eric F.

@ef1j95

Parallelizer problems.

5/9/2020, 9:54:21 PM

Favs: 2

Retweets: 1

eric F.

@ef1j95

Olivetti mechanics with the cover off. I adjusted the EM a little, which seems to have helped. Might require a deeper dive into a magnet adjustment at some point.

5/10/2020, 10:17:00 AM

Favs: 4

Retweets: 1

eric F.

@ef1j95

EM and parallelizer in slow motion. At 240 fps, just squeaking by Nyquist sampling for each bit.

5/10/2020, 11:38:05 AM

Favs: 9

Retweets: 0

eric F.

@ef1j95

9 bytes: CR NULL NULL CR LF NULL NULL $ SPACE or 8d 00 00 8d 0a 00 00 24 A0 with even parity bits set.

5/10/2020, 12:14:32 PM

Favs: 1

Retweets: 0

eric F.

@ef1j95

This is pretty cool... here's a binarized image of the electromagnet armature. You can use ImageJ to track the armature moving up and down.

5/13/2020, 7:43:56 PM

Favs: 2

Retweets: 0

eric F.

@ef1j95

A little noisy, but the bits become visible as movements in the electromagnet. Up is "marking" and down is "space", or "lavaro" (working) and "riposo" (resting) for the Olivetti.

5/13/2020, 7:48:39 PM

Favs: 3

Retweets: 0

eric F.

@ef1j95

Each bit is 1/110th of a second, and each byte lines up neatly on a 0.1s mark.

5/13/2020, 8:07:53 PM

Favs: 1

Retweets: 0

eric F.

@ef1j95

The armature extends out of the electromagnet and causes the demodulating guide blade to tap in (1) or move out (0). As the selector disk spins (it’s rotation is triggered by the start bit) the serial input is parallelized, like a mechanical shift register.

5/13/2020, 11:37:25 PM

Favs: 5

Retweets: 0

eric F.

@ef1j95

così

5/14/2020, 3:41:10 PM

Favs: 1

Retweets: 0

eric F.

@ef1j95

More dark mode notes on the selector disk operation. Movement of the mechanics between the demodulating guide blade and the entry wedge, and the timing and position of the wedge relative to the entry needles.

5/17/2020, 11:16:49 AM

Favs: 8

Retweets: 0

eric F.

@ef1j95

This latest thread had me diving in to the electromagnet and parallelizer mechanics to try to understand these reception errors. I think I’ve got it.

5/24/2020, 10:30:47 AM

Favs: 0

Retweets: 0

eric F.

@ef1j95

In normal reception, the parallelizer selector disk starts spinning when the start bit triggers the release crank. The entry wedge spins around, moving the entry needles in and out depending on the bit value. The stop bit holds the disk until the next start bit is received.

5/24/2020, 10:37:12 AM

Favs: 1

Retweets: 0

eric F.

@ef1j95

The pattern of the reception error suggests that the disk misses the stop bit hold and continues to spin. This means the entry wedge will be out of sync with the incoming serial signal. It picks up the start bit as the first bit and continues spinning, shifting the byte pattern.

5/24/2020, 10:41:26 AM

Favs: 1

Retweets: 0

eric F.

@ef1j95

That seems to explain most of the subsequent bytes, such as this case. After a few bytes, the pattern gets confusing, but this could be an aliasing effect due to a slight difference in the selector disk rotation timing versus the signal. That’s probably why we like two stop bits.

5/24/2020, 10:47:37 AM

Favs: 0

Retweets: 0

eric F.

@ef1j95

My movie of the EM armature made me confident that the signal coming in was ok, and the magnet was behaving as expected. So, maybe the release crank was getting triggered inadvertently?

5/24/2020, 10:55:53 AM

Favs: 0

Retweets: 0

eric F.

@ef1j95

Here’s the release crank mechanics. The crank is activated by the downward movement of the EM armature by an incoming start bit. The crank releases the parallelizer mechanics. A counter rotating lever should catch during the stop bit to hold the parallelizer for the next byte.

5/24/2020, 11:00:17 AM

Favs: 0

Retweets: 0

eric F.

@ef1j95

The hold mechanics looked fine. Nothing seemed sluggish or stuck. I made an adjustment, loosening the screws in figure 271 and rotating the crank slightly counter-clockwise. This should make it less sensitive to the EM armature. I also had to adjust the EM unbalancing.

5/24/2020, 11:20:46 AM

Favs: 2

Retweets: 0

eric F.

@ef1j95

Weather isn’t great, so I might as well putter around down here.

5/24/2020, 12:54:58 PM

Favs: 3

Retweets: 0

eric F.

@ef1j95

Logging into miss piggy. Took a a little work to get the right stty setting.

5/24/2020, 1:11:40 PM

Favs: 3

Retweets: 0

eric F.

@ef1j95

The coda to this chapter: each dive into the Olivetti brings a new puzzle, just a little more understanding, and slightly better operation. Until the next time it breaks down.

5/24/2020, 8:06:26 PM

Favs: 0

Retweets: 0