Annex A
JUMA Frequency Step Accuracy
The Juma TRX2 using firmware revision 1.06 has an inherent frequency resolution of 10Hz.
This means that the actual frequency to which the transceiver is tuned is within ±5Hz of the
displayed frequency, if we neglect the reference oscillator errors. Or is it?
If we carefully examine the frequency display, we can observe some interesting quirks. Try
this, tune the transceiver to, say, 10.00000Mhz, and select the fast tuning rate of 100Hz, now
rapidly rotate the VFO knob to increase the frequency and carefully observe the least
significant (10Hz) digit. After a large number of steps the digit will suddenly change by 1.
Select the slow (10Hz) tuning rate and carefully go back and find the exact frequency at
which the 10Hz digit changed, and note it. As an example, in my case it was 10.20840MHz.
Now re-select the fast tuning rate and increase the frequency, and in my case at 10.46510 the
10Hz digit changed again and with a single increment from the tuning knob the display
changed from 10.46510 to 10.46519, and the next increment was to 10.46529.
The reason for this anomaly is not hard to find. The output frequency of the synthesiser is:
local oscillator
f
= N * 180000000 / 2^33 Hz
Where N is the 32-bit binary word used to load the DDS chip. In fact, the actual output
frequency is twice this, but the synthesiser’s frequency is divided by two to obtain the phase-
quadrature local oscillator signal.
The frequency steps are obtained from an array of steps and for a step increment of 100Hz the
value is 4,772. This is the binary increment that is added to the existing frequency word for
every step generated by the VFO encoder at the 100Hz tuning rate.
The actual frequency increment is therefore:
increment
f
= 4772 * 180000000 / 8589934592
= 99.996104837 Hz
There is thus an incremental step error of
!
0.003895164 Hz for each step, and eventually
these step errors accumulate until there is sufficient error for the next step to cause the 10Hz
digit to change.
In fact, after every 2,567 steps this anomaly will appear, corresponding to a frequency change
of about 256Khz.
Now, one could reasonably say, “So what?” and I would entirely agree that for all practical
purposes this is of no real significance, the actual frequency of the transceiver is always
within ±5Hz of the displayed frequency, which is more than adequate – even on the lowest
20