I'm sure anyone who has serviced or restored a number of auto-reverse walkmans will have come across this problem: azimuth is better in one direction than the other, and if the head azimuth is adjustable at all, then you can only get it adjusted perfectly in one playback direction, and you're forced to choose which direction to prioritise. I'd say it noticeably affects about 25% of the auto-reverse models I've worked on. I'm aware that there are a handful of walkman models (some Sony and some of the high-end Sanyos) that feature independent head azimuth alignment screws for each direction. For all the rest, the head alignment is fixed at a constant angle (sometimes adjustable with screw, sometimes not) to the chassis. For a 4-track auto-reverse head, the pole gaps of each of the four magnetic cores should be, ideally, in a perfectly straight line. Azimuth adjustment is correct when the gaps are exactly perpendicular to the direction of tape travel. Nominally, the tape travel would be expected to be exactly parallel to the cassette shell, but, evidently, there is often some slight change in the angle of tape travel between the forward and reverse directions, which causes the azimuth alignment to differ. Why is this? This is generally put down to some unspecified misalignment of the tape path. What I've discovered is that the misalignment is in fact dominated by one single factor, and that this can often be corrected with a fairly simple adjustment. This idea originates in a feature I first spotted in a Toshiba KT-4168 - and have since also discovered in two other Toshiba models with completely different mechanisms. (I'm guessing Toshiba may have had a patent on the feature, since I've not seen it on any other makes). The principle is quite simple, yet I've never seen it mentioned in any of the repair guides or forum discussions on the topic... The key point is as follows: the angle of tape travel through the cassette shell, past the head, will be pulled into line by the capstan, until the tape is exactly perpendicular to the axis of the capstan. Therefore, when adjusting the head azimuth, what you're doing is aligning the head gap, not (necessarily) to be perpendicular to the cassette shell, but to be parallel to the capstan. For the azimuth to be perfect in both directions, this means the forward and reverse capstans must be exactly parallel to each other. Azimuth alignment is sensitive to errors of a small fraction of a degree, so it only takes a small amount of manufacturing tolerance, or bearing wear, or flex or warpage of the chassis, to disturb the azimuth alignment of one or both capstans, relative to the head. Misalignment of the pinch rollers seems to be fairly inconsequential, since rubber is compliant and will mould to the shape of the capstan at the point of contact. Therefore it is almost entirely the capstan that defines the alignment angle of the tape travel. This is, presumably, why Toshiba included a feature in (at least some of) their auto-reverse models to allow the angle of one capstan to be adjusted. The mechanism is easiest seen in the KT-4168, below - the adjustment screw for the angle of the reverse capstan is arrowed, and is accessible through a hole in the rear panel. The same type of adjustment is also possible, this time on the forward capstan, in this KT-4035 (not a great photo, but the arrow is pointing to a black screw head): Also featured on the KT-4177 (KT-GS3) below. The azimuth adjustment procedure is actually specified in the service manual for this model. The head azimuth is first adjusted in reverse playback mode, then the forward capstan angle is adjusted in forward playback mode to optimise forward azimuth. And it works!! This feature gave me an idea: could other makes and models benefit from adjusting the capstan angles? One of the units in my collection that was suffering alignment issues was a Panasonic RQ-JA2. This model has the capstan bearing housings screw-mounted to the chassis. I did some azimuth measurements in each direction, and worked out that the capstans must have been angled very slightly inwards towards each other. So, with a bit of trial and error, I adjusted the angle of the reverse capstan outwards a little by slipping a few layers of kitchen foil under the inner edge of the bearing housing. And this worked, too! I now have perfect azimuth alignment in both directions! Pic below... Similarly on a Sharp JC-785 - this one already had a thin spacer under one side of the capstan bearing housing (presumably factory-fitted), and I got just the right amount of adjustment by moving the spacer to the opposite screw. Again, it worked!! More recently, I've extended the idea to align couple of Panasonic RQ-S series models... these do not provide any adjustment, so I've made a tool for the job. It's simply a metal bar with a hole drilled in one end, equal in diameter to the bearing shank, and deep enough to accommodate the height of the capstan. I slot the bar over the capstan so that the hole is a close fit over the bearing, and use gentle force to bend it in the required direction. The brass bearings are press-fitted into a thin steel backplate, which bends easily enough with a little pressure. Both the units I've restored have needed the forward capstan bearing tilted slightly inwards towards the head. Potentially, I believe this idea could be applied to almost any auto-reverse model of any make, with a bit of experimentation... I hope someone reading is sufficiently interested to try it for themselves!
This is a really interesting topic. I currently have in my possession a Sony EX9 that sounds dramatically different in forward and reverse. (Example here if interested! I had already recorded this in conversation with the seller.) Do you think this might be a classic azimuth issue? I had an EX2000 that sounded similarly poor (except in both directions), and I'm wondering if some other players also have this issue. Previously I'd chalked it up to poor W&F, but the EX9 has tremendously low W&F for a portable (around 0.10%) and I just recently listened to a Fiio CP13 with significantly worse W&F (0.3%) that didn't have this type of issue whatsoever. So...hm.
Super interesting topic indeed. This is indeed an annoying problem with some auto-reverse players. It's good that it's also about the RQ-JA2, because I still have it here for a new belt. Then I can try it out too.
Side A sounds really bad: lots of flutter and also muffled sound. Side B is much better in both aspects. Given the super bad flutter it is possible there's more than just capstan misalignment. This problem of different FWD/REV azimuth can exist in many autoreverse walkmans. Pinch roller angles do affect this also, it's just the deviations have to be way more severe (like 5° or more) to cause problems. Horizontal deviations have more impact than vertical deviations, which can be compensated to some degree by fact rubber is compliant. The tool showed by @nutronic is very useful. Thanks for sharing ! Wouldn't recommend to use anything else, as with a pair of pliers it's easy to deform the bushing (which would ruin the walkman). That Toshiba KT-4168 design is excellent with that capstan angle adjustment. Wish more walkmans had this, but unfortunately most don't even have adjustable azimuth. One interesting aspect I noticed: most (if not all) Dolby C walkmans have a second set of tape guides (apart from the ones on the head) to compensate for this problem, since Dolby C is way more sensitive to minor tape path misalignments.
