|
New FW release
Version v1.1-2
Changes:
- The console command "Marker {number} {value}" now accepts either a index (if below sweep points) or frequency as value
Examples:
>marker 1 100
sets marker 1 to index 100 (if sweep points > 100)
>marker 1 100M
sets marker 1 to 100M (if within sweep range)
If frequency is outside sweep range nothing happens.
Setting the marker to an index or frequency disables tracking
|
Hi Erik,
I just tried the modulation on the Low output.
The modulation frequency is correct from 1KHz up to about 9KHz. When set below 1KHz the modulation frequency is >1KHz
FM Wide produces an RMS deviation of around 42KHz regardless of modulation frequency
FM Narrow produces an RMS deviation of around 2.8 KHz with 1KHz modulation frequency
FM Narrow produces an RMS deviation of around 4.1 KHz with 2KHz modulation frequency
FM Narrow produces an RMS deviation of around 5.0 KHz with 3KHz modulation frequency
AM produces an RMS value of of around 14% regardless of modulation frequency
The FM Narrow deviation increases as the modulation frequency is increased.
For typical receiver testing it may be useful to have some fixed settings in addition to the current adjustable values.
e.g.
AM 30% modulation 400Hz
AM 30% modulation 1000Hz
NBFM +/- 2.5KHz peak 400Hz for 12.5KHz channel spacing
WBFM 50KHz RMS 400Hz
Etc.
400Hz is a particularly useful audio modulation frequency because it's on the flat part of the pre-emphasis curve.
Incidentally it may also be useful to show folks how to measure transmitter deviation with a spectrum analyser using the Bessel Null technique.
We used to use this method to check FM broadcast transmitters for compliance 'back in the day', before we had decent deviation meters.
Regards,
Martin
|
Martin,
Thanks for the feedback.
It should be possible to increase the AM modulation depth, will look into this
Weird you can not set the modulation frequency below 1kHz. Kurt did some testing and he could use till at least 100Hz.
Not sure why the NFM deviation should increase with frequency. Could be PLL overshoot as I am modulating inside the PLL loop. If so there is not much I can do to improve.
The Bessel Nul technique is referred to in the tinySA? wiki measurement examples:?
You are most welcome if you want to provide some better text for the wiki.
|
On Tue, Sep 29, 2020 at 10:56 AM, Erik Kaashoek wrote:
Weird you can not set the modulation frequency below 1kHz.
Hi Erik, I just rechecked and it is OK with modulation frequencies below 400Hz. For some reason my measurement kit didn't like the low deviation 500Hz RMS (it may have been locking to something else) that is produced with this modulating frequency, but I can confirm that it does indeed work with modulating frequencies down to 100Hz. My apologies for not spotting this. Are you using phase or frequency modulation, could this be the reason for the variation in deviation vs modulating frequency ? Regards, Martin
|
Hi Erik,
Is the AF modulation a square wave?
It looks like it could be when observed with a scope on the demodulated output from my deviation meter.
Regards,
Martin
|
As the components in the tinySA do not support FM modulation I change the fractional divider of the PLL to modulate the frequency. As this is done inside the PLL loop the PLL loop has to follow the change in register setting and the speed at which this happens will depend on the loop parameters which are unknown to me.
So I guess the dependence of the? deviation in NFM on the modulation frequency comes from the PLL loop having some problems (overshoot?) following the changes of the fractional dividers.
|
The modulation is a rather crude 5 point approximation of a sinus. Yes, there should be harmonics.
|
Hi EriK,
Ah OK that could explain a lot of things :-)
I hadn't realised you were doing it this way, but it makes sense now that I think of it.
Because of the non-sinusoidal modulation, I get variations between modulation values depending upon what method I use and what filters I apply.
I've dug out my old Racal 9008 modulation meter, and it's giving different readings to my Spectrum Analyser modulation measurement option.
I think this is because of the TinySA modulation wave shape and the type of detector used to measure it.
Regards,
Martin
|
Hi Erik, Demodulated output from my Racal deviation meter. AM 1KHz modulation  WB FM 1KHz modulation  NB FM 1KHz modulation - not so good, wide excursion spikes.  I tried a HP signal generator just to check the demod process was OK. 5KH deviation 1KHz modulating frequency. Note the deviation meter scale is set a bit more sensitive that the previous capture as the peak deviation was much smaller.  Regards, Martin
|
On Tue, Sep 29, 2020 at 05:35 AM, Martin wrote:
NB FM 1KHz modulation - not so good, wide excursion spikes.
Many thanks for the measurement.?Indeed, the PLL loop has some severe overshoot. Would a 7 or 9 point approximation help? That should halve the frequency steps The WFM and AM look correct.
|
On Tue, Sep 29, 2020 at 12:48 PM, Erik Kaashoek wrote:
The WFM and AM look correct.
The AM is a bit asymmetric but otherwise OK. The Racal modulation meter says it's about 100% modulated. I'm not sure if 7 or 9 point approximation would help. It may simply be the loop bandwidth, or something associated with it, that is the issue as you suggest. If you wish to send me test firmware to see if it makes a difference, that would be OK with me. Regards, Martin
|
On Tue, Sep 29, 2020 at 01:22 PM, Martin wrote:
The Racal modulation meter says it's about 100% modulated
Correction 50% modulated.
|
Hi Erik and Martin I just checked the new firmware and it is as when I tested it.
With my Stabilock the Am modulation is 27% and distortion some 15%. Due to the modulation with only 5 steps it is very difficult to measure any ¡°sensible¡± deviation. E.g. the NFM has a deviation of 17KHz and pretty high distortion so using it with a HAM radio transceiver the deviation at too high. Erik mentioned to me that some extra pulsing noise is reduced by not having the USSB cable connected which is correct Modulation can be set as low as 20Hz, despite the input table says 10Hz, but not to detect nether at 50Hz by Stabilock or SDR radios or output from a small ICOM general AM, NFM and WFM radio, due to squarewaves on top of squarewaves in one big mesh ? Kind regards Kurt ????? ? ? ? ? On Tue, Sep 29, 2020 at 05:35 AM, Martin wrote: NB FM 1KHz modulation - not so good, wide excursion spikes.
Many thanks for the measurement.?Indeed, the PLL loop has some severe overshoot. Would a 7 or 9 point approximation help? That should halve the frequency steps
The WFM and AM look correct.
|
On Tue, Sep 29, 2020 at 01:41 PM, Kurt Poulsen wrote:
the NFM has a deviation of 17KHz and pretty high distortion
Hi Kurt, The NB FM 'spikes' are peaking up to 22KHz according to the graticule on my scope. If I switch in the 3KHz LPF on the Racal 9008 modulation meter , the results are a bit more sensible, but still not good enough to use. Erik, is the Narrow and Wide FM modulation produced in the same way ? Can you reduce the 'size' of the modulation pseudo sine wave 'steps' in FM wide mode (or increase then in FM narrow mode) to see if we can find the point at which the stepped modulation turns into sharp spikes. This may give clue about what's happening. I also tried modulation in High mode and it's showing the same effect at 300MHz. The RF output level in Low mode is pretty close to the stated value, but in High mode it's good at -30dBm, but varies considerably from the selected value at other settings (typically 6dB out, but sometimes 10dB or more) it doesn't always step up or down by the same amount. If the attenuator is binary switched, I'm wondering if two of the bits have got crossed somewhere. Regards, Martin
|
WFM and NFM are done in the same way but NFM only makes small adjustments to the fractional divider.
The high output level steps are using some tricks to get to the levels. These tricks are not formally documented and the actual output levels can indeed deviate considerably from the target level
The spikes probably come from an overshoot of the PLL loop filter
In a new FW I switched to an 8 step approximation and only make the smallest possible steps NFM
Will send to you and Kurt for testing.
|
Hi Martin Aha, that reminds me that I have forgotten I also have an old Racal deviation meter but where is it ? Kind regards Kurt ? ? On Tue, Sep 29, 2020 at 01:41 PM, Kurt Poulsen wrote: the NFM has a deviation of 17KHz and pretty high distortion
Hi Kurt,
The NB FM 'spikes' are peaking up to 22KHz according to the graticule on my scope.
If I switch in the 3KHz LPF on the Racal 9008 modulation meter , the results are a bit more sensible, but still not good enough to use.
Erik, is the Narrow and Wide FM modulation produced in the same way ?
Can you reduce the 'size' of the modulation pseudo sine wave 'steps' in FM wide mode (or increase then in FM narrow mode) to see if we can find the point at which the stepped modulation turns into sharp spikes. This may give clue about what's happening.
I also tried modulation in High mode and it's showing the same effect at 300MHz. The RF output level in Low mode is pretty close to the stated value, but in High mode it's good at -30dBm, but varies considerably from the selected value at other settings (typically 6dB out, but sometimes 10dB or more) it doesn't always step up or down by the same amount. If the attenuator is binary switched, I'm wondering if two of the bits have got crossed somewhere.
Regards,
Martin
|
Kurt Poulsen