Tweet
I recently finished working on my 1940 Model D, which still had its chorus generator intact. I thought I'd share some of the finer points of why I did what I did to the generator. This organ had an 82 note generator, and its original paper and wax capacitors. Most of them were E.U.C. brand, and were saturated with oil. About 7 of them had no brand, and looked like what you'd see on a 50s era Hammond. They were noticeably cleaner, less saturated with oil, and based on the solder joints, looked like they were replaced at some point in the organ's life. These replacement caps were actually pretty close to accurate. I'll keep them in the parts bin for later, but I still replaced them for longevity sake.
I wanted to recap the generator because it sounded really inconsistent. With the 1' drawbar out and walked up the scale, about 10 or so notes were noticeably quieter, and a few were very harsh sounding. This effect was made worse when engaging the chorus generator. The peak to peak voltage of each tone, as measured at the generator manual wiring harness, is graphed below. This graph does not include the chorus generator.
I wanted to do this job without moving any magnets in or out. That's because I did not want to disrupt the relationship between the volume of the main tone, and the volume of its corresponding chorus generator tones. For capacitors, I used Panasonic ECQE(F) plastic film capacitors. These have a dissipation factor of 1%, similar to paper and wax caps. I used 0.1uF, and 0.22uF caps from this series. After the recap was finished, this is the result. (Blue = before, Green = after)
Not surprisingly, many big peaks existed. If this were a Hammond without a chorus generator, I would have probably pulled out the magnets of the high gain tones, but I decided not to in this case. I used trim capacitors on those tones instead. This technique worked well for several of the tones that were too high, but it did not work for the tones that were too quiet. I could not get much more gain out of those tones when testing with a capacitor decade box. At most, I could squeak out an extra 10th of a millivolt of more gain, but no more. So I left the tones that were too quiet alone, and just trimmed the ones that were too loud. In deciding how MUCH trimming mattered, I had to use my ears, and some math. Most would say the human ear can decipher gain differences between two neighboring tones starting at about 2 dB, so I stayed between 2 to 3 dB as the maximum acceptable tolerance tone-to-tone. So in the resulting graph, some of the tones are 6 mV apart from their neighbors, which is a difference of about 3 dB. Overall, the difference tone to tone volumes stays within about 2 dB. Considering that I started this process with deltas of 12 mV (about 7 dB), this is progress.
I am well aware of Kon's tone generator data curves too. However, there is not much raw data from organs of this era. Only one data set was taken from a 1940 Model D, and it still had wax caps. If I superimpose the "theoretical" factory curve from Kon's data for organs of this era, here is what we have, compared to my finished curve:
So, considering maybe my oscilloscope was at fault here, I am overall happy with where I ended up on the main generator.
To be continued...
I wanted to recap the generator because it sounded really inconsistent. With the 1' drawbar out and walked up the scale, about 10 or so notes were noticeably quieter, and a few were very harsh sounding. This effect was made worse when engaging the chorus generator. The peak to peak voltage of each tone, as measured at the generator manual wiring harness, is graphed below. This graph does not include the chorus generator.
I wanted to do this job without moving any magnets in or out. That's because I did not want to disrupt the relationship between the volume of the main tone, and the volume of its corresponding chorus generator tones. For capacitors, I used Panasonic ECQE(F) plastic film capacitors. These have a dissipation factor of 1%, similar to paper and wax caps. I used 0.1uF, and 0.22uF caps from this series. After the recap was finished, this is the result. (Blue = before, Green = after)
Not surprisingly, many big peaks existed. If this were a Hammond without a chorus generator, I would have probably pulled out the magnets of the high gain tones, but I decided not to in this case. I used trim capacitors on those tones instead. This technique worked well for several of the tones that were too high, but it did not work for the tones that were too quiet. I could not get much more gain out of those tones when testing with a capacitor decade box. At most, I could squeak out an extra 10th of a millivolt of more gain, but no more. So I left the tones that were too quiet alone, and just trimmed the ones that were too loud. In deciding how MUCH trimming mattered, I had to use my ears, and some math. Most would say the human ear can decipher gain differences between two neighboring tones starting at about 2 dB, so I stayed between 2 to 3 dB as the maximum acceptable tolerance tone-to-tone. So in the resulting graph, some of the tones are 6 mV apart from their neighbors, which is a difference of about 3 dB. Overall, the difference tone to tone volumes stays within about 2 dB. Considering that I started this process with deltas of 12 mV (about 7 dB), this is progress.
I am well aware of Kon's tone generator data curves too. However, there is not much raw data from organs of this era. Only one data set was taken from a 1940 Model D, and it still had wax caps. If I superimpose the "theoretical" factory curve from Kon's data for organs of this era, here is what we have, compared to my finished curve:
So, considering maybe my oscilloscope was at fault here, I am overall happy with where I ended up on the main generator.
To be continued...
Comment