Removing "key clicks" from relay switch
How can i remove or limit the key clicks in the RF signal caused by the relay action?
No idea what exactly mean with "key clicks" in this regard. Is it contact bouncing? If so, it's an unavoidable effect of mechanical relays.
If you are just annoyed about relay sounds, use a different relays or mount it differently.
The best solution is not to use a relay, but to perform low voltage electronical keaying of the oscillator.
What I mean is that, if you see the spectrum of the transmitter that is keyed (tx signal dounconverted to audio, using a receiver) in the waterfall display of an fft program, you are going to notice little glitches where the beginning ang the end of the keyed signal parts (TX on). http://www.dxatlas.com/Rocky/KillClickOFF.JPG
They happen because of the rapid switching of the relay contacts or is an electronic switch is switched rapidly.
If a relay is used, I think a shunt capacitor from the relay output to the GND will reduce the ending clicks, but I am not sure about the click at the beginning of the signal.
Update,
I think I have found what I was looking for, if it works...
(w w w.ipadebooks1.com/244/index-77_2.png)
I believe the first post is rather misleading. You are talking about switching transients in transmitter spectrum which are an unavoidable supplement of OOK (on-off keying) modulation. You can reduce it by ramping the PA output over several millisecónds. The suggested low-pass filters may work for it, if dimensioned suitably.
There is a page that I have found http://www.9h1mrl.org/ukrae/arc_cd/full/html/c8-1-3.htm which suggestd values too. I think this will work, I'll let you know.
The first circuit will not affect the key clicks. The second would if placed directly on the electronic end of the keyer. i.e. if the TX is keyed by grid blocking then the circuit must be connected directly to it, not to the relay coil, but to its contacts.
I would think if the time constant of the R and C, should be 1mS. Adding the L improves its performance but at a great cost.
If you show how the TX is keyed perhaps we could suggest something.
Frank
Thanks,
Here is my keyer so far. It works, but I am trying to remove the key clicks.
If I use the relay contacts to drive the base of another NPN (12v at the base when switched on), then if I add a capacitor from the base to the GND I can remove the ending click, but not the starting click.
I use the relay because I cannot make the transistor saturate fully, so as to provide 12v keyed to the transmitter. The relay solves this problem, but I woulr really like to make it pure electronic switch for costs reasons.
The other problem is the starting click edge as I said...
I would really appreciate some help on this, I cannot find many discrete circuits on the net about it.
No, no, no, its the not RF bit, we have seen the relay energising circuit before. Does it switch the Vcc on an RF amp, block a stage with a cut off bias or what?
Frank
The circuit is used to cutt off the VCC from this kind of amplifier and the relay contacts are directly connected to the vcc point, just before the RF choke.
To my surprise, I have found that the keying circuit in a transmitter is not obvious at all if an output LPF is not used.
You see, with the method I used (cutting the VCC), any try to slowly rize the vcc and slowly bring it back to zero (waveform shaping), resulted in increased harmonics. This is because, at the short time periods of the rising and falling edges, the amplifier vcc slowly rizes from 0 to vcc and then falls from vcc to 0v. But if the vcc gets much lower, then the output waveform becomes much distorted in its positive edges. This causes huge amount of harmonics at these time intervals. I do not use an output LPF for various reasons, so this would be a problem.
So my thought about switching the vcc seemed obvious but it was bad idea!
Since my amps are running in class-A and they can be operated without an antenna with no problems of damaging the transistors, I thought that by having them running continuously and not cutting their vcc off, I could use the relay to disconnect the antenna (tx output) instead of disconnect the final amplifier vcc.
This solution worked for me ok, but key clicks are still a problem.
An idea I had, was to use an LDR in the output of the transmitter instead of a relay switch (tx output is 100mW). Then I could use a tiny incandescent lamp switched on and off, facing the LDR. the slow switching of the lamp would give perfect and similar rising and falling edges without clicking.
But my vt43n1 LDRs seem to attenuate the signal too much, and I am not sure if they can be used on HF.
So if the Vcc is reduced on your class A amplifiers, you get the clicks. Have you tried opening the emitter lead?, switching in a lot of base current (series collector resistor required) to saturate the transistor? Replacing a series feed capacitor with two putting a diode at the intersection to earth and switching the diode on to short out the RF path (several diodes in series may be required, depending on RF level).
Frank
If the Vcc is reduced on this class A amplifier, I do not get the clicks. The clicks are caused by rapid switching of the the VCC. If you try to smooth this, at the time intervals that you slowly reduce/increase the VCC (key click removal), the waveform is distorted and you get lots of harmonics, but this is not due to the key clicks but by the reduced VCC to the amplifier during these VCC transitions.
I have not tried the diode solution you refer to. However I fear that the diode will have an unexpected behaviour in this stage, introducing signals from it's non-linear behaviour. I have done special care to prevent inducing signals from non-linear elements throughout the whole transmitter stages.
The only "clear" solution I have found was to use a relay to cut-off the RF output instead of the VCC and having the transmitter continuously operate. But key clicks are not eliminated this way of course.
Since the TX output is only 100mW, I have also tried to cut-off the RF using a series LDR instead of the relay contacts. This worked but even at the maximum illumination the maximum output RF signal was about 70mW instead of 100mW.
I do not know if this is because of the minimum resistance of the LDR, or if the LDR cannot respond to the times of the RF signal, but If I could find an LDR with very tiny low-resistance then this could work to eliminate key clicks and in fact the rising and falling edges would be symetrical is an incandescent lamp was used!
Any ideas?
If your 100mW is unstable when Vcc is reduced, then that must be fixed, then transistor with one shot on key pulse can be used.
When fixed then to use 1st order LPF at 300 Hz or a time decay dv/dt of 1.5ms then Ic=Cdv/dt so choose C = Ic/1.5ms or so.
For improved spectral stop band, 2nd or 3rd order LC LPF filter with Q~1 based on impedance of LC at 300Hz and effective load impedance of 100mW Tx.
How can the output RF output signal be undistorted when VCC is instantaneously reduced? I am taling about on/off keying. The only way is to cut-off the input RF, not VCC.
You don't switch instantaneous because of supply time constants (capacitors) so the PA will go through "bad" operating points with heavy distortion. And even if you could switch instantaneous, you don't switch at zero crossing of the RF, so that causes some (smaller) distortion.
I think switching the supply of the PA is about the worst place for keying.Why don't you key at smaller RF levels ealier in the TX chain? There are all sorts of RF switches.
I am glad you confirmed this first point about switching the PSU, because my experiments confirm that too and I was not aware of this point up to now.
About your second point, I do key at smaller levels indeed, the transmitter is a 3 stage one, and I key at the final amplifier (100mW RF out). I cannot key on the oscillator buffer, that is not a good technique.
You mention that there are all sorts of RF switches for low powers. What do you mean by that, a simple relay? Currently I use a reed relay to disrupt the RF (disconnect/connect the antenna, thus key).
What are the other options you mention please?
Thanks a lot!
I was thinking of PIN diodes or FET switches.
http://www.qsl.net/va3iul/RF_Switches/RF_Switches.pdf
The CW wave shaping theory (with examples) I think is presented in all ARRL handbooks back to 1930'ties. And when I said all handbooks, I mean all of them.
go to page 14.50
http://mxh.strefa.pl/pliki/tech/book2006/14.pdf
The main rule is that never ever switch a high current or a high voltage stage.
The top-end transceivers switch the CW transmitter signal in more stages. They switch the DC bias of buffer stages, and sometime switch simultaneous the TX IF signal path.
All these switching use a careful timing, providing at the end a nice shape of the CW signal.
Using five HG 5007 diodes as an attenuator, the receivers I worked on were good for i/ps of less then -120 dB at 1V Rf input, hence my comment about more then one diode
Frank
This information has been very helpful. Indeed it shows that the on/off keying is not a cheap solution like used in most qrp CW transmitters worldwide! However, such kits use an output LPF and they can reduce harmonics during the rising and falling edges, so simple shaping can be achieved by coils and capacitors. My scheme is somehow different though because I do not use an LPF, but I see the difficulties after this document read.