Paperless Chart Recorder: Specs, Memory and Compliance

By Wu Peng, Senior Process Instrumentation Engineer · Last reviewed September 10, 2026

A paperless chart recorder is a panel instrument that takes process inputs, shows them as trends on a screen, and writes the samples to internal memory instead of onto a paper chart. Sixteen universal channels, an accuracy of about ±0.2 percent of span and a few hundred megabytes of storage are typical of the mid range.

The interesting numbers are not on the front panel. They are the ones that decide how long the instrument records before it wraps. They also decide whether the record satisfies an auditor, and why a paper chart could never have met the same test.

This guide works those out. Memory becomes recording days, a millimeter of pen travel becomes minutes, and the two compliance regimes that name real numbers are read at clause level.

Contents

What paperless recorders do

Three jobs sit inside one case. The instrument measures, usually from thermocouples, resistance thermometers or 4 to 20 mA loops. It displays, as real time trends, bar graphs, digital pages and alarm lists. It records, to flash memory that survives a power failure, with export to a USB stick or over a network.

Only the third job is the reason to buy one. A local indicator on a pipe tells an operator the value now and forgets it a second later, which is the gap a recorder fills.

Insulated process line with field transmitters and small local display panels mounted on stands
Field transmitters feeding local display panels. Every one of these shows a number now and keeps no record of it. Turning those readings into something that can be looked at later is what the recorder is for.

Chart recorder background

A chart recorder drives a pen across moving paper. Circular models turn a disc under one to four pens; strip models pull a roll past them. Rotation is selectable, commonly 8 or 12 hours, 24 hours, 7 days, and on some models out to 744 hours, which is 31 days.

The performance of that arrangement is published and it is modest. One vendor specification for a 10.24 inch, 260 mm disc gives chart rotation accuracy of ±0.5 percent of rotation time. Chart recording accuracy on the same sheet is quoted at up to 1.5 percent of chart span. On a 24 hour disc, ±0.5 percent of rotation is ±7.2 minutes of timestamp error.

Mechanical models are quoted differently again, at one chart graduation, with no percentage attached at all. None of those figures survive a modern calibration requirement, which is the real reason the standards moved.

Paper still wins in three narrow places. A spring wound circular recorder needs no power. It keeps working through an outage, and inside areas where an electrical instrument would need a certificate. And some inspectors are simply used to leafing through discs.

Channels inputs and accuracy

Channel count is the first specification anyone quotes, and the migration from paper is blunt. A two or four pen recorder is replaced by a 6, 12, 16, 24 or 48 channel instrument. Channels cost far less than pens ever did.

The table below is a real mid range datasheet rather than a catalog summary, and it shows what an actual 16 channel instrument carries.

Parameter Specification
Analog inputs Up to 16, fully isolated universal
Input types 0-5 V, 1-5 V, 0-20 mV, 0-100 mV; 0-10 mA, 4-20 mA, 0-20 mA; Pt100, Cu50, Cu100; S, B, K, T, R, E, N, J
Accuracy ±0.2 percent F.S. displayed and recorded
Cold junction error ±2 °C, additional to the figure above
Recording interval 1 s to 240 s, 11 steps
Memory 64, 128, 192 or 248 MB flash, plus FRAM for configuration
Outputs 6 relay alarms at 1 A, 4 retransmission channels into 750 ohm, 3 transmitter supplies at 24 V
Communication RS232C and RS485, Modbus RTU, 1200 to 57600 baud; USB export
Clock error ±2 seconds per day
Power 100-240 VAC 50/60 Hz, or 24 VDC in the 18-36 V range

Figures are from a 16 channel universal input recorder datasheet. High end instruments go to 48 channels or beyond, with 125 ms recording and 12.1 inch screens.

Two things in that table decide arguments. The accuracy is quoted against full scale. A 0.2 percent figure on a 0 to 1200 °C range is ±2.4 °C anywhere in the range, not 0.2 percent of the reading. And the cold junction error adds on top, which matters when the comparison is against an RTD rather than a thermocouple input.

Scan versus record interval

Two intervals appear on every datasheet and they are not the same thing. The scan or sampling interval is how often the converter reads a channel. The record interval is how often a sample is committed to memory. A recorder can scan at 100 ms and still store one value every 10 seconds.

The scan figure is also the one most often quoted without its condition. A high end instrument that advertises 1 ms scanning may deliver that on a single channel, 2 ms on five and 10 ms on ten. Another vendor quotes 125 ms across a full 48 channels, which is the more useful way to state it.

Ask for the fastest scan at the channel count you are buying. On thermal processes the answer rarely matters, since a thermocouple in a well responds in seconds. On pressure surges and pump trips it decides whether the event appears in the record at all.

Memory and recording days

Memory is quoted in megabytes, and nobody buys megabytes. The number that matters is how many days pass before the file wraps or has to be pulled off.

Bytes per day = channels × (86,400 / interval in seconds) × bytes per sample

Take 16 channels at 4 bytes per sample. At a 1 second interval that is 16 × 86,400 × 4 = 5,529,600 bytes, or 5.53 MB per day. Everything else follows from dividing memory by that.

Record interval Per day 64 MB 248 MB
1 s 5.53 MB 12 days 45 days
10 s 0.55 MB 116 days 448 days
60 s 0.09 MB 694 days 2,691 days

16 channels, 4 bytes per sample, 1 MB taken as 1,000,000 bytes. Real instruments add file headers and event logs, so treat these as upper bounds.

Two conclusions come straight off that table. Dropping from 1 second to 10 seconds buys a factor of ten, which usually turns a monthly chore into an annual one. And on slow thermal processes the largest memory option is often unnecessary, because 64 MB at a 60 second interval already covers nearly two years.

Application example

Industrial tender, Caspian region. A multi channel process recorder was specified for a tender, calling for 16 channels, a 5.7 inch LCD, 500 MB of memory, 24 VDC supply and compliance with EN 61326-1. On the arithmetic here, 500 MB at 16 channels holds about 90 days at a 1 second interval. At 10 seconds it holds well over two years.

Paper chart equivalence

Anyone replacing a circular recorder eventually asks what resolution is being given up. The geometry answers it. A published 10.24 inch disc is 260 mm across, and only a 100 mm wide band of it is calibrated. For one full rotation the pen therefore travels a 189 mm arc at the inner edge of that band, or an 817 mm arc at the outer edge.

Divide a 24 hour rotation by those arcs. One millimeter of pen travel is 1.8 minutes at the outer edge and 7.6 minutes at the inner edge. On a 7 day disc the outer figure stretches to 12.3 minutes per millimeter.

Now set a recorder beside it at a 1 second interval. That same millimeter of paper corresponds to between 106 and 458 stored samples. The paper is not storing a coarser version of the same record. It is storing a drawing of it.

Circular chart geometry showing the calibrated band and the time value of one millimeter of pen travel Disc 260 mm, calibrated band 100 mm wide One rotation of the pen Outer edge arc 817 mm Inner edge arc 189 mm One millimeter on a 24 hour disc 1.8 minutes at the outer edge 7.6 minutes at the inner edge A 1 s record interval stores 106 to 458 samples there

The timestamp comparison is just as wide. Chart rotation accuracy of ±0.5 percent puts ±7.2 minutes of uncertainty on a 24 hour disc. The electronic instrument specifies a clock error of ±2 seconds per day. That is a factor of 216.

Compliance under Part 11

Vendor pages label a recorder “21 CFR Part 11 compliant” and stop there. The rule itself is short and names four things worth reading before signing a purchase order.

Section 11.10(e) requires secure, computer generated, time stamped audit trails. They log the date and time of operator entries and actions that create, modify or delete records, and record changes must not obscure what was there before. Section 11.50 requires a signature to show the printed name of the signer, the date and time it was executed, and its meaning.

Section 11.70 requires signatures to be linked to their records so they cannot be excised, copied or transferred. Section 11.200(a)(1) requires at least two distinct identification components, such as an identification code and a password.

Read that list and the boundary becomes clear. An instrument can provide the audit trail, the signature fields and the two component login. It cannot provide the written procedures, the user administration or the validation record, and those are the parts inspections usually turn on. Buying a Part 11 option is necessary and is not sufficient.

Heat treatment and AMS2750

Aerospace heat treatment is the other regime that names numbers, and it is where version drift causes real trouble. AMS2750 has moved through E, F, G and H, and vendor pages currently in search results cite three different letters, or none at all.

Per revision G, digital recording instruments must produce permanent records with a minimum readability of 0.1 °F or 0.1 °C. Whole number displays were given only a one year grace period after that revision. The calibration accuracy allowed for control, recording and data acquisition instruments is ±2.0 °F or ±1.1 °C or 0.2 percent of reading. The greater of the three applies.

Put that beside the paper figures from earlier and the standard explains itself.

A chart recording accuracy of 1.5 percent of chart span cannot meet 0.2 percent of reading. A pen line cannot be read to 0.1 °C either. Note also that the instrument calibration tolerance above is not the system accuracy test tolerance. That one is set by furnace class in a separate table and is a different number.

Write the revision letter into the specification. A recorder bought against revision F and audited against a later one is an argument nobody wins.

What the price buys

Published prices span more than a factor of ten across this category. The steps are fairly orderly once they are sorted.

Published price What sits at that step
Around 600 to 800 Two channels, small screen, USB export, no network
1,200 to 2,200 Six to eighteen channels, 4.3 to 5.7 inch screen, Modbus, relays
2,100 to 4,100 Larger screen, Ethernet and web server, more memory, batch functions
3,700 to 6,300 24 to 48 channels, 10.4 or 12.1 inch screen, fast scan options
Above 7,000 Compliance packages, expansion racks, redundancy, aerospace options

US list prices seen on distributor catalogs in September 2026, rounded. Steps are indicative and vendors overlap heavily.

The last step is the one that needs explaining. Channels and screen size are cheap by comparison. The audit trail package, the electronic signature option and the heat treatment option carry the top of the range.

We can quote a recorder against a written specification, including channel count, input mix, memory, and communication. What we cannot supply is a compliance status. Part 11 and heat treatment approvals rest on the procedures and validation around the instrument, not on the instrument alone.

For the sensors that feed one, see our temperature sensor range and the notes on thermocouple types and RTD sensors. Where inputs arrive as current loops, the wiring rules are in our guide to the 4-20 mA current loop. Digital instruments on the same pair are covered under HART.

A recorder is also not a totalizer and not a plant historian. The distinction between accumulating a quantity and trending a value is set out in flow totalizer versus flow meter.

Against a control system historian the argument for a recorder is independence. The record is sealed and separately powered. It does not go down with the control network, and it is not written by the same software that runs the process. Its own calibration follows the same logic as transmitter calibration.

FAQ

What is a paperless recorder?

A paperless recorder is a panel or field mounted instrument. It samples process inputs, trends them on a screen and stores the samples in internal flash memory. Data leaves by USB stick or over a network instead of on a paper chart, and the instrument keeps the record through a power failure.

Are chart recorders still used?

Yes, though mostly in narrow situations. Spring wound circular recorders still run where there is no power or where an electrical instrument would need a hazardous area certificate, and some audit routines are still built around paper discs. New installations in regulated process plants are almost entirely electronic.

What is a paper chart recorder?

A paper chart recorder drives one or more pens against paper that moves at a set speed. Circular models rotate a disc, usually over 24 hours or 7 days; strip models pull a roll past the pens. The trace is the record, so its resolution is limited by pen width and chart geometry.

What are the different types of chart recorders?

By medium there are circular chart, strip chart, hybrid printing and paperless types. By mounting there are panel, field and portable versions. By input there are dedicated temperature recorders and universal instruments that accept thermocouples, RTDs, millivolts, volts and current together.

Request a quote

Tell us the channel count, the input types, the record interval you need and how long the record has to sit in the instrument. We will size the memory and the outputs around that. Reach our application engineers or use the form below.

Contact Form Demo

Written and technically reviewed by Wu Peng and the Instranova engineering team.