4
\$\begingroup\$

I acquired an old Eistar SJ-1 Digital pulser, however the frequency it generates is always 66% of what its supposed to produce. Turns out that the frequency coming out of the frequency stage (at Pin 11) is 66% of the 16MHz crystal or about 10.6MHz. (The 74LS38 is an open collector quad NAND gate)

I opened it up and traced out the layout. The frequency stage looks like to that I have now learned as a Pierce type NAND gate oscillator. I have drawn an abbreviated schematic of the digital pulser showing the frequency generation stage.

In the Pierce type oscillators I have researched, there is a capacitor in between the nand gates where resistor R1 is; however in this design there is a 2.2k resistor (R1).

What could be causing the wrong frequency to be generated? How I proceed to troubleshoot and correct the pulser? Should I:

  1. Replace the crystal with another and check?
  2. Replace resister R1 with a capacitor as I have seen in a Pierce type oscillator. I am guessing since the design would have worked at one point in time, R1 is not an issue here.

Any guidance would be appreciated.

Eistar SJ-1 Schematic

\$\endgroup\$
  • 1
    \$\begingroup\$ The ground connection you show at U1 pin 11 is almost certainly incorrect. \$\endgroup\$ – Dave Tweed Jan 12 '13 at 21:41
  • \$\begingroup\$ In any case, this circuit looks like it will drive a lot of current through the crystal and the crystal may have been damaged over time as a result. R2 in particular should probably be much larger. \$\endgroup\$ – Dave Tweed Jan 12 '13 at 21:43
  • \$\begingroup\$ Oops, I fixed the incorrect ground. \$\endgroup\$ – Doc Jan 12 '13 at 22:04
  • 3
    \$\begingroup\$ 66% sounds suspiciously like 2/3. Are you sure there aren't some multipliers and a PLL in there, and they aren't set as you think? \$\endgroup\$ – Olin Lathrop Jan 12 '13 at 22:47
  • \$\begingroup\$ 2/3 is what I thought too, but I(google) has been unable to find a harmonic for a crystal at 2/3. Could it be a crystal failure mode? For the whole device, the 16MHz from the pierce oscillator feeds into a 74LS93 for a divide by 8, then into three successive CD4518BE for divide by 10 each. All the separate inputs go into a mux (74LS151) which gets selected by the switch and then its back into 74LS93 for a divide by 2. Finally its buffered through a nand gate in the 74LS38. So no multipliers or PLLs, but there are some 5 band inductors/resistors(blue above) which I have been unable to decipher. \$\endgroup\$ – Doc Jan 12 '13 at 23:09
4
\$\begingroup\$

With the details provided, it does seem like the crystal may be damaged - it's the most sensitive component on the board, and there's a good chance the pulser has been dropped a few times during it's long life.

The circuit seems reasonable and if it was originally designed with the resistor (rather than the capacitor), then I'd leave it like that. There are similar circuits out there that do use a capacitor as you say:

NAND Oscillator

I'd go for replacing the crystal - choose a crystal designed for series operation such as the FOXLF160 (as opposed to the FOXLF160-20) While you're grabbing the crystal, getting a couple of 74LS38s wouldn't be a bad idea either, given they are only around 50 pence/cents each.

Also, given the price of decent CMOS oscillators nowadays, the other option is just to buy a 16MHz CMOS oscillator (5V supply tolerant) and replace the first stage with that.

\$\endgroup\$
  • \$\begingroup\$ I got a new 16MHz Crystal. Took the old crystal off the pulser and tested both the crystals clocking an Atmega328p, both give 16Mhz. Put the new crystal in the pulser and I get the same 66% frequency rate. Very interesting. \$\endgroup\$ – Doc Jan 30 '13 at 12:35
  • \$\begingroup\$ That is interesting, I agree - there must be more to this circuit then. At a very rough guess, I suspect the circuit is oscillating in an LC mode somehow (did you work out what the inductors are doing?) I'd be interested to see if the circuit oscillates (and at what frequency) without the crystal, or with e.g. a 4MHz crystal in place, or even a small capacitor (e.g. 10-1000pF) in place of the 16MHz crystal. \$\endgroup\$ – Oli Glaser Jan 30 '13 at 15:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.