First of all, an honest warning - I am a brave (and stubborn) idiot, and a self-taught idiot to boot. I know my way around the basics and have repaired some non-trivial circuits (various complex acid-damage trace repair and diagnosing a partial short on a RAM address line of a pinball CPU board) but essentially, I am learning electronics completely backwards and learning about things after I need to know them to complete a repair. My formal background is software programming and I'm learning the hardware side of life as a very useful (and fun) hobby.

Of course, this 'learn backwards' approach is not great but it's okay for most kinds of devices and designs I run into. Except now, I've smashed into the learning curve... well, learning wall that are crystal oscillators. I'm a bit stumped.

For the actual problem, I'm attempting to repair an old CRT, a Panasonic WV-CM1000. It's not especially great but it's a cute, small CRT with great picture quality for what it does. I'm repairing a fault in it not because it's economical, but as a point of pride and a learning experience.

The fault is that for NTSC video signals, it normally supports the very common 'normal' NTSC with a 3.58MHz subcarrier, and the weird type of NTSC that uses a 4.43MHz subcarrier instead. PAL signals and NTSC 4.43 signals work perfectly on the monitor. Plain NTSC signals do not show any color. Support for NTSC 4.43 or 3.58 is selected via a physical switch on the back of the monitor.

I have the service manual for the monitor and I've worked out that changing the NTSC mode with the switch logically swaps out the crystal used to generate the 3.58MHz and 4.43MHz frequencies. The 3.58MHz by all means I could observe (I quickly learnt the hard way how tricky it is to actually test a crystal without changing its behaviour) was completely dead. All load capacitors, resistors etc. appeared to be identical between the two - save for some extra circuitry that enables the 4.43MHz crystal regardless of NTSC mode selected when the CRT is receiving PAL video. Specifically these are two-pin HC49/U crystals.

The obvious and idiot test to do that would confirm it was the crystal and not any other part of the supporting circuitry - or the auto-PAL-selection circuit - was to just physically swap the crystals and change nothing else. That actually worked in that now PAL and NTSC 4.43 signals lost color, and 'NTSC 3.58' mode was now displaying color as if it was NTSC 4.43. Which means 'psychedelic and totally wrong for normal NTSC.'

I ordered some replacement 3.58MHz crystals - same form factor, 'fundamental' operation and not 'overtone' operation and I thought totally appropriate. Dropped one in circuit and... nothing. No effect. The new oscillator is not oscillating. Old one still wasn't, either.

Touching between the video processing IC and the crystal with anything metal coaxes out some partial, very incorrect color operation, but I know this could be any number of side effects I'm introducing into the circuit that is making the crystal partially oscillate.

From what little I understand, my replacement crystal is not close enough to original specification and I need to make adjustments to suit the new crystal. The below schematic is the relevant portion of the whole thing. The signals from the other side of the diodes come a mess of circuitry that includes the selection switch, the color and brightness knobs, and obviously the video signal itself. The crystals are connected to pin 16 of the LA7680 video IC, which is (obviously) the XTAL pin. X202 was dead and is still dead with a new crystal. X201 works happily sitting in X202's place.

So my questions are:

  1. What is the general workflow for replacing a crystal in an existing circuit with another one - what adjustments should I be making to existing parts?
  2. Am I making some fundamental misunderstanding that is obvious to better educated individuals, so far? Is this repair even sensible? Am I just looking at adjusting the below section of circuit to match the new crystal, do I need to instead provide a better-matching crystal, or am I totally dead in the water without an exact replacement for reasons I'm ignorant to?

Ideally I'd like pointers as to how I should be working this one out for myself - this is no homework question!

Schematic of the relevant part

  • \$\begingroup\$ The intentions behind you being self-deprecating are noble but don't sell yourself short :-) A very warm welcome to the site and thank you for a well-written question, which should attract discussion and well-written answers. \$\endgroup\$
    – TonyM
    Commented Mar 1, 2020 at 10:31
  • \$\begingroup\$ Crystals may have different levels of "activity" - ie need different amounts of drive or feedback. Try paralelling the series 15pF capacitor with another capacitor (the two capacitances add when in parallel). IF this is the issue something as small as 5 pF extra may help, 15 pF extra is a probably safe value and more again may be necessary and, if not necessary, may still work. | A 15 pF cap has about 3000 Ohms reactive impedance at the cb frequency which demonstrates how a change in value may be significant. \$\endgroup\$
    – Russell McMahon
    Commented Mar 1, 2020 at 12:11
  • \$\begingroup\$ @RussellMcMahon I did try that before (another capacitor along with the existing) and got no benefit from it. I've also tried a trimmer capacitor that went up to 50pf and it couldn't coax a reaction from the new crystals either. \$\endgroup\$ Commented Mar 1, 2020 at 21:50

1 Answer 1


Two points worth looking at:

1) Switching between crystals is accomplished by applying DC voltages to D207/D213.When the diode is forward biased, it is low resistance to AC and therefore "switched on"; when reverse biassed, it is switched off. It is worth checking DC levels around both diodes in both modes, looking for differences or anything odd.

2) In addition to fundamental/overtone (which will always be fundamental below 10 to 20 MHz) crystals actually have 2 resonances - parallel and series resonance. Any given crystal can be made to oscillate in either mode if you try hard enough, but oscillating in the wrong mode will be at the other resonant frequency - somewhere around 1000 ppm off station. (I have seen some semi-pro audio gear about this far off its nominal sample rate; presumably because someone bought crystals cut for the correct frequency in the wrong mode!)

Your circuit, where the crystal is connected in series with a 15pF capacitor, would appear to be designed for series resonant crystals. Can you check if the ones you bought were parallel cut, i.e. parallel resonant?

A side effect of using a parallel resonant one in a series resonant circuit is that it may be harder to get the thing going. You may be ably to coax it into starting by increasing C238 - at first up to 100 pF, then reducing it until it stops (and backing off the last change!) It will then be slightly off frequency : if that matters, you are looking for a series resonant crystal.

enter image description here

This Stack Exchange Q&A shows the two resonances, and example oscillator circuits suggesting yours is series resonant. I've pinched Andy's image from it to illustrate the two resonances...

Parallel resonance is tunable (within a limited range) but series resonance is determined more accurately by the cut of the crystal, which may be why it is preferred for colour subcarrier.

  • 1
    \$\begingroup\$ Presumably the cheque is in the post as usual LOL \$\endgroup\$
    – Andy aka
    Commented Mar 1, 2020 at 14:00
  • 1
    \$\begingroup\$ @Andyaka I'm actually old enough to understand that comment! \$\endgroup\$
    – user16324
    Commented Mar 1, 2020 at 14:01
  • \$\begingroup\$ "Your circuit, where the crystal is connected in series with a 15pF capacitor, would appear to be designed for series resonant crystals. Can you check if the ones you bought were parallel cut, i.e. parallel resonant?" That would be the immediate problem, I reckon. The ones I got are described as parallel. Along with the rest of the fantastic detail - thanks so much for getting me going on this one again! \$\endgroup\$ Commented Mar 1, 2020 at 21:49
  • \$\begingroup\$ It turned out to be exactly that - a parallel resonant crystal wasn't working in that circuit. Dropped in a more appropriate crystal and it works a charm. \$\endgroup\$ Commented Mar 10, 2020 at 14:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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