Quartz reference



Page setup: 19 June 2012

Status: 7 July 2012


Time and again a repeating discussion on the stability of the Nachtfee internal quartz controlled time base is taking place. As to clear this discussion, I have dedicated some extra time in accurately watching what the actual frequency drift versus time is. I cannot say whether the better performance is due to frequent operations versus its initial non activity for say 67 years. Things apparently run smoother than was encountered when we begun with starting up the Nachtfee experiments in the beginning of this year.


We will have to deal with accuracies which were in the 1940s out of the question! The measuring steps will be 1 : 1000.000 or even higher. Normal frequency counters may causing problems when reading off their minor significant digits. I therefore opt for a means which I used already in the early 1970s. Having two signals feed each of them onto the X and Y channels of an oscilloscope and painting a Lissajous circle. It becomes possible with the combination of Z-modulation to measure down to ≈ 10-6, which I regard is a quite high accuracy. Getting the results is a rather time consuming process, as phase shift is occurring very slowly. There exist also another means by using phase-measuring setup. In combination with our experimental Nachtfee reconstruction, using a Lissajous figure is very convenient.


 This is the way we do it comparing both the Nachtfee signal phase against an external signal reference. In our case employing a PM 5193. This is a synthesiser which allows frequency steps of  Δ 0.001 Hz (a single sine wave passing through 0° to 360°, then repeating itself every 1000 seconds). This sounds impressing, but we will see that even this figure is still too coarse


On 19 June I approached in the Klooster the Nachtfee system setup. Starting from 13.11

f = 500.176    Cold start

f = 500.171    13.15

f = 500.160    13.20 I had to switch off Nachtfee shortly for technical reasons

f = 500.093    13.48

f = 500.090    13.49

f = 500.075    13.55

f = 500.068    13.59

f = 500.061    14.01

f = 500.056    14.03

f = 500.046    14.06

f = 500.043    14.7/8

f = 500.011    14.17

f = 500.001    14.23

f = 499.996    14.27    It was noticed that the oven is switching off and on

f = 499.991    14.30

f = 499.990    14.33

f = 499.989    14.34

f = 499.987    14.38    Quasi synchronism started, stayed for minutes on the HRP2/100/1.5(A) 0° stable

f = 499.986    14.40/41

f = 499.985    15.07   This is just a bit too low, whereas 499.986 likely is a slightly bit to fast. Both only measurable after some time of observation. The main reference is the actual 0° signal-phase (Bladwijzer44) on the dual trace CRT (HRP2/100/1,5(A). As noticed below, the actual figure is a bit influenced by the fact, that two goniometers is to be dealt with. Knowingly, the main 'Phase' and the 'order' or command phase shifter. The common phase shifter is in the red line chain, the 'order' phase shifter is in the upper blue line chain.    (Bladwijzer43)


We have noticed that some of the apparent drift is owing to goniometer drift inside the Nachtfee apparatus. Still tiny, but definitely going into the above results. On the other hand cancelling their drift, as only phase-shift versus temperature is the cause and no frequency drift is concerned.

It was found that 499.985 is likely slightly too low whereas 499.986 is slightly too high. For this a resolution steps of 0.0001 Hz is needed, which is out of our scope.

 It is also found, that the previous finding that the Nachtfee quartz deviated 0.57 Hz is no longer valid. Today I measured 0.19 Hz, when we consider that 499,985 is the most valid figure. Hence, we know that Q5 at 15000 Hz is being divided by 30 making a frequency drift between room temperature and 60° C oven temperature 5.7 Hz. This drift continued still a bit for a while, although the oven switched on and off in the meantime. It finally stabilized which is visible on the above list. We have nevertheless, also taking into account, that I recently, thus say a week ago, measured that frequency stabilised at about 499.993 Hz. Why this is occurring I don't know. Hans Jucker just e-mailed me some data on LORAN B-1 time base quartz which oscillates at 50 kHz. This is, however, a flex-mode quartz type, and is far more favourable than is the X-mode vibrator used in our Nachtfee apparatus. Decisive may have been, that Nachtfee was a Luftwaffe project and may not have been regarded as providing the best apparatus possible. Nachtfee, from various perspectives, is showing an experimental nature. I have, nevertheless, never seen, in the 45 years that I deal with German wartime electronics, such a special quartz controlled system. The industrial capabilities of the US were far more advanced than was possible in the German wartime hemisphere.


I used also the opportunity to distinguish what the interactions between retarding and advancing phase shift or drift, against 'Phase' control and their signal movements on the two CRT screens is. It is very handsome knowledge when interpreting actual CRT screen interactions.


Shown is the case when the simulated aircraft time base frequency is too high against the Nachtfee's internal reference. When operating 'Phase' control the comparable response is also shown 



Shown is the situation that the Nachtfee internal (quartz) time base is running faster than the one in the simulated aircraft. When operating 'Phase' control the response is also shown




On 21 June 2012

the longest day of the year


Following the many discussions, it has become time to draw in a proper manner the principle schematic of the Nachtfee quartz oscillator modules. 



The principle schematic of the Nachtfee quartz modules. Please notice, that the RV12P2000 pentode is wired as a triode


Please notice also the principle block diagram of the Nachtfee quartz oven module (Bladwijzer46)


The only transformer data I have is their ohmic values between the windings (please notice  for more data: Nachtfee notebook report 1, pages IV and VII). It is, however, not exactly known how their (mutual) inner transformer coupling is. I trust that the winding ratios may be more or less correctly drawn. The transformer numbers are the ones belonging to the actual tags. It is clear, that the transformer is not in resonance. Also that the quartz resonator is being operated in series resonance, as most X-mode piezo-electric vibrators do. The dotted line and the 100 pF capacitor is used occasionally. Some modules were modified this way, though, most were without. Also the 2 k resistor is sometimes bridged, thus simply omitted. They must have encountered quite some difficulties getting the quartz and their according modules operating. Module to Q5 is having the 100 pF modification. It is also found, that Q5 is likely the least active quartz, as its output signal is lower compared to the other ones.  It was also found, that a module runs in its appropriate slot, whereas it refuses to do so in some of the another mountings. Three quartz modules refuse working at all. This must be caused by their quartz resonators, as these modules do operate when being plugged into another slot. This is, however, not very essential and have been kept this way.  


The transformer taken out of its metal can



On 22 June 2012


New thoughts came up about the purpose of the the potentiometer circuit


During the MLK experiments, about a half year ago, it was experienced that maximum signal output was obtained when the potentiometer was having its highest value - thus 2 kΩ. This will, however, not imply that this is its optimal working point. Maximum signal may well cause that the next cold starting up may face a striking quartz oscillator. It was also found, that likely owing to the very high Q of such kind of quartz vibrators, starting up need some time, Q5 takes, my guess, about 1 s before reaching a stable signal amplitude. After Nachtfee's cold start, all oscillators will starting up, a channel selector switch is then connecting a particular oscillator signal onto the Nachtfee system. Helwig Schmied suggested once, that more than a single quartz channels could be operated at the same time. This is out of the question, as the divider stages will not be able to cope (locking upon) two, or even more, slightly different signals; as cannot the Nachtfee signal processing.   However, in my understanding the potentiometer is part of a negative feedback circuit (countering in some respect rising signal amplitude), which has to limit (contravening) the mechanical loading of the quartz vibrator. From experience, I know from our Ring Quartz (375 Hz) demonstrations (year 2000), that the Dutch PTT designers had  taken utmost care of controlling the maximum mechanical movements (> 3 mm) of the quartz, be it a bar or in the form of a ring; both being X-mode vibrators (A ring quartz may be regarded being a wound-up or circular bar). It takes >> 5 minutes for its starting up!



On 23 June 2012

It is found that it would make sense to add some details on the quartz modules


It has to be noticed (please see the next photo): that the feedback potentiometer is having a slight influence on the actual signal phase. But it has a clear downside when it is operated, albeit slightly, that the signal phase starts drifting for quite a while. I therefore doubt, that this potentiometer was meant for Nachtfee phase control; as the 'Phase' control available on the Nachtfee front panel is not showing this negative behaviour. My guess, is that the feedback circuit is changing and controlling the loading of the quartz circuitry. A test button left of the dual trace CRT (Frequenzkontrolle) is for checking its proper operation; including the wave form of the oscillator signals. The parallel resistor and potentiometer chain is also damping (loading) the transformer. What the implication of it is I cannot judge. On the other hand, oscillators were mostly operated in conjunction with valve grid current, hence some HF grid current is conducted through the piezo-electric vibrator, whether its actual value is inflicting, remains open? Please remember: that a series resonance device is constituting a virtual resistance R1 only. Often in the range of a few ohms up to say 100 ohm or even more. I don't know the actual 'R1' values of our Nachtfee channel quartz. Viewing the fact that they may have encountered problems, noticing their various module modifications, their figures may not have been all very favourable. On the other hand, nevertheless, modifications could also have been implemented as to prevent overloading of the quartz vibrators.     I am afraid that these open aspects may never be lifted entirely.




Photo taken of the module Q6

Please notice the not connected open end of the potentiometer tag (the one in front). Viewing its position, this shows that when the potentiometer is turned clock wise its resistance is increasing. It was already observed, that the quartz module provided its maximum signal amplitude when the potentiometer was set at its open end (highest ohmic value). This is symbolically indicated by the according numbers on the potentiometer scale (0 - 11).


V/6 may stand for: quartz channel 6 of Nachtfee series V (Roman figure 5), as all other quartz oscillator modules carry the same number V, whilst 6. is a unique number. Where NF 13 and BL may stood for I don't know. The 2 kΩ potentiometer is down right of this module. My first impression was, that the exclusive number 6. was due to its mechanical fitting onto the according slot. Which after all proved to be incorrect. It stood for special adaption onto the actual quartz channel parameters. Each quartz is having its own particular Q factor and herewith its ability of starting up oscillation. Please notice just next to the green ceramic capacitor (200 pF) the vertically mounted roll-capacitor of 100 pF, which had been adopted as modification between the anode and ground; most quartz plug-in modules operate without it. From some information we may derive that 5 Nachtfee sets had been build. Whether our apparatus is the last one of this series, may never be answered. Although, viewing the fact that when the Nachtfee package arrived from the US we discovered that most valves had been accepted (Abnahme) in October 1944. It may thus well have entered experimental service on a later date.



   The way of mounting the quartz modules is easy to understand. Channel Q5 is in its slot


In the forefront is the, formerly called 'double sized module', because its purpose was not yet understood; which constitute the 1 : 15 frequency divider stages 1 : 5 + 1 : 3 → 15. Hence 15000 : 15 makes 1000 Hz. In the Nachtfee main chassis additionally division by 1 : 2 is making 500 Hz (counting for Q5). The black vertical box is containing the thermostatically controlled quartz oven which is kept at 60° C. The divider module is vertically mounted just next to the contact strip.

Please notice for it (Bladwijzer47) as well




On 30 June 2012

I was since recently confronted with strong opposition by Günter König and also Hans Jucker, both men contravening my views that the Potentiometer in the quartz module is meant for limiting the mechanical stress of the quartz crystal. Their meaning is: that its purpose is fine tuning of the actual frequency of the quartz module.


The only means underlining my perception is measuring what the consequence of the potentiometer setting is.



A special facility is the existence of a special measuring window, where the direct oscillator output can be picked up 



The two measuring points. The lower one is ground or chassis and the upper one is the actual oscillator signal



The potentiometer being turned anti clockwise set at position 1

The simulated aircraft time base used as frequency reference showed quasi synchronism at: 500,129 Hz



1 Potentiometer set at 6

Quasi synchronism at: 500,119 Hz



Potentiometer set at 9 (90 ° of the scale 1 - 11).

This is having a severe impact, as after less than a minute the quartz vibrator became overloaded and stopped its oscillations! Weak oscillation occurred when the negative feedback signal was at its maximum, thus at potentiometer scale 1 - 3 maximally. When a quartz vibrator is being overloaded, thus operated under too much stress its loss resistance is also increasing; which parameter is known as: R1

The Nachtfee quartz oscillator is with no doubt a series resonator. In such a concept ωL1 = 1/ωC1. In case of resonance both equal and only the loss resistor R1 remains. Such a circuit is pure ohmic and no complex component exist.

However, quasi synchronism at 500,091 Hz


We have nevertheless, to reduce this deviation owing to the fact that the Nachtfee quartz oven is still heating up. We may draw however that the total deviation is 0.037 Hz Let us estimate that 0.007 Hz is due to the heating up of the oven. Remaining 0.03 Hz. But this deviation is mainly caused by the fact that the quartz crystal is being loaded over a critical driving level. 

An interesting aspect is, that quartz vibrators tend normally to rise in frequency over time. We have seen that the Nachtfee oscillator is responding contrary. May this mean that the actual operational condition owing to the oscillator concept is loading the quartz vibrators too much? As long as the actual temperature versus frequency curve isn't known, we must be precautious with definite conclusions. We only know that their actual quartz loading is quite critical.



This is why the Germans must have introduced a driving level control on their quartz oscillator modules and for no other reasons!  



This is the typical control signal of Q5 on the dual trace scope HRP2/100/1,5(A) (Frequenzkontrolle). Please notice, that painted picture consist of two (equal) signals traces, each originating from one of the two CRT systems. Apparently their time axis may have shifted a bit up and one downwards. Maybe you will recognise that the sine wave is showing distortions


All the Nachtfee quartz channels Q1 - Q10 (leaving out the three striking ones) are showing a different kind of output signal on the CRT. My guess, owing to their different parameters (L1 - C1 - R1)




On 1 July 2012


I quite annoying discussion came to a virtual closure


Time and again I had to respond onto self repeating e-mails. The response was, that his views are correct, as the above figures show clearly that he is right. What can we, however, grasp and should we bearing in mind?     That a further clockwise rotation of the potentiometer setting exceeding scale division 6 a bit will cause that the quartz vibrator is stopping its oscillations; after overloading the quartz for not yet a minute. For practical reason, the quartz power loading should be kept fairly low. My estimation after viewing what happened after the quartz oscillator struck, is that the internal dissipation of the quartz crystal may cause that the quartz-end may touch a single or both of their electrodes, owing to its mechanical expansion. Be it mechanical stress or increasing temperature or both. This is dangerous and when this happens too often the quartz may break down for ever. It is well shown above that the quartz stage is becoming totally overloaded.

His concept understanding is: that by means of the controlling potentiometer it is possible to fine tune the actual quartz frequency. We have just noticed that this is not the purpose of this controlling device (Poti). We have also emphasised upon the fact that longitudinal or X-mode vibrators can mechanically extract and contract a mm or more easily. In our ring-quartz-case it is >> 3 mm! Killing for a piezo-electrical-vibrator. Open ring quartz. Please notice, that a ring quartz is a X-mode vibrator, where a virtual quartz bar is wound up to a ring (for better understanding). Nevertheless, vibrating in the X-plane only. The only way obeying to this, it has to contract and extract, hence changing its diameter in accordance to its frequency as there is no other means.

Resuming:    It is out of the question that frequency fine setting or tuning should be accomplished by overloading a quartz vibrator. Some mentioned that R1, thus the virtual loss resistance of the quartz resonator, will change. Indeed, R1 will chance - but only in the direction of greater loss (its value will increase). Causing an internal heating-up due to the product of: R1 x quartz current. Neglecting the mechanical limits due to its very mounting. Very precise quartz oscillators will have a power feedback, as to keep the quartz loading within limits. Nachtfee accomplished this manually. My guess, they had to cope with 10 quartz crystals, all having their particular parameters. This is reflected in the individual adaptations in nearly each quartz module apart. The modifications were often bridging the fixed 2 kΩ resistor and/or adding a capacitor. It is likely that such quartz crystal was having a too high Q-factor, thus a low R1 value. Hence, the negative signal feedback should be stronger, in other words: providing more compensating signal (the feedback signal is in counter phase thus shifted 180°). What may cause the quite high level of signal distortion? Maybe owing to the transformer, or is it entirely the quartz loading? For the latter speaks, that I saw for a flash after the quartz crystal struck that when it started up, or was it just about breakdown, that a nice sine wave was painted on the CRT screen. Only for a very short while and with considerable lower signal level. May this indicate that the circuit concept is having shortcomings? Maybe. We have, however, to take into account that the Nachtfee designers may have been under great stress, coping with the scarce availability of  components. Also does it matter that the quartz module output was (is) not sound? No. It works and the first frequency divider is responding appropriately. I cannot judge whether 'controlling the quartz loading' was a great issue during the 1940s.





On  6/7 July 2012


Digesting the awful confrontations of the past weeks, it is felt necessary to explain, in more details, what the actual purpose of the quartz feedback control is


Recent measurements stressed the necessity of limiting the load of the quartz resonators. Owing to its very high Q-factor, which I guess may exceed 100.000, the X-plane motions in the longitudinal plane may endanger the quartz bar; as it may touch their (counter) electrodes. It was encountered, that when the feedback control potentiometer of Q5 was set a bit above 6 (say 7-8 out of 1 - 11) lowering the amount of feedback signal, that after not yet a minute oscillations vanished for a while. Only restarting after some cooling off time and then only at the lowest driving level possible. Thus with max. feedback signal (set at scale number 1).  



The variable feedback control is the potentiometer in the previous schematic; R1 is constituting the equivalent loss resistance of the series resonance quartz crystal; the current through the 2 kΩ potentiometer and the 2 kΩ fixed resistor (the latter not drawn and in some modules even not existing) is reducing the actual signal or current level supplied onto the quartz crystal. The 200 pF capacitor in series does not matter much, as at 15 kHz it constitutes |53|, whereas R1  may be count for a few ohms up to several 100 ohms.


Supplying two signals being in counter-phase, though of which one is having a variable amplitude is easing signal amplitude control, by means of  amplitude summation (subtraction). When both are of equal amplitude, but in counter phase, the result would be null. Consequently, the feedback signal should be of a lower magnitude, otherwise oscillation is impossible.

In oscillation terms: the round going circuit amplification is known to be 1. Of course, not the valve amplification factor.





On 7 July 2012


As usually, after having contributed additional information, it became apparent that some of you may be confused by the consequence of signal summation or subtraction. This explanation is omitting the mathematical equations, those familiar do not need this graphical tour, anyway



Hypothesis 1 (not in line with what appears in Nachtfee): the feedback as well as the oscillator signal is having both an amplitude i and being exactly in phase. This will after summation resulting in a signal having a magnitude of 2 i




The next hypothetical example is (not in line with what appears in Nachtfee): when signals are still in phase though, the feedback signal is being reduced to, for example, ½ i, providing a resulting signal of 1½ i


Please bear in mind that the previous two figures do not represent what actually is happening the Nachtfee oscillator stage




Hypothesis 3, phase difference 180°, in line with what is happening in the Nachtfee oscillator circuit: The Oscillator signal amplitude is being countered or reduced in amplitude by means of summation (or call it subtraction). In this case resulting in a signal  magnitude of i - ½ i = ½ i




  Hypothesis 4 (not in line with what appears in Nachtfee, as oscillation is impossible): Both signals are in counter phase (180°) and having equal magnitudes. Hence, i - i = 0 (null); thus resulting in a vanished signal


This latter example is only showing what the full consequence of signal summation or subtraction is. Is this ever used in practice? Oh yes, see for example when cancellation of signals is concerned, like for directional findings. But also in many other applications.



Not yet dealt with, we have seen in the three previous screen shots taken of the Nachtfee oscillator output, that the signal is apparently having quite some distortion.




Odd as well even harmonics are ultimately resulting in this kind of signal

Whether this is due to its circuit design or being caused by transformer short comings I don't know yet. But, a normal  iron-sheet-core transformer operating on 15 kHz and according the screen picture it is having a considering value of harmonics, thus 15 kHz - 30 kHz - 45 kHz and even higher, may also contribute to what is shown (again, owing to various harmonics summation and subtraction). It is also likely, that the feedback signal is adding additional phase shifts (for each harmonic differently), which may also change as function of its loading onto the transformer. The absolute value of the 200 pF capacitor in series to the quartz crystal, may also having a variety of values rather lower than |53| for a pure 15 kHz sine wave, owing to harmonics.




In case of Q5, where the feedback magnitude apparently is far too low, a square wave like signal is provided (causing overloading). Whether this is owing to valve saturation due to grid current or that the transformer is becoming saturated or both, I cannot yet judge



How should we interpret this HRP2/... screen shot? Shown is the quartz stage output, but now watched by means of the 'Frequenzkontrolle' push-button on the front panel. Distortion, however, is still visible but not as severe as on the previous screen shots


The only explanation I have, is the fact that its signal is passing a low-pass filter section, consisting of a 350 kΩ resistor and some lengths of screened cables and other circuitry. My conclusion: it reduces the harmonics content quite well. Maybe, the bandwidth of our HP scope is > 100 MHz whilst the Nachtfee 'Frequenzkontrolle-bandwidth' may not exceeding some hundred kHz, is to take into consideration as well. 



Maybe to be continued in due course


Arthur O. Bauer

Since August 2012


Please don't forget to use the handsome: Nachtfee Chronology page


And, the PowerPoint progress page (converted into PDF)





Please go back to, or continue with: Nachtfee hypotheses