The explanation of the tension, linked to from the other site seems very unscientifically worded. we have a belt around two identical pulleys, driven with a single motor, and the 'explanation' states that the section between the right pulley and the motor spindle is 'stronger' than the return to the left pulley from the spindle. because of rotational forces and friction, the tension in the return path is clearly lower than on the right hand side where tension is higher, but the belt speed is constant, and if the pulleys are the same diameter, then rotational speed will be the same - UNLESS - continuous slippage in the system takes place. I wonder if it is slippage that is causing the issue. The most likely location would be on the right hand pinch wheel. If it slips, at any point, and the left one does not, then the tape between goes slack?
The thing is, any mechanical constant tension system I’ve encountered is always based on some degree of “slippage” to function. I’m surprised the pulleys are the same size, and the explanation of how it works doesn’t make sense looking at the pulley and motor configuration…the takeup capstan is at the end of the drive chain, the supply capstan is at the beginning because of the rotational direction of the motor and capstan shafts. If I had one here I’d be curious to measure the diameter of the capstan pulleys with a dial caliper. I don’t see how it works unless there is a slight difference in the diameter of the pulleys, though I also acknowledge on an auto-reverse consumer model that wouldn’t work because then the pulleys would have to change size. We know that’s not the case. They other way this works is if there is some slight variation in the pinch roller pressure, slightly greater at the takeup side, that would cause more wrap of the tape on the takeup capstan and create that constant tension. So maybe that’s how it works. But, unfortunately, at least as far as the X-10/X-10R service manual goes (I have nothing on the 32-2 in my library), it’s a poor service manual because it covers no theory of operation information, which IMO is essential for understanding a machine and being able to troubleshoot.
@flyingace does your 32-2 service manual include any operational theory information?
There is some fairly critical information in the X-10/X-10R service manual about setting pinch roller stroke and pressure, and it says if the pressures are out of spec you have to replace the whole assembly as it is not adjustable…!

This is the epitome of a consumer-grade machine if I can be a little snooty…so if the 32-2 is essentially a Tascam-badged X-10 series machine I’d call it one notch below “pro-sumer”.
@flyingace does your 32-2 service manual include this information about checking pinch roller stroke and pressure?
With all this discussion my hypothesis is reinforced: you need to replace the drive belt, clean and lubricate the capstan shafts and bearings, check pinch roller stroke and pressure for each side…basically execute some standard maintenance procedures and go through the setup in the manual and see where things are at. I think it may be a combination of issues with all of those points.
As a related aside the term “closed loop” is being used. I don’t know if Teac used that term or not, or whether that’s used on the interwebs…I don’t typically see that term used regarding a dual capstan drive, it’s usually associated with the geometry of the tape path…most audio tape machines are open loop, where the loop beginning is at the supply reel, and the loop end is at the takeup reel, vs a “closed-loop” tape path where there is a second loop inside that greater loop. I think this started in the instrumentation field. The 3M M23/56/64/79 machines are a good example of a closed-loop machine…here is my M64:
3M’s Mincom Division adapted their instrumentation transport for audio in the early 1960s with the M23. They monikered the tape path their “isoloop” design based on the isolated closed-loop design, closed-loop because of the geometry, and “isolated” because of the mechanical constant-tension design…the top black circle in the center is the capstan shaft. It’s about 2” in diameter. Tape enters the path through a set of fixed guides at the top left of the loop and then pass over the capstan shaft, then across the erase and record heads, around the reversing idler at the bottom, then over the playback head, and then again over the capstan shaft and out the exit guides. On the 1/4” machine the capstan shaft surface is comprised of a series of a flat plateau with flat valleys on either side, like a really shallow square wave. On the wider format machines there are multiple plateau and valley sets. There are pinch rollers on the left and right of the capstan shaft, but they are different. The left roller presses the tape against the valleys, and the right roller presses the tape against the plateau, so effectively the tape is being pressed against a slightly greater diameter shaft on the right. This creates a constant tension over the heads and reversing idler, and isolates tape in the closed loop from the supply side holdback tension and takeup side takeup tension as well as any mechanical noise or scrape flutter before or after the isolated loop. It works…and works well…it’s genius. Here are closeup views of the capstan shaft surface and the dissimilar pinch rollers:
My point is that the “closed-loop” term isn’t related to the dual capstan contact, it has to do with the geometry of the tape path. The dual capstan contact creates an isolated loop or section of tape. So the Teac X-10 series and Tascam 32-2, if my understanding is correct, are open-loop, but they *do* have an isolated tape path section through the headblock assembly because of the dual capstan.