What to Check Before You Buy a Tape Machine (And What a Dying Capstan Motor Sounds Like)

A Revox PR-99 quarter inch reel to reel tape machine, heads and transport visible

A steel tape head is rated at roughly 2,000 hours of tape. The machine you are looking at is forty years old and the seller has no idea how many of those hours are gone, because nobody wrote it down.

That is the real problem with what to check when buying a used tape machine. Almost every checklist on the subject is written for someone who wants to play back pre-recorded reels: thread a tape, listen for hiss, watch the meters. Fine, if playback is the whole job. This one is for the other job, where the machine is asked to record.

A worn reproduce head loses treble you can partly claw back with EQ afterwards. A worn record head bakes the loss into the tape and nothing gets it out. That one difference reorders the whole inspection.

A Studer A810 tape machine photographed from the front, BASF metal reels fitted on both spindles, meter bridge raised above the transport controls and tape counter.
A Studer A810, the kind of machine this checklist is written for. The meters and the counter will tell you plenty, and none of it is how much life is left in the heads.

What changes when you buy a machine to record on

  • The record amplifier path, a second set of electronics per channel. A machine can play beautifully and record badly.
  • Bias and calibration, both specific to the stock you intend to use.
  • Record head wear specifically, plus the erase head, which nobody remembers until a track will not wipe.
  • The sync path. You overdub by listening to the record head acting as a playback head. No sync, no overdubs in time.
  • Whether spares still exist in 2026. Not whether it works today, but whether it can be made to work in three years.

What to check when buying a used tape machine: the ten-minute inspection

In this order, so the expensive discoveries arrive first.

  1. The back, before the front. Find the voltage selector. Japanese-market Otari and Tascam machines are often 100V, US ones 117V, and 230V mains ends the viewing and the machine together. A fishy smell at the vents is a mains filter capacitor cooking, which on Revox and Studer means RIFA.
  2. Power up with no tape on it. Listen for hum, a buzzing transformer, chattering relays. The reel motors should sit silent.
  3. Spool a tape end to end, both ways, exercising reel motors, tension arms and end sensors at once.
  4. Stop from full fast wind. Both reels should arrive together. A thrown loop or snatching take-up means the brakes are done. Linings are cheap; a stretched master is not.
  5. Watch the tape path, not the reels. Tape should sit centred in every guide and flat across the heads.
  6. Play a known factory-recorded tape, testing the repro path with no record-side variables.
  7. Record something and play it back. Nobody does this, and it is the only step that tests the record amplifiers, bias oscillator and erase head.
  8. Only now look at the heads, with a loupe, machine off.

Reel to reel red flags that should end the viewing

A channel recording quieter than the other with trims matched. A recording with the previous one faintly underneath, which is a dead or misaligned erase head. Black lines across a head face. Burnt resin. And a seller who will not let you record, which usually means they know.

Heads: how to read wear, and when a head is finished

A new head has a rounded taper so tape wraps evenly over the gap. Tape is an abrasive: thousands of hours grind that curve into a flat which grows. Reeltoreeltech quotes a Revox rule: a flat wider than 4mm is a worn out head, under 4mm has life left.

Two other signs matter as much. A ridge at the top of the tape path, worn in by decades of tape at one height, lets the tape edge wander out of its groove and gives level and treble variation like a bad cable. A black line along the gap means the gap is opening, which is where a head stops being worn and is finished.

Material changes the arithmetic. Steel heads (Otari, early Teac) are rated near 2,000 hours; reeltoreeltech puts 1950s and 1960s Sony machines nearer 500. Ferrite and the Akai glass heads were advertised at 150,000 hours but fail by pitting microscopically, losing treble without showing a flat, which is why an Akai M-7 wants magnification, not a glance.

The sources disagree. Fletcher, on Gearspace:

you can't tell much about tape heads or a machine by looking at the heads with the naked eye [...] it won't tell you the "remaining life" on the head.

He is right, and it does not cancel the 4mm rule. Visual inspection is a screen, not a measurement: it catches a finished head and never certifies a good one. Posters in that thread put a JRF assessment at around USD 125.

Relapping still exists: JRF have done it in New Jersey since 1979, and Summertone in the UK give the number, two or three reprofiles in a head's life and past saving once the gaps open. A worn head is a bill. An open gap is a funeral.

Transport: capstan, pinch roller, brakes and tension

Reel to reel wow and flutter causes, as a decision path

The definitions do half the work. Wow is slow variation, conventionally 0.5Hz to about 6Hz. Flutter is faster, roughly 6Hz to 100Hz, heard as roughness rather than pitch bending. Above 100Hz is scrape flutter, the tape vibrating stick-slip against a stationary part. Sources place the boundary differently: Wikipedia notes the historic split was simply above and below 4Hz.

What you are really doing is timing the wobble. At 7.5 ips tape moves about 190mm a second, so a capstan shaft a few millimetres across turns ten or twenty times a second while a pinch roller twenty-odd millimetres across turns about three. Different bands, which is why the symptom names the part.

What you hear When What it points at
Slow wobble, a few times a second Play and record Pinch roller: flat spot, out of round, hardened
Roughness or graininess, no pitch bend Worse at slow speed Capstan: bearing wear, glazed shaft, servo instability
Rate changes as the reel fills or empties Across the reel Tension: arm, felts, reel torque, dragging guide
Rate unchanged switching 7.5 to 15 ips Either speed Nothing rotating in the path. Motor control, servo, supply
Speed drifting over minutes, or warming in After ten minutes Capstan servo, tacho sensor, failing electrolytic
Clean on a factory tape, wobbly on your own recording Record, then play The record pass: different tension under record
A fine hash on sustained notes Any speed Scrape flutter: dirty or worn guides, seized flutter idler

Those last two rows are what playback-framed guides never cover. Wow and flutter also compound: record with a wobble, play with the same wobble, and you hear the damage by the third generation.

Ten seconds of a machine actually running, shot in portrait. This is the one thing a listing photo can never show you: tape sitting steady in the guides, a pinch roller running true, speed that holds. Ask to see it run, or assume it does not.

Capstan motor symptoms, and how to tell if a tape machine needs a new capstan motor

A dead capstan motor is rarer than the forums suggest. Rule out the cheap causes first: corroded speed-switch contacts, which on the Otari MX-5050 can stop the capstan engaging at all; dried bearing grease after long storage; and a glazed capstan shaft, a surface problem not a motor problem, which Revox machines develop from use. Teac and Tascam use direct-drive reel motors with a belt to the capstan, and reeltoreeltech describes the belt going gooey, seizing and burning out the DC motor behind it. The belt kills the motor, in that order.

Tape machine speed drifting slowly while everything else behaves is almost always electronic. The Revox B77 and PR99 set capstan speed from a tacho sensor, removing the dependence on mains frequency and adding a servo loop that can misbehave on its own. Drift appearing as it warms is a supply fault; constant from cold, a speed reference fault.

Pinch rollers, brakes and a tape-eating transport

Press the roller with a fingernail. A healthy one gives slightly; a hard, shiny, glazed one does not, the commonest cause of speed instability in a machine that has sat. Flat spots come from years left engaged under spring pressure, and lithium grease migrating out of the bearing hardens the rubber from the inside, which is why Tascam rollers go off. Athan supply new rollers, Terry's re-rubber old ones, and Nagravox list kits from AUD 135.

The same Studer A810 seen from the side, wooden cheeks visible, with the head block and the capstan and pinch roller area in clearer view.
The same machine from the side, because this is the angle that shows you the head block and the capstan. Press the roller, look for a flat, and get the loupe on the head faces while you are down there.

A transport that eats tape is telling you about tension and braking, not the capstan. Too little take-up torque and tape loops past the head block; too much supply drag and it stretches; uneven brake release throws a loop when you stop. All of it is cheap to fix, and all of it destroys a master first.

Electronics and alignment: a machine nobody has aligned since 2005 is not broken

It is also not usable for mixdown yet, and most sellers cannot tell the two apart. A machine can be mechanically perfect and still print a mix that is dull, lopsided and 3dB down, because nobody aligned it to your tape.

The procedure runs in a fixed order and needs a calibration tape. Clean, demagnetise, then play the calibration tape and set azimuth, reproduce level at 1kHz and reproduce EQ at 10kHz, repeating both for the sync path on a multitrack. Only then load working tape for the record side: bias first, recording 10kHz and advancing the bias past the peak until level falls back by a set amount (Ampex 456 practice is 1.5dB past the peak, and it differs for every stock), then record level at 1kHz and record EQ at 10kHz and 100Hz.

So "it was aligned" means nothing without the words "to what", and alignment is the last thing to judge a machine on at a viewing, because it is the one thing guaranteed to be adjustable. Calibration tapes are still made: a quarter inch MRL full program tape carrying thirteen frequencies from 32Hz to 20kHz is USD 170, wider formats USD 335 to USD 590.

For the measuring, the period-correct answer is a test set, and the Sound Technology 1500A here is exactly that: generation and analysis in one box, built to test tape recorders rather than to be one. Radiomuseum's entry for the related 1510A lists azimuth, frequency response, channel separation, speed, flutter and noise, to NAB, ANSI and CCIR. The difference between "sounds about right" and a number.

On the electronics, listen channel to channel at matched settings. Crackle from a trim pot is dirty contacts, which is good news; a channel quiet on playback but noisy on record is in the record amplifier. Assume the capacitors are old: reeltoreeltech puts a full Revox recap at USD 600 to USD 1,500, roughly what we say about recapping and servicing costs generally, and the Revox reason is the RIFA and Frako parts, which fail without warning.

Which reel to reel for a home studio in 2026

Judged on parts and running cost, not on nostalgia.

An eight-track reel to reel recorder with eight illuminated meters across the meter bridge and a row of record-ready switches along the lower panel.
An eight track, all eight meters lit. Eight tracks on tape is enough for a band playing live in a small room, and it is also eight record amplifiers, eight sets of trims and eight alignments to keep honest.
Machine Parts in 2026 What that means
Revox A77 Good: Nagravox, Premium-Hifi A rebuild, not a service
Revox B77, B77 MkII Very good: new butterfly heads, PCBs, rollers The sensible entry point. RIFA capacitors
Revox PR99 MkI to MkIII Best here. Sound On Sound: "virtually every part is still available" Balanced XLR, NAB or IEC. The MkIII dropped the sync buttons, so MkII is the one
Studer B67, A807, A810 Costly. NOA took the last AM Belgium head run The A810 is microprocessor controlled: a dead battery loses the calibration
Studer A80 Specialist only, a handful of techs worldwide A bill that assumes a commercial studio
Otari MX-5050 B, BII, BIII Thin but real. Otari's own line: "for possibility of servicing, please contact Otari" Three direct-drive motors, hard to stop. Jammed sensors, dried grease
Otari MX-70 Heads turn up used. The audio cards are scarce Buy one with spare cards, or not at all
Tascam 32, 38 Reasonable: Tascam Ninja, Audio-Parts Cheap eight tracks. Cracked solder joints are the signature fault
Tascam 388 Awkward. The mixer doubles the failure surface Same solder joints, plus a console
Tascam BR-20, BR-20T Best Japanese machine, built to 2004 +4dBu balanced XLR. Youngest here by fifteen years
Teac A-series, X-series Good for common models Unbalanced: tape sound, not mixdown. X-series switches burn out

On the question people actually type, Revox B77 vs Tascam: the B77 has the better transport and parts supply, the Tascam is cheaper and gives more tracks, and neither is balanced. If balanced connections and a real +4dBu record level matter, and for mixdown they should, the answers are the PR99 and the BR-20, whose published figure is 0.06% peak DIN weighted at 38cm per second.

What to bring with you

  • A torch and a 10x loupe. The loupe is the head inspection. Your phone camera is not good enough.
  • Your own tape: a fresh reel to record on, and a known factory-recorded tape for the repro test.
  • A sustained 1kHz tone on your phone, and a lead to get it in. It shows wow and flutter in a way music does not.
  • Headphones you know, a multimeter for the voltage selector, and isopropyl and buds, because half of what sounds like head wear is oxide.
  • A number in your head for a rebuild, so you can walk away.

What we could not verify

Parts availability is the hardest thing here to state honestly, because it changes monthly and a supplier's website is not an inventory. We could not verify current JRF Magnetic Sciences pricing, whose main page would not load, and the USD 125 figure comes from forum posters rather than JRF. Summertone's relapping prices sit on a page we could not read. We found no committed supply for the Otari MX-70's audio cards, and would treat a Gearspace thread titled "Otari MX 70 PB16xa Cards and other parts wanted" as fair description of that market.

Nor could we find a published specification sheet for the 1500A itself; the list above is from Radiomuseum's entry for the 1510A of 1981, the later machine in the same family. The 4mm flat-spot rule is reeltoreeltech's summary of Revox practice, not a figure we have seen in a Revox document. And the rotation rates above are arithmetic from typical dimensions, not measurements.

If you would rather buy one already sorted

All of that is the argument for a machine somebody has already sorted. The Revox PR-99 is here for the reasons the table gives, the Otari MX-5050 for the reason it earned its living, and the Sony TC-510-2 for anyone who wants a portable built to be carried. There is a standalone TEAC record and playback amplifier for rebuilding around known-good electronics, and the TEAC AN-300 for multitrack work where hiss is the problem. For tape on the way into a session you keep in software, the Endless Analog CLASP does that: audio hits the record head, the repro head feeds the converters, and the plug-in compensates the head gap so the take lands time-aligned.

If the machine is overseas, importing vintage studio gear into Africa covers the duty, VAT and voltage arithmetic that decides whether a cheap machine abroad is actually cheap. And the PR99 mark question is the trap we unpick in which dbx 160 version to buy: one model name, three machines, and the cheap one is missing the feature you wanted.

Sources

Tags: Buying Guide Maintenance Revox Studer Tape Machines
Back to blog

Leave a comment

Please note, comments need to be approved before they are published.