TCXO Testing

[NZ0I]

As discussed in an earlier email exchange, the original TCXO component we tried (Fox Electronics FOX924B-25.000) did not have acceptable performance. When used with the Si5351 it would result in frequent, and obvious, jumps in frequency when outputing a VHF (~112 MHz or above) clock signal. The jumps were on the order of 10Hz or so, and would occur as the FOX924B would make frequency adjustments to compensate for temperature changes. The jumps were not apparent when the Si5351 was set to lower frequencies. The frequency jumps did effectively keep the oscillator within tolerance of 25MHz, but the sudden and frequent corrections were multiplied by the Si5351 and would make for a bad user experience were the clock used to generate a VFO signal for a SSB receiver, since the jumps would propagate directly to the mixer products and down to the operator's ears.

After discovering how durable the OSH Park PCBs are, I became confident enough to desolder the Fox TCXO and replace it with an alternative clock component. Desoldering involved using a 150-watt soldering gun to heat up the TCXO until all four pads reached the temperature of molten solder, freeing the device from the board. The process worked well, leaving all four pads in clean undamaged condition. The Fox TCXO was discarded, since it was likely damaged by the excessive heat.

The next part I tried was a MEMS oscillator, the DSC6101CI2A-025.0000 by Microchip Technology. It claims to have frequency stability of about ±25ppm and "... exceptional frequency stability and jitter performance over temperature" according to the spec sheet. The cost of the MEMS device is about the same as purchasing a crystal of about the same stability. The current consumption of the MEMS device is only 3mA, or roughly half the current of a TCXO. It seemed like it might be a reasonable compromise between a TCXO and a crystal.

Testing conducted on the device was subjective. It involved listening to the signal output by the Si5351 using an Icom IC-R10 handheld receiver set to CW mode, and making qualitative observations regarding its stability. Although the test results fall far short of the precision one expects from a more formal (and properly instrumented) test procedure, the qualitative tests should be adequate for judging the relative stability performance of the parts, and in some cases (like the Fox TCXO) it might help rule out some devices as being unsuitable for this application.

DSC6101CI2A-025.0000 Results

This MEMS component has a smaller footprint than the FOX924B-25.000 that it replaced. But luckily, the DSC6101's pads just overlapped with the inside corners of the original pads, and was soldered in place without any issues. The first observation of the 144.42 MHz signal coming from the Si5351 after installing the DSC6101 was that the "jumps" were gone. But small apparently-random fluctuations in frequency were observable: a slight, but noticeable flutter of several Hz.

Using a steady stream of compressed air from a "dust remover" compressed air can, the MEMS device was selectively cooled. The temperature was not measured, but the device was likely more than 20 degrees cooler than ambient, but still above 0C. The 144.42 MHz signal generated by the Si5351 dropped in frequency by approximately 1kHz as a result of the cooling. The frequency recovered almost immediately (within ~1 second) after the compressed air flow was stopped, suggesting that the device has little thermal mass and is closely coupled thermally to the PCB, allowing heat from the PCB to quickly flow back into the device bringing the oscillator back to ambient.

Placing the tip of a 780F soldering iron a few millimeters from the DSC6101 caused the 144.42 MHz signal to rise by several kHz. As with the cooling experiment, the clock returned to its original frequency within ~1 second after the soldering iron was removed. The temperature of the part during high-temperature testing probably did not exceed 50C or so. As a check, the soldering iron was allowed to cool and was then placed near the MEMS oscillator - no discernable change in frequency was detected by bringing the room-temp soldering iron close to the chip.

The experiments were repeated several times with identical results. The DSC6101 responds so quickly to temperature changes that it would not be surprising if the small observed flutter is the result of tiny localized temperature fluctuations at the DSC6101.

Summary

The Si5351 coupled with the DSC6101CI2A-025.0000 results in a smoother output signal at 144.42 MHz. But using the DSC6101 as the clock for an Si5351 to generate a VHF signal for direct use in a receiver's signal chain would, it seems, result in marginally-acceptable temperature stability, and would likely require measures to stabilize the MEMS oscillator's temperature. 

Next up: the Abracon ASTX-H11-25.000MHZ-T

[NZ0I]

Abracon ASTX-H11-25.000MHZ-T Results

Removing the DSC6101CI2A-025.0000 by Microchip Technology, and installing the Abracon ASTX-H11-25.000MHZ-T in its place went smoothly. The ASTX-H11 has the same small footprint of the DSC6101, so it will result in a little additional free board space compared to the original Fox device. The first observation of the 144.42 MHz signal coming from the Si5351 after installing the ASTX-H11 was that the flutter was gone, and no more jumps. Hooray.

The same subjective seat-of-the-pants tests were conducted on the ASTX-H11 as were conducted with the previous two candidate components. Results were excellent. Only a subtle shift in frequency (less than 100 Hz) was evident while the ASTX-H11 was cooling or heating. But the slight shift would disappear after a few seconds once the temperature stabilized (high or low), and the TCXO successfully caught up to the changing temperature. Otherwise the output of the Si5351 sounds solid and stable under all conditions tested: ambient, hot (~50C) and cold (~5C).

Summary

The Si5351 coupled with the ASTX-H11-25.000MHZ-T exhibits no obvious deficiencies at 144.42 MHz. It appears that using the ASTX-H11 with the Si5351 to generate VHF signals for use in a receiver may not require any special measures in order to stabilize the frequency over temperature. This appears to be a very viable oscillator device. 

Currently no other devices are slated for testing. If cheaper, or less power hungry, components are identified then more testing can be done.