Nachtfee tuning fork experiments
Page created on 2 June 2012
Status: 8 June 2012
After quite vivid discussions with Günter Koenig, where we time and again were discussing the technical possibilities using quartz controlled oscillators in the aircraft system, thinking of alternatives was bearing in my mind since. The problem with these kinds of (quartz) devices is their dependency to operational temperature; say its environment. As phase equality between the ground based Nachtfee system and the system in the aircraft should all the time being assured, a circumstance which have proved being unrealsitc. A quartz crystal operating in the extremely low spectrum (below 15000 Hz is not frequency controllable, particularly not in respect to 1940s techniques. For this, even today, we need at least a Rubidium standard operated on both sides of the information chain.
I have already mentioned, that a tuning fork oscillator would be a possible replacement (substitute) of the rather clumsy low frequency quartz crystals arrangements.
As so often, when something is constantly bearing my mind, I remembered during a virtual dream that we possess a tuning fork oscillator, albeit not operating at 500 Hz though at 1000 Hz. Necessary is thus a frequency divider. Not a digital one, as these provide square waves, often of a non symmetrical nature. Please bear in mind, that this 500 Hz should be having a rather good sine wave shape; because its signal should be fed onto the X-Y channels of our HP scope; which is simulating the FuG136 aircraft display; as to paint a quite nice circle (painting a Lissajous figure).
Our tuning fork was once donated in 1948 to the Dutch people by the Rockefeller Foundation
It is mechanically driven and manufactured by General Radio. It is kept within a wooden box and the tuning fork output is being fed onto a filter section providing 50 - 500 and 5000 Ω. When I started with opening the box I saw a tuning fork with strange interrupting contacts. Which I have seen a few weeks ago when I asked Günter Hütter whether he possesses a 500 Hz tuning fork. He showed me one having similar kinds of contacts, though operating between 20 and 72 Hz. His one must thus also originate from General Radio.
To investigate this device I had to supply a dc voltage onto it and also linking our counter and oscilloscope onto it
After feeding about 5 V dc onto the A and B contacts of this apparatus - after a while I heard a 1000 Hz tone and also a rather nice sine wave was shown on the scope screen (please see below). A downside was that it gave a signal of 1002 Hz. How, can we lower it and how can we regulate its output frequency or phase a bit? (maximally 1 ‰)
The General Radio Type 813 tuning fork oscillator principle is quite easy understandable
After some experiments, it is found that two tiny plastic tubes, which are folded a bit, is lowering its resonance frequency. Like is done by fine-adjusting quartz crystals, by loading them by means of a tiny liquid drip or a tiny pencil dash onto the quartz surface. We have, however, to notice that this is also lowering its overall Q-factor. However, in our case it worked rather well. It is also found, that the tuning fork frequency is dependable on to the supplying voltage.
My first attempt was to insert tiny metal rod, but this lowered the Q-factor too much as to be of any practical value. The small plastic tubes worked better, but still not entirely
I have, however, no idea what its actual impact may be. I would not wonder, when it mainly will rely upon the stability of the power supply. It is found that we may expect that it will operate between say 4.5 and 11 V. Frequency is then operating between 998 and about 1000.1 Hz.
The next day after I got my camera from the museum, as these tests are being accomplished in our MLK Lab., it proved that what I had measured the day previously wasn't repeatable. So a new series of experiments had to be undertaken.
After a series of experiments a solution proved to be loading the two tuning fork legs by other means. Don't forget, loading a tuning fork additionally is always having the consequence that the Q-factor is decreasing. Hence, its capability of starting up and oscillating.
The two so-called oil-sock-tubes gave a better result. After having cut a series of tiny bits its length proved to fulfil my desire
I do not understand how the brass interrupting contacts work, when these are being operated when the arrangement is kept in a vertical position the lower contact is warmer than the upper one. Whether his is also the case when it is build within its housing horizontally I can not yet judge. As the internal wiring is kept too short as to allow it to be operated horizontally outside the wooden box.
A rather curious effect is occurring
That the tuning fork resonates at a higher excitation current (thus at a higher voltage) at a lower frequency than with a low excitation current
I expect that this may be caused by the dc current loading of the tuning fork-body; which is causing an alternating flux loop through most of the tuning fork legs.
At about 5 volt supply its fork frequency is just above 1000 Hz
In the beginning I used our lab. power supply, which provides various voltage and/or current sources, but it is having a 'switched power' system. It proved later that an analogue, series regulated, power supply works better. As to separate the power supply from the tuning fork output signal, an additional choke has been implemented.
It takes, by the way, quite some time (maybe 15 to 30 seconds) before the tuning fork starts oscillating, and it keeps running also for quite awhile (although not exactly measured, my guess is 5 seconds) after the feeding power is been interrupted. This is directly the consequence of dealing with a very high Q device. On the scope screen it is visible that it takes seconds before the signal output reaches maximum.
Visually the signal distortion is quite low
My next move is, building a frequency divider similar to the one used in the Nachtfee apparatus.
Its principle is that the driving signal of a higher order is being fed onto the non tuned grid windings whereas the anode circuit is tuned at the lower order frequency. In our case being 1 : 2, thus at 1000 : 2 = 500 Hz
On 6 June 2012
I continued experiments which I have started yesterday 5 June. My aim first is, to build myself a frequency divider, which is capable to divide the existing tuning fork signal of about 1000 Hz into a proper 500 Hz signal.
After some first attempts, I realised yesterday evening that such a circuit must be of its very own an oscillating system. The supplied higher order signal is locking the 500 Hz signal just in a factor where n = 2, 3, 4... . I must admit, however that I haven't checked yet whether it works with higher order signals. But the way it responds onto the second order signal is showing that a higher order signal will also lock onto the lower order signal.
Its circuit is quite simple to understand, it is an oscillator, where a higher order signal is synchronising the lower order signal. Watching the output signal at the scope, the potentiometer is the element which is harmonising (adjusting) the optimal locking condition
My experimental divider stage
I always like these kinds of flying wiring, as change of components is being made very easy. The circuit is build around a RV12P2000 valve, as to keep it in German like circuit design. I must admit, that the divider has much in common with the three divider stages inside the Nachtfee frame. Although, not in all respects
Watching what happens on the scope screen, when the potentiometer pointer is being set to ground first, the oscillator is running freely, at a frequency a bit lower or a bit higher than the wanted signal-output. Tuning the potentiometer up, various situations can be observed. First significant is the the sine wave is showing two signal components, hence the 500 Hz as well as the 1000 Hz signal components. At a certain potentiometer sector the sine wave is becoming stable and this is the capture range where the circuit is following the input signal, be it divided exactly by a factor 2 (or a higher order division). The general capture range is often 100 Hz. Quite an achievement when we regard that we operate at 500 Hz. There is, however a dependency on the wave form of the triggering (input) signal. It may be that pulses do it a bit better than sine waves. But nearly all experiments have been accomplished with sine wave signals.
My first trial was using an existing LF output transformer with primary 3500 Ω and secondary 5 Ω. I took it because it was just at hand, and why not giving it a proper trial? The secondary windings is being employed as feedback supply, and the primary section is being placed in the anode circuit of a RV12P2000 valve. The anode circuit is tuned as to oscillate freely (unlocked) at the wanted frequency, in our case being 500 Hz.
Divider being locked upon 1000,80248 Hz providing after division by an order of 2 500.40124. One may ask why not 500 Hz exactly. The answer is quite simple, because I used HP RC tone generator, which was manually set at about 1000 Hz (please notice the previous photo)
This actually is the unlocked, thus free running frequency of my experimental divider stage. It may also indicate the quite wide locking range of these types of frequency divider stage. The free running frequency may be lower or higher than the to be locked upon frequency spectrum
Within all together say an hours and a half all worked as is desired experimentally (as it has to be build within a proper chassis frame). The only practical fact encountered is, that when I connected it onto the GR 813 tuning fork oscillator, that when I changed its supplying voltage (as to pull its tuning fork frequency) which caused a change in its overall output signal, that the working point of the system had to be realigned by means of a small re-adjustment of the input potentiometer. A rather good sine wave output is being achieved.
On 8 June 2012
I continued with a second experiment, whether this circuit would also function when semiconductors being employed?
About 40 years ago I used a pentode like circuit in a special signal mixing convertor. I used then for input a field effect transistor as to have a high impedance loading of the input circuit. Galvanically followed by an RF transistor. Such a circuit is also known as 'Cascode'. Such a principle is in some respect acting like a pentode.
Would it be possible to replace the existing RV12P2000 by a semiconductor substitute directly, thus virtually 1 to 1.
The upper schematic is already shown previously on this webpage. The lower schematic shows the semiconductor substitute
k = directly wired onto the P2000 cathode, thus using the same 1 kΩ resistor as well
a = is wired onto the P2000 anode, using its 10 kΩ resistor too
g1 = directly wired onto the P2000 grid
However, the only change is implementing a potentiometer as to enable varying the current flow. It is, nevertheless, replacing g2 of the P2000. The screen-grid current is normally lowering is operating voltage due to its initial current flow. This is of a different value in the experimental circuit.
What I simply did, is switching the HT power supply off, thus also cutting off the P2000 filament supply. Pulling the two HT lines out of it and connecting them on to the low voltage power supply, which is also used for the tuning fork oscillator. Even the 10 kΩ anode resistor of the P2000 is being used.
Conclusion it principally works.
Please notice the two BC547s hanging in the air
The formerly operated RV12P2000 is still kept in its socket. First I checked whether it oscillates, which wasn't the case. Then I checked whether the upper emitter is responding on to the setting of the basis potentiometer. Yes it does. The emitter of the lower transistor was not showing a sign of voltage drop. Then I realised, that I use a transistor which needs a bias current. For this I extra implemented a 100 kΩ resistor between collector and basis. Now the transistor is conducting, and is showing oscillations. I don't know why, but its free running frequency was about 800 Hz instead of about 500 Hz when a P2000 is being operated. Would the P2000 circuitry represent such a high capacitive loading component?
Dividing 1000.9451 Hz by 2 is creating 500.47255 Hz
On the scope screen on the right-hand side we see the divided by a factor 2 signal. It is showing distortion, but getting the best out of this circuit is not its aim. I only would like to know whether it is possible, doing it also by means of a semiconductor substitute?
516.95521 Hz seemingly is its free-running (un-locked) frequency
Conclusion: It is possible to use such a kind of substitute circuit. However, the RV12P2000 circuit works smoother and is providing a quite nice sine wave signal. I have to admit that its design could have been improved a lot. But I have decided to build a sound chassis for the RV12P2000 version. I have, nonetheless, first to find a second Uniframe (Amroh), which have become these days a real collectors item. I also need to find a small HT transformer having also a 12.6 V filament provision.
Since August 2012
Please don't forget to use the handsome: Nachtfee Chronology page
And, the PowerPoint progress page (converted into PDF)
To be continued in due course
By: Arthur O. Bauer
Please notice also our recent new discoveries: Nachtfee new findings (Status: 12 May 2012)
Please go back to, or proceed with: FuG136-Nachtfee starting page! (Status: 5 March 2012)
Please go back to, or proceed with: Nachtfee survey page 2 (status: 8 December 2011)
Please return to, or proceed with: Nachtfee survey page 3 (status: 21 December 2011)
Please return to, or proceed with for the survey pre-phase to: Nachtfee 3a (status 8/1/2012)
Please go back to, or proceed with: Nachtfee MLK Lab. Survey (status: 13 December 2011)
Please go back to, or proceed with: Nachtfee-Inbetriebnahme (status: 5 March 2012)
Please go back to, or proceed with: Nachtfee evaluation and conclusion page (status: 1 March 2012)
Please continue or proceed with: Nachtfee-FuG25a concideration page (status: 10 March 2012)
Please go back to, or proceed with: Handbooks papers and product information
