Things to be accomplished in September and/or the following time
Page initiated on 21 July 2012
Status: 27 August 2012
Please consider also the newly made webpage called:
Nachtfee things done 1
Responding to this page of various proposals, which never can be accomplished at once, I have created a new web page showing our progressing process.
Rather annoying discussions stresses me to measure the actual loss resistance of the 10 Nachtfee quartz in due course
Has been commenced (Bladwijzer183)
Measuring the quartz parameter R1, being its series loss resistance. This can easily be measured when the quartz crystal is operated in series resonance. L and C being equal and the series loss resistance remains. What I would like to do is: connecting the two crystal electrodes onto a kind of π network. Consisting of two resistors of say 12-20 ohms or that like. The quartz crystal is bridging these two resistors. Onto the most left resistor (Ri) the digital signal generator is being connected. On the most right resistor (Ro) we connect an a.c. millivoltmeter. Care should be taken so that we supply a quite low signal level onto this setup. By carefully stepping up or down the expected frequency in +/- Δ 0.001 Hz until max. signal is measured on the millivoltmeter scale. This should mean that we are just measuring at the exact series resonance frequency. Maybe our 0.001 Hz steps is not providing an optimum, consequently we should need a higher resolution. But we don't have such an extreme precision synthesiser, it is even doubtful it exists. However, when we have found a specific frequency, say for example: 500.151 Hz. We notice the exact millivoltmeter scale reading. Now the quartz being disconnected and it is replaced by a small variable resistor. We tune this such that the scale reading is exactly equal to the one we measured with the quartz involved. When this is done properly, with some tolerance we have found the substitution of the quartz R1 parameter. Theoretically this could be done at any frequency, but the higher it becomes, the more the parasitic parameters will count (L of the tuneable resistor and its wires, as well as the capacitances involved). But, we are measuring only at 15,000 Hz (15 kHz), which may be regarded as a low frequency signal. Measurements have to be accomplished without dismantling the entire quartz module again. For this I would prefer to measure the best accessible quartz stages. For this maybe Q4 - Q5 - Q9 and Q10. We have, however, to consider the fact that Q9 and Q10 are refusing to operate. When measurements will show that Q9 and Q10 are indeed faulty, I could access Q8 as well. Maybe useful, measuring the R1 parameters under different temperature conditions. For this, the thermostatic oven has to be mains operated; but the rest of the Nachtfee system can be powerless. It might be interesting to know what the temperature influence versus R1 actually is.
The setup for measuring the series quartz loss resistance R1, which is determining the Q-factor of a crystal.
Just at series resonance, the quartz acts as a 'zero order' device, where there does not exist a phase shift (= 0°) between in - and output (because its L and C virtually do not exist)
Repeating: The measuring procedure is quite simple. First: we determine what the exact series resonance frequency of a Nachtfee quartz crystal is. This has, however, to be done carefully, as to prevent overloading of the quartz device. It has also to be accomplished accurately, so that the exact frequency is determined in Δ 0,001 Hz synthesiser steps. The detecting device is a sensitive mV meter. At a certain setting the max. signal is measured. Be always aware that you really have found a single (series) frequency. When there exist a max at two frequencies, the generator step resolution may be too low. However, we only can determine in steps of 0,001 Hz, whereas steps of 0,0001 Hz may be then necessary (it proved in the meantime, that our PM5193 is capable of handling even Δ 0.0001 Hz steps up or down). Read off the mV scale as accurately as possible. Disconnect now the quartz device and bridge the π like network with an adjustable variable resistor, which value should be in the range of the to be expected R1 values. I expect that 2.5 kΩ will do. Tune now the variable resistor such that you get exactly the same mV scale reading. After this has been determined, disconnect the bridging variable resistor and measure its resistance by means of a digital ohm meter. This technique can easily be used at least up to 1 MHz. For higher frequencies a different bridge is a must. But we are only measuring at 15 kHz, where parasitic L and C can be neglected. Ri and Ro should be quite low, 12 to 20 ohm will do well.
Would it be possible measuring the Q-factor of a piezo-electric vibrator in the same measuring configuration?
For it we need to measure the bandwidth of the crystal device (also valid for any kind of resonator). Theoretically, we measure the frequency deviation necessary by lowering the passing through signal amplitude till we get down a factor 1/√2 = 0.707, or, with minimal error, being -3dB, on both sides of the series resonance frequency f0. This measurement result is also valid for its parallel resonance (vibration) mode; as these differ only slightly, owing to its parasitic capacitances. However, the way we have set up this measurement is causing less ambiguities. When we would like to measure the Q-factor accurately, (without bringing here the proof) it is advisable keeping the resistors Ri and Ro at about 10 - 12 ohm or even less.
Graphical explanation of measuring the Q-factor of a resonator by means of determining its 3 dB bandwidth, also known as B√2 (= 0.707)
The simple equations are:
Q = fo/B = 1/d
B = f2 - f1
Q = (1/R1) • √L/C (for a series resonance circuit)
The loss in L as well as in C is accumulating in R1, which latter value also incorporate its mechanical loss
d is known as the decrement, a figure used in the old days
We count with signal amplitudes, what is, however, also possible is measuring the actual phase-shifts either way; in this case the signal phase should rotate over + or - 45° (sin 45° = 0.707). But this would need a more complex test setup. As always, I prefer to keep it simple, constituting a for nearly everybody reproducible method. One may wonder why the resistors Ri and Ro are being kept at a low ohmic level? Simply, because its value is influencing negatively the quartz parameters otherwise too much.
What we have to do is: after having determined the series resonance, call it fo or centre frequency fc, we step up or down frequency until we get a remaining signal amplitude down a factor 0.707, which is 3 dB on both side of the resonance curve. For this our synthesiser PM5193 is quasi made for it. The ratio between fcentre and the value B is providing directly the quartz Q-factor. I regard this as an elegant way of measuring the Q of a piezo-electric vibrator. It has, however, to be done carefully, as we are likely dealing with very high Q-values my guess > 105, where maybe each Δ 0.001 Hz step is having (I don't hope too) great influence on the measuring results.
On the back of an envelope: when, for example, f2 - f1 = 0.004 Hz and fc = 500 Hz we get 500/0.004 = 125,000, a Q exceeding 105. This example is only giving you an idea of the dimensions we likely have to cope with.
Let us consider briefly, what kind of test setup the IEC 444-1/4 standard advised
Test setup for measuring the quartz parameters R1 and Q (vibrating in series resonance mode)
On the left is the signal synthesiser, on the far right the so-called Vector - Voltmeter. This latter instrument is starting from 1 MHz onwards. Hence, it cannot be applied for our purpose. Theoretically, one should determine 0° degree phase difference between the A and B channel. We have to overcome this downside by measuring its least transition loss; maybe a tiny bit less accurate, though sufficiently for our purpose.
We luckily possess such an original, rather accurate, Pi-network, which in fact is capable operating up to 200 MHz, with minimal errors. However, for our purpose it would be too clumsy using it for the Nachtfee quartz parameter measurements. The to be measured quartz device was normally loaded at a < - 30 dB signal level, my guess for us a too low level. By the way, a most rare device, as it was only used in quartz production. Donald Prins once told me, that its Pi-network principle originated from Philips (Nat.Lab. Eindhoven?)
A new thought came up how the Nachtfee system can be displayed optimally (most functionally)
I would like to set it up just adjacent to the Freya power supply and its attached receiver. On the other side of the gangway between the FuG10 setup and the stairs to the basement, the Nachtfee apparatus with attached Gemse receiver. When this space is not optimally matching, it would be possible to change both sides of the gangway. The advantage of having the Nachtfee/FuG136 system on either side of this corridor, is that one can monitor what is happening on the opposite side. They are communicating by means of wireless over a distance of say 2 metres.
Right of the FuG10 corner, in front of the red stone wall, may be a suitable place for setting up either the Nachtfee or the simulated aircraft system replacing the hypothetical FuG136 system
On the top of the centre Al frame section is still visible is the FuG25a with in the middle the Gemse ground receiver. The FuG25a as well as the Gemse have been demounted and being part of the hypothetical Nachtfee system reconstruction. We have to find another suitable place for displaying the Jagdschloss transmitter (on the left) as well as the X-Gerät of the Seetakt system (on the right-hand side).
Whether the Nachtfee consol will be placed about here, or on the other side of the corridor is to be decided later. Next to it will be the Gemse receiver. Whether on the left- or on its right-hand side is not yet decided (photo taken on the day Nachtfee arrived on 12 November 2011)
It appeared that I do not have a good photo of the opposite space (next to the Freya receiver/power supply)
It has been successfully accomplished (Bladwijzer180)
Another thought is about the simulation of the aircraft display (FuG136)
In my current perspective, it should be tested what happens when the time base is operating at a different frequency than does the Nachtfee 'order' signal (n = 1/10, 1/5, 1/2, 1, 2....)So far we have experimented with a PRF of 500 Hz (Q5) (n = 1). What would be the consequence of using a deflection frequency of, say, 250 Hz? During the time base paints a single circle, the Nachtfee has transmitted in the meantime 2 signal blips. This can also be done with a time base frequency of 100 Hz. The advantage would be, that only a fifth of the number of time base pulses have to return to the Nachtfee controlling screens. Hence, a trial series have to be arranged. The only change necessary is that we need to adapt each time the 90° phase shifter accordingly. It is of course also possible to test a time base PRF of 1000 Hz (n = 2). My preference is to test first series, thus a lower PRF, but it may well be proven that 1000 Hz or even higher is favourable.
When 100 Hz time base frequency would operate sufficiently, we also may have solved my considerations handled in the last system block diagram. Maybe 50 Hz will do as well. I believe it is worth setting up trials as to distinguish what fits best into our hypothetical system reconstruction. Using a synthesiser is very handsome, as we only have to divide or multiplying the Nachtfee Q-channel frequency accordingly.
All the time bearing in my mind, that they must have used a slightly adjustable time-base-phase in the aircraft as well. In my perception a tuning-fork might have been, at least hypothetically, a realistic option. As nearly everything in Nachtfee is different from what one is thinking in the beginning, its ultimate solution may also be an entire surprise to all of us.
Why not implementing experimentally the available, rather old fashioned, General Radio electrically driven tuning-fork type 813A?
According measurements done some months ago, it operates at about 1002 Hz
Which would be in the spectrum of Q5 500 Hz x 2 = 1000 Hz
When we only could lower its number of vibrations per second a tiny bit
This photo is showing foregoing experiments as to lower its frequency a tiny bit
Maybe changing its driving current may causing some phase adjustments.
Whether its output voltage is sufficient is to be answered when this experiment is started. When necessary an additional amplifier have to be adopted.
This subjects should stressed forward, an interesting experimental job for Jaap Keijzer, designing a range of tuning-fork oscillators
On 25 August 2012
Time may, Deo volente, come soon arranging a new series of experiments
Up until now, we have used channel Q5 (500 Hz) as being the Nachtfee signal source. Which, theoretically is correct. Though, Freya-EGON incorporated several functions. Acting like a regular EGON controlling station, providing vector and distance of an aircraft under control. Secondly, it also was transmitting the Nachtfee data-signal, and most likely, receiving the returning aircraft time-base-signal-phase. Both may be called "secondary radar" signals.
EGON used the standard Freya PRF of 500 Hz. Whereas Nachtfee could be operated in channels below or up from its 500 Hz centre.
Each successive Nachtfee channel differs 2 Hz up or down. Thus: Q6 = 502 Hz and Q4 = 498 Hz
My aim is, to operate first a quartz channel of 2 Hz difference. When we operate it as was done in our previous experiments, we will not encounter a difference, as long as, the hypothetical aircraft display substitute is controlled accordingly. The next step, is to substitute to the HF signal modulator a second pulse with a PRF of 500 Hz; substituting what originally came from the EGON system. My main concern is, that I should prevent that the TTi pulse generator output-stage is being overloaded (its output might be dc coupled with their output transistors). I will use, for this experiment, our second LF synthesiser PM 5190X.
What will happen now on the aircraft 'order' or command screen?
We will see two signal blips, of which one stays in concert to the Nachtfee system (thus resting at a certain 'order' vector), the second blip (pulse) will rotate either anti-clockwise- or clockwise for each quartz-channel-step additionally 2 times per second faster or slower.
I prefer using first 2 Hz difference, as the human eyes can easily distinguish its motion. I believe, that both signals will not make conveying the various 'orders' difficult on the aircraft display substitute. As the EGON signal is never staying in concert because it rotates constantly
The red lines and square belong to the hypothetical Freya-EGON system operating on 500 Hz fixed. Whereas the Nachtfee system will be operated at will on a quartz channel other than Q5; in the Q5 case distinguishing will be made impossible, as one does not know which blip actually is the real Nachtfee 'order' signal. Both blips will be, nevertheless, virtually synchronised, although, likely having a different signal signal phase
Double click on this drawing as to get it in PDF
On the above drawing it seems that there exist a mutually stable wireless pulse pattern, but this is not the case; the E (EGON) pulses will move either for- or backwards for every successive Q-channel-step 2 times per second additionally. Of course, when we trigger onto the Nachtfee signal.
When we would set the aircraft time base at the EGON (PRF) signal being 500 Hz, the Nachtfee 'order' signal will rotate accordingly against the now quasi stationary EGON blip (pulse).
We should also implement experiments with a lower rate of returning time-base-phase signals. Called 'periodical timebase interrupter'. Its implication should be determined empirically. We may likely never know whether it was done this way when Nachtfee was developed.
A second plan is, getting a video camera, so that we can show the moving results via YouTube. I have to purchase one; but also have to find out what the best way is uploading it.
Has been accomplished too, see our YouTube testfilms
On 26 August 2012
A new thought came up
Why not giving the following circuit concept a try?
Please consider for the main description the flag: (Bladwijzer176)
It has to be experimentally discovered, whether this modification is hypothetically workable. The disadvantage is, that the first order differentiated time-base pulse will also be painted on the aircraft CRT. Making it three blips that are appearing on the CRT screen. Albeit, the EGON pulse will be spinning in either direction anti- or clockwise. The striking advantage is, that the FuG25a IFF transponder does not need a modification.
This has also been accomplished, but it proved not being very successful, because the retransmitted pulses are rather like spikes, very difficult to adjust upon. Please notice: Nachtfee things done
On 27 August 2012
New thoughts came up during my early morning waking up
It may be hypothetically possible to modify the previous system concept
When it proves that the three pulses is a bit too much confusing, that the aircraft CRT may be blanked for the time-span of the differentiated time-base-feedback pulse. Please notice for additional details also flag: (Bladwijzer177)
This concept was finally adapted at the end of 2012 and early 2013. Please notice Bladwijzer202
It is still quite exhausting going through all possible system options of Nachtfee time and again!
Especially in respect as to how the aircraft system might have been commenced in the wartime days. Still bearing in mind the extreme difficulties encountered keeping both: the Nachtfee on the ground in concert to aircraft deflection system.
It is likely, that we will never find anything original on the FuG136 aircraft system
This consideration is also a personal drive envisaging, be it hypothetically, all the imaginable circuit options
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