Solutions for obsolete transistors

[czuhars]

Hi,

In my retro project, I’m going to tackle clock generation, ala Apple II, with a 14.3MHz crystal. I’ll need to divide it down to various frequencies for video, disk drive, CPU clock, etc.

I’m working from existing schematics. But it’s all based on the 14-ish MHz clock frequency. To make this frequency, the schematics ask for a couple of 2N4258 PNP transistors. As these are obsolete and hard to come by, Id like to use modern equivalents. Any suggestions? I know the NTE106 are a similar replacement. But since I primarily have 3906s on hand, could I use those? Would I need to change the capacitor and resistor values to compensate for the change? Or would any changes to the supporting hardware values be necessary?

I don’t have my schematics handy, but can post them later.

Thank you
Chris

[GARTHWILSON]

It looks like the PN3640 is the same thing, and is still available (but not for long).

[czuhars]

Thanks. My test board currently has the crystal across a couple of 1k resistors and 74LS04 with 68uf capacitor, similar to the NMOS oscillator in the 6502 Primer. Works, but how accurate will it hold? In my research, the transistor/ capacitor arrangement is said to be more accurate.

[Chromatix]

To a large extent, the frequency of an oscillator is determined by the resonant filter made of passive components, not the active driver which merely injects energy into it. As such, there are many ways to build an oscillator that will produce the same frequency from the same crystal.

The transistor-based versions probably originate from radio practice, where very high frequency accuracy is desirable for band-sharing reasons. In computer practice, such high accuracy is not normally required; an error of say 0.1% won't cause your TV to lose lock on the display signal, which is the main reason for using such an "odd" frequency as a reference. In clock terms, 0.1% corresponds to about a minute and a quarter per day, so you would want something much better than 0.1% to drive your RTC's reference clock (usually at 32768 Hz).

[czuhars]

I knew the subject of clocks and oscillator design had to have been discussed here previously. One such thread included this link: http://www.ti.com/lit/an/snoa676/snoa676.pdf
It answered a few questions with regard to pros and cons of various clock designs. A good thread was this:
viewtopic.php?p=17401#p17401

I’ll try substituting the PN3640 in my base frequency design.