Und dann seien auch nicht alle Kondensatortypen gleichermassen geeignet:
Ich Paste hier einfach mal meine Informationen, die ich zusammengesucht habe:
Einige Hinweise zum Tongenerator Recapping
Empfohlen: Polyester Film Caps (Mylar), angeblich besser als Polyetylene Film.
- Orange Drop Polyester Film (225p)
Anbei einige interessante Posts aus dem Organ Forum:
David Anderson:
„I've done a lot of research on TG recapping. Capacitor Dissipation Factor does matter in terms of the output and the Q of the series resonant filter, particularly on tones 73 and higher. The originals were either paper or Mylar, and the DF of new paper and new Mylar capacitors is virtually identical. Polypropylene has a DF around 1/10th that of paper or Mylar, and that does make tuning the exact value of the capacitor between 90 and 110 nanofarards more critical because the low DF of polypropylene narrows the bandwidth of the filter.
If there is any error, one should err on the high side of this capacitance value as erring on the low side thins the timbre; IMO, this is part of the reason that recapping has been controversial and that highly varying outcomes have been reported. This comes from direct listening tests as the capacitance value is varied via a decade capacitance box. I also use 716P caps for a number of applications, just not this one. The tight tolerances of modern caps (usually ~2% on 716Ps) actually work against us for this job.
It's more complicated than that. Hammond turned the wider tolerances of these capacitors to their advantage. They tested the capacitors for their individual values and then divided them up into bins. They brought the TG to a station at the factory that had a device that allowed them to find the capacitance value that peaked the output of each filter, exactly like tuning a radio tuner onto a particular frequency. They then picked the closest value available to that ideal value from the bins. After this process, final TG calibration was done. I've found that there were usually some "outliers" in inductor values that could not be peaked with the range of capacitors available. Those issues were fixed in calibration.
Taking advantage of the tolerance range of the capacitors made winding the inductors to precise values less critical. Two wrongs made a right, one could say.
The capacitance value on the 6 tones with the 0.255uF capacitors is the least critical because the Q/bandwidth of those filters is half that of the filters with 0.105uF caps. It's all in the math. The resonant frequency of the filter is determined by the product LC, but for any given value of LC, Q decreases as L decreases and C increases. My theory is that Tone 55 represented the largest practical inductor they could wind, given the form factor, so they had to jump to a larger value of C below that.
The reason that the old capacitors still work--sort of--is that two things are going on: capacitance value drift and wildly increasing DF/ESR (often 10-20x or more what it should be). That R (resistance) value added to the filter as a result of DF/ESR widens the bandwidth, preserving some of the output level, but simultaneously making the filter less selective/effective. (Based on the 1949 TG I used for my measurements, Q had dropped to around 1/6 of what it should have been at the factory.) In other words, the increasing DF/ESR counteracts the capacitance drift, to some degree; if not for this, the filter outputs would be far lower than what they are. Whether Hammond knew this or whether it was serendipity . . . I have no idea. That they would have known in 1935 exactly how capacitors would age is somewhat doubtful.
Based on my measurements and calculations, filter Q should peak around tones 79 and 80.
I never said that polypropylene was "wrong," per se. My goal in the recapping debate has always been guided by one idea: trying to reverse engineer the factory capacitor selection process by figuring out which parameters they were focusing on. Using paper or Mylar caps, to my mind, helps to keep the variables under control.
Using polypropylene will have two effects: higher peak output from the filter and narrower bandwidth/sharper filter slope, increasing the effect of being above or below the peak output capacitance. I've seen this in my own experiments. My EICO decade capacitor box was built with polystyrene caps (very low DF), and it places them in parallel to get different capacitance values. No single capacitor would equal the peak output obtainable with the substitution box.
I think the single most important thing I can add to the discussion is: If you err, err on the high side of capacitance in this circuit. This is a result of listening to individual tones while dialing in different capacitance values. To my ear, something happens when you go under the peak value that seems to thin out the tone, and I've hypothesized (for about eight years now) that part of the controversy over recapping has to do with this effect. So, if I tone peaks at 94.5nF, then I think it's fine to go with 96nF. I don't think that the Hammond employees went OCD about getting the exact value that peaked the filter. They picked one that was close, and it was all fixed in the final calibration.”
John Mihevic
“I have other posts on this forum on my TG cap replacement in my 1955 B3. I’ll summarize below:
I have recapped my TG twice. The first was with a purchased “Cap Kit” of “matched caps”. In the Cap Kit the capacitors are matched close in value to each other (0.22uF and 0.1uF). However, they are not matched to the filter coils as was done at the Hammond factory. The Cap Kit was a big improvement over the old wax caps. However, I decided to go one step better and match capacitors to the filter coils. I used “Orange Drop” 225P polyester film for the capacitors. At every frequency, the capacitor to coil matching produced a higher millivolt output than with the Cap Kit capacitors that I replaced. The highest change was +1.72 mV RMS. The tone with the matched caps is crisp & clear.
To get the resonant peak you need a True RMS DVM (Digital Voltmeter) or scope, and a way to change the capacitance such as a capacitance box.
This was my procedure:
With the TG running, and the DVM monitoring a tonewheel output, the correct value of capacitance was obtained by increasing the capacitance until the maximum mV output was achieved. At the maximum output I knew the filter was at the resonant peak. More or less capacitance reduced the mV output. All work was done from the front of the organ with both manuals raised.
The capacitors I used were:
0.22uF(220nF) for Freq. 49 to 54
0.082uF(82nF) for Freq 55 to 91
I then used the following for parallel trim capacitors:
0.015uF(15nF)
0.01uF(10nF)
0.0047uF(4.7nF)
Note that 1000nF = 1uF
The following is the marked values and actual range in measured capacitor values:
Marked Value**Actual Range In Value
220.0nF*******215.0nF–241.0nF
82.0nF*********82.3nF-91.4nF
15.0nF*********15.1nF-16.6nF
10.0nF**********9.42nF–10.56nF
4.7nF***********4.79nF–05.30nF
For those interested, the following is the value of capacitance that produced a resonant peak at each frequency for my tone generator:
Freq.****Capacitance
No.********(nF)
49********241.0
50********233.0
51********230.0
52********239.9
53********240.6
54********240.9
55*********88.6
56*********91.1
57*********91.4
58*********93.4
59*********92.0
60*********91.4
61*********88.3
62*********89.3
63*********91.8
64*********90.6
65*********92.9
66*********91.6
67*********88.3
68*********92.5
69*********95.4
70*********98.3
71*********97.7
72*********90.9
73*********88.6
74*********90.2
75*********90.2
76*********92.3
77*********90.2
78*********95.5
79*********97.0
80*********94.4
81*********97.9
82********100.2
83*********96.5
84*********97.5
85*********96.2
86*********97.4
87*********98.7
88*********96.1
89*********94.8
90*********92.9
91*********93.0
The process is not difficult, just time consuming.
John M.“
und 
und 
