits.
- Test Stands and Rigs: Live readout at the rig with an analog output feeding a recorder or data acquisition system at the same time.
- Overload Protection: A relay trips a drive, a brake or an alarm before a load reaches a level that would damage the machine or its tooling.
- Tension Control: Continuous web, wire or cable tension displayed and regulated through the analog output or setpoint outputs.
- Packaging and Conveyor Lines: Check weighing and reject signalling at line speed, with relay response measured in milliseconds.
- SCADA and PLC Integration: Modbus over serial, Ethernet or WiFi allows a supervisory system to read the value and the setpoint status.
- Remote and Awkward Locations: WiFi and Ethernet boards bring readings back from equipment where pulling a signal cable would be the expensive part of the job.
The DPM-3 works with our full range of load cells, torque sensors and pressure transducers, and its plug-in option boards are field installable so a meter can be reconfigured as an application changes.
Frequently Asked Questions
What does the base DPM-3 meter include, and what arrives as an option board?
The base meter is the chassis, the display, the converter and the sensor excitation — everything needed to read a load cell and show the result in engineering units, with front panel setup and automatic scaling from a TEDS sensor. What arrives on plug-in boards is everything that connects the meter to the rest of the world: relays, analog output, serial, Ethernet and WiFi communications, and an alternative power supply. The boards are field installable, so the configuration is not fixed at the moment of purchase. That is a deliberate design decision and it changes how you should specify one.
Which input range should I select, and does it affect resolution?
It affects it by a factor of ten. The 20 and 50 millivolt ranges resolve to one microvolt; the 100, 250 and 500 millivolt ranges resolve to ten. So the rule is to choose the smallest range your full scale signal fits inside. Work out what that signal is before selecting: sensitivity multiplied by excitation voltage. A 2 mV/V sensor at 10 volts gives 20 millivolts, which lands exactly on the most sensitive range and yields twenty thousand counts across the sensor's working range. Selecting 500 millivolts for that same sensor would still work and would throw away most of the resolution you paid for.
What excitation voltages are available, and how much current does each supply?
Five, ten or twenty-four volts DC, selected by jumper, isolated from the power and output circuits. The current available differs considerably: 100 mA at five volts, 120 mA at ten, and only 40 mA at twenty-four. That last limit is the one to check against your sensor rather than assume, because the highest voltage setting supplies the least current and the two facts together decide what you can actually connect. Ten volts is where most of our load cells and torque sensors are specified and where the resolution works out best.
How many load cells can the DPM-3 excite at once?
Four 350 ohm bridges in parallel at ten volts, which is a calculation rather than a rule of thumb and you can repeat it for your own sensors. Four 350 ohm bridges in parallel present about 87 ohms; ten volts across 87 ohms draws roughly 114 milliamps, which sits just inside the 120 milliamp rating. Higher resistance bridges allow more of them, lower resistance fewer. If your arrangement needs more current than the meter can supply, the answer is a separate excitation supply rather than accepting a sagging one, because excitation that droops under load makes every reading wrong in a way that looks like a calibration error.
How accurate is the DPM-3?
0.01 percent of full scale plus two counts, with zero drifting by 0.2 microvolts per degree Celsius and span by 0.0015 percent of reading per degree. Those are laboratory figures on an instrument designed to live in a machine cabinet, and in practice the meter is very unlikely to be the limiting element in your measurement — the load cell, the mounting and the mechanical arrangement will all contribute more uncertainty than this does. The two counts matter mainly at the very bottom of a range, which is another argument for choosing the smallest input range your signal fits.
What does the meter do about electrical noise in a control cabinet?
Two separate defences. Common mode rejection of 130 dB — better than three million to one — deals with interference that appears equally on both signal wires, which is what a difference in ground potential between the sensor's location and the panel produces. The meter withstands 250 volts AC of it. Separately, because each conversion integrates over exactly one power line cycle, interference at the mains frequency and its harmonics largely cancels within the measurement itself, which is where the 90 dB normal mode rejection figure comes from. Between them they cover the two ways mains noise gets into a low level signal.
What can the relay boards switch, and how quickly do they respond?
The changeover relay boards are rated at 8 amps, at 240 volts AC or 24 volts DC, in sets of two or four. That is enough to operate a contactor coil, a solenoid valve, a warning beacon or a small motor starter directly, without an interposing relay. The solid state boards handle 120 milliamps at higher voltages and suit logic inputs and indicators. Either way the outputs update fifty-six times a second, so a setpoint acts within about twenty milliseconds of the reading crossing it — considerably faster than the display refreshes, which is deliberate. There is also a programmable delay of one to 128 readings if you want the meter to be sure before it acts.
What does the analog output board give me?
A choice of 0–20 mA, 4–20 mA, 0–10 V or ±10 V, at sixteen bit resolution, tracking the meter's own accuracy to within a further 0.02 percent of full scale. Current output will drive up to 600 ohms of loop, voltage output wants 5 kilohms or more. In practice this is how the meter talks to a controller, a chart recorder or a data acquisition system that expects a process signal rather than a digital message, and 4–20 mA remains the sensible default over any distance because a broken wire reads as zero current rather than as a plausible number.
Which communication option should I choose?
Work backwards from what is asking for the data. RS232 is simple and short range, one meter to one device. RS485 suits several meters sharing a line over a long distance in a plant, and is the usual choice for a Modbus network. USB is convenient for configuration and for a PC alongside the panel. Ethernet puts the meter on a plant network. WiFi exists for the case where the cable is the problem, and comes with an internal antenna or an external one for cabinets with enough steel to block the signal. Several boards combine two of these. All of them speak plain ASCII or Modbus in RTU or ASCII form, so most supervisory software will talk to them without anything custom being written.
Do I have to configure the meter by hand?
Only if the sensor cannot tell it what to do. Connect a TEDS sensor and the meter reads the calibration data out of it and scales itself, confirming with a front panel indicator that it has. For a sensor without that memory, setup happens through the front panel buttons, or from a PC using free downloadable software if a communication board is fitted — which is the easier route when several meters need the same configuration, since it can be prepared once and sent to each of them.
Questions From The Field
Can I use the 24 volt excitation setting with my 350 ohm load cell?
Check the current before you do, because the arithmetic says no. Twenty-four volts across a 350 ohm bridge asks for nearly 70 milliamps and that rail is rated for 40, so the supply is being asked for more than it can give. The 24 volt setting is there for higher impedance sensors — 700 ohms and upward — and for powering transmitters. For a standard 350 ohm load cell, ten volts is the correct choice, and it also gives you more signal and better resolution than five volts would. If you have inherited a panel wired for 24 volts and a 350 ohm cell, that is worth investigating rather than leaving alone.
My reading jumps whenever a motor or contactor switches in the same panel.
The meter's rejection is designed for exactly this, so a visible jump usually means something in the installation is defeating it. Check that the sensor cable runs separately from motor leads and contactor wiring rather than in the same tray or duct — crossing at right angles is fine, running in parallel for any distance is not. Check the shield is grounded at one end only, so it is not carrying current between two grounds. Check the excitation and signal wiring is the twisted, shielded type intended for the job. If the disturbance survives all of that, suppression across the offending coil is usually the answer, and it is cheaper than anything else you might try.
The display seems slow for something that reads sixty times a second.
It is meant to be, and the fast part is still working. The converter runs at the mains frequency, the outputs and relays update fifty-six times a second, and only the display refreshes at about three and a half. A number changing sixty times a second cannot be read by a person, so the display is deliberately the slowest thing in the instrument while control response stays fast. If you need to see a fast event rather than control on it, use the peak or valley capture, which follows the input at full speed and holds the extreme for you to read at leisure.
My setpoint chatters when the reading sits right on the threshold.
Normal behaviour for any threshold, and there is a setting for it. The output time delay can be set from one to 128 readings, which requires the condition to persist before the relay acts and is the usual cure — at fifty-six updates per second even a substantial delay costs very little time. Latching mode is the other approach where a condition should be captured and then deliberately cleared rather than following the signal back down. If the chatter is violent rather than marginal, look at whether the mechanical arrangement is vibrating, because the meter may be reporting something real.
I bought the meter with a serial board and now the plant wants it on Ethernet.
That is precisely the situation the modular design exists for. The communication boards are plug-in and field installable, so the meter takes the new board rather than being replaced, and the sensor, wiring, scaling and panel cutout all stay as they are. Tell us which meter you have and what the plant network needs, and we will confirm the right board — there are versions combining Ethernet with USB and with RS485, so a change of mind upstream does not have to mean losing the connection you already use.
The meter reads correctly but my controller sees a different number.
Almost always a scaling mismatch at one end rather than a fault at either. On an analog output, confirm which endpoints of the reading were mapped to which endpoints of the signal, and that the controller has been told the same thing — a 4–20 mA loop configured for one span at the meter and another at the controller produces a smooth, believable, wrong number. On a digital link, check the register, the data format and whether decimal places are implied rather than transmitted. Also check the loop resistance against the compliance figures if the output is current, since an overloaded loop will read low without complaining.
My load cell is a long way from the panel. Does that affect the reading?
It can, and in a way that looks like a calibration error rather than a wiring problem. Excitation current flowing down a long cable drops voltage in the wires, so the sensor receives slightly less than the meter is supplying, and since output is proportional to excitation the reading comes out low by a consistent percentage. Distance also increases the chance of picking up interference and of the two ends sitting at different ground potentials. Use decent gauge, properly shielded cable, keep the run away from power wiring, and if the distance is substantial talk to us about the arrangement before installing — there are ways to handle it and they are all easier to design in than to retrofit.
This panel gets washed down. Is the meter going to survive?
From the front, yes, provided it is mounted properly. The instrument is sealed to NEMA-4X, equivalent to IP65, once installed in the panel cutout with its gasket seated correctly — that rating covers the front face only and depends entirely on the installation being done as intended. The back of the meter and everything plugged into it lives inside the enclosure and relies on that enclosure being sealed. Where hosing is frequent or aggressive, a splash proof cover is available for the front. What the rating does not cover is condensation forming inside a cabinet that heats and cools daily, which is a common cause of trouble in wet areas and is worth thinking about separately.