Thar be dragons! The Chinese 1Hz-50MHz “Crystal Checker” and why it spells trouble

The “1Hz-50MHz Crystal Counter Meter with Housing Kit” available various retailers. There are other versions of this kit in other colors that use the same circuit.

TL/DR: It is a poor and misleading circuit for testing crystals in the low HF region (~7 MHz and below). Relying on this device to screen old pressure-mounted rocks from the from the 40’s and 50’s will result in the discarding of many already hard-to-find good crystals. This post explains why.

It is billed as a DC-50 MHz frequency counter and crystal checker. The idea of a crystal oscillator coupled with a frequency counter is a good one–plug in an old crystal you got in the flea market and it will not only tell you if it is oscillating, but it will also tell you at what frequency it is oscillating. Some might question why you need to measure the frequency when its written on the case, but this is often not true when it comes to old amateur crystal holders or WWII surplus that was later modified; labels fall off, cases are switched, quartz blanks are reground, etc. Sometimes the case is empty or the crystal has been cracked. We often want to know if there is a good chance that a crystal will work in our gear, and the best test always seems to be in the equipment its intended for. That said, one can’t walk around the flea market with a BC-610 under his arm or take the case off a Viking II just to spot-check a crystal or two, so something like this for $10 and about an hour of your dexterous soldering skills is kinda handy.

Besides, kits are fun.

This adventure started when a friend of mine with an interest in restoring WWII communications gear was using one of these testers to screen crystals. He commented to me that he had a hard time finding good 80-meter FT-243 crystals. While a little harder to find, their reliability generally pretty good as these things go, as they thicker(less likely to be cracked) and the manufacturing tolerances are looser. All his 40 meter FT-243’s tested fine, but he said he had “gotten burned” on a few 3885 KC crystals that ended up being duds. He had more faith in a modern crystal oscillator that had an apparent track record of service more than the BC-611 that might or might not be working right. His one 80m crystal–a modern HC49/u design, seemed to work okay in this tester.

Was this bad luck or is something else going on? Suspicious, I ordered a couple.

False advertising?

I won’t bore you with the assembly–you can watch glowing reviews of the this device online and you can even watch someone put it together on youtube (it’s slightly more interesting than watching someone play a video game, but you’ll have wasted a small portion of your life all the same). I ordered a couple kits–one from “Banggood” (perhaps not the best onomatopoetic name for an electronics retailer, but I’m not a marketing guy) and another from the back of a van in one of Shanghai’s darkest allies Ebay. The kits were identical as to the parts and went together in under an hour. The parts placement diagrams were all you needed and they were correct!

The Ebay package came with instructions in absolute pidgin English, along with a photocopied schematic that could fit in a fortune cookie. Despite its lexicographic shortcomings, it had one piece of helpful information:

Before measuring crystal please switch the J1 jumper cap, in JP1 insert 4 m to 40 m crystals can measure the crystal frequency

(sic)

Frequency counter installation instructions

Mirabile dictu!

Well, someone admits that the crystal measuring range more limited than that of the counter, regardless of whether this note refers to wavelength or frequency. It seems this is a pretty big sin of omission (or ignorance) in the online advertising for this thing; I can’t find any references to the limited crystal range anywhere.

Going for a test drive…

The Radio-Idiot Lab
The Radio Idiot’s lab. I’ve been told I have quartz disease and that there is no cure.

I broke out the rock collection to get some idea of how useful this thing is for screening ham-band crystals on 160, 80, and 40 meters–the old “novice era” stuff. Of the thirty or so known-good FT-243 in the 3-4 MHz range I tested, perhaps 25% registered their marked frequency some of the time. Often I got nothing on the display and the scope confirmed no oscillation, or I got the third and sometimes fifth overtone (9-12 MHz), which is usually a little below the arithmetic harmonic of the fundamental…Sometimes I got a random number generator. None of my crystals below 3 MHz would oscillate at all. Everything above 6 MHz behaved much better, although it was a rather stingy tester at 40 meters. Many of my 1930’s Bliley BC3’s didn’t want to start there, but X-cut crystals are always a little less active than the more modern AT or BT cuts. My aging WR-50B doesn’t always start them, but they do just fine in the transmitters. Like I said, these are all known crystals that work wonderfully in any of the transmitters I have and I even use with some regularly in the old Adventurer in the picture.

The oscillator

The kit uses a Colpitts crystal oscillator with a bipolar transistor whose output goes directly to the RA4 pin of the PIC by way of a DC blocking cap (C6). The PIC simply counts waveform peaks each second and shows you the number divided by a million to get MHz (more on that in a bit).

The Colpitts oscillator in the kit. Test crystal is across pins 1 and 3 of JP1.

The choice of a bipolar transistor, bias resistors (R1, R4), and low-value capacitors all points to one thing: this circuit is designed more likely to oscillate with crystals whose fundamentals are above 10 MHz.

Circuit analysis

Let’s consult Matthys 1983 for Colpitts design parameters. This is a fabulous reference for transistor crystal oscillators that discovered as part of this project. He says that at any given frequency there is a minimum shunt resistance to the transistor that will allow oscillations to start. He even gives us a handy chart with empirically derived values:

From Matthys, Robert J. ” Crystal Oscillator Circuits”, Revised Ed., Krieger Publishing, 1992, p.34.

The shunt resistance in our circuit is a function of the bias resistor (R1) the emitter resistor (R4) and the gain of transistor. The actual gain of a bipolar transistor isn’t that well controlled from what I understand, but if we assume a beta of 100 it is probably in the neighborhood of a few kilo-ohms. This explains why <4 MHz crystals won’t start: too little shunt resistance to get going. It also explains why the more modern vacuum-sealed crystals tended to score a little better on average. Packaging in air (like a pressure mounted FT-243) increases the equivalent series resistance of the crystal about 3 times on average (Matthys, p. 7), and that makes the already-to-low impedance of the transistor base that much more mismatched.

To prove I was on the right track, I changed R1 and R4 to 470K and 4.7K to raise the shunt resistance. I found I could get crystals to oscillate reliably down to 900 kHz, but they still wouldn’t count. The scope showed the strength of the oscillations to only be about 1 or 2V P-P. The PIC requires at least 3V P-P, according to the datasheet (o.2Vdd and 0.8Vdd for the quasi-Schmitt trigger input).

Once again, let us ascend the mountain to consult the oracle…

Matthys, p. 38. Same circuit as the tester.

We see that the bottom capacitor (C1 in Matthys, C3 in our oscillator) should be about an order of magnitude larger than the 22pF that came in the kit if we want to work at the low HF range. Indeed, changing the capacitors to 47pf/200pF brought the oscillation amplitude up significantly–still stingy when it comes to triggering the counter, but at least the most of the crystals started counting and I don’t get any overtone oscillations. In my opinion, it’s still too stingy for screening crystals for the old boat anchors, but perhaps further improvements could be made. I don’t think that is worth doing, as there are better circuits to drive the counter than a unbuffered bipolar Colpitts.

This circuit is really terrible for this purpose, but don’t take my word for it!

Prophetically, Matthys warns against the exact properties our circuit has when used with crystals in the low HF region:

In the [bipolar] transistor-Colpitts circuit parasitics will occur at some nonoptimum circuit values. In contrast, no parasitics of any kind have been found in the FET-Colpitts circuit. The parasitics turn out to be third harmonic oscillations or a combination of fundamental and third harmonic oscillations. The circuit values are rather critical for obtaining this harmonic oscillation. The effect can be enhanced by decreasing the crystal’s shunt resistance down to a point where the fundamental frequency is discouraged from oscillating while still keeping the shunt resistance high enough to permit oscillation at the third harmonic. Setting the time constant R,C, for the third harmonic frequency also helps. Both third and fifth harmonic oscillation have been reported by Bahadur and Parshad [16]. The amplitude of oscillation obtained this way is rather low, and there is a better harmonic Colpitts circuit available, which is discussed in the following paragraphs.

Matthys, p. 36

I think that covers the bases: bipolar transistor, low shunt resistance, short time constant from C2 resulting in low oscillation amplitude, harmonics and spurs….The spurs and multiple frequencies which in our case manifest as random digits on the display as we confuse the software.

Conclusion

I wouldn’t use this circuit as a go/no-go crystal checker for anything except modern > 10 MHz fundamental crystals (e.g., computer clock crystals)…for that purpose it is kinda fills a useful niche. But even then you may run into trouble with older third overtone types in the 9-15 MHz range where both the fundamental and overtone are encouraged more or less equally. Of course, crystals marked in this range usually don’t tell you if they are fundamental or overtone so your are still guessing with these old cans.

This oscillator design simply can’t work reliably over more than about two octaves while supplying enough output to drive the PIC’s input. Look again at Matthys’s Table 5.1 above: you can’t pick a shunt resistance value that works at 160 meters and still gives sufficient output at 20 MHz for >3V P-P. The original designer of the counter firmware even mentions the desirability of a buffer amp to present a consistent signal to the counter input…it seems the Chinese didn’t heed his advice. One variant of this board in the wild puts the full supply voltage on the transistor (9V? 12V?), which drives a much larger output voltage. I’m quite sure this counts over a wider range…until it fries the input of the PIC!

Not all is lost…I think you can make a much better crystal oscillator to drive a PIC- or arduino-based counter input for these purposes and do so very cheaply. That’ll be a future post.

I just wish my friend hadn’t trashed $100 of perfectly good and hard to find amateur-band crystals because of this…that kinda makes me angry.

Leave a Reply

Your email address will not be published. Required fields are marked *