Go to the Tech Index page Back to Home

Basic Information On Temperature Compensation and Modulation Capability for Channel Elements, ICOMs, and TCXOs Compiled from contributions from Eric Lemmon WB6FLY, Jeff DePolo WN3A, Robert Meister WA1MIK, Kevin Custer W3KKC, and Mike Morris WA6ILQ Edited and HTML'd by Mike Morris WA6ILQ

Concept:

This page deals with the subject of replacing the crystal in any type of Temperature Compensated Crystal Oscillator (TCXO), be it a Motorola "Channel Element", a GE "Integrated Circuit Oscillator Module" (commonly called an ICOM), or an RCA TCXO. In this article we're going to call it an "element" just to make things easy, and describe the problems that can arise by working on them yourself. Most hams are frugal (some are downright cheap) by nature, always wanting to save a nickel anywhere they can. Replacing the crystal in the precision frequency determining element is a necessary task when taking a crystal controlled radio and moving it from one frequency to another. Money can be saved by recrystaling the elements yourself, but there are problems that can arise when doing so. Many folks don't know about the issues when buying a $10 crystal and stuffing it into the element themselves. Sometimes the issues are with the crystal itself, sometimes they are with the element, and lastly, both can contribute to the end result.

A Very Perceptive Comment on Rebuilding Elements, from an email to Repeater-Builder from Eric Lemmon WB6FLY:

...There still seems to be a tendency to seize upon temperature compensation as the primary reason for sending the element to a crystal house, and that is not the case. In fact, the temperature compensation is not the most important issue at hand.

It should be obvious that few, if any, of the above actions can be performed on a crystal by itself, since most of the components that directly affect the operation of the crystal are not in the crystal. In other words, whether or not the radio is in a temperature-controlled environment is hardly the point; it is more important that the crystal perform properly in the oscillator circuit. That's hard to verify if you don't send the oscillator circuit to the crystal house!

Notes on Temperature Compensation:

The first thing to realize is that the crystal frequency varies with ambient temperature. Period. It's a simple law of physics. The frequency the crystal oscillates at is also determined by the circuitry around it, and the values of the components in that circuitry. The crystal manufacturers even have a name for this characteristic - you will see references to a crystal "operating into a 32pF load" (or some other number). If the load value changes, the frequency changes. The method used in temperature compensating a crystal is to deliberately use other temperature sensitive oscillator components to vary the capacitance across the crystal and the load on it, thereby counteracting the temperature-caused change in crystal frequency.

A bit of physics: Crystals are made by putting a rectangular slab of polished quartz between two conductive plates. Picture a sandwich made from two pieces of bread and a slice of cheese in the middle - the slices of bread are the plates, and the cheese is the quartz crystal. Raw quartz has a grain, just like wood, and the angle of the cut relative to the grain has a lot to do with the performance of the finished crystal. There are several preferred angles to cut quartz for crystals, and the companies that do communications crystals use the cuts that are optimized for frequency stability. Companies that make crystals for digital circuit clock oscillators use cuts that maximize the number of crystals per slab (i.e. make them to a price target) and rely on the fact that any frequency error is divided as the crystal frequency is divided down to a target frequency. A drift of 100 Hz on a 1 MHz crystal divided down to 10 pulses per second is one millisecond. You will never see a 1 ms drift (even if it was 10 times that - a full kilohertz) if that crystal is in the digital LCD clock on the wall.

This situation is the exact opposite of the communications world where crystal frequencies are multiplied up to a operational frequency and every frequency error is multiplied as well. This is how the modulator in an FM transmitter works - a UHF transmitter using a 12 MHz crystal multiplied 36 times only has to wiggle the crystal frequency about 140-150 Hz to get a full 5 kHz deviation. A 900 MHz transmitter with a times 48 multiplier needs a 50 Hz wiggle to get a full 2.5 kHz deviation. 50 Hz is more drift than most crystals have during warmup!

In short, you need to get your crystals from a company that understands the technical requirements of the communications crystal as opposed to a microprocessor (or timepiece) clock crystal, and from one that operates to a quality target, not a price target.

Note from WA6ILQ:
By the way, if your crystal is in a temperature stable environment, then no temperature compensation is needed. This is why two-way radios in the 1940s through the 1970s used crystal ovens - small plug-in housings that kept the crystal at 80 degrees C (about 175 F). This is also why so many people "get away" with changing the crystals themselves on their home base stations - the radios are in a temperature stable environment (the average heating/air conditioning thermostat is pretty good at keeping an area to within 5 degrees of a selected temperature). Repeater sites are a whole different situation - I've personally seen cabinet interior temperatures as low as -20 degrees C (-5 F) and as high as +50 degrees C (+130 F) at sites that had no environmental controls. I've also seen sites that have air conditioning / heat pump systems (i.e. cooling, heating and humidity control) that kept the equipment within a 10 degree window year around. And note that just because the building has a nice environmental control system that does not mean that you can scrimp on the crystals or the elements - the temperature inside the radio, where the crystal or element is located, is often much warmer and more widely variable than the room temperature. Those RF shields make great airflow shields too.

So from the above, you can see that the performance of the element during extreme temperature swings is only as good as the crystal installed in it and the compensation that was done at the time the crystal was installed in the element. The element only achieves its specified stability if the element is matched to the crystal. If the crystal tends to swing upwards in frequency as the temperature rises, the element has to swing lower, and by the same amount. Likewise, if it swings down in frequency, then naturally the element has to swing upwards. Another way to think of this behavior is "temperature to frequency tracking". The desired end result is a crystal and element combination that maintains the specified frequency and stability, and operates properly in the circuit for which it will be used, and does it from one temperature extreme to the other. And the previous discussion is only concerned with the oscillator - and has not touched on if the crystal can be modulated, or if the modulation is equal above and below the center frequency, or having a stable output level, or a pure signal.

When placing a new crystal in an element yourself, you take the chance that it may or may not work correctly (and the odds are stacked against you). When someone re-crystals an element, it's a game of chance if the values of the existing compensation parts that were installed when the element was manufactured (or was last recrystaled) will accurately track the new crystal to specification. There's a chance it could, there is also a chance it could make it worse! If it does work out, it's only by blind luck. And the average ham does not have the test equipment to validate the new combination of crystal and element.

Many folks will choose to recrystal their elements themselves to save money, and, in some situations, that's fine. However, just because you have an element rated at 5 Parts Per Million (PPM) (i.e. a GE "5C" element) doesn't mean the element with your new crystal inside it will provide you with 5 PPM stability. As you now realize, when you change the crystal in an element, you will almost always need to make other component changes inside the element to ensure its designed temperature stability. And it's more difficult with a 2 PPM element (the ones required in the GMRS service), and even worse on a 1 PPM element used at 800 MHz, 900 MHz or 1200 MHz.

Note from WA6ILQ:
And what was the source of the element?

I've seen the following scene too many times: Joe Ham finds a surplus GE or Motorola station in surplus, and buys it. He carts it home then discovers that the station has no elements. None of his friends have any in stock. So he shops on eBay, or has a friend pick up a set at Dayton. He orders crystals from Driftal Inc (or some other lowest-price place) and mounts them in the elements. He fires up the station and has nothing but troubles.
So what's wrong?
1) He does not know if the station had problems before he bought it. It may have had more bench time than in-service time. I've seen used stations where every major assembly had problems (i.e. it had been cannibalized to fix other stations, and all the bad assemblies put into that chassis that was then sold to an unsuspecting ham).
2) He does not know if the elements are good. They may have been at Dayton because they were bad and someone wanted them off of his shop shelf.
3) He does not know if the cheap crystals are temperature stable and if the transmit element can modulate adequately and symmetrically.

This is why time and time again you see the comment on the Repeater-Builder mailing list: always get the old elements with the station and verify the operation of the station on the old frequencies. If you can get an extra set of elements so much the better. A half hour with a dummy load and a service monitor (or a good signal generator and a spectrum analyzer) will let you know if you have a good radio BEFORE you go setting it up on amateur channels. Then take some KNOWN GOOD elements and send them to the crystal house for recrystalling. They will be tested for operation, output level and spectral purity before your new crystals are installed, and then tested again afterwards before the compensation process.

Back to elements and compensation...

Note that the compensation components in the element are dependent on the characteristics of the individual crystal, and those characteristics have to be measured (with extreme temperatures) after the crystal is made. The components are then selected by hand and installed. Hence, this has to be redone whenever the element is recrystaled. Most of the cheaper suppliers build a crystal, install it, and change the primary parallel capacitor so the channel element resonates "on frequency" with the trimmer device centered. This one component has nothing to do with temperature stability, modulation capability, modulation symmetry, output level or output purity.

Speaking of modulation capability...

A crystal with the wrong cut may not exhibit enough rubberiness for full and undistorted deviation to occur (yes, we invented the word. Can you think of a better one?). This situation happens when a crystal vendor cuts a crystal for absolute stability (by making the crystal stiff) and doesn't realize that it will be used in an FM transmit element. Even if the crystal does make enough deviation, it may not swing the same amount on each side of the center frequency, or the modulation may otherwise be distorted. International Crystal and Sentry are the only companies (and ICM bought Sentry in January 2005) that I know of that actually test the modulation capability and ensure modulation symmetry in their finished elements. ICM and Bomar are the only companies that I know of that actually change the temperature compensation components (capacitors and resistors) to properly temperature compensate the element. And we use the term "modulation symmetry" - that means that for a given modulating audio voltage that the carrier shifts an equal amount above and below the idle frequency. Poorly made or poorly compensated crystals don't have a symmetrical response to deviation voltage, causing the average center frequency to shift during modulation.

The actual schematic of the GE ICOM is not available (the schematic in the LBIs is simplified, and on it the frequency compensation area is just a block diagram), but the schematic of the equivalent Motorola Channel Element is below, courtesy of the reverse engineering capability of Scott Zimmerman N3XCC, of Repeater-Builder:

Look carefully at the schematic. Q1 is the oscillator. Q2 is a tripler-buffer. X1 is the crystal). Go north from there, and across to C3, the trimmer. Then find L1 and C7. Nothing north of the crystal but C3, L1 and C7 is needed to make an oscillator. Now find the AFC pin, C2, D1, R1 and R2. Those components are all that is needed to make the frequency move with the AFC voltage. Everything else that is north of the crystal is compensation.

To deal with temperature stability, the crystal and element must be heated and cooled under strictly controlled conditions, and the frequency must be measured accurately at each temperature extreme. New compensation components are then chosen and installed in an attempt to minimize the overall frequency change of the assembly. This must be repeated until the desired stability is achieved. It's a trial and error process; each heat/measure/cool/measure/component-swap cycle could take several hours (so the worker is testing another element while the first one is heat-soaking or cold-soaking... in fact he or she may be rotating as many as a dozen across his/her bench and through the chiller and the oven at any one time). And remember that in most circuits the intent of the designer is to select components that have a minimal effect from temperature changes - here it is the exact opposite in that the components in the compensation area are specifically and carefully chosen for their changing characteristics at varying temperatures.

Once the temperature compensation is done the element is tested again for proper RF output level and spectral purity on the new frequency. If it is an AFC receive element they test it for proper direction of shift with AFC voltage, plus the sensitivity to that voltage and proper range. If it is a transmit element and has an internal FM modulator they test it for modulation sensitivity, symmetry and distortion under ambient temperature and at temperature extremes.

When you are ready to recrystal an element and are calling around to get prices, you need to ASK what each company does when "compensating" an element. Not all will go through the time-consuming process of hand-selecting components and thermal cycling in the two environmental chambers (one for heating, one for cooling, and both use air drying equipment so that there are no condensing humidity droplets to cause problems). Considering what you are going to spend on the entire repeater, paying a little extra for a professional top quality crystal house to manufacture the crystals and do the compensation is definitely worth it. If you are serious about your repeater, especially if it's going to be located in a hilltop building with no climate control, do yourself a favor and send the element(s) back to the crystal house and pay to have them compensate the element to the crystal. After all, those people have the thermal chambers, the air drying equipment, have a wide variety of the proper temperature sensitive capacitors and thermistors in stock, and have the expertise to use them to their best effect because they do it 8 hours a day, 5 days a week, for 52 weeks a year.

Crystal and Element Specifications:

Crystals and elements are built to certain stability specifications, measured in Parts Per Million (PPM), and refers to how many Hertz a crystal or element can drift (vary in frequency) with changes in temperature from a known reference point. In a GE Mastr II, that's 80 degrees Fahrenheit. A 5 PPM crystal / element combination can drift 5 Hertz for every million Hertz of operating frequency. So, a 146.000 MHz element can drift 146 times 5 which equals 730 Hertz. This means a 146.000 MHz element can vary plus or minus 730 Hertz and still be within the plus or minus 5 PPM specification. Holding to this same specification, a UHF transmitter operating on 445.000 MHz can vary 2,225 Hz and still be within specification. And remember that the plus and minus 2,225 Hz tolerance at UHF is only 61.8 Hertz at the 12 MHz crystal frequency (due to the times-36 multiplication)... and it's even tighter if the exciter uses an 8 MHz or a 6 MHz crystal (about 30 Hz).

Yes, that means that your 12 MHz crystal oscillator signal can drift at most about SIXTY-TWO HERTZ each direction of the center frequency from the coldest winter night to the hottest summer day and stay legal at UHF. On 220 MHz it is 360 Hz. On VHF it's relaxed to 730 Hz. That's still not much. It's even less at 1200 MHz... about 20 Hz.... Warmup drift in most equipment is more than that.

The tighter frequency tolerances and 2.5 KHz devaition limit on 900 MHz makes it even worse - one 927.5 MHz transmitter I worked on has a multiplier of 48. One PPM at 927.5 MHz is 927.5 Hz. Divide that by 48 and you have 19.3 Hz. This means that your 927 MHz repeater transmitter oscillator has to stay within 20 Hz of frequency !   And the 2.5 kHz deviation at 927 with a times 48 multiplier only needs about 50Hz of wiggle for full deviation. Yes, a drift of 50Hz will push the transmit signal right outof the bandwidth of the receiver that is hearing that 900 MHz signal.

From the above, the FCC rules notwithstanding, it's plain to see why people choose 2C elements for UHF repeaters and mobiles (resulting in a maximum of about 900 Hz variation) and 1C for 800 and 900 MHz (800 Hz or 900 Hz variation). It's also plain to see why you should spend the extra money and have a reputable crystal house redo your repeater elements so they are properly temperature compensated. Also remember, you get what you pay for. I have heard the complaint many times "ICM wants $50 to replace the crystal in the element". Now you know why, and that it's not all that simple. It's more than just mounting and soldering the crystal, it's many hours of heating, measuring, cooling, measuring, installing compensation components, and repeating the cycle until it's absolutely right. And then testing for modulation symmetry, modulation sensitivity, minimal modulation distortion, adequate RF output level, output purity, AFC sensitivity and range.

So go ahead and buy the crystal for $20... Can you do the rest at home for $30 more ?

Note from WA1MIK:

I recently sent two Motorola MICOR receiver channel elements to ICM. They charged $19.95 for each crystal and $30 to install, temperature-compensate, and test each element. The entire process should take about a month, and more than half of that would be devoted to the compensation and testing phase.
(Follow-up: elements sent in 1/19/07, received back 2/15/07. Worked perfectly.)

Don't take the above quoted price as an absolute - redoing a 0.0005% element is the least expensive, an 0.0002% element costs more, and a 0.0001% element (used at 800 MHz, 900 MHz or 1200 MHz) costs the most.

Notes on Ordering Crystals: