These are modifications I developed for MCI and Sony JH-16 and JH-24 tape machines. I maintained many of these for years and, watching them age, I saw behaviors that the factory had not anticipated. Usually the fix involved replacing an expensive part, such as a tach generator because it’s got dropouts in the commutator or a capstan motor because it’s worn smooth and can’t start the tape. But what happens when these parts are no longer available? Like now. I developed these work-arounds to help bring cranky machines back to life. Some of these mods only apply to the earliest MCI machines, or were developed to deal with borderline conditions that arose when component tolerances conspired to produce a circuit that did not work as the designer intended. I also include some maintenance pointers that I have found to help make noises go away and keep this particular type of machine running.


– To reduce sensitivity to tach generator dropouts, which manifest as repetitive jumps in tension correlating with reel position, do these steps:

1) On the Analog Torque Board connect .5uf 35volt non-polar electrolytic caps (these can be made of a pair of 1uf 50 volt caps in series plus to plus) across resistors R13 and R37. These reduce the bandwidth of the absolute value stage in the PLAY direction. There is no need to have the servo run wideband when the motor cannot respond to the high frequencies; there is only noise up there.

2) On the Analog Torque Board connect 2) 50uf 35 volt nonpolar caps (these can be made of a pair of 100uf 35 volt caps in series plus to plus) to the center pin of 3 pin Molex connector J-13. The other end of each cap connects to one of the outer pins. This mod “holds” the speed voltage value through tach dropouts since the caps are directly across the tach generators. A tach generator can be almost dead and this mod will bring it back to life. There is no downside.


-To correct remote varispeed jitters:

On the Phase Lock Loop (PLL) board add a .01uf ceramic disk cap connected between the reference select jumper common and pin 7 of IC19 (which is a ground). This dumps the high dv/dt overshoots in the reference frequency waveform back into the ground of the signal source IC. These overshoots have been seen to make the one shot double trigger, which makes the PLL hiccup.


– Mod for correct operation of locate mode, autolocate FET on Analog Torque board and autolocate command on logic board. Old MCI 16 and 24 tracks.

Find resistor R1 on the transport mother board and solder a jumper across it.

This resistor is in series with the Autolocator logic signal and serves no useful purpose. The time constant that it is part of is not needed and it causes only trouble. This is a place where a borderline condition will manifest. One end of R1 is being fed, via 3 connector pins, by a 750 ohm pull-up to +15V and the collector of an NPN transistor pulldown. The other end of R1 is feeding diodes that clamp the signal between 5.7V and -.7V. The signal then branches out through 2 Molex pins to the Analog Torque Board and the Logic board. On the Analog Torque Board the signal terminates into a 1000 ohm pulldown and on the Logic board it terminates into a 330 ohm pulldown. The logic gate input also loads a bit. A little math shows that the resulting “logic 1” signal is only about 1.5V. This is not a TTL “1”, which needs to be above 3V. This mod raises that signal to about +3.75V.


– To shorten startup tension pulse:

On Phase Locked Loop board change C2 from 15uf to 4.4uf (two 2.2uf in parallel, voltage same or higher than existing). This shortens the time it takes the tape speed to stabilize from about 4 seconds down to about 1 second. There is no good reason to have such a long start pulse.

There is a 100k resistor connected to one end of C2. Solder a 4.7k resistor in parallel with this resistor. This strengthens the start pulse.


– Adjustable supply reel startup tension ducking. This should work for both MCI and Sony machines.

This mod really helps when the capstan is old and smooth, when the capstan is seen to be scuffing the back of the tape, or when the tape rides up and down during play startup. It makes the speed stabilize very quickly and lets you reduce pinch roller pressure.

Keep in mind that in this transport the capstan was intended to do no work, only meter the tape. On the stock MCI this excellent design philosophy kind of falls apart when the machine enters play for the following reason. The output torque of the motor is calculated by an analog divider as: slower turning = larger tape pack radius = more torque to keep tension constant. When the stock MCI transport enters play the supply reel is stopped and the servo saturated so the tension is at a maximum and the capstan is doing quite a bit of work to bring the tape up to speed.

On the Analog Torque board solder a 10k ohm  multiturn trim pot across resistor R96, with the adjustment screw faced for best trim access. Make mounting as strong as possible. From the wiper of this pot feed IC16 pin 3 (accessed at top of R79) through a silicon diode and an 8.2k 1/4watt resistor in series. The arrow of the diode points toward the R79 connection, this is to say the banded end is nearer to R97.

On the Phase Locked Loop board pin-1 of connector P44 feeds 3 things, a 1k resistor in series with C2 and a diode and resistor in parallel feeding a 22uf cap. This cap gets replaced with 6.6uf, 35V (three 2.2uf-35V in parallel). This shortens the ramp-up time of the capstan motor.

After balancing play tape tensions (see below) start with equal pack on both reels and, with screwdriver on new “supply duck” adjustment and while holding pressure roller away from capstan, enter play, stop a few seconds, try again. Adjust until tape comes up to about the right speed without the aid of the capstan. Spool to each end of the tape and make sure that startup is smooth. You can then reduce the roller pressure to just enough to maintain speed when you drag fingertips with moderate force on the supply reel flange. It may be necessary to fine tune the time constants a bit. When set up correctly startup is very quick and repeatable, so you can keep the pace up during multiple punch-ins, without the recorded pitch wandering.


– The screws on the power transistors in the power supply and motor drive amp work loose because of thermal cycling. Star toothed lock washers will help to hold them still.

Even so, it is good to look at the power supplies every month or two and make sure things are tight, connectors clean and that no fractured solder joints have developed on the Molex connectors. In my experience, Molex connectors do not take solder well to start and the joints do not age gracefully, especially where they are carrying enough current to produce heating. They may be objectively evil, I don’t know. Temperature cycling and oxidation are the cause of most MCI ills and mostly manifest in the Molex’s.

Be sure to check the fans and replace them when they start to drag; if they stop, solder joints on the power transistors and mid sized transistors on the circuit board can melt due to thermal runaway.

Sometimes pins of the Cinch-Jones connectors that connect to the power supply and motor drive amp have been seen to run so hot that hard crusty deposits have formed on contact surfaces and the solder joints melted. Check these & keep them clean.


– Intermittent and grainy sounding Sync Head and Bias/Erase relays can be brought back to life by unplugging then sharply rapping all four sides several times, hard enough to make the contacts bounce and expose fresh gold. This trick cost much less than replacing the relay and almost always works. This step can be made part of routine maintenance, as a matter of course. To find relays that need treatment play the 100Hz tone from a test tape and gently tap the relays with your fingers, listening in both sync and repro. Test the bias/erase relays in a similar way with the machine in record and listen to a 100 Hz tone going through the rec/play cycle. The sync relays are usually filled with nitrogen, but somehow the contacts still get contaminated. The head signal is very small indeed, so a little non-linearity goes a long way. The Sync relay can cause problems for the normal Repro mode too. The sockets like to be cleaned with a light oil lubricated cleaner, which holds the oxygen away. This may sound odd, but in my experience Radio Shack cleaner with lube works well. I have found that cleaning with a “zero residue” spray leads to connector related problems sooner. Contacts don’t seem to like to be dry. I also stay away from the colored sprays that smell like rancid olive oil and claim to remove oxides. I have found that they gunk up in time and actually cause problems, which kind of defeats the purpose, I think.


– Mod to MCI JH-16 audio control card to increase headroom of the record electronics.

In older MCIs the record pre-eq and phase correction stage was found to clip at +9dbm to +15dbm, substantially before any other stage in the chain. This made the headroom on some channels harsh and electronic, not warm and tape-ish. The preceding op-amp stage, biased to +13 to +15VDC, is DC coupled to the all-pass filters phase splitting transistor, which wants to be biased at about +10V and has it’s own bias network to make it so. This mod blocks the DC from the preceding stage and raises the record headroom to +20, so that the tape overloads well before the electronics.

Find the 5.6k resistor R57 and determine which end connects to two 100k resistors, one end of capacitor C35 and the base of a transistor. (The other end traces back to an op-amp output.) Using a solder sucker, lift this end of the resistor and clean the hole. Solder a 10uf-35V cap in series with R57, with the negative end connected toward the two 100k resistors and the transistor base.


Leave a Reply

Your email address will not be published. Required fields are marked *