Early in 2018 I purchased a BX-300 from marktplaats.nl. Don't ask me why. The price was not particularly low. The few photographs were poor. Details on the state of the deck were scant.
What arrived looked poor and behaved poor. The legends on the transport, power and eject buttons were faded to invisibility. The front top and case were scratched all over. Attempting to play a tape resulted in the reel motor not starting, and the deck disengaging after two seconds (to protect the tape, at least that was working well).
Putting it into FF for a few hours loosened up the reel motor enough to play a tape. Sort of. Now it disengaged half-way through. More FF, more play trials. After a while it could play a whole side. I put it on repeat. After a few sides it started eating tape. Of course. But at least now I could do a quick calibration and run a record/play frequency response on a worn UR, and lo!, both channels easily got beyond 22kHz. There was still life in those heads ...
A bit more of experimentation revealed that the transport control was erratic. Sound occasionally was imbalanced between the channels, and often crackly. The input and output level potmeters also crackled. The bias pot seemed to be fine, though. Not that this means a thing, as these will fail eventually too.
So armed with the service manual I decided to give this deck the full treatment. Over the next three months I put all of my spare time into refurbishing the transport, repairing and upgrading the electronics, and beautifying the cosmetics.
This started with totally dismantling the deck: removal of the outer and inner front panels, loosening of the vertical circuit boards, and removal of the transport. This looks scary, this is scary, but it is really necessary in order to perform the work this deck needs. Just take note of which screw exactly goes where, and in which order. This is important for putting it back together.
Cosmetics
I no longer have a picture of the entire deck in its state of arrival, but here is a photograph of the plastic inner front panel, i.e. what sits behind the external metal front, carrying many of the buttons.
As the metering scale differs between BX-100 and BX-300, the latter's had to be removed. It is taped in place, very carefully prying under it with a sharp knife and a thin screwdriver got it off. Then it got glued to the BX-100's frame.
The scratches on top of the BX-300's front I filled in with a black felt tip. Normally this looks bad, but in this case it worked out acceptably. Or perhaps this is just in my eye, knowing what state this machine came from.
The cover I sanded lightly, then spray-painted, with a rattle can, into a shade of matte black very similar to that used on the CR series of decks. I had never done this before, but it came out nicely. Of course I practiced first on the BX-100's cover.
Transport
The failures mentioned before, the reel motor stopping, the erratic transport behaviour, the eating of tapes, ... all are normal for a dual-capstan deck of this age based on the so-called Sankyo transport. These things demand some serious attention by now, there is no way around that. The most thorough refurbishment is explained here http://www.tapeheads.net/showthread.php?t=39955. This is not always required, though. Here is what I normally do with these mechanisms.The entire transport has to come out. This is not easy with the BX-300. I know that some get it out without dismantling the whole deck, but I failed to do so. Not a problem, as I wanted access to all of the circuit boards anyway. Once out I pulled out the flywheels, cleaned them with acetone and IPA, lubricated their bearings with Mobile One 0-W-40. I replaced the capstan belt, the idle tire, and the back tension belt. The latter solves the tape-eating-issue. It is best to do this with the transport out, allowing one to look behind the supply reel table to ensure that the belt is running properly on its pulley.
The cam leaf switches were thoroughly cleaned with IPA-wetted paper. The cam motor was removed and then I let it run on a 5V supply for a couple of hours, once in a while changing its direction. Some say it is better to replace this motor, others open it up to clean the commutator and brushes. I've done both in the past, but so far all three methods have given me equally-good results, curing the erratic mode changes or refusal to do anything. (Addition 2022: here is a better method for rejuvenating this motor.)
Likewise with the reel motor: it develops bad spots on the commutator and then gets weak in play mode, which is then detected by the deck, that switches off in order to save the tape. So far I have always had success with putting the deck in FF mode for a few hours, or even an entire night. Sometimes I spray Contact 61 through the motor's unused front mounting screw hole, but I am not sure this really does much good. Here is more information on motor care: http://www.tapeheads.net/showthread.php?t=61988, http://www.tapeheads.net/showthread.php?t=40311.
The real tables were removed, their axles cleaned and lightly oiled, and the running surfaces of the reels cleaned. All gliding surfaces in the transport that were accessible after removal of the motors and flywheels were cleaned of their original grease, and then got a dab of fresh lithium grease.
After reassembly speed and take-up torque were set with a TEAC MTT-211 speed tape and a Sony torque measurement cassette.
WRMS wow and flutter is 0.056% on the average. This is out of spec for this transport, but still reasonable enough, given that I used a capstan belt of unknown provenance.
Later I renovated a CR-4E, which has the same transport. Here I used an original belt (really!) and got a much lower w&f. Something noteworthy is the following modification I added to the transport. Normally speed is set with VR101 on the motor PCB. This is a single-turn 50k pot, used as a variable series resistor. With some aging and wear speed setting is a very twitchy process, making it very hard to hit the correct speed. But on the CR-4 I noticed that the resulting series resistance at the right speed was 8k or thereabouts. So I turned VR101 to give 50k (ensuring this with a multimeter), and then soldered a 100k multi-turn pot in parallel over it. Then I set the speed with this new pot, which gave a much smoother and more repeatable experience. When I find the time I will adapt the BX-300 similarly.
Electronics
My work on the electronics encompasses repairs, replacements of wear-sensitive components, and small modifications that might improve performance. I started by mapping out everything that could usefully be replaced in an excel sheet, together with the specification of the new components.
As you know the deck produced uneven levels and crackling sounds in replay. Even though it has not been confirmed that the BX series suffer from Orange Cap Disease, I replaced all of them anyway. I also dismantled the input and output level sliding pots and cleaned them with Contact 61. This solved the issues.
A lot of the old electrolytic capacitors were replaced. I generally used new Panasonic FCs and new-old-stock Rubycon ZA and ZL series for the power supplies and control functions. The NOS capacitors were first (re)formed on an external supply, and then measured for excess leakage current, this to ensure their health. Some of the supply capacitors had their values mildly increased, and where possible their voltage rating bolstered: this deck comes from the period when Europe was on 220V, nowadays we are at 230V.
The electrolytics in the audio circuits were replaced with low-distortion Nichicon Muse SE bipolars (i.e. the green ones), and Elna Silmics.
Then operational amplifiers. Some people prefer to replace these blindly with the favourite (expensive) opamp of the day. Not me. The right opamp choice is determined by its surrounding circuit and its function in it. In some cases small improvements can be had, but often not. Or worse.
So I left the opamps in the supply regulator and the active ground in place. Their stability is key, and I am sure the original designers did their homework. The prospect of swapping these for something 'better', followed with hours of debugging with a high-speed scope did not look very appealing. Likewise, the NJM2043 remained as the record amplifier. After all, it fulfills this task in nearly every post-Classic Nakamichi, there cannot be much wrong with it.
The TL072 opamps in the playback chain were replaced with the newer, slightly faster, slightly less noisy precision OPA2132s. I had them in stock anyway, and they are a safe improvement over the 072. The 4558 input amplifier and the buffer of the same type had to make room for industry stalwart NE5532.
All opamps in the deck obtained a 100nF film capacitor between their supply pins. This is a practice advocated by Douglas Self in his books and articles, after he observed how some opamps outright demand this. It does not hurt in other cases, and anyway, the power supply bypassing in Nakamichi decks traditionally is pretty simple and crude, so this may indeed do something good.
The bias potmeter (VR750) appeared in good shape. But as this is a known point of failure, buried deep into the deck's innards, I chose to replace it anyway. I used a Bourns PTD902-2015K-A104 dual log 100k pot, which is a perfect fit. It lacks the original's center detent, though.
Another issue was that I thought that the deck got pretty hot inside, even with its bonnet open. There is a small metal plate to which the power regulator transistors are mounted, acting as a heatsink. This plate is screwed to the deck's casing. Using this screw, and thermal paste, I added a small heatsink to the back of the deck. I also glued another small heatsink onto the plate inside.
After all of this the deck was in a pretty good shape, working without crackles, hum, or noise. I aligned it (see further) and started using it.
Then, one day, when monitoring a recording using headphones, I noticed that one channel was audibly slightly down. On the workbench there was a loss of 2dB, and increased second-order distortion, the latter indicating a problem of a electronic nature. I replaced the output mute transistors, but to no avail. Probing around revealed that the problem existed on the output volume control PCB, which is strange because that board only contains passive components. At any rate, the output volume potentiometer seemed to be shot. Not having a replacement for this one, I had a look at the output amplifier board, which feeds this potmeter, and found a suitable spot at the output of the driver opamp to tee-off the signal, after a 200R resistor, and feed it to a new pair of RCA terminals I made in the deck's back panel. I also rerouted the collector connections of the mute transistors to these new terminals, otherwise there were >7V switch-on transients, powerful enough to damage downstream equipment. So now I have a deck with a low-impedance fixed-level output, free of channel imbalances and distortions, and a 2k-impedance variable-level output, albeit with some signal issues.
The next problem to tackle was the deck's playback equalisation. This thread will tell you that the BX-300 has the wrong type I equalisation, leading to a hole in the upper midrange. It is not clear if this was done deliberately. The deep bass also is a bit excessive. I verified this with a TEAC MTT-356 70us playback frequency response tape:
Changing R108 (R208) from 12k4 to 13k6 and R107 (R207) from 316k to 270k brings up the midrange 0.8dB, and reduces the deepest bass with 1dB, yielding a slightly flatter result, more in line with what later Nakamichi decks produced:
With this done I could embark on the final calibration and alignment: playback level with A.N.T. L04 (200nWb/m ANSI), azimuth with TEAC MTT-114NA (12.5kHz), track alignment (playback head height) with A.N.T. L10. Record head height and azimuth where then set using a first-generation BASF TPII, a cassette known for its accurate housing. Then the BX-300 was aligned to my standard tapes: Maxell UR 1994, XLII 1994, and Sony Metal XR 1995.
The resultant Maximum Output Levels (i.e. the recorded level referred to Dolby level, at 400Hz, for 3% third harmonic distortion) for these Type I and Type II tapes were +5.3dB. With tapes like TDK AR or SA-X, and Maxell XLII-S a MOL of +7dB is attained, which is excellent. However, with the metal tape the deck only managed +6.5dB or so, 1 or 2dB below what should be expected from a 3-head Nak and a budget metal (although I have found that the three CR-series decks I have refurbished so far fared even worse, but that is another story).
I also verified the response of the upper 4 segments of the, oh so crude, signal meters. Nominally these are labelled 0, 'between 3 and 5', +7, and +10dB. In reality the LEDs light fully at 0, +4.5, +7.5, and +10dB, with 0 corresponding to Dolby level (after me aligning it so).
This particular BX-300 is now the machine for my cassette characterisation series of articles on this blog, at least for Type I and Type II. In the past four years it has proven to be very reliable and consistent. For measuring Type IV and for music recording I use a modified CR-4.