Crystal calibrator / Frequency standard

[Skywave]

Intro.
These days - as a hobby in radio / electronics - I tend to specialize in test gear: the overhaul & design of. My latest creation is this crystal calibrator / freq. standard unit. It's quite basic: it arose when I discovered a substantial collection of TTL logic i.c.s in my stock, which date back to the mid-1970s when I used to design logic circuits. At first, I thought "I'm never going to use these now - perhaps it's time to chuck them out" - then I had an idea
Overall scheme.
10 MHz crystal oscillator produces a 1:1 square wave. Subsequent TTL logic consists of a chain of dividers using 7490 decade counters, 7474 flip-flops and a few inverters / buffers: nothing exotic.
Next, a simple resistor chain potential divider provides a range of different amplitudes of signal to the chosen EUT. Various freqs. and voltage levels are each switch-selectable from the item's front panel. O/P is on a BNC socket.
Finally, a conventional a.c. - d.c. linear regulated PSU is fitted, featuring two 3-pin voltage regulators.

Intended uses.
Checking time-base of oscilloscopes; checking calibrations of signal generators; freq. counters; radios. Can be used as a signal injector for fault finding on radios & audio equipment. Use as a selectable freq. source for developing various logic ccts.,etc. There are probably other applications as well . . .
The O/P voltage level arrangements are a bit crude, but adequate for intended purposes. Those levels are not designed to be used for calibration purposes: they are for 'convenience only', whereas the freq. O/Ps are intended for accurate indications.

Construction and mechanical details.
Much of this is evident from the attached photos., q.v.. During development, it became quite clear that meticulous attention was required towards good R.F earthing (of course) - hence the copper tape strips seen in the photos. The number on the mains xfmr. has no relevance here: it's simply my in-house catalogue number. The case was one I found in my assorted salvaged metalwork collection. The only item which was necessary to purchase was the 10 MHz TTL osc.: cost about £5.

I haven't included the cct. diag. here - but can easily be added if requested to do so.

Al.

[Skywave]

An addition to the above post.

The main reason I sent in that post was simply to demonstrate that by using bits and pieces of electronics hardware that most of us have kicking about - especially old stuff that we've forgotten about - can be used to make something useful.
This item could be made by almost anyone here. Apart from the obvious requirements for any project build, care is needed in its physical assembly on account of the high frequencies involved. And I'm sure someone could produce a better attenuator than my effort.

The cct. diag. is available on request: there is no patent on this design!

Al.

[Joe_Lorenz]

Dear Al,

this seems to be a very useful project idea. In the 70s I got a shopping bag of Siemens DM 5400 and similar from their test lab. Wondering what to do with these, at last I built a digital control unit for the central heating system of our house. All output was formed by BD 115s and this proved to be bullet proof. The unit was in continous use for more than 30 years without any problems, but nowadays just outdated.

Regards, Joe

[Skywave]

Hi Joe,
Thanks for your post. You state that your project is now 'outdated'. O.K. - but does it still work? I've have many an item here which is also 'outdated', but still works and, moreover, is in constant use.
Just curious . . . .

Al.

[CambridgeWorks]

Nice piece of kit Al.

Back in the late 70s, I made a calibrator, mains powered, 1Mc/s xtal osc, 7490 dividers and 1Mc/s, 100Kc/s and 10Kc/s outputs on a bnc.
In the early 80s, with a young family to look after, I had a clear out at a rally or two of stuff I no longer used.
Fast forward to around the year 2000, at a rally rummaging under a club junk stand I picked out a distinctive blue and white metal box with mains lead attached and recognised it. However, I couldn't figure out where I had seen it before though.
Then, I saw my own callsign in letraset on the front panel! I just had to buy it! 50p I think it was?
It now sits, pride of place on top of a box of radio items in my garage!!!
Despite this, I doubt I would ever part with it again though.
Rob

[G6Tanuki]

That's really rather neat: I admire the attention-to-detail of using copper foil to provide a really-good low impedance ground connection to the TTL and so avoiding each of the divided-down frequencies having a nasty 'comb' of switching-products alongside.

[Skywave]

I must admit that arranging for a 'clean, low inductance earth' presented the biggest challenge in building this item. Several attempts of layout were tried until the one used (and shown) was chosen. How to determine the cause(s) of unacceptable waveforms when looking at very fast rise and fall times of square waves with a 'scope + probe that have a combined B/W of about 100 ~ 130 MHz doesn't help either. And, of course, for this task, the use of the probe's 'wandering earth lead' is an absolute no-no.

As for the O/P attenuator, that's where the biggest compromises are. I tried the classic capacitor-compensated attenuator approach with unsatisfactory results: ultimately, I settled for a simplified version. The net result is waveform shapes that gradually degrade as the attenuation increases. However, this item is primarily for generating accurate frequencies: the variable attenuation is really a necessary convenience.

Finally, initially I did contemplate using a TCXO, but for this item's intended purpose, I decided that the cost / benefit ratio of that choice was not justified.

Al.

[G0HZU_JMR]

Ideally, to prevent overshoot and undershoot when connecting the BNC output to a high impedance scope input via 50 ohm RG58 coax the source impedance at the BNC connector needs to be maintained fairly close to 50 ohms at all attenuator settings. Otherwise things can degrade quite a bit. If the source impedance is too low there will be overshoot and ringing.

If the source impedance is too high the risetime can get very droopy and rounded when viewed on a scope.

How to determine the cause(s) of unacceptable waveforms when looking at very fast rise and fall times of square waves with a 'scope + probe that have a combined B/W of about 100 ~ 130 MHz doesn't help either

If it helps, it's possible to predict the overshoot (or undershoot) and the subsequent ringing with a simple excel spreadsheet as long as the source impedance is known and the length (delay) of the RG58 coax cable is known.

This simply calculates the reflection coefficient at the far end and at the (non 50 ohm) source and predicts how the waveform will appear on a scope. The alternative is to simulate it.

A reasonably good output source would be a classic line driver chip with maybe a 27 ohm resistor in series at its output. This is a common setup and can work well as a 50 ohm source. It might be able to drive a series of 50 ohm attenuator sections in your attenuator switch but it may require a beefier line driver chip to drive this.

[G0HZU_JMR]

I dug out a basic line driver chip, I think it is an old SMD 74HCT244 but the logo is very faint. This isn't the greatest line driver in the world but when I add a series 27R resistor at the output it does a pretty good job of driving a 2MHz square wave into a decent JFW 50 ohm step attenuator and this is then fed via about 4 feet of RG58 coax into a 1Meg scope input.

The square wave looks good at all attenuator settings with nice and crisp risetime and a flat top to the square wave. In other words, I can attenuate down from 5V pkpk in 1dB steps to just a few mV pkpk and I still see a classic square wave at the scope.

The scope is not terminated in 50R and the reason I get this good performance is because of the line driver + series 27R resistor and the 50 ohm step attenuator after it. They provide a good 50 ohm source impedance to the RG58 cable at all attenuator settings.

[G0HZU_JMR]

Note that the 74HCT244 line driver chip with the 27R series resistor isn't really beefy enough to use as a permanent solution to drive a 50 ohm load. It is the only line driver chip I could find here in a hurry. It might be possible to parallel more than one driver input together with suitable resistors on each output pin but I think a beefier line driver designed for 50 ohm use would be a better choice if anyone wants to experiment with this arrangement.