Calibrating the MX-5050

The complete calibration of a first-generation MX-5050 — playback alignment, record alignment, and the factory performance tests — using two MRL test tapes, a MOTU M2 audio interface, a Mac running free software, and a multimeter. Written for a first-timer, mapped step-by-step to the factory procedure in the maintenance manual's digital edition. The operation manual — OIM-1005E, the book handed to the operator rather than the technician — carries its own, simpler version of the same trims in Section IV-3: standard-reference-level, bias, and record-EQ adjustment, no azimuth or MRL calibration step, no split by pot pair or speed. The two coexist: this page still works from the maintenance manual's Chapter 3, IV-3 just being the manual Otari expected an operator, not a bench technician, to run it from.

What you need↑ TOP

This is the kit that gets a first-generation MX-5050 through every performance test and both alignment procedures in Chapter 3, wired into a MOTU M2 audio interface and a Mac running free software. The case for why this interface and why free software covers three of the manual's six instruments lives on the test equipment page — this list is the shopping list and the wiring, not a repeat of that argument.

Two MRL tapes, two different jobs. MRL — Magnetic Reference Laboratory, the tape maker — is not the "MRL" the manual means when it names its own Maximum Reproduce Level calibration step (Otari's own abbreviation, nothing to do with the tape company; more on that when reproduce alignment gets there). Say it once, here, so the two never get tangled:

• A 15 ips NAB multifrequency tape sets reproduce azimuth, reproduce EQ, the manual's own "MRL" pots, and SRL — the deck's own reference-level setting, more on that in "The level story" below — covering everything in 3-5 Electronic Alignment. The multifrequency tape I bought is a fresh order straight from MRL, placed through their store's ¼″ Full Program Calibration Tape listing at 15 ips, NAB EQ, 180 nWb/m. What arrived — August 14, 2026 — is labeled MRL Multifrequency Calibration Tape, catalog 243-101-450-110, serial 149 454, made July 2026: 15 in/s, "NAB (IEC2) Equalization," 180 nWb/m reference fluxivity, all tones at 0 dB, 5.5 minutes, wound tail out under an orange sticker that says it plainly — "PLAYBACK ONLY — Wound Tail Out, Mount on Takeup Side, Rewind to Use — DO NOT ERASE." (That "(IEC2)" is not a different standard sneaking in: IEC 60094-2 is the international designation for the NAB curve, just as the flutter tape's "IEC1" is the CCIR-family curve. Two label dialects, same two standards.) The tape is full-track with fringing compensation disabled, which is why the correction table appears later on this page. And that "180" is good news: it is the manual's own reference level wearing its 1 kHz name. MRL's guide explains that the old Ampex operating level — the manual's "185 nWb/m," measured at 700 Hz — reads as roughly 180 nWb/m once re-referenced to 1 kHz, a small difference from NAB equalization not being perfectly flat between those two frequencies. Same flux, two names.

  Check at the bench — done, 2026-08-14: every MRL tape ships with its own calibration graph, and reading it was the first thing that happened when the box opened. Mine (form G-1, hosted here) confirms the reference fluxivity — 180 nWb/m at 1 kHz — and plots the full tone sequence: 1 kHz reference, then 500 Hz, 8 kHz and 16 kHz (the azimuth block), then the thirteen response tones from 32 Hz to 20 kHz, then a closing 1 kHz, every one within the graph's ±1 dB staff. The reference tone lands at exactly 0 VU in "The level story" below — no offset arithmetic needed, confirmed on the tape's own paper.

• The flutter and speed tape, MRL 241-570-480-104 — 15 ips, a single 3150 Hz tone, four minutes. Mine is serial 149 456, made July 2026, and its own calibration graph (form G-0) plots the tone flat across the reel. It measures speed and flutter only. MRL notes the level and EQ it's recorded at aren't important for that job, which cuts both ways: it also means this tape is useless for anything level-related. Don't reach for it there.

• Handle both the same way. Never record on either. Rewind before use and store tail out — the multifrequency tape's own documentation says so explicitly (Publication 101); treat the flutter tape the same way, it's a single precisely-calibrated tone, not something to bulk-erase by accident. Once a tape is loaded, the manual adds its own rule: "the rewind and fast forward modes should not be used" during alignment (3-2 Test Tapes) — you only ever wind a test tape by playing it through.

A MOTU M2 audio interface. The same unit the test equipment page picks for the mid-tier bench, standing in for the oscillator, the AC voltmeter, and (for the distortion test) the wave analyzer, all through REW. Not repeating that case — just the wiring it needs:

• 2× XLR-female-to-TRS-male cables, deck LINE OUT to the M2's combo jack, TRS side. Don't plug the deck's XLR output straight into the combo jack's XLR hole — that side feeds a microphone preamp, not a line input, and MOTU's own manual says so plainly: "Do not connect a +4 (line level) XLR cable to the inputs (because of the preamps). Use a quarter-inch input instead." The combo jack is pin-2-hot and tip-hot, so a standard XLR-F-to-TRS-M cable carries polarity correctly.

• 2× TRS-male-to-XLR-male cables, M2 monitor out to deck LINE IN — the stimulus path for record-side alignment. Same tip-hot/pin-2-hot convention, the other direction.

• Possibly an inline pad or attenuator. MOTU's own manual rates the TRS input at +16 dBu before clipping — dBu, a decibel scale referenced to a fixed voltage rather than to impedance like dBm, gets its full definition in "The level story" below — but an independent Audio Science Review measurement found the real clip point nearer 2.5 V RMS — about +10.2 dBu, roughly 6 dB under the datasheet figure. The deck's own peak recording level runs 9 dB above its nominal +4 dBm reference. Whether that leaves enough headroom is a bench question, not a shopping-list one:

  Check at the bench: feed the deck's TEST OSC at 0 VU into the M2, then push it up to the deck's own peak program level — 9 dB above SRL, see "The level story" below — and watch the M2's front-panel overload indicator. If it lights, pad the input before trusting any level reading downstream of it.

A true-RMS DVM. Its job is narrow but load-bearing: it's the one instrument in the whole chain that reads volts directly, which is what turns a REW dBFS number into something traceable back to the deck's own +4 dBm spec — the mechanics are in "The level story," next. Models and prices: the test equipment page's multimeter section.

Head cleaner and a demagnetizer. Cheap, unglamorous, and non-negotiable before any test tape goes near the heads — the manual makes it the first line of both 3-2 and 3-3. Options and prices: the test equipment page's consumables section.

Blank stock for the record side — RTM LPR35 or SM900, whichever reference level this machine ends up standardized on. That decision, and the honest gap between RTM's modern datasheets and the manual's Ampex-era assumptions, belongs to record alignment later on this page, not here.

REW, free. The software that turns the M2 into an oscillator, a voltmeter, and — for the distortion test — a wave analyzer. Setup specifics, including selecting the M2 explicitly rather than letting macOS pick a default device, are in "Set up the bench," next.

The level story↑ TOP

Everything else on this page comes back to one idea: a "0" on the deck's VU meter, "0 dBu" on a voltmeter (decibels referenced to a fixed 0.775 V RMS, unlike impedance-relative dBm), and "0 dBFS" in REW (decibels below digital full scale — the ceiling a digital recording can't exceed without clipping) are three different zeros. The tape itself carries a fourth, a physical magnetization level with its own unit. Getting this chain straight before touching a trim pot keeps the rest of the procedure honest.

Fluxivity is the physical quantity underneath all of it: how strongly a stretch of tape is magnetized, in nWb/m — nanowebers per meter, a unit of magnetic flux per unit of tape length, existing on the tape itself independent of whatever meter reads it.

The manual defines its own zero against that quantity. SRL — Standard Reference Level — is the flux level the deck is calibrated to read as 0 VU. On a 15 ips Ampex-family alignment tape that's the first tone, 185 nWb/m at 700 Hz; setting the deck to it is 3-5-(6) SRL Adjustment: trim the SRL pots until the VU meters read 0, which — with the LINE OUTPUT LEVEL switch at HIGH — also reads +4 dBm at LINE OUT.

There is no such thing as "NAB level." NAB names an equalization curve (the 50 µs/3180 µs time constants), not a reference fluxivity — MRL's own guide is blunt about it: "there are no published engineering standards … that tell how to choose the reference fluxivity." 185 nWb/m is an Ampex-era American convention this manual inherited, not a law of physics.

That figure also comes in two flavors depending on where it's measured. The manual's 185 nWb/m is read at 700 Hz; MRL references its own tapes at 1 kHz, where the identical physical tape reads roughly 180 nWb/m — a small difference from NAB equalization not being perfectly flat between those two frequencies. Same tape, same flux, different measuring frequency — which is why the multifrequency tape's actual identity (see "What you need," above) has to be settled before trusting a VU reading against it.

That whole cascade, sourced end to end:

Stage	What "operating level" means there	This deck's number
Flux on tape, 700 Hz reference	nWb/m — the physical magnetization	185 nWb/m (3-5-(6))
Same tone, re-referenced to 1 kHz	nWb/m	≈180 nWb/m (MRL)
Deck meter	VU — the needle	0 VU
LINE OUT, LINE OUTPUT LEVEL at HIGH	dBm — voltage into 600 Ω	+4 dBm (3-5-(6))
Same point, electrical reference	dBu / volts RMS	+4 dBu ≈ 1.228 V RMS (0 dBu = 0.775 V RMS)
REW, M2 input at a fixed, marked gain	dBFS	your own bench anchor — see below

Peak recording level sits above all of that. The manual's own signal-to-noise test states the relationship directly: "peak recording level—520 nWb/m—is 9 dB above SRL—185 nWb/m" (3-4-(1)) — check it yourself, 20·log₁₀(520/185) works out to just under 9 dB. That same 9 dB of headroom above 0 VU is what the M2 clipping check under "What you need" is testing against.

My tape's own calibration graph confirms the ordered 180 nWb/m, so on this bench the reference tone goes to exactly 0 VU and the whole chain above holds as written. For a tape at any other fluxivity, MRL's own conversion table gives the VU offset to use instead, all referenced to 1 kHz:

Tape reference fluxivity	dB re 180/185 nWb/m
180 (= 185 at 700 Hz)	0
200	+1
250	+3
355	+6
500	+9

Two more traps live in this same territory. First: every tone on a 15 ips MRL tape sits at 0 dB — there's no −10 dB pad, unlike the 7.5 and 3¾ ips versions. The manual's own 3-2 states plainly: "On slower speed tapes, all tones are recorded 10 dB below operating level except for the last tone" — describing those Ampex-era slow-speed tapes specifically. Apply it to a 15 ips tape and every level in the procedure comes out wrong. It does, genuinely, apply further on: the 3-5-(5) MRL calibration step reads its first tone at 10 dB below operating level, and record azimuth sets its 15 kHz tone at −10 dBm — a level the manual's own text also states as "14 dB below SRL" in the same sentence. Those aren't two different numbers, just the same one said two ways: with LINE OUTPUT LEVEL at HIGH, SRL sits at +4 dBm on LINE OUT, and +4 − 14 = −10 exactly. The step's own "at least 10 dB below SRL" framing holds too — 14 dB below is at least 10. Worth running that arithmetic once here, so neither phrasing reads like a misprint at the bench.

Second: VU meters lie below about −6 dB. They're only accurate roughly +3 to −6 dB around reference — MRL's own guide flags that limit, and recommends its higher-fluxivity tapes for VU-metered machines partly because −10 dB tones fall outside it. That's most of why a true-RMS DVM is on the kit list — the precision readout for the −10 dB tones above and the manual's own −10 VU steps, where the needle can't be trusted.

Which leaves the last link: dBFS. REW's own calibration routine is built for acoustic SPL — sensitivity in dBFS per 94 dB SPL — not volts, so it publishes no documented dBu-anchoring procedure. This bench uses plain arithmetic instead: with 0 dBu defined as 0.775 V RMS, play the deck's TEST OSC at 0 VU, read the AC volts at LINE OUT on the DVM, and note whatever dBFS REW shows at that instant, at a fixed, marked M2 input gain that doesn't get touched again. That reading becomes this bench's own +4 dBu anchor, good until the gain knob moves — "Set up the bench," next, walks through making it stick.

Set up the bench↑ TOP

Wire the kit from "What you need" above, then get REW talking to the M2 — in this order, before any test tape goes anywhere near the heads.

1. Cable it exactly as above: TRS into the combo jack, never XLR. Deck LINE OUT to the M2's combo jack, TRS side, with the XLR-female-to-TRS-male cables; M2 monitor outs to deck LINE IN with the TRS-male-to-XLR-male cables for the record side, later. The combo jack's XLR hole is off-limits for a reason — MOTU's own manual says so directly: "Do not connect a +4 (line level) XLR cable to the inputs (because of the preamps). Use a quarter-inch input instead." (M Series User Guide, p. 18)

2. MON button off. It's true analog direct monitoring — hardware that routes whatever's hitting the input straight to the output, regardless of what REW or the Mac thinks is happening (M Series User Guide, Quick Reference p. 7). Leave it engaged and every level reading downstream is contaminated by a path that never touched the software at all. Off, for the rest of this page.

3. Set the M2's input gain to minimum, then mark the knob. It's an unmarked analog control with no detent and no memory, so I put a paint pen dot on the knob and the housing the first time I set it, and don't touch it again mid-session. Then prove there's headroom rather than trusting a datasheet: MOTU rates the TRS input at up to +16 dBu before clipping (M Series User Guide, Appendix B, p. 27), but an independent bench measurement from Audio Science Review found the real clip point nearer 2.5 V RMS — about +10.2 dBu, roughly 6 dB under that figure. The deck's own peak program level runs 9 dB above SRL (see "The level story" above). Feed the deck's TEST OSC at 0 VU into the M2, then push it up to that 9 dB peak, watching the M2's front-panel overload indicator the whole time.

   Check at the bench: if the red overload box lights at the deck's own peak program level, the M2 is clipping before REW ever sees the signal. Add an inline pad (see "What you need" above) and check again — don't move on to a soundcard cal or a DVM anchor on an input that's already clipping.

4. Point REW at the M2, not whatever macOS picked. In REW's Preferences → Soundcard, select the M2 explicitly as both the input and output device — REW's own help page is direct about why: it "needs direct access to the controls on the interface." Before measuring anything, open Audio MIDI Setup and confirm the M2's sample rate matches whatever REW has selected — the same help page warns that a mismatch here silently resamples every measurement. Then run REW's loopback soundcard calibration: a physical cable from an M2 output back into an M2 input, phantom power off, hit Calibrate, and expect near-flat response, comfortably under 1 dB of ripple. REW's calibration is per sample rate, not universal — save a cal file for each rate before relying on it.

5. Anchor REW's dBFS to a real voltage. "The level story" above already flagged that REW's own calibration routine is built for acoustic SPL, not volts, with no documented procedure for tying dBFS to a line-level signal — so build the anchor from the definition of the unit instead: 0 dBu is 0.775 V RMS, so dBu = 20·log₁₀(V / 0.775). Play the deck's TEST OSC at 0 VU, read the AC volts at LINE OUT on the true-RMS DVM — expect close to 1.23 V RMS, since 0 VU here is +4 dBu — and note whatever dBFS REW's level meter shows at that same instant, at the input gain marked in step 3 and not touched since. That dBFS reading is this bench's own +4 dBu anchor, good until the gain knob moves.

6. Give the M2's output knob the same discipline, for the record side. Once REW is driving the deck's LINE IN as an oscillator, the M2's physical output knob sets how hot that stimulus is — another unmarked analog control with no memory. Set it once, mark it the same way as the input knob, and treat it as fixed for the session.

   Check at the bench: if the output knob gets bumped mid-session, don't guess — re-run the DVM anchor from step 5 before trusting another reading.

7. Don't trust the frequency counter's absolute number yet. REW can resolve a 3150 Hz tone to about 1 Hz from a one-second FFT — plenty of resolution for the manual's ±0.2% speed window. What no interface maker publishes is the sample clock's own accuracy in parts per million, and the test equipment page's software-bench discussion found no independent measurement of one either.

   Check at the bench: cross-check REW's frequency readout against a known reference before trusting an absolute speed number from it. The speed test later on this page leans on this same caveat.

Before you touch anything↑ TOP

Everything above gets the measurement chain ready; none of it has touched the deck's heads or a test tape yet. Before either does, the manual's own preflight applies, verbatim.

Before a test tape is ever loaded (3-2 Test Tapes): "Clean and demagnetize the heads and tape guiding components before installing the test tape." Do it every session, not just the first.

Before running any test (3-3 Test Conditions):

• Clean and demagnetize the heads — restated here, deliberately, on top of 3-2's version above.
• Bulk-erase a reel of high-quality low-noise blank tape (Ampex 406, 407, 456, Scotch 206, 207, 250, or equivalent) for anything that records.
• Turn power on and "allow unit to warm up for 20 minutes" before any measurement.
• Be sure "head cover is installed for all tests" — no exceptions.

Once a test tape is loaded, handle it gently. The rewind and fast-forward modes "should not be used" during alignment — wind only by playing. And pull it off the transport "only after a normal play run (never after a fast-winding mode)" — run it through in play at least once before removing it (3-2).

Worn tape, honestly: the manual's own caveat, worth knowing before blaming the machine for a soft top end — "After extensive use, high frequency tones may drop as much as 2 dB. In addition, flutter indications may rise even though actual flutter remains unchanged" (3-2).

With heads clean, the deck warm, and a test tape ready to thread, set these before touching a single calibration pot (3-4 Performance Tests):

Control	Setting
REEL	appropriate reel size
EDIT	out (non-edit)
LINE OUT	H
RECORD, each channel	in
SEL/REP, each channel	out
MONITOR, each channel	SOURCE (out)
SRL	SRL
PITCH CONTROL	in (fixed)

Find the azimuth screw before there's a tone anywhere near it. Azimuth is the tilt of the head gap relative to the tape path — the alignment that decides whether high frequencies arrive in phase across the track — and trimming it on the reproduce head carries a hard limit, stated twice: "DO NOT ADJUST ANY OTHER SCREWS EXCEPT AZIMUTH ADJUSTMENT SCREW (LABELED "A" IN FIG. 4-6)," and then, in plain text: "If a maximum reading cannot be achieved within one full turn of the azimuth screw... a full head alignment may be required. Refer to section 4 of this manual" (3-5-(3)). Record azimuth (3-6-(2)) carries the identical one-turn ceiling and the same referral to a full head alignment, later in this page. Screw "A" lives where the digital manual's component-replacement section diagrams it, Fig. 4-6 — I found it on my deck, head cover off, nothing energized, before either azimuth step got anywhere near a screwdriver.

Check at the bench: the whole preflight, confirmed in one pass before a screwdriver goes anywhere near a pot — heads cleaned and demagnetized this session, the full 20 minutes of warm-up on the clock, head cover installed, and the Fig. 4-6 azimuth screw located with the deck cold and nothing energized.

Check at the bench — done, 2026-08-14: before turning a single pot on the level side of this procedure, confirm the multifrequency tape against its own MRL calibration graph. Mine arrived and its graph (form G-1, serial 149 454) settles it: 180 nWb/m at 1 kHz — the manual's own 185 nWb/m-at-700 Hz reference under its 1 kHz name — so the reference tone lands at 0 VU, no offset, exactly as the manual states — and that graph is hosted here, serial number and all. For anyone else running this page: the graph is always the authority over the label or the listing. If yours shows a different fluxivity, the conversion table in "The level story" gives the VU offset to trim to instead of 0 — get it right once, here, rather than chasing a phantom offset through every step that follows.

Speed↑ TOP

The manual's speed test rides the flutter/speed tape behind a full reel of blank stock, not threaded bare — tape tension affects tape speed, so the tape gets spliced to the end of a full-wound blank reel exactly as Fig. 3-2 shows, then loaded so the flutter tape itself is what runs across the heads. PITCH CONTROL fixed, every RECORD switch out — the manual's own first step here, and cheap insurance with an irreplaceable test tape threaded, since the preflight table above had them in — MONITOR to TAPE, SPEED to HI, REEL set to match whatever reel size is loaded. In place of the factory's HP 5300A/ 5301A counter, this is REW's frequency readout on LINE OUT — the same M2 input already wired and anchored in "Set up the bench" above.

Play the tape and read the frequency. The manual's own tolerance:

"Reading should be 2994 ~ 3006 Hz (3000Hz ± 0.2%)." (3-4-(5) Tape Speed Measurement)

That window was written for a 3000 Hz reference tone. Mine isn't one — the MRL 241-570-480-104 tape on the kit list above is 3150 Hz. The manual's ±0.2% is a percentage of the reference tone, not a fixed 6 Hz, so it has to be rescaled: 3150 × 0.002 = 6.3 Hz, which puts the real window at 3143.7–3156.3 Hz. Read this tape against the manual's printed 2994–3006 Hz band and every good reading looks like the deck is running 5% fast, when nothing has moved.

Repeat the reading at the start of the reel and again near the end, on the 7-inch EIA reel as well as the 10½-inch NAB reel, and at low speed — the manual's own repeat list, same section. At low speed the same tape just plays at half rate: a 3150 Hz tone cut for 15 ips reads 1575 Hz at 7.5 ips, and the ±0.2% window rescales with it — 1575 × 0.002 = 3.15 Hz, so the low-speed pass band is 1571.85–1578.15 Hz. Expect the counter to land there, not anywhere near 3150.

Before an out-of-window reading gets pinned on the capstan, price out where else that number could have come from — the manual jumps straight to capstan servo, capstan motor, pinch roller and brake, but the tape itself carries its own error budget:

• Tension. MRL's own Pub 570 worked example — an Ampex 350 reading around 3145 Hz at otherwise-correct speed — shows a tension mismatch alone shifting the reading by about 0.17%. On this tone that's roughly 5.4 Hz, nearly the entire ±6.3 Hz window, before the capstan has done anything wrong.
• Age. MRL's Pub 570 rates a new flutter/speed tape's wavelength accuracy at ±0.1%, at its own reference tension — but only guarantees ±0.5% once a tape has been sitting in storage. On this tone, ±0.5% is 15.75 Hz, more than double the whole tolerance window. A stored tape that's aged past its guarantee can fail this test on its own, with the deck doing nothing wrong at all.
• The counter. REW resolves the tone far finer than this test needs, but its absolute accuracy rides on the M2's sample clock, whose parts-per-million spec no interface maker publishes — the same caveat flagged in "Set up the bench" above, cross-check against a known reference before trusting the number outright.

Check at the bench: bracket a borderline reading against the flutter tape's own condition before touching a capstan or brake. A used 241-570-480-104 carries an unknown gap between its ±0.1% birth wavelength accuracy — rated at MRL's own 0.8 N reference tension — and its ±0.5% guarantee after storage, and this deck's own hold-back tension against that 0.8 N reference hasn't been independently checked here. A reading a few Hz outside the window is tape-condition territory first, capstan-fault territory second.

Flutter↑ TOP

This is the honest section. What the factory asks for, what this kit can actually produce, and the gap between them.

What the factory asks. The manual's test-tape method patches a flutter meter to CH-2 LINE OUT, with one step that matters more than the rest: "Set selector switch of flutter meter to NAB weighted position" (3-4-(6) Flutter and Wow Test). Every number in Table 3-3 is read through that filter — 0.05% at 15 ips, 0.06% at 7.5 ips, 0.15% at 3¾ ips, all NAB weighted, none of them comparable to a reading taken through a different curve. (A small manual quirk to know about: for the "standard flutter test tapes" this method needs, it says "See Table 1-2" — but Table 1-2 is the alignment-tape list, with no flutter tape in it. The tape that actually does this job is the MRL flutter/speed tape from the kit list.) The manual's alternate record-playback method — recording the meter's own 3000 or 3150 Hz tone, then playing it back — carries its own warning: "Since it is possible for the record mode flutter to either add or subtract from the reproduce mode flutter depending upon the phase relationship, it is necessary to make several passes over the recorded section of tape and average the flutter meter readings" (same section). MRL's own Publication 570 says the same thing about the same trap, more bluntly: "you should not measure flutter while simultaneously recording and reproducing." One element of the tape path at a time, never record and monitor together, and average multiple passes when the loopback method is the only option.

What this kit can do. Play the 3150 Hz tape back through the same REW session used for the speed test above, and it's straightforward to turn that tone's frequency wobble over time into a percentage-deviation figure — useful for a before-and-after comparison on this deck, a real number that moves when a bearing dries out or a pinch roller goes flat. What it is not is a NAB-weighted reading. REW runs it through no weighting filter at all, standardized or otherwise, so whatever number comes out belongs next to nothing in Table 3-3. That's the same trap the specs page draws around this deck's own signal-to-noise figures — don't put two numbers side by side without asking what filter each one went through, and that page says outright that NAB flutter and DIN flutter are "different curves." A diagnostic, differently-weighted flutter figure from this bench is real and useful for tracking this deck over time; it is never printable beside Table 3-3 as if it answered the factory's question. The software candidates fare no better on the weighting itself — the test equipment page's flutter section already worked through why: WFGUI's own documentation offers only DIN-family weighting, and Virtins Multi-Instrument implements AES6-2008/IEC/DIN with no NAB mode and is Windows-only besides. Neither clears the manual's NAB requirement on this Mac, full stop.

The rigorous path. Closing that gap means a hardware flutter meter with an actual NAB position — the test equipment page's Meguro MK-668 pick is the budget route to one, a used unit with a dedicated NAB average-value mode and resolution well under the 0.05% target. Short of that meter arriving on the bench, Table 3-3 stays factory-only-verified on this deck.

One last number, from the tape rather than the machine: MRL rates the 241-570-480-104 at ±0.03% residual flutter when new — a figure Pub 570's own table labels as weighted peak, while the manual, from its side, says standard flutter test tapes are recorded on equipment built "to produce less than 0.03% rms flutter" (same section as above; "rms" is the manual's word, and by the weighting rule this section keeps hammering, a weighted peak 0.03 and an rms 0.03 are not the same measurement). Either way it's read, that residual is the noise floor any flutter reading taken from this tape rides on top of. And a caution the weighting trap above doesn't cover: IEC-family flutter ratings (Pub 570 also quotes a ±0.10% IEC figure for 15 ips machines) are a different spec family from Otari's NAB Table 3-3, not a stricter or looser version of the same number. Seeing an IEC flutter spec quoted for some other deck says nothing about how it compares to this one.

Reproduce alignment↑ TOP

3-5-(1) through 3-5-(6) are the reproduce half of Chapter 3's electronic alignment — six steps, factory order, running from bench setup to the SRL trim. Four of them read tones off an alignment tape: azimuth, EQ and the "MRL" calibration, which the manual writes against a 7.5 ips tape, and SRL, which it explicitly says to set at whatever speed the machine usually runs. I only own the 15 ips multifrequency tape from "What you need" above, so every tape-referenced step below runs at 15 ips, against that tape's own tone sequence, not the manual's 7.5 ips one — same pots, same order, different tape underneath them. The other two steps — the setup and the VU calibration — never touch a tape at all.

1. Set up the reproduce chain (3-5-(1)). Clean and demagnetize the heads again — every session gets its own pass, per "Before you touch anything" — then set LINE OUT to HIGH, SRL to "off" variable (not SRL yet; that switch flips back in step 6 below), REEL to match the reel loaded, EDIT out, every RECORD switch out, SEL/REP out, MONITOR to SOURCE (out) on each channel, and the reproduce head selector to 2T — this deck's two channels, half-track. Connect REW to LINE OUT in place of the manual's VTVM — its vacuum-tube voltmeter, the AC-voltmeter role REW and the DVM fill on this bench. Reproduce alignment only wants to hear what comes off the tape; a RECORD switch left in, or a MONITOR switch still routing the input, feeds the meter something other than the tape, and every trim downstream inherits the mistake.

2. VU calibration (3-5-(2)), R86 (CH-1) and R186 (CH-2). The only step in this section that never touches a tape: depress TEST OSC and adjust INPUT LINE CONTROL until the DVM (or REW's anchored dBFS reading, from "Set up the bench" above) reads +4 dBm, then trim R86 and R186 until each VU meter reads 0. This is what makes "0 VU" mean the same electrical thing for every step that follows — azimuth's reference tone, EQ's 10 kHz null, the MRL step's 0 VU, SRL's own 0 VU all read off this same needle. Skip it, or trust an un-anchored needle instead of the DVM, and every later tape-referenced trim quietly inherits whatever drift was already sitting in the meter.

3. Reproduce azimuth (3-5-(3)). The manual's version plays a 7.5 ips tape, sets a 700 Hz reference tone to 0 dBm, then maximizes a single 15 kHz tone against azimuth screw "A." My tape doesn't carry that sequence — MRL's own Publication 101 gives the 15 ips multifrequency tape's own azimuth section as three tones in order: 500 Hz (a low-frequency tone standing in for the level-setting role the manual's 700 Hz plays — the tape's actual reference tone is the 1 kHz that opens and closes it), then 8 kHz for coarse azimuth, then 16 kHz for fine azimuth. Set OUTPUT on the 500 Hz tone the way the manual sets it on 700 Hz, then work the azimuth screw against 8 kHz first and 16 kHz last.

   Check at the bench — done, 2026-08-14: my reel's own calibration graph plots exactly this order — 1 kHz reference, then 500 Hz, 8 kHz, 16 kHz, then the response run — so the three-tone azimuth procedure above matches the physical tape, confirmed on MRL's own paper (the BENCH-1 check in "Before you touch anything"). For anyone else: if your graph lists a different sequence, follow whatever azimuth tones are actually on the tape, in whatever order they actually play, not this list.

   REW's peak-hold display is the maximize indicator here — hold on the number and nudge the screw until it stops climbing, rather than eyeballing a needle. The one-turn ceiling on screw "A" is already stated in full in "Before you touch anything"; it applies here exactly as written — if 16 kHz won't peak within one full turn, that's a full head alignment problem, not a case for a bigger turn.

4. Reproduce EQ (3-5-(4)), R60 (CH-1) and R160 (CH-2). Reproduce the 10 kHz tone and trim R60/R160 until the meter reads 0 dBm, then check every other tone on the tape against 0 dBm ±2 dB. A full-track alignment tape read on this deck's half-track head runs low frequencies artificially hot — fringing, not a real response error — so the low end needs a correction before it means anything:

   Frequency	Subtract this from the reading
   32 Hz	1.4 dB
   63 Hz	1.3 dB
   125 Hz	1.1 dB
   250 Hz	0.6 dB
   500 Hz	0.3 dB
   1 kHz	0.1 dB
   ≥2 kHz	~0 dB

   (MRL's own guide, Table 6 — same source as "The level story" above; MRL calls the corrections "only approximate, especially at the lowest frequencies.") The manual arrives at the identical warning from its own side, in a NOTE at this exact step, quoted verbatim:

   NOTE

   Many test tapes are recorded full track. When reproduced by a half-track or multi--track head, the fringing effect produces invalid response at frequencies below 700 Hz. This effect, which results in high indications in the lower frequencies, does not occur when tapes are recorded and reproduced with heads of the same configuration. Do not adjust the low frequency reproduce equalizers for flat response from a full track standard tape.

   Trim R60/R160 to flatten the low end against an uncorrected reading and the fix is worse than the problem — it bends the EQ curve to cancel an error that only exists because of the tape, not the head.

   One honest gap: the manual closes this step by checking response "for the other tape speed" too, and trimming at 15 ips with a 15-ips-only tape leaves that 7.5 ips response check unverifiable on this bench — a labeled gap until a 7.5 ips alignment tape joins the kit, not a box quietly ticked.

5. Otari's "MRL" calibration (3-5-(5)), R63 (CH-1) and R163 (CH-2). Say this loudly, because the acronym collision is real and the manual does nothing to warn about it: this "MRL" is Otari's own abbreviation for Maximum Reproduce Level, not Magnetic Reference Laboratory — the tape company whose products are on the kit list above. Nothing about this step involves the tape maker. What the factory step actually sets is the reproduce chain's maximum gain: with both OUTPUT controls at full clockwise, it plays the tape's first tone and trims R63/R163 until each VU meter reads 0 — and the manual's own NOTE on that tone, quoted verbatim, is the part that matters: "The first tone is a reference level which is recorded at 10 dB below operating level (185nWb/m)." Run that arithmetic forward: if a tone 10 dB below operating level reads 0 VU at full gain, then a tone at operating level through the same chain reads +10 — the factory target is maximum reproduce gain at SRL + 10 dB.

   That −10 dB pad is real on the manual's own 7.5 ips tape; a 15 ips MRL tape carries no such pad — every tone on it, including the reference tone, sits at the tape's own operating level with nothing subtracted. Trim to 0 VU against that tone and the maximum reproduce gain lands 10 dB below where the factory put it — quietly, since the needle still reads a reassuring 0 — and the error surfaces later, in the record-EQ step on this page whose slower-speed branches assume this gain is exactly where the factory left it. The adapted target for a pad-less tape: reproduce the operating-level reference tone, OUTPUT controls still full clockwise, and trim R63/R163 until LINE OUT reads +14 dBm on the DVM or REW's anchored meter — 0 VU's +4 dBm, plus the 10 dB the missing pad no longer subtracts. The VU meter cannot referee this trim: +10 VU sits far past the +3 dB ceiling the needle is honest to — exactly the meter-accuracy territory "The level story" above already fenced off — so this one is read on the DVM or REW, never the needle.

   Check at the bench: the +14 dBm target is this article's own adaptation of the factory step to a tape without the −10 dB pad — the manual never prints it, because the manual never imagined this tape. The arithmetic, in one line: factory trims a −10 dB tone to 0 VU, so operating level through maximum gain = 0 VU + 10 dB = +4 dBm + 10 dB = +14 dBm at LINE OUT, LINE OUT at HIGH. My tape's calibration graph confirms the 180 nWb/m reference this arithmetic rests on; a tape at any other fluxivity shifts this target by the same offset the conversion table in "The level story" gives.

6. SRL adjustment (3-5-(6)), R65 (CH-1) and R165 (CH-2), done at 15 ips — the speed this deck usually runs, which happens to be both the manual's own recommendation for high-speed models and the only speed my multifrequency tape supports. Flip SRL back to the SRL position, reproduce the tape's operating-level reference tone, and trim R65/R165 until each VU meter reads 0 (or +4/−10 dBm, matching whichever LINE OUTPUT LEVEL setting is in use). That plain 0 VU is correct exactly as the manual states it, because this tape is 180 nWb/m — the manual's own 185-at-700 Hz reference under its 1 kHz name — confirmed by its calibration graph on arrival (2026-08-14). This is where the BENCH-1 finding gets spent: a tape at any other fluxivity would need the reference tone trimmed to the conversion-table offset instead of 0, because a wrong target here doesn't just misread the meter — it moves R65/R165 until the whole reproduce chain is recalibrated off every number the manual prints, silently, while the needle still looks right.

   This is the adjustment the manual calls out as worth getting right the first time: SRL "affects signal to noise ratio, distortion, and tape saturation level" for every reel played back afterward, not just the alignment tape currently threaded.

Record alignment↑ TOP

3-6 Record and Bias System Adjustment is the record half of Chapter 3 — five steps, factory order, done after reproduce alignment above, never before it. The manual's own setup for all five: VTVM on LINE OUT, an audio oscillator on LINE IN, a blank bulk-erased tape threaded, POWER on, SPEED to HI, MONITOR to TAPE, LINE OUT LEVEL to HIGH, SRL to SRL. REW stands in for both instruments here exactly as it has everywhere else on this page — its signal generator drives LINE IN through the M2's monitor outs wired in "What you need" above, and its level meter or the DVM stand in for the VTVM on LINE OUT.

Every target in these five steps is a null, not an absolute number: "0 VU" or "maximum output," read straight off the deck's own VU meters, which is what the manual actually specifies as the readout at every single trim in 3-6. None of them ask the meter to agree with an outside voltage reference — they ask the deck to agree with itself, tape against source, or a pot against its own peak. That sidesteps the entire M2-input-accuracy question "The level story" spent so much effort on: the dBFS anchor from "Set up the bench" exists for reading absolute levels off REW, and nothing in record alignment reads an absolute level off REW. What still matters here is the M2's output side, since REW is now the stimulus generator feeding the deck rather than the thing being measured. The same discipline "Set up the bench" put on the input gain knob applies to the output knob: set it once, mark it, leave it alone for the whole session, and re-anchor with the DVM if it gets bumped.

Blank stock, before any of the five steps. The kit list above named two candidates for the record side — RTM LPR35 and RTM SM900 — without saying which one this machine should be standardized on, because that decision belongs here, against the manual's own reference-level logic, not there. MRL's own guide to choosing blank tape gives a class mapping by predecessor stock:

Blank stock	MRL's predecessor-stock class	Reference fluxivity
RTM LPR35	406-class, no noise reduction	250 nWb/m
RTM SM900	499 / GP9 / BASF 900-class, no noise reduction	500 nWb/m

That table is MRL's, not RTM's — it names the predecessor stocks (Ampex 406, and 499/GP9/BASF's 900-class) that today's RTM tapes are understood to continue, not RTM's own datasheets. I have not sourced RTM's LPR35 or SM900 datasheets for this article, so no RTM-specific bias current or maximum output level figure appears anywhere on this page — printing one without having read RTM's own numbers would be exactly the kind of unsourced bridge this site doesn't build.

Check at the bench: pull RTM's actual LPR35 and SM900 datasheets before standardizing this machine's SRL and bias on a specific number. Until then, treat 250 nWb/m and 500 nWb/m as MRL's class-derived starting points for each stock, not as RTM-verified figures — and remember from "The level story" above that either one sits well above the manual's own 185 nWb/m SRL, so choosing a stock here also means choosing how far off the manual's factory numbers this machine ends up living.

1. Bias calibration (3-6-(1)), R38 (CH-1) and R138 (CH-2). Bias is the inaudible high-frequency current the record head mixes in underneath the audio — magnetic tape is hopelessly nonlinear without it — and how much of it a tape wants varies by stock. The manual states the reason for the whole step plainly: "Record bias current affect signal-to-noise ratio, frequency response, distortion, and recording level. Then it is important to adjust bias current for the using tape in order to make perfect recording." That last phrase is the point — bias current gets tuned to this reel of tape, whatever it is, not to a number printed for some other stock.

   The method is a peak search, not a target number: record a 1 kHz tone, trim the INPUT LINE control for 0 VU, then turn R38/R138 while watching the VU meter for the point where the reading stops climbing and starts falling back — that point is the bias peak. Nothing about this asks what tape is threaded; it chases whatever peak that reel produces, so it works identically whether LPR35, SM900, or Ampex 406 is loaded. That's what makes it usable today, ahead of the RTM datasheet gap above — it needs no reference-fluxivity number at all, only the tape's own peak.

   REW's peak-hold display, the same tool that maximized reproduce azimuth earlier on this page, reads that peak more precisely than eyeballing a VU needle through a broad, shallow maximum — patch it in ahead of the VU meter and use the number, not just the needle.

   What this step is not. A separate, more modern practice exists — find the bias peak at a high frequency (often 10 kHz), then deliberately overbias a further few dB past that peak, trading a little high-frequency output for lower distortion and better headroom on today's higher-output stocks. It is a real and commonly used refinement, but it is not in this manual — this manual's 1 kHz maximum-output method is what Chapter 3 actually specifies — and this article does not teach the overbias-by-N-dB method. It would need its own sourcing, not a borrowed paragraph here.

2. Record head azimuth adjustment (3-6-(2)), screw "A," at 7.5 ips. Record a 15 kHz tone and trim the INPUT LINE control until the meter reads the level the manual specifies, quoted exactly: "Start record mode and adjust INPUT LINE control for -10 dBm on VTVM (14 dB below SRL)." That sentence states one level two ways, and — as "The level story" above already worked through — they agree exactly: with LINE OUT at HIGH, SRL reads +4 dBm at LINE OUT, and +4 − 14 = −10. The step's opening line, "at least 10 dB below SRL," is satisfied as well, since 14 dB below clears it with room to spare. Three phrasings, one level. Set −10 dBm on the VTVM or REW and move on.

   With that level set, work azimuth screw "A" for a maximum reading — REW peak-hold again, same technique as reproduce azimuth. The one-turn ceiling on that screw, and the referral to a full head alignment if a peak won't come within one turn, is already stated in full in "Before you touch anything"; it applies here exactly as written, on the record head this time rather than the reproduce head.

3. Record level calibration (3-6-(3)), R30 (CH-1) and R130 (CH-2), at 15 ips, SRL position. Record a 1 kHz tone with MONITOR set to TAPE and trim INPUT LINE for 0 VU — that's the deck reading its own tape-monitor path. Then, without touching the oscillator or the tape, flip MONITOR to SOURCE and trim R30/R130 until the meter reads 0 VU there too.

   That flip is the entire point of the step: SOURCE monitors the signal going in, TAPE monitors what's actually coming back off the tape a moment later. Matching the two means that whatever reads 0 VU going in also reads 0 VU coming back — the deck's own record electronics stop adding or subtracting level between source and tape. Skip the SOURCE check, or trim R30/R130 while still on TAPE, and every later step that assumes "0 VU in = 0 VU on tape" — record EQ, SEL/REP, any actual recording made on this machine afterward — inherits whatever offset was left uncorrected here.

4. Record equalization adjustment (3-6-(4)). This machine is a high-speed model (15/7.5 ips), which decides which pot pair applies at which speed — the manual splits the same step three ways:

   • 15 ips: SRL position, MONITOR to TAPE. 1 kHz to 0 VU, then 18 kHz, trimming R29 (CH-1) and R129 (CH-2) for 0 VU.
   • 7.5 ips: SRL to "off"/"variable," OUTPUT controls full clockwise. 1 kHz to 0 VU, then 15 kHz, trimming R28 (CH-1) and R128 (CH-2) — the high-speed-model pot pair — for 0 VU.
   • 3¾ ips (a speed this 15/7.5 machine doesn't have — the manual's branch for low-speed models, kept here for completeness): same SRL/OUTPUT setup as 7.5 ips. 1 kHz to 0 VU, then 10 kHz, trimming R28/R128 again for 0 VU.

   At both slower speeds the manual notes, in the same parenthetical each time, that the 1 kHz reference itself is set with "recording is made at 10 dB below SRL" rather than at SRL. That's deliberate, not an oversight: pushing SRL to "variable" and OUTPUT to full clockwise, then recording the reference and the high-frequency trim tone 10 dB down, keeps those top-of-band checks clear of tape saturation and record-electronics headroom — a high-frequency tone recorded at full SRL sits close enough to where a tape's high-frequency output ceiling and the record amp's own headroom start rolling off that the EQ trim would be fighting saturation rather than measuring EQ. It's the mirror image of the trap flagged in "The level story" above about 15 ips MRL tapes carrying no −10 dB pad at all — here, on the record side at the slower speeds, the −10 dB pad is exactly what the procedure wants. And the premise only holds because reproduce alignment step 5 set the full-clockwise maximum gain to the factory's SRL + 10 dB target — that adapted +14 dBm trim is precisely what makes "0 VU" in this SRL-off, OUTPUT-full-clockwise setup mean a recording 10 dB below SRL, keeping the 15 kHz trim tone out of the saturation territory this paragraph just described.

   Each of the three branches closes with a sweep check, done in REW as a stepped-tone sweep rather than a single spot frequency: 35 Hz–22 kHz at 15 ips, 30 Hz–18 kHz at 7.5 ips, 30 Hz–12 kHz at 3¾ ips, each checked against ±2 dB around the 0 VU reference and the relevant pot readjusted if any step in the sweep falls outside that window. A sweep that won't settle inside ±2 dB no matter how the pot is trimmed is also where the blank-stock choice above comes back around — a reference level chosen for the wrong stock's sensitivity can push a sweep's high or low end outside the window even with the EQ pot correctly peaked.

5. SEL/REP level adjustment (3-6-(5)), R367 (CH-1) and R387 (CH-2). Record a 1 kHz tone at SRL for about a minute on both channels, rewind, and reproduce it — both VU meters should already read 0 VU at this point, since that's exactly what record level calibration in step 3 above set up. If they don't, that's a reason to revisit step 3 or the bias peak in step 1 before trusting anything trimmed here. Only once that checks out, depress SEL/REP and trim R367/R387 until each VU meter reads 0 VU in that mode too — the point being that switching into SEL/REP mid-session, to monitor the record head as a playback head, shouldn't jump the level up or down relative to normal reproduce.

   The manual's own footnote on this step: "If this adjustment is made by using standard alignment test tapes which are full track recorded, there will be a little error" — on account, it goes on to say, of the fringing effect. The procedure above records its own 1 kHz tone rather than reaching for a factory alignment tape, so that error doesn't apply to it directly — but the same fringing logic from reproduce EQ earlier on this page (full-track tape, half-track head, low end reads high) would apply the moment an alignment tape gets substituted for this step instead of a freshly recorded tone.

Prove it↑ TOP

Reproduce and record alignment above get the machine into spec, pot by pot. 3-4 Performance Tests is the check that it actually landed there: four measurements — signal-to-noise, distortion, crosstalk, erasure — each with its own factory target, run once both alignment procedures above are done, not instead of them. Same bench as everywhere else on this page: REW standing in for the manual's VTVM and wave analyzer, the true-RMS DVM behind it for anything the software can't be trusted to read on its own.

Signal-to-noise (3-4-(1)). What it proves: how far below the tape's reference level the noise floor sits once the program material is gone — bias hiss, head hum, tape noise, whatever's left. How the manual runs it: an ASA "A" weighted filter ahead of the voltmeter on CH-1 LINE OUT — never sold as a product even in 1976; the manual has you build it yourself, a small aluminum box housing the circuit in Fig. 3-1 (more on sourcing that filter on the test equipment page). TEST OSC to 0 VU, note the filtered VTVM reading — it won't read +4 dBm, the filter has its own insertion loss; MONITOR to TAPE, both inputs full counterclockwise, record one minute of silence; rewind, reproduce, and read the noise floor on the same meter. S/N is the difference between those two readings, plus a fixed offset — the manual's own words, the same sentence "The level story" above already checked by hand: "peak recording level—520 nWb/m—is 9 dB above SRL—185 nWb/m" (20·log₁₀(520/185) comes out just under 9 dB). Target: Table 3-1, 68 dB, 2S at 15 or 7.5 ips — labeled ASA-weighted, the same rule the specs page already holds this deck's other S/N numbers to: never print it next to an unweighted or differently-weighted figure without saying so.

Check at the bench: REW's SPL meter offers A, C, and Z weighting toggles, but that tool is built for acoustic sound-pressure calibration — sensitivity in dBFS per 94 dB SPL — not a documented line-level electrical noise mode. Whether its A-weighting curve is close enough to the manual's ASA "A" filter to stand in for Fig. 3-1 hasn't been checked against a real reference here. Until it is, report the noise floor unweighted, labeled as such, rather than quoting an unverified A-weighted number next to Table 3-1's 68 dB as if the two were the same measurement.

Check at the bench: there's also a decision this test can't make on its own. The 9 dB in the manual's formula is anchored to this machine's SRL — 185 nWb/m, the manual's own reference — and to a 520 nWb/m peak level that sits 9 dB above it. Record alignment above left the blank-stock choice open: RTM LPR35 (MRL's 250 nWb/m class) or SM900 (500 nWb/m), either one well above 185. Standardizing this machine's SRL on one of those numbers instead of the manual's 185 changes what "9 dB above SRL" means — and with it, whether a measured S/N figure is even comparable to Table 3-1's 68 dB, or has to be re-derived against a different peak-level assumption first. That's not a bench measurement, it's an editorial one, and it's mine to make — once, deliberately, before running this test for a number I intend to keep.

Distortion (3-4-(2)). What it proves: how cleanly the tape path reproduces a 1 kHz tone — second and third harmonic content, read separately rather than lumped into one THD figure. How: an oscillator under 0.1% distortion into CH-1 LINE IN, a wave analyzer with adjustable bandwidth on CH-1 LINE OUT — the manual is specific about why a fixed-bandwidth THD meter won't do: "A wave analyzer with too narrow a bandwidth may make it difficult to measure harmonics due to flutter components within the transport," and separately, "Instruments that measure total harmonic distortion will be affected by tape and modulation noise in addition to harmonic distortion." MONITOR to TAPE, oscillator at 1 kHz, analyzer tuned to 1 kHz at 100 Hz bandwidth in AFC mode, record at 0 VU, analyzer calibrated to 100% full scale on that tone — then retune the analyzer to 2 kHz for the second harmonic and 3 kHz for the third, without touching anything else. REW's own distortion measurement is the modern version of that same selective retuning — individual harmonics, not one blended number. Target: 2nd harmonic ≤0.5%; 3rd harmonic typ. 0.5% on low-noise high-output tape, 0.8% on standard professional-grade stock — both at 0 VU, 1 kHz.

Crosstalk (3-4-(3)). What it proves: how much of CH-1's signal bleeds into CH-2's output on a blank, bulk-erased tape — head alignment and contact, not electronics, being the usual cause when it fails. How the manual runs it: the same setup as the distortion test's first five steps, then the wave analyzer moves from CH-1's output to CH-2 LINE OUTPUT and reads the leakage in dB. Target: Table 3-2, 60 dB, 2S. REW has no documented crosstalk mode — the test equipment page already found that "No REW page documents a crosstalk or erasure mode" — so this bench's version is a DIY improvisation: drive CH-1 with the oscillator, and meter CH-2 LINE OUT on a second REW input channel at the same time, watching what shows up where nothing should. Label it that way in any result this bench produces — an improvised two-channel reading, not a documented REW feature standing in for the wave analyzer.

Erasure (3-4-(4)). What it proves: how completely the erase head removes a previously recorded tone before new material goes down. How: record a 1 kHz tone at SRL; then SRL to "off," OUTPUT trimmed for −10 VU, INPUT trimmed for 0 VU — which the manual notes records that same tone "10 dB above SRL" (a deliberately hot test signal to give the erase head something worth erasing, not a peak-level claim — it isn't meant to line up with the S/N test's "9 dB above SRL" peak-recording figure, so don't read the two against each other) — for a few minutes; pull the oscillator, rewind, reproduce (0 VU should come back); then press RECORD to erase while the wave analyzer reads whatever residual 1 kHz tone is left. Target: >65 dB erasure effect. The same DIY-improvisation label as crosstalk applies here — no documented REW erasure mode either, per the same test-equipment verdict — so this bench meters the residual tone on a second REW channel by hand, not through a purpose-built mode.

None of this replaces the tape. REW generates a clean tone and reads back a clean voltage — it measures electricity, accurately, at both ends of the deck. What it can never do is stand in for a properly calibrated reference: the multifrequency tape defines what 0 VU actually means in nWb/m, the flutter tape defines what "no speed error" sounds like, and no amount of software rigor on the electrical side substitutes for either one. Software replaces the instruments that read what a tape produces — never the reference tape itself.

And the numbers above are still just targets, not results. The same honesty clause that closed out recap complete applies here too: every part on this machine went in on its datasheet guarantee and the derating physics, untested, with a plan to measure it properly once the right bench existed — "the numbers get logged here whatever they say." This section is the same promise, aimed at Chapter 3 instead of a parts bin: the procedure is ready, the targets are sourced and labeled, and what's missing is a bench day. When it happens, the results go in the table below, and then in a log entry of their own — whatever they say.

Results↑ TOP

A copyable form — the same measurements above, each with its factory or site-derived target already filled in, ready to paste the real numbers into when they exist.

Measurement	Spec target	Measured	Notes
Speed — start of reel	3150 Hz ± 0.2 % (3143.7–3156.3 Hz), rescaled from the manual's 3000 Hz window		15 ips, 10½" NAB reel — see Speed
Speed — end of reel	3150 Hz ± 0.2 % (3143.7–3156.3 Hz)		Same reel near tail-out; repeat on the 7" EIA reel too
Speed — low speed	1575 Hz ± 0.2 % (1571.85–1578.15 Hz) — the 15 ips 3150 Hz tape played at 7.5 ips		See Speed
Flutter (diagnostic)	0.05 % NAB — not measurable with this bench		REW's reading is unweighted; not comparable to Table 3-3 — see Flutter
Reproduce frequency response	0 dBm ± 2 dB per tone, fringing-corrected below 700 Hz (Table 6)		15 ips, MRL tone sequence — see Reproduce alignment
Record frequency response	± 2 dB re 0 VU, 35 Hz–22 kHz at 15 ips / 30 Hz–18 kHz at 7.5 ips		See Record alignment step 4
Signal-to-noise, CH-1	68 dB, ASA-A, re 520 nWb/m peak		Table 3-1, 2S at 15/7.5 ips
Signal-to-noise, CH-2	68 dB, ASA-A, re 520 nWb/m peak		Same target, other channel
Distortion, 2nd harmonic	≤ 0.5 %, at 0 VU 1 kHz
Distortion, 3rd harmonic	0.5 % (low-noise/high-output) – 0.8 % (standard) tape, at 0 VU 1 kHz		Depends on which blank stock is loaded
Crosstalk	60 dB, 2S — DIY two-channel reading, no documented REW mode
Erasure	> 65 dB — DIY two-channel reading, no documented REW mode

Before any of it goes near a pot, the same preflight this whole page has been building toward, one line each:

• BENCH-1 — Done 2026-08-14, on arrival: the tape's own calibration graph (form G-1, SN 149 454) confirms 180 nWb/m at 1 kHz and the printed tone sequence — reference tone → 0 VU, MRL-cal target +14 dBm, no fallbacks needed.
• BENCH-2 — Feed the deck's TEST OSC at 0 VU into the M2, push it to the deck's own 9 dB peak program level, and confirm the front-panel overload indicator stays dark; pad the input if it doesn't.
• BENCH-3 — Cross-check REW's frequency readout against a known reference before trusting an absolute speed number from it.
• BENCH-4 — Verify REW's A-weighting against a real ASA "A" reference before trusting it for the S/N test; report unweighted and labeled as such until it's checked.
• BENCH-5 — Pull RTM's actual LPR35 and SM900 datasheets before standardizing this machine's SRL and bias on a specific number.
• BENCH-6 — Decide which reference fluxivity this machine gets standardized on going forward, and accept what that decision does to the S/N test's comparability against Table 3-1.
• BENCH-7 — Bracket a borderline speed reading against the flutter tape's own condition — tension, age — before touching a capstan or brake.
• BENCH-8 — If the M2's output knob gets bumped mid-session, don't guess: re-run the DVM anchor before trusting another reading.
• BENCH-9 — Confirm heads clean and demagnetized, 20 minutes of warm-up done, head cover installed, and the Fig. 4-6 azimuth screw located, before a screwdriver goes anywhere near a pot.

Bench day hasn't happened yet — this page is the procedure, sourced and ready, not the result. When it does, the numbers go in the table above, and then in a log entry of their own: real reels threaded, real VU needles read, whatever this fifty-year-old deck actually does against these targets. Filed here whatever it says.