Beogram 4002 RIAA Preamplifier: A New Power Supply Design

This is a follow up to my original post about the RIAA preamplifier that I am currently designing for a Beogram 4002. This will enable this Beogram to play directly into a standard AUX or TAPE input on a modern amplifier. At first I was very happy with my initial prototype design and I enthusiastically redesigned the board for a more orderly and systematic appearance and sent it away for manufacturing. Well, I got the boards and I assembled a preamp. Here is an impression of the nice looking board:

I installed it and gave it some more testing:

And it quickly occurred to me that I should have done some more testing before getting the boards made. In the redesign I had replaced the original 7824 voltage regulator that steps down from the 31V rail of the Beogram with a smaller SOT-89 package to save some board real estate for the other components. Well, it turned out that this was an interesting mistake. The smaller package was not only smaller, but also had a smaller ripple rejection capability, which immediately brought a slight hum and some annoying hiss to my attention. The amp otherwise seemed to perform nicely, but it was noisy. Another shock occurred when I accidentally slowed down the platter with my hand while I was doing measurements: A very loud crackling hum was emitted from the speakers in that moment. Very un-Beolovely! This indicated that the DC motor, which was working a bit harder during the episode caused the 31V rail of the Beogram to cave in a bit, low enough to inject some juicy ripple into the 24V regulator, which it happily forwarded to the rail splitter of the single supply (the Beogram only has a positive rail...) amp design.

After all this it was obvious that this design was not ready for prime time yet: Back to the drawing board. I realized that the power supply of the Beogram is pretty dirty due to the DC motor, which is the main load during operation. The back EMF of the motor causes a lot of disturbance, and that is of course not a great thing for an amplifier that is supposed to reliably amplify <10mV signals. It was now clear to me that the power supply needed a lot more attention. I read up on the subject and I designed a new supply based on the Linear Technology LT3063 low noise LDO. This marvelous little package has a ripple rejection of 85dB, which is about 25dB better than any standard 78xx regulator and 35dB better than the SOT-89 unit I unfortunately had selected. Unfortunately, this lovely performance comes with the need of a few external components. This meant I had to design a small additional board that I could solder to the input header of my prototype board. Here is an impression of the piggyback board:

In this design I first go into the standard 7824 regulator to make 24V and then comes the LT3063 to drop the voltage to about 17V. Serializing regulators gives an additional 3dB of ripple rejection. I replaced the original regulator of my prototype board with this board

And then implanted it into the Beogram. Anxiously I fired up the 4002 and dropped the arm next to the platter to get an impression of the noise floor before playing a record. And the result was amazing. Much less noise than before, practically inaudible. Then I slowed down the platter until it stopped. No transmitted ripple at all. So the LT3063 was doing an admirable job! Time to enjoy one of my most recent additions to my vinyl collection and write this post. I am listening to Miles Davis' "Sorcerer" album (excellent quality "two eye" CBS pressing CS9532). How much better can it get! Now this is Beolovely!

The next step with this project is to do some more characterization of the amp and then redesign the board one more time (hopefully...;-) to get a final version going.