Broke the Crystal off an Osborne 1

[MGaddict]

I need to know if there is a modern replacement part for the Nymph 15.9744Mhz SRX1460 crystal in the Osborne 1 computer.
When I google just 15.9744 crystal, I find one modern part but its an SMD part with 4 contacts. I have no clue how I would modify that to fit, or even if that was possible.

Now if you want to know the story, because I love to talk, here you go. But the rest of this is really irrelevant to the question:
I was repairing the disk drives on an otherwise working Osborne 1 computer that I own. My kid ran into my leg while I was handling the mother board and *snap*, the crystal sheared clean off. I'm sure it still works but I have no idea how I would attach bodge wires to those tiny tiny specs on the bottom of it. I've been searching and searching and the only 2 sites that claim to have it. Both are aviation parts sites and I have to request a quote. I am not an engineer. I find trashed old computers that I want in my collection and restore them. 99.9% of the time, the early 80s stuff is excellent because I am able to follow schematics on them. I can repair bad traces. I don't need a magnifying glass to work on them, nor do I need a college degree. I mean, I have a college degree, but it's in cyber security, not electrical engineering.
Any assistance is appreciated. I hate the idea of having to create something fully custom to fix this computer.

[Whales]

A standard 16MHz crystal is 0.16% off from 15.9744MHz.

Would the difference matter?  Even if the crystal needs to be exact for timing against mechanical parts (eg spinning HDDs) then do you need timing tolerance better than 0.16%?  Perhaps it would matter if there are other crystals on the board that need to stay in ratio with this one, but that would be a fiddly design.  Worst cast you can probably mod a 16MHz crystal by opening it and drawing on it with a pencil (adding mass) until it reduces down a little bit.

Do you know what this crystal controls?  Schematics might be here?

[james_s]

0.16%? Can't you just tweak one of the loading capacitor values to nudge it that far?

The 4 pin devices are not just crystals, they're a complete oscillator. Sometimes it's very easy to use one of those in place of a crystal, sometimes not. Either way I'd bet if you stick a 16MHz crystal in there it will work just fine.

[MGaddict]

You might be right. "The basic timing originates with a 16MHz crystal oscillator. The 15.9744 MHz signal is divided down to provide nominal 8, 4, 2, and 1 MHz signals to control other circuits of the system." However, call me a noob, thats fine, I don't know how to figure out what pF or ppm to look for. At 16Mhz, there are dozens of through hole options which is great, but options are bad some times. Honestly, I doubt the system requires a ton of tight tolerances, it's from 1981, and it's a portable computer. I suppose as long as I get that 8, 4, 2, and 1 MHz signals, I should be good.

[edpalmer42]

Did you see this one?

https://www.ebay.com/itm/323829078506

Based on the part #, it's an HC-45 case.  One reference I found quotes the size of that package as 8.40 X 7.90 X 3.20mm.

The price is ugly, but ......

Ed

[MGaddict]

Where are you looking up this part number? Google doesn't pull up anything.
On the other hand, $42 with shipping is cheaper than having to purchase 200 custom crystals.
I think I'll try a 16MHz crystal 1st.

[T3sl4co1l]

If that's going into something like a composite video output, and it uses color, it's going to be off enough to matter (rainbow columns or something like that).  I doubt it as it's not a harmonic of the colorburst frequency.  Anything else, the timing shouldn't be tight enough to matter: CRT deflection systems have much wider lock ranges (and are often adjustable i.e. V, H HOLD), video amps have more than enough bandwidth, serial ports only need to be within a percent or so, etc.  Drives shouldn't care, either being synchronized to the same rate, or adaptive (nonsynchronous spindle, self clocking data).  Networking I guess might (but did any kind of network card even exist for that thing?).  If it's got a tone generator, it's not even off by a cent, you can tune an orchestra to it.

One wonders why they chose such a particular frequency in the first place... but these are hardly exhaustive uses, who knows..?

Tim

[T3sl4co1l]

0.16%? Can't you just tweak one of the loading capacitor values to nudge it that far?

Nah, not quite that much.  Pencil might do it though.

Tim

[fzabkar]

The crystal frequency is an integer multiple of common modem and RS232C baud rates (38400, 19200, 9600, 4800, 2400, 1200, 600, 300). Is that why it was chosen? If that is the only reason, then there is plenty of tolerance in RS232C for the Tx and Rx clocks. The UART should resync on every start bit in each frame.

 15974400 / 38400 = 416

The main frequency could be divided by 16 to produce a 1MHz clock for an SIO ...

 15974400 / 16 / 38400 = 26

This still results in an integer multiple.

[pigrew]

My guess is it's because of the standard modem/UART frequency of 9600 baud. 15.9744Mhz is 9600*52*32, and is conveniently close to 16 MHz.

For the 4-pin SMD crystals, two pins are ground (for shielding and mechanical support), and the other two are the two terminals of the actual crystal.

The main things to match about the crystal are its load capacitance (which can be tweaked) and if it's parallel or series resonant.

I'm attaching the Osborne's oscillator circuit schematic. C3=33pF C17=1nF.  UA11 is 74S04 inverter

Instead of finding a series-resonant 15.9744 MHz crystal, I probably would substitute a surface-mount 16 MHz 5V oscillator. The circuit would need to be reworked a bit. It's less authentic, but would be much easier to source.

Then again, maybe a parallel-resonant 16 MHz crystal would oscillate close to 15.9744 in a series-resonant circuit? Anyone have experience with this? I'm not familiar with the oscillator topology used. :/

[edpalmer42]

Where are you looking up this part number? Google doesn't pull up anything.

I didn't look up the part #.  The case style and frequency are included in the part #.

Ed

[MGaddict]

Thanks to the suggestions, first about trying a 16MHz crystal, then somebody suggested if it's too high I could use a pencil to potentially lower it, I started looking into videos on how to do that. Found a good one showing how to cut open the pot and how it actually works. This got me thinking. So, I opened it and using my finest iron tip, I managed to get badge wires connected to the leads and was able to close it back up. It works! It breathes life again! Thank you guys!

[T3sl4co1l]

Hah, madlad!  Well done.

Tim

[Whales]

Excellent!  Enjoy the good vibrations.

[abdulbadii]

usually crystals fixed tolerance values are typical as being ±10 ppm, ±20 ppm, ±50 ppm (part per million)
let it have an output frequency of 16 MHz (16 million Hz) and it has a freq. stability 10 ppm, so it'll vary in freq. by 160 Hz: 15.99984 - 16.00016 MHz

[Andy Watson]

If you can tolerate a HC-49/U package:
https://www.ebay.co.uk/itm/282133825997