Your browser is out-of-date!

Update your browser to view this website correctly. Update my browser now


Maintenance: The Hidden Cost

Two years ago, I wrote a three-part series about analog tape machines that generated a surprising amount of feedback from people I wouldn't have expected

Two years ago, I wrote a three-part series about analog tape machines that generated a surprising amount of feedback from people I wouldn’t have expected to be Mix readers. This month, we’ll return to that theme and explore an MCI JH-24 with a tension headache, along with the usual subliminal messages that are always an undercurrent in this column. I don’t expect “everyone” to devour these articles and actually do all of the work, but hopefully enough information comes across for troubleshooting purposes. It may still be necessary to call a technician, but at least you can speak the language.


Used analog tape machines are temptingly affordable until maintenance and tape costs are factored in. Maintenance is “The Hidden Cost” in every business and especially for a “great deal” that seems too good to pass up. Replacing a headstack, motor, pinch roller or obsolete electronic part could (temporarily, at least) put you out of business, and that’s before labor is factored in.

I once flew to Eureka Springs, Ark., to rebuild a pair of 3M analog multitrack decks. It is not practical to ship an analog machine for repair, although there are companies such as ATR Service ( that specialize in overhauls, upgrades and making travel arrangements to and fro. Modular digital multitracks (and their DAT cousins) ushered in a new era of mail-order maintenance, but to live in the vintage analog world requires a knowledge base of parts suppliers and experienced technical specialists. Before you buy, check your local technicians’ availability and rates.


Most vintage analog recorders include a combined operations and service manual. Ampex documentation, in particular, provides step-by-step descriptions of each electronic and mechanical subassembly, a mini-engineering course. Contact used-equipment dealers such as Mercenary Audio at, or Blevins Audio Exchange,, for documentation.

The best way to learn how to read schematics is to draw them; those reproduced on the page are often squeezed to fit, which can severely hinder translation. I could take the easy way out by using block diagrams, but here at “Tech’s Files,” I have simplified schematic excerpts to improve readability. Two examples are included here; full-sized versions are available at my Web crib.


People say it in the Midwest all of the time. No one ever bothers in New York City. (I kinda miss that!) Tape machines can be divided into many categories; in this case, the question is tension control: Oui ou Non? For machines sans tension control (Ampex AG-440 and Otari 5050 Series), tape speed is typically a little fast at the head and a little slow at the tail. Optimum tension occurs only when the amount of tape on the supply and take-up reels are nearly equal. This limitation does not stop anyone from making great records, but it does insert a little if-then logic loop into the creative process. Subtle speed variations may not be noticed until a piece of a take is moved from one end of the reel and spliced into a take at the other end.

All of the various tension-control methods minimize headwear, improve speed accuracy and handling, as well as improve tape path stability, especially at reel extremes. In the most general of terms, Ampex, Fostex, Otari, Studer and Tascam chose tension guides/arms while MCI placed a tachometer under each reel table, eventually adding a “dancer arm” on the supply side. The primary goal is to quickly get the tape up to speed. The secondary goal is to minimize “tension blips” that might be caused by accumulated splices, poor threading or poor tape wind. Let’s look at the challenges.


If this question isn’t silly enough, did you miss the lecture on the purpose of capstan drive and (in most cases) that puck-ish roller called “pinch?” The Capstan Motor — just to the right of the head area — drives the tape at a constant speed. On ancient transports, the Capstan Motor was of the synchronous variety, which means that its speed was determined by, and locked to, the AC line frequency (60 Hz at the outlet in America and some other places). At the other end of the capstan shaft is a large flywheel that smoothes speed variations.

On modern machines, the flywheel is gone. Instead, capstan speed is precision-controlled by a Servo System (often referred to as a Phase-Locked Loop, or PLL) and is referenced to a crystal oscillator. An example from an MCI machine is shown in Fig. 1. The more sophisticated transports add a tension servo and a tape-speed Tachometer (via roller guide) so that the machine can get the tape speed in the ballpark before engaging the pinch roller. With the flywheel now in an electronic form, the capstan speed is now easily slaved to timecode.


After the capstan, the other major influence on tape motion across the heads is the torque applied to the supply and take-up reels. If motor torque is servo-locked, then a tension malfunction can easily break the lock, allowing tape speed to run away or slow down without recovery.

From the Stop position, all machines apply a “Start Torque” signal to the take-up motor to overcome reel inertia. (MCI calls this “acceleration.”) The intensity and duration of this momentary boost correspond to the amount of tape on the take-up reel. For example, Start Torque is almost unnecessary when the take-up reel is empty, but an absolute must when trying to move a nearly full 2-inch reel.

The tension on either side of the Capstan Motor should be nearly the same, with just a little more take-up tension to move the tape at approximately the desired speed. A gross imbalance of supply and take-up tensions can aggravate subtle tape-path anomalies. Under the worst conditions, too much take-up tension and too little supply of hold-back tension may cause the tape to ride very high or very low in the guides, edge-curling at those extremes if not eventually slipping out and causing more serious damage.


The two tension-measuring tools are an old-fashioned spring gauge and a Tentelometer (, a pricey but effective device that slips over the tape. The spring gauge is useful to measure brake tension and pinch-roller pressure; the Tentelometer is best used for measuring tape tension under operating conditions. Most often, absolute tension is measured in two locations: just as it spools off the supply reel and just as it winds onto the take-up reel. Even at center reel, these two measurements will not agree because of the added friction as the tape passes over the heads.

The manual is a good place to start to learn how to adjust reel tension, but the ultimate test, when applicable, is the relative tension on either side of the capstan. Because most versions of the Tentelometer will often not fit in the space between the capstan shaft and the head block, another technique is employed. Once the machine is in play, the pinch roller is disengaged and then either the supply tension is adjusted down or the take-up tension is adjusted up so that the tape runs as close to speed as possible; this method is detailed in both the MCI and the Studer manuals. Try to minimize supply tension whenever possible to extend head life. If performance suffers, then have the heads lapped.


Justin Morse — at here in the Twin Cities — contacted me regarding a problem with an MCI JH-24 16-track. The machine was in remarkably good shape, but tape speed at the reel’s head and tail was out of whack as if tension seemed to be the culprit. After a few phone calls and multiple e-mails, he finally requested a service call.

It had been awhile since my tool case and parts bins last traveled. Ages ago, they negotiated the New York subway system on a luggage caddy with giant bungee cords that held everything together. Now a car is required. I walked through the doorway of Roll Music, we chatted briefly, and the machine was threaded.

Ah, the sweet aroma of analog tape as it passes from one reel to another, constantly changing the hub diameter and along with it, torque (large hub = low torque; small hub = high torque). The machine behaved almost normally until the end of the reel, at which point either the tape would not go forward or it took too long to achieve full play speed. Cheating the tension a little higher on the take-up side almost solved the end of reel start-up problem, but it also made the tape run too fast at the head of the reel.

Note: MCI machines have three torque-limit switches in order to safely limit total torque (and minimize tape stress) for 7-inch plastic, 10-inch or 14-inch reels. Under normal conditions, even the lowest setting should have worked.


When I pivoted the transport for service, the power connectors fell out, which provided a good opportunity to check for corrosion. The connectors were surprisingly clean, so I reseated and secured them using the factory-restraining loops. Justin had already measured all power-supply voltages, but beyond this, I assumed nothing. The Vulcan Mind Meld was applied to the machine and then autopilot took over. It’s best to start at the beginning, so here’s the checklist:

Under each reel motor, a Tachometer generates a DC voltage corresponding to reel rotation. With the tape wound to mid-reel while in play, each Tach should (and did) generate the same voltage. Had the machine failed this test, the Tachs would have been cleaned or replaced.

The Capstan’s PLL board has a test point for observing duty cycle via an oscilloscope. In Fig. 1, Test Point 3 is highlighted using “reverse video,” indicating a square wave that is 30% “on” (70% “off”) when optimized. During operating conditions, it’s possible to “see” how reel-tension variations can push or pull the duty cycle, which is totally cool because that’s how the start-torque signal is derived. Observe the PLL with no tape threaded: The capstan gets right up to speed and the duty cycle is rock-solid. However, if a finger is dragged on the supply reel — or the take-up reel is helped along — the duty cycle will change and the PLL will attempt to correct the tension.

Go to the analog torque board (Fig. 2) to confirm the start-torque signal at pin-6 of IC-22 and IC-23. In this case, start-torque signal was wimpy, so the signal was traced through the motherboard and back to the PLL board (pin-7 of J-61). Simulating stress on the Capstan Motor helped to generate a more substantial error signal. However, no stress on the Capstan Servo means that an error/compensation signal was not generated; a similar condition exists when pinch-roller pressure is too light. To test my theory, the pinch-roller tension was increased by hand whilst going into Play mode and, voila, the start torque increased dramatically.

This may appear to be a labyrinthine way to discover that the pinch-roller pressure was too light, but it is through experience that technicians learn to go easy on this adjustment to avoid damaging the Capstan Motor bearing. Too much pressure makes the bearing hot, increasing friction until the PLL can no longer maintain speed. For a machine on its third owner, the “slippage” could be as simple as 20-year-old spring fatigue, or as complex as having been tweaked by who knows how many technicians.

Pinch-roller pressure is a function of tape thickness as well as pinch-roller diameter, combined with any “play” in the capstan-mounting holes. Any past operation to remove or shim the Capstan Motor might have resulted in a position that relaxed the pinch pressure. I gradually increased pressure while applying a little supply-reel drag (by hand) until it became more difficult to slow or stop the tape. Once satisfied, the pressure was confirmed via spring gauge and found to be spot on. Woof!

Eddie thought this was going to be an easily written article because the sleuthing went so well. Instead, he pored over the schematics to prove the convergence of logic, technician’s intuition and reality (finding at least two instances where a pair of signal lines were mislabeled). Now he’s gonna take a break from the analog world and delve into the DVD/r/rw/ram domain.