BeoloverRIAA Phono Preamplifier: Further Improved Power Supply - 87 dB Noise Floor with Running DC Motor!

Inspired by Sonavor at Beoworld.org, I recently acquired a QuantAsylum QA400 audio analyzer. This neat little piece of hardware allowed me to dramatically improve my BeoloverRIAA phono preamp design.

This preamp is supposed to go right into the Beogram 4002 (551x, 552x) enclosure and be powered via the Beogram internal power supply. Basically a plug and play solution for those customers who do not have an amplifier with a phono input anymore. I also think that basically it is best to put the preamplifier as close to the cartridge as possible to avoid potential noise issues. While on paper this sounds great, the practical implementation of this concept has been quite challenging since the Beogram internal power supply is very noisy due to the DC motor induced EMF that permeates the entire system when a record is being played. My initial design resulted in a perfect RIAA deemphasis but with a quite audible noise floor. And of course that was not acceptable. I finally figured out how to quiet down the power supply sufficiently to supply a sensitive amplifier without disturbing its amplification performance. Here is an impression of the current prototype as implanted in a Beogram 4002 (5513):

And this are the curves that I measured with the QA400:

These curves were measured with the platter motor on and the QA400 directly connected to the RCA plugs of the Beogram via gold plated RCA/BNC adapters.

The green curve on top shows the original noise spectrum with my initial 'primitive' power supply based on a simple 7824 regulator connected to the 31V rail of the beogram. The spikes are mainly related to EMF coming from the motor. The highest spikes reach -30dBV, which is quite audible. Since there are many spikes one hears a 'hiss'. 

The red curve at the bottom is the corresponding curve measured after improving the power supply with a more professional setup based on a high ripple rejection regulator device from Linear Technology and a better noise control at the rail splitter (that is the big capacitor on the photo above). The spikes are basically gone and the floor is now at about -87dB. This corresponds to a  unnoticeable noise level when the output relay opens after the arm drops.

The blue curve finally shows a noise floor measurement while playing the 'rumble track' on my Analogue Test LP. It is obvious that the noise coming from the surface roughness of a (high quality) vinyl is more than 30 dB higher at the low end of the frequency spectrum, and about 10 dB (~3x) higher at the crucial 1kHz point. This means that the amplifier does not add significantly during playback of a record and that most audible noise comes from the record itself.

You may wonder why my spectra end at 12kHz at the upper end. Well, unfortunately, while the QA400 is a lovely piece of hardware, the software right now leaves a few areas of potential improvements. At this point I am not able to export spectra that are longer than 4096 points, and that means that everything above 12kHz is right now cut off in exported spectra. I am waiting for a response from their tech support how to alleviate this, but they are a bit sporadic with their attention. At any rate, the spectra were quite unspectacular at higher frequencies. Essentially they simply continue smoothly further down the dB scale following the RIAA curve as would be expected.

Comparing these curves with the specs of 4002-contemporary receivers such as the 4000/4400 or 6000 4-Ch Beomasters reveals that the BeoloverRIAA is considerably better than their discrete transistor based preamps, which have noise floors of 58-60dB for Phono and ~75dB for Tape. And this is quite obvious when I connect a 4002 directly to the phono preamp of my 6000 4-Ch for direct comparison: The BeoloverRIAA outfitted 4002 is noticeably more quiet via the Tape4 input.

So I think this design is nearing a point where it can be implemented. Stay tuned for the next PCB iteration!