Seeing DCF77 radio on a wire antenna using FFT

[petert]

Hello,

I am trying since a long time to get a reasonable signal of the DCF77 time signal radio emitter/station.

First I used a magnetic antenna from a radio clock (that is made for DCF77), made out of a coil around a ferrite rod, and a matching capacitor.
I generated a signal of 77.5 kHz with a DAC to which I attached a wire, and tested that indeed the magnetic antenna picks up the signal (when reasonably close to the "DAC wire", as could be seen on my scope. When generating signals of other frequencies, they appear much more attenuated, as you would expect. So it seems the antenna is tuned properly.

I have not managed to pick up any radio signal of DCF77 however with this antenna, no matter how I turn it. Surprisingly, the alarm clock did work (picked up the DCF77 signal), however only in certain locations.

So I tried an 30 m wire antenna that is horizontal to the ground, attached to a tree. Using the FFT function of my scope, I can indeed see radio stations at 164Khz, and some around 88kHz and sometimes around 60kHz, 38Khz, or 20kHz.

Yet, I still cannot pick up anything on 77.5kHz (or close by).

Is the radio station maybe emitting a too narrow-band signal so that the FFT doesn't identify it reliably? Especially the 164kHz station is pretty wide band (I checked this with a WebSDR).

I am not sure really how to manage to finally receive this DCF77 station.

Any ideas?

P.S.: I tried to attach it to a TV antenna on the roof (directly to the scope using FFT again, but it didn't seem to be better than the long wire antenna).

[ogden]

Is the radio station maybe emitting a too narrow-band signal so that the FFT doesn't identify it reliably? Especially the 164kHz station is pretty wide band (I checked this with a WebSDR).

FFT indeed shall "see" signal even if it falls perfectly between two bins. In such case it's power will be equally distributed between two bins.

[edit] 1) Your scope "sensitivity" most likely is too low. 2) Usually magnetic antennas of DCF77 receivers are tuned circuits to increase sensitivity and selectivity.

First google hit (note capacitor parallel to magnetic antenna):

https://pe2bz.philpem.me.uk/Comm/-%20Receivers/-%20Misc/Tim-102-Simple-Dcf77/rcvr-e.html

[petert]

Thanks for your reply. But I am not sure what to make of your link. I have a DCF77 module from Conrad, but I'd like to build my own, starting with getting a clear signal of the radio station (well clear so that at least an FFT would show it).

Are you suggesting to put a capacitor in parallel to the wire antenna? The magnetic antenna with ferrite rod from the clock already has a capacitor in parallel.

1) Your scope "sensitivity" most likely is too low.

Tried using an opamp, but it didn't seem to change much. Is there maybe a minimal simple circuit for such an application?

2) Usually magnetic antennas of DCF77 receivers are tuned circuits to increase sensitivity and selectivity.

You mean tuned in the sense of having a resonant frequency at 77.5kHz? The clock antenna itself should be, but maybe more stages are needed.

[awallin]

FWIW here are spectra from both a mini-whip antenna and a long-wire antenna, measured on a HP FTT analyzer:
http://www.anderswallin.net/2017/02/lf-time-signals/
with this setup the max signal is at around -60dBm to -50 dBm, which is just about 2mVpp - so not easy to measure with a scope... perhaps you could do better with a good soundcard?

[ogden]

Are you suggesting to put a capacitor in parallel to the wire antenna? The magnetic antenna with ferrite rod from the clock already has a capacitor in parallel.

Yes, capacitor in parallel to rod antenna. If you did not remove it and use together with rod - fine. I missed that you mention "matching capacitor" in your initial post. Then I would say - scope is not good enough DCF77 receiver. Why don't you use one from radio clock?

[NivagSwerdna]

First I used a magnetic antenna from a radio clock (that is made for DCF77), made out of a coil around a ferrite rod, and a matching capacitor.

In the UK I used DCF77 to synchronise my lab wall clock as it is easily received here.   The ferrite rod is very directional so make sure you have it pointing at 90 degrees to the source also be aware that around PC equipment the switching in PSUs, monitors etc will completely drown out the signal.  My ferrite rod currently sits in a corner of the lab away from main equipment and works reliably most of the time.

[PA0PBZ]

I just had a try with a scope and a SA, the scope was not able to show me anything and the SA.. well, maybe there was something about 3dB above the noise, but it was marginal. I have one of those DCF77 modules we used to sync Novell (yes kids, that was before the interwebz) and had a look inside. There's no less than 3 77.5KHz crystals which I guess are used as filters, a 741 and a blob. There must be some critical loop because it takes at least a minute to stabilize after you switch it on. So what you need is a high gain, small bandwidth amplifier with a slow AGC to get anything useful.

[petert]

FWIW here are spectra from both a mini-whip antenna and a long-wire antenna, measured on a HP FTT analyzer:
http://www.anderswallin.net/2017/02/lf-time-signals/
with this setup the max signal is at around -60dBm to -50 dBm, which is just about 2mVpp - so not easy to measure with a scope... perhaps you could do better with a good soundcard?

Thanks, that already gave me a better understanding of what is going on. The noise floor on my scope (with detached probe) is pretty high (roughly -25/35dBV, not sure exactly how that would translate to dBm). But if I switch the FFT scale to V instead of dBV a 4 mV signal is just one pixel tall, at the highest possible resolution. So it would be pretty hard to discern or distinguish a 2mVpp wave/pulse from a glitch. Seems like I need to amplify the signal enough before I can display it, assuming the antenna even captures the signal.
Now I hope that the antenna doesn't pick up too many unwanted signals so that amplifying actually helps.

How long is the aerial you are using? Have you tried measuring with a ferrite antenna?

Why don't you use one from radio clock?

You mean a receiver or antenna? I used the antenna from a radio clock, but the signal was too weak/is not detectable (except when artifically generated as mentioned in the first post). I want to understand how to make a simple receiver step by step. I made a crystal radio, this is the next learning step.

Thanks also for the noise from PSU etc.. I tried eliminating that by putting the scope outside in the garden.

[ogden]

I just had a try with a scope and a SA, the scope was not able to show me anything and the SA.. well, maybe there was something about 3dB above the noise, but it was marginal.

I just checked signal in the online SDR, http://websdr.ewi.utwente.nl:8901/. It's very strong, -64dBm. Maybe it is worth to consider using SDR reciver, with preselector bandpass filter obviously.

You mean a receiver or antenna? I used the antenna from a radio clock, but the signal was too weak/is not detectable (except when artifically generated as mentioned in the first post). I want to understand how to make a simple receiver step by step. I made a crystal radio, this is the next learning step.

I did mean receiver.

[petert]

There's no less than 3 77.5KHz crystals which I guess are used as filters, a 741 and a blob. There must be some critical loop because it takes at least a minute to stabilize after you switch it on. So what you need is a high gain, small bandwidth amplifier with a slow AGC to get anything useful.

Perfect, knowing the exact opamp used already helps.
Some questions though. If I look at the FFT only to discern the 77.5kHz signal, are the filters really relevant? For later processing it makes sense, but before that, just to see it in the FFT plot, is it needed?
What would a slow AGC achieve?

[PA0PBZ]

Perfect, knowing the exact opamp used already helps.

I think the blob is the frontend in this case, the 741 could be used as a threshold detector or something like that. I can make some pictures if you want.

Some questions though. If I look at the FFT only to discern the 77.5kHz signal, are the filters really relevant? For later processing it makes sense, but before that, just to see it in the FFT plot, is it needed?
What would a slow AGC achieve?

You have to amplify the DCF signal a lot to get a meaningful signal to send (in this case) to the pc, you need a reliable on/off detector. To not overload your amplifier and detector you have to filter out all/most of the unwanted signals. You need a slow AGC because the modulation is already slow and you don't want the AGC to track the modulation, only the varying signal strength. I think part of the 'slowness' is the big amplification (you don't want the loop to go instable) and the small bandwidth.

[PA0PBZ]

Pictures
blue:      GND
Green:  -5..12V
White:  +5..12V
Orange: 'Data' out

[Kleinstein]

In principle the FFT should be able to see even rather small signals, if the FFT uses enough samples and thus a low bandwidth per bin.
The DCF77 signal  (especially the carrier wave) is (very  ) low bandwidth.

Just a wire may not be a very efficient antenna in this frequency range. Usually the ferrite rod or a reasonable size air coil (frame type antenna) should work. However due to resonance the output impedance is relatively high. A 1:1 scope probe would load the antenna quite a lot and a 1:10 probe adds quite some noise. So it would really help to have at least just a JFET wired as a source follower an a first amplification stage.

When using the FFT for detection there is not that much need for extra filtering. Usually the resonance of the magnetic antenna is already quite some filtering.

[petert]

To not overload your amplifier and detector you have to filter out all/most of the unwanted signals.

Overload in the sense that the voltages become too high and you get clipping? I am not entirely sure what overloads means here.

Also thanks for the pictures, any clue what that blue part is close to the antenna leads? Is it a tuning cap or is that on the antenna coil already?

However due to resonance the output impedance is relatively high.

Adding a link that explains why (for future reference): https://www.quora.com/Why-impedance-is-high-in-parallel-resonance-circuit

[Kleinstein]

The blue part looks like a film capacitor, likely to make the antenna resonate. Fine tuning is usually be moving the coil on the rod.

Overload here would be to avoid clipping or getting extra signal from inter-modulation possibly before real clipping. With the scope having to much other signal from other sources might force to use a larger range and thus have higher noise. This can especially be a problem with a non resonant antenna.

[petert]

So it would really help to have at least just a JFET wired as a source follower an a first amplification stage.

Since a MOSFET has higher input impedance compared to a JFET, wouldn't it work better than a JFET (reduced load on the resonant circuit) or is there more to it?

[ogden]

For inspiration: http://www.creative-science.org.uk/MSF3.html

[PA0PBZ]

The blue part looks like a film capacitor, likely to make the antenna resonate. Fine tuning is usually be moving the coil on the rod.

Overload here would be to avoid clipping or getting extra signal from inter-modulation possibly before real clipping. With the scope having to much other signal from other sources might force to use a larger range and thus have higher noise. This can especially be a problem with a non resonant antenna.

Yes, the blue cap clearly says 6N8 in the picture, so this is the tuning cap.

Apart from overload you really want your AGC to only act on the wanted signal, so there is another reason for sharp filtering.

[PA0PBZ]

So it would really help to have at least just a JFET wired as a source follower an a first amplification stage.

Since a MOSFET has higher input impedance compared to a JFET, wouldn't it work better than a JFET (reduced load on the resonant circuit) or is there more to it?

Just guessing here but a MOSFET and a thunderstorm…. hmmm…

[petert]

A last question. Since the DCF77 signal power is lowered regularly to actually transmit information, this change should show up in the FFT as well, as a lobe that regularly gets less tall, right?

The AM stations I could see on the FFT plot were curiously pretty stable (tallness of the lobes). Wouldn't an AM music or speech channel also have some variation in the height of the lobe (around the frequency of that radio station)? Or does it somehow average out that this is not noticeable?

The reason for asking this is to know how to distinguish a radio station from (random) noise/fluctuations on an FFT plot.

[PA0PBZ]

A last question. Since the DCF77 signal power is lowered regularly to actually transmit information, this change should show up in the FFT as well, as a lobe that regularly gets less tall, right?

Right, but the transmitter is turned on and off to modulate the signal (named OOK), so it will not get less tall, it will disappear for a short while.

The AM stations I could see on the FFT plot were curiously pretty stable (tallness of the lobes). Wouldn't an AM music or speech channel also have some variation in the height of the lobe (around the frequency of that radio station)? Or does it somehow average out that this is not noticeable?

When a signal is AM modulated the power from the signal will be divided between the carrier and 2 sidelobes, both with a distance from the carrier equal to the modulation frequency.
What you can or can't see depends on your FFT function, the width of the bins and the time it takes to gather the samples that are needed for the FFT.

The reason for asking this is to know how to distinguish a radio station from (random) noise/fluctuations on an FFT plot.

The modulation frequency of DCF77 is 1Hz, so you will not easily see the sidebands, it will show up as the signal disappearing. See a screenshot of the SDR mentioned earlier in this thread:

[awallin]

A last question. Since the DCF77 signal power is lowered regularly to actually transmit information, this change should show up in the FFT as well, as a lobe that regularly gets less tall, right?

The AM stations I could see on the FFT plot were curiously pretty stable (tallness of the lobes). Wouldn't an AM music or speech channel also have some variation in the height of the lobe (around the frequency of that radio station)? Or does it somehow average out that this is not noticeable?

The reason for asking this is to know how to distinguish a radio station from (random) noise/fluctuations on an FFT plot.

for DCF77 the side-lobes created by the faster phase-modulation are easy to see. this paper has a figure:
https://www.researchgate.net/publication/282761542_Time_and_frequency_broadcast_with_DCF77/figures?lo=1

for the 1 Hz modulation you need to think about how long the FFT-widow is compared to the ~1Hz modulation.
usually the FFT or waterfall updates faster than 1 Hz, so you don't resolve side-bands but rather modulation of the carrier itself (e.g. on/off).
if you can tweak the FFT averaging time to be really long, with 0.1 Hz or so RBW, then you should see side-bands at multiples of 1 Hz.

similar but slightly different shapes would be expected for MSF and TDF also.

any of the web-sdrs out there will show you how a voice AM radio station looks like! 

[petert]

I finally decided to try and fix my soundcard mic/headphone jack in my laptop (at least enough so it would detect a cable is plugged in) to try and sample up to 192kHz, as suggested on the web and other posts in this thread.

And ... I finally could make out the DCF77 station!



When I lower the bits/sample to 8, the station is drowned out in noise. At 16 or 24 bits, it is just discernible. That explains why it was so hard to find on the scope.
The spike of the station in the FFT is barely noticeable and fluctuates a lot, but the waterfall display allows to see the very narrowband signal. It seems also much weaker than other stations, which explains also why I couldn't detect it properly on the scope.

This is with the long 30m wire antenna, the ferrite antenna did not work yet. But now that I have a working base line, I have something to start from, and compare to.

Thanks to all again, I learned a lot, and much more to learn yet.

A screenshot is attached.

[madires]

The DCF77 signal is too low for a scope, but an SA would be able to see it. I can pick it up easily with a short random wire attached to a shortwave receiver (S5).

[petert]

I am in France currently, so that may affect things as well. The ferrite antenna didn't work (tried again).
With the long wire it works, though, and the software can even decode the time information properly.

[petert]

For future reference, or others that are interested in building a complete receiver, the following two links have good hardware building advice (and links to good ferrite antennas and crystals).

http://arduino-projects4u.com/dcf77/
http://duinolab.blogspot.com/2009/06/summertime-rf-propagation-and-cmmr-6p.html

Make a DCF77 sender with a crystal oscillator:
http://www.novatech-instr.com/grandadselectronics.html

Make a DCF77 receiver with a crystal (German):
https://www.mikrocontroller.net/topic/78005