For test points you need to find the ones that carry the amplified head signals. Only the very early drives have TPs for direct head signals, these are too weak to be useful in most cases. Most modern drives would label the head TPs with lower numbers so look for TP1-3, keep in mind sometimes one is GND point meant to be used for measurement. But to be honest any GND nearby will work since you won't have the exact specs for the signal levels anyway - you just try to maximize the amplitude seen on the scope as you move the head assembly. Obviously there needs to be a formatted floppy in the drive, and it needs to be spinning.

Another issue is head 1 pressure. So in PCs (and many other systems as well) head 0 is the bottom and 1 is top. Gravity works in favour of head 0 but even so it needs a pressure from head 1 to read correctly. But it often happens that dirt or improperly assembled head mechanism might not provide enough pressure (say, a missing spring on the top head). In that case head 0 might still pick up something but head 1 probably wont. Don't always assume this to be the problem though, since the head 0 has it easier it might still be misalignment but the amp can still compensate for it, but not for head 1.

There are also a few things to keep in mind - many drives "latch" the position of track 0 sensor on power-on or first seek, and after that moving it makes no difference to drive operation until power is cycled. A bit annoying as it's often the sensor that gets knocked around, or just partially dirty so it triggers in a different spot.

You might want to get a more standard 3.5" PC floppy drive and experiment on that, the laptop ones are not much different but more tightly packed and more difficult to get right. But unless you really hit and bent something badly it's all down to patience and steady hands. Don't give up and you will fix it eventually - in the end it's a pretty simple mechanical issue, just very touchy due to high track density.

Thank you for the info and encouragement, Deunan, what you said makes perfect sense.

I was indeed focusing on the low-number test points (T1 to T3). There are some great hi-res images on Wikimedia (link
). I soldered short wires to all the test points and tried different grounding combinations (ground clip on the chassis, ground point on the board, etc.), but couldn’t get anything meaningful on either of my scopes, a small yellow FNIRSI 100 MHz digital one and an older analog Tek that goes up to 25 MHz.

I’m powering the laptop with a bench supply to rule out noise from the original power brick, but since everything else works without any issues, I doubt that’s the problem.

Regarding pressure on H1, I suspected that early on because the read signal improved when I applied light pressure. I fully disassembled the mechanism and increased the spring tension, which gave me a consistent read on track 0 without any extra pressure. Reinstalling all the case covers also helps, but not enough. That makes me think the issue might be a very slight angular misalignment, since the inner tracks effectively form a tighter circle.

I’ll keep trying.

It's probably not an alignment issue anymore. Get the head to one of those high tracks (e.g 70) and toggle between H0 and H1 in IMD. If H0 is succeeding but H1 is failing, veeeerry gently touch on the top head with your finger. If it starts reading sectors again, then you know for sure it's a pressure problem as Deunan mentioned.

Solving that, however, is another story... It's perhaps a warped plastic piece, or a bent metal piece, or a weak pressure spring for the top head. The pressure spring sometimes has different options for pressure control.

wbahnassi, I tried what you suggested. I used a tiny crowbar as a controlled weight, and it adds about 6 grams of extra pressure directly at the head assembly. With that amount of pressure, the drive reads noticeably better. However, applying even the slightest additional pressure at the spring completely stops the disk from spinning.

This makes me think the issue might be the head angle and/or its position relative to the cradle rather than the spring tension, but I’m hesitant to bend the head assembly since that feels like an irreversible step.

I’ll see if there’s any other way to adjust the tension or alignment before resorting to bending anything.

Nice hi-res images. So if you touched any of the screws visible on this photo your heads will be misaligned: https://commons.wikimedia.org/wiki/File:Citiz … _motor-9710.jpg

That's two screws holding down the stepper motor, two holding the upper head to the sled, and one just barely visible holding the track zero sensor. If you did unscrew the upper head then the alignment process is bit more involved - you start with head 0 (lower) and ignore head 1 (upper) until you get H0 working properly. Only then you loosen the screws and adjust H1 as necessary - usually by bumping it tiny bit to see if it made it better or worse. Sometimes there is so much adjustment possible (the screw holes are big) that the head 1 can be on a different track to head 0.

It seems test points T1 and T2 are GND, and I would guess you want to tap T5 and T6 as on this photo: https://commons.wikimedia.org/wiki/File:Citiz … en_W1D-9372.jpg
The signal after the amp should be strong enough that any modern scope can pick it up no problem, and the frequency is also <1MHz so nothing to worry about - except for those super-cheap DIY scope kits that max out at audio frequencies. Your TEK should handle it just fine but I see a lot of noise there, possibly wrong TP or bad trigger level (and the scope is free-running instead of triggering). Here's what a 500k MFM signal should look like - note this is two channels because usually the amp only selects one head (depending on the control signal) and the test points provide differential output. If you only connect one channel to one TP you'll see either the yellow or the blue, not full "eye" as shown.

I'm surprised you have to re-align after a belt change - the only time you should have to re-align is after changing the stepper position.

All the drives I've changed belts on have not needed the stepper to be moved - you say "I must have knocked it" - how hard would you have knocked it?
Drive head alignment is pretty secure - it has to endure the vibration of near constant stepping back and forth, so it should tolerate a gentle-to-moderate
knock...

My ImageDisk has a align function which I've had good results with - you do need an "alignment disk", best is an actual one manufactured to very close tolerances, but if not available, you can use a disk formatted on a known good drive (best to use one from the drive you are most likely to want to exchange floppies with)

It basically works by letting you step to any track, and reading the disk, it sounds a beep which gets higher in pitch the greater number of sectors read with IDs matching that track in each revolution - basically you step to a track and adjust the head alignment position for the highest beep, note the positions where it begins to drop, and lock it 1/2 way between them.

Repeat to several tracks over the disk, and adjust for the best compromise in all tracks.

DaveDDS, thanks for sharing your experience. This was my first FDD repair and I was rushing it. I flipped the drive while the covers were unscrewed, so the top assembly fell and dangled by the head.

I’ve spent another couple of hours trying to align the heads. At this stage, H0 (bottom) tracks well and reads Tracks 0 to 80 without a problem. H1 is the trouble-maker. While Track 0 is relatively easy to align by feel using ImageDisk, the reads become unreliable past Track 55.

I decided to use the microscope and a grid to be more precise with the direction and distance I moved the top head. I got it to read reliably up to Track 70 and was even able to run Prince of Persia direct from the disk. However, when I tried other disks, the results were inconsistent. I’m going to leave it there for now, as I'm starting to see diminishing returns.

I was hoping to use the analog oscilloscope to test the signal from the heads, but since it was another 'project' purchase, it seems it has its own issues! 🤷‍♂️

So, I will leave it at this for now. I'm hitting diminishing returns, so I'll revisit it when inspiration strikes.

PurpleOzone wrote on 2025-11-23, 21:56:

I’ve spent another couple of hours trying to align the heads. At this stage, H0 (bottom) tracks well and reads Tracks 0 to 80 without a problem. H1 is the trouble-maker.

Sounds like your heads are misaligned with each other - trickier to fix and can mean a lot of trial and error.
Normally you keep them locked together and only adjust the stepper position. If they got "knocked" hard enough this could still happen.

While Track 0 is relatively easy to align by feel using ImageDisk, the reads become unreliable past Track 55.

Hence "Repeat to several tracks over the disk, and adjust for the best compromise in all tracks".