The 17th harmonic of a 35 kHz erasing signal would produce a 1 kHz beat tone with a 594 kHz carrier. For it not to affect reception of weak 594 kHz signals, its level at the antenna would have to be less than the thermal and atmospheric noise measured in the critical bandwidth of the ears of the listeners. That means you can allow only a very small amount of 17th harmonic if you don't switch the frequency.
OK, great! Now show me how a bias oscillator will have any 17th harmonic. And include an analysis of how that oscillator system can generate a moderately clean 35KHz and still have the bandwidth to pass 500KHz.
This is now getting silly. The switch is for 19KHz leakage, which will definitely beat with the bias fundamental. That's the fundamental, not harmonics. And it's always a known frequency. No question. In later products the switch was labeled MPX Filter (for Multiplex). That's an FM function only.
The beat tone you might or might not get from a distant harmonic of the bias oscillator can't be filtered anyway.
What do you think that switch was for?
Done now?
It would be more useful if you showed me a way to make an oscillator that doesn't generate any 17th harmonic. I can't think of a way to do so.
Even oscillators with a variable gain amplifier and a slow amplitude control loop generate some 17th harmonic. The oscillators in cheap radio-cassette recorders are no doubt of the type that use a clipping amplifier stage for amplitude control. The clipping amplifier generates 17th and higher harmonics, and those are filtered by the LC tank, but with a finite suppression.
For example, in this circuit, the differential pair is driven into hard clipping and essentially produces a square wave output current (17th harmonic at 1/17 of the amplitude of the fundamental). In the voltage across the tank, the 17th harmonic is 17 Q times smaller, if I'm not mistaken, so 1/(289 Q) times the fundamental. The current through the inductors has again a 17 times larger ratio between fundamental and harmonics, so a 17th harmonic at 1/(4913 Q) times the fundamental. That looks perfectly clean on a scope, but I haven't a clue if it is clean enough to reduce the harmonic in the reception channel to below the noise floor. Besides, the erasing head will not be perfectly linear itself, chances are it generates more 17th harmonic than the oscillator does.
The remark about 500 kHz bandwidth makes no sense to me. You just need to have some 17th harmonic leaking out, enough to be higher at the antenna than thermal and atmospheric noise measured over the bandwidth of an auditory filter, roughly 100 Hz.
How do you know it's the same type of switch controlling the same type of circuit?
That's exactly why you need to switch the bias oscillator's frequency to move the harmonic out of the channel you are receiving. We do the same nowadays with switched-mode power supplies in car radios: make sure they have at least two well-known operating frequencies and when you listen to AM, choose the frequency that doesn't interfere with the channel you listen to. (Of course the waveforms are more distorted, but then again, car radios have no built-in ferrite antenna.)
No idea, I wasn't designing radios yet when radio-cassette recorders were made. I've read some interesting hypotheses in this thread, from Galu, Mark and phase_accurate, I hope the answer will be clear when @tiefbassuebertr has either done measurements or dug up the service documentation.
No, as tiefbassuebertr hasn't done the measurements or dug up the service documentation yet, as far as I know.
Not necessary. It would be only a factor if the following were true:
1. A harmonic of the bias oscillator was within 15KHz of the AM carrier
2. The energy in the harmonic were sufficient to generate a beat note by entering the first RF stage at a level near enough to produce the beat
3. That beat note were positioned exactly where the pre-set filter could deal with it (if not, then the filter's purpose cannot be to deal with this kind of beat)
I doubt it matters much. Assuming a sine wave is the goal at less than 3% THD, the 17th harmonic, already likely more than -80dB from the bias signal, and must also couple to the antenna. There's no point in trying to analyze this further because we get into layout, shielding, and so forth, none of which information we have. Specific design issues, and not having one of these AM radio recorder beasts in hand, the theoretical discussion just can go anywhere, we can't prove a thing.
Again, see above. No point in going further with the analysis because, as you said about the head, we have no idea. It would be a relatively simple matter to shield the entire erase and bias path to limit coupling to the antenna. Which, if this were a problem, is what they'd do. They would never have been able to address it with a filter.
That's the entire point here, the purpose of the filter.
No, you can't now just jump into arbitrary quantities after trying to be so specific. Atmostpheric noise isn't the same as a signal causing a beat note. The antenna may have that bandwidth, the IF strip does not.
Well, I hate to admit it, but I'm old. I have actual hands-on experience with the filter, though not the radio recorder.
They never did that.
Great. But they never did that in the era of that filter.
Look, it's simple. For the filter to be useful to attenuate an AM carrier beat it would have to be frequency agile (and automatic!), and/or the bias oscillator frequency would have to be variable and coupled to the AM tuner's frequency somehow, as well as bias level, record EQ, etc., none of which was done. These were very simple machines.
It's a simple, fixed filter to notch the 19KHz FM stereo pilot leak. That's it, that's all.
Except that some had 3 settings - its definitely intended for AM: my speculation was wrong about the mechanism, the issue was that the radio would receive the harmonics of the tape recorder bias oscillator as interference, being able to choose different bias frequencies gives a chance for landing the harmonics away from the station being recorded. Some units had a 3 position switch to increase the possibilities.
See here:
https://www.vintage-radio.net/forum/showthread.php?t=71590
https://boomboxery.com/forum/threads/beat-cancel-switch.30091/
https://boomboxery.com/forum/threads/beat-cut-switch.25683/
http://www.stereo2go.com/topic/index.php?content_oid=193392314111958937&board_oid=193392314111653319
Having read the first of these links I do recall more about the issue, and I'm pretty certain it is about harmonics of the bias oscillator.
Whether it might have some effect on stereo FM I don't know, but all the sources quote AM explicitly.
I agree that a filter wouldn't help for the AM interference, but if phase_accurate and Mark are correct, it is not a filter at all, just a bias oscillator frequency switch. I guess ejp is right and both filters (for FM) and switchable bias frequencies (for AM) were used.
That should be two times the critical bandwidth, as noise from the upper and the lower sideband ends up at the same audio frequency.
The usual whistle requirement is that a human observer shouldn't hear any whistles. That requirement is met when the whistle is masked by the noise. Whether it is masked by the noise depends on the critical bandwidth of the ears of the listener.
Hi , i would like to add what i think to this discussion. There are two types of erasing oscillators ,used in casette recorders , ones are push pull , symetric type , and with single transistor , whose can produce much more harmonics , because small core can saturate easily. Cores used in single ended were similar to shielded AM IF filters inductors , unsure if they had air gap , required to prevent saturation. Initially i was sure that switch is for beat frequency was to prevent AM problems , but talk about 19khz pilot tone ... Remember my tuner was diy and without notch filter, and you know what ?? Cassette recorder easily recorded those 19khz pilot tone on Fe cassete , and in playback noise reduction switch didn't worked... It was so strange why .