I recorded the replay of this tape into DAW software and then used the tool's amplitude statistics feature to note the relative level of each tone. These were then plotted with Excel. The result, prior to any attempt to alignment, was this:
The left channel rolls off early and deep. The right channel looks OKish, but the droop between 100Hz and 1kHz does not belong there. Apart from a rising bass due to fringing effects of the full track, MTT-356 should either play back flat (on ordinary heads), or with a rising treble (on narrow-gap heads). It is clear that we are not there. But at least azimuth seems fine ...
Above is the diagram of the playback amplifier. The input from the head is at the left hand side. The output to the Dolby IC is at potentiometer R12. The nominal 3180/120 microseconds turnovers are realised by feedback network R11/R110/R111/C107. The 70 microsecond turnover for Type II is then added by shunt network R113/C109. None of the circuit values match 3180, 120, or 70, but that is to be expected, as the playback network must be seen strictly in conjunction with the playback head and its losses. Variable resistor R11 is used to tune the amplifier to the head and/or compensate for wear. Tweaking it for maximal channel symmetry over the mid band after some effort yielded this:
Better. But the lack of high treble in the left channel suggest some wear, and the overall drooping character tells us that the A-550RX differed significantly from the (admittedly later) standard, presumably in a bid from its designers to obtain a higher signal to noise ratio.
Then followed a very long period of attempting to modify the replay circuit for a wide response, with less treble droop, and identical in both channels. In the end nothing really worked very well. The best compromise consisted of adding 100pF to the head resonator circuit (C101) for the left channel, to regain some high treble, replacing R110 with 15k, and then tweaking R11 for both channels independently. After a lot of head polishing, which really helped, the following result was obtained:
Mind, this is from an ABEX TCC-260 tape, and goes only up to 12.5kHz. This result is not bad at all.
Anyway, doing a few quick tests allowed me to balance more or less for Maxel UR and XLII, but not for MX. When the bias was raised sufficiently for MX to bring the treble down, the tape's 400Hz sensitivity tanked. I had to reach back to a contemporary of the A-550RX, the one Scotch Metafine I own, to return to a semblance of normality. But even so, both type II and type IV gave a very large bulge in the treble, rising many dBs above 400Hz over the 3-14kHz range. I suspected the record equalisation being responsible for this (remember the treble-cutting replay eq TEAC used, which I now had removed).
The rec eq is set by three switchable networks, one for each tape type, in the emitter leg of a single-transistor amplifier. The graph below shows a simulation, with type IV blue, type II green, and type I red. (Note that these simulations are only indicative, as the actual curves vary a lot with the source impedance of the input signal, in this case the per-tape type record level settings resistors!). We see that metal boosts treble most, then type II, and finally type I gets a lot less treble gain.
The plots below are actual record/replay curves at -20dB (ref. Dolby level), after a first alignment trial. Blue is XLII (1994) and is not too bad, although there is a 4dB peak at 5kHz, and -3dB is reached at 15kHz or so. Green is UR (also 1994), peaking ridiculously at 10kHz before plummeting. Orange is that old Metafine, with eq in the Type IV/Metal position. Pink is the same tape, now recorded with eq in the Type II/Co/Cr position.
As the deck has independent switches for bias and equalisation I decided to try and turn the Metal position into something more suitable for Type II: changing C142 for 22nF brings the entire treble range 1dB below that of the Type II curve. Certainly a move in the right direction, but still not enough to allow making recordings with anything resembling neutrality. Ii appeared that the entire thing is severely underbiased for modern tapes.
In the end I added a treble shelf by means of an socketable RC network between the output of R155 (not shown in the diagram above) and ground, allowing me to reduce the entire treble area relative to lows and mids. First I did this independently for left and right, trying to force both channels individually flat (see picture), but after while I thought it wiser to keep the re-equalisation circuit identical to both channels, accepting a small channel difference in the treble due to remaining head issues.
After this I calibrated bias and level for Maxell UR (1994), XLII (1994) and Sony Metal XR (1992). The following graphs show the results at levels of -20dB, -10dB, and 0dB relative to Dolby level:
The glitches in the curves are caused by synchronisation problems with the measurement software, not by the deck. These frequency responses are not outstanding, but they are perfectly serviceable. Let's not forget that this is a 40-years old deck with essentially basic ingredients.
I put it in the rack and occasionally used it for erasing cassettes and re-packing badly packed tapes. But it mostly served as a bulky ornament. Until I got enough of that and decided to let it go. After all, what use is a cassette deck when it is not playing or recording?
It was quickly sold. Prior to delivery I embarked on making a couple of recordings with it, on Maxell SXII. And truth be told, these recordings, when played back on my Nak CD1, were unexpectedly really rather good!
Part 1: http://audiochrome.blogspot.be/2017/12/restoring-teac-550rx-cassette-tape-deck.html
Part 2: http://audiochrome.blogspot.be/2017/12/restoring-teac-550rx-cassette-tape-deck_30.html
Part 3: http://audiochrome.blogspot.be/2018/01/restoring-teac-550rx-cassette-tape-deck.html
