Debouncing

[GM0WEZ]

I note G0KVL's Lionel bug has wee bit of sponge on the dit contact U spring. I did this too when I had one and it worked well, but it was perfectly useable even without it.

But I have also heard of some bug users using electronic de-bouncing for their bugs.

Any views or experiences?

 

[G0KVL]

Bug debounce is a can of worms, some are adamant that all that's required is a correctly set up bug.

My own finding are it's very radio specific, some you can hear scratching more than others, some are very clean and others just won't have it.

I also built a thingymejiggy on my Vibroplex (forgot what you call them) as per picture, seems okay but not totally convinced, it seemed to speed the dits up and operated it at such a balance I don't know whether it's touching or not.

What I use now is electronic intervention, it's fabulous, a 555 timer IC (see attached) by the way if anyone decides to go from this circuit the top pin next to 8 is a 4 not a 7 .

The sponge in the J-36 seems to offer me some 'fine tuning', and are surprised at what a slight adjustement can offer,  that said I tried the same with both my Vibroplex bugs and thought it didn't do a lot at all from what i recall, maybe the spring metal and tension differs somewhat.

This may definately be a case of set up correctly and i could ditch said sponge, but I don't like to touch things once set up as I seem to take forever to get back to a state of 'happy again' with them. And not only that but a local Amateur friend curses me for my piece of sponge so provides entertainment  ;D

Craig

 

[GM0WEZ]

Many thanks for the circuit.

I particularly like circuits like this when all the components are ones you have readily to hand!

 

[G0KVL]

pleased you like the circuit, I was going to upload my build but I couldn't managed to get the picture down to 500k or less without trimming it to nothingness on my phone.

yes there's not much to it, but it works brilliantly

 

[G0KVL]

Here is my build of the circuit, it took a Linux moment and image editing software.

This gets used regularly, if you hear me on HF , I'll most likely be using this

Craig

 

[Ham4CW]

While reading this thread it occurred to me that I could measure the bounce on a Vibroplex key using one of the logic analysers I have to hand.

The setup was very basic, but tht's all I needed at this point. I used an old Vibroplex key that I had purchased a few years ago now for the very purpose of working on a debounce circuit. The key had been chosen to be the poorest looking one I could find on eBay, which would hopefully have lots of oxide on the contacts and not be in the best of health. My purchase did not let me down!

The keying 'circuit' was simply a 9v battery to give a voltage bias across the contacts of the key, the current was limited with a 4k7 resistor in series with the battery.

The logic analyser was connected across the key terminals, along with the 9v battery/resistor combo (see the first picture below).

What showed up was quite interesting. When the key was allowed to run for a few seconds before checking the recorded levels, it was noticed that most noise was generated as the contacts were closing. Oddly, in nearly every case, there was one spike before the contacts opened! :-*

This was repeated over and over, using up to 60 second runs on the key, but always the same. One spike before opening, lots of spikes as the contact closed.

However, I also found that at the beginning of each burst of dits I sent, the first dit was extremely noisy, but the following dits dropped in to the pattern mentioned above.

The noise burst at the start of a dit sequence seemed to tie in with a noise the key made each time it was triggered, a kind of "doink"sound (sounded like the entire arm assembly may have been resonating with the mechanical shock of being keyed). I have a few other Vibroplex I can try when I dig them out, and it will be interesting to see if they are the same, or if this 'resonance' is just perculiar to this key.

The two remaining pictures below then show a burst of dits after the key has 'settled', and a cluster of groups of dits sent to show the noise at the beginning of each group. The noise can be seen as a thickening of the trace.

In all cases the keying line was being pulled high by the bias voltage, so when the trace is low (logic 0), that is when the key contact is closed.

73, Mark...

 

[Ham4CW]

I note G0KVL's Lionel bug has wee bit of sponge on the dit contact U spring. I did this too when I had one and it worked well, but it was perfectly useable even without it.

I thought I would test this out. So, using the same 'circuit' as before I tried a 'run' of dits both with and without a piece of sponge wedged in the dit contact.

The first image below shows the keying WITHOUT any sponge fitted, the noise is quite noticeable.

The second image shows the same key but with a small chunk of sponge gently wedged in the hairpin dit contact. The contact bounce was reduced considerably! When zoomed in some slight bounce was visible each time the contacts closed, but the opening noise had gone.

I also noticed that the initial burst of noise at the start of each run of dits was also pretty well non-existant.

Who would have thought such a simple mod would work so well! bd

73, Mark...

 

[Ham4CW]

I came across this video by AA4OO showing a simple method of debouncing bug keys using a capacitor. He also appears to have the 'sponge mod' done too.

https://www.youtube.com/watch?v=lUPqf1bIpAA

73, Mark...

 

[GM0WEZ]

I think this is the first scientific evaluation of the benefits of sponge dampening of bugs!

It did occur to me that there is potentially room for improvement in bug design, but as doing the same thing electronically with a keyer is so much easier, it's never been pursued in the last 40-50 years. I guess it's a bit like mechanical watches, which are complicated and expensive compared to the electronic versions, but still quite fascinating as they are tiny precision machines.

I think GHD make an opto-electronic bug, but there must be other ways of doing it mechanically. And I think there is a bug that generates the dits mechanically as well?

 

[Ham4CW]

I think GHD make an opto-electronic bug, but there must be other ways of doing it mechanically. And I think there is a bug that generates the dits mechanically as well?

I'm guessing you meant to say dahs?  :-[

The GN209FS-P by GHD can do both dits and dahs, but it also uses opto-couplers rather than mechanical contacts (I think!).

The older one that comes to mind is the Melehan Valiant, which was full auto and totally mechanical.

Also, way back, Dunn & Co. use to make one called the "DOUBLE-LEVER DUNDUPLEX".

And the "Automorse" by Hitchcock Bros. Co. Adelaide, Australia.

I think there were a number of electro-mechanical ones too, very early on, mainly to get around Vibroplex patents.

Going back to testing the mechanical ones, I have a copy of the first model H.G. Martin patented, which is generally known as a "Patent 457 Key".

On that early key Martin did not use a hairpin contact support for the dits, but instead it uses a straight metal strip, with a few other 'odd' variations too.

I got the key out of my 'museum' yesterday with the intention of testing it against the hairpin spring versions to see if it fares better or worse. I haven't got around to wiring it up to the analyser yet but what I did notice was that when allowed a prolonged run of dits the dit rate (wpm) increases the longer the key is allowed to run for. Maybe that's why the hairpin contact arm was introduced?

Like I said, it is only a copy or clone, made from the original patent drawings. Interestingly it was only a year or so ago that a genuine 457 was discovered, until then it was believed that no original ones existed. I doubt if all that many copies have been made either.

I'll report more when I've tested it.

73, Mark...

 

[Ham4CW]

Well, to say I'm impressed in an understatement!

I connected up the "Patent 457" bug key copy as with the previous key (9v battery + 4k7 resistor). The sample rate was set on the highest for the analyser (24 mega samples per second so as not to miss any 'action'!).

Incredible results. Although I mentioned in my previous post that there was a noticeable increase in dit rate if the key was left to 'run' for three or four seconds, which in normal use this would not occur anyhow, other than that the results of contact bounce measurements are excellent!

All of the images below show the key without any sponge mods, or capacitors across the terminals etc, this is 100% as measured with bare contacts only.

The first image shows a bunch of dit groups. As soon as I saw this come up on the analyser the lack of contact bounce was noticeable straight away. Just now and again (over repeated runs) could be seen the occasional bounce, but by and large the key was bounce free to all intents.

The second image shows the start of a typical dit group, again the lack of 'noise' is really noticeable when compared to the earlier Vibroplex key tested.

The third image shows a long run of dits, and the change in rate and mark-space ratio is very apparent towards the end of the run. I found that if the key was set to give the longest run of dits (up to about nine seconds), the mark-space ratio was not equal. When the key was adjusted to give a 1:1 mark-space ratio on dits, the key would only produce dits for around four seconds. Having said that the longest character might only take a second or even less to send, so this die off after four seconds is neither here nor there.

The fourth and fifth images show contact bounce (where it did occasionally occur), really zoomed in. The equivalent frequency was always around 4.5kHz, so this I would guess shows why various low pass filter type debounce system work as well as they do. At the speed the key was set for (its slowest, around 20-25wpm), the dit frequency was around 8-9Hz as shown by the analyser.

The remaining images are of the 457 key itself, and a closeup of the contact arrangement.

73, Mark...

 

[G0KVL]

Interesting tests, and as noted the first and probably only time the sponge has been observed for it's effect.

Well i finally moved on to an alternative method, my 'bug stabilizer' on this key was straight forward unlike on the other varient of vibroplex, and lends itself quite nicely to a bent paperclip attached, and seems to work rather well!

Craig