modifying the rega planet

I have worked a bit on my first generation rega planet. it had been used as a transport up to now (and received a better clock crystal for that reason) but now it is playing—straight—in the living room system.

I had looked at the pcm1710 datasheet and noticed that it could easily drive the input of the tube amp after it (100k). this would completely bypass the la9216 (no datasheet, see la9215), a filter/amplifier/buffer—yes, an op-amp is used for each step…

so it was time to try that out. looking for more info on rega planet mods, I really only found two worthwhile sources: on acoustica.org.uk and on pinkfishmedia.net—have a look especially at the pics at the latter. a big thank you to the flat-earth society is in place.

great info there, but there were still some things I had to figure out by myself by tracing my planet:

1. the 5v supply that comes in via the ribbon feeds both of the dac digital supply pins (8, 21) and some mode pins. this supply is decoupled by both c9 caps (left side, right side of the chip), each in parallel with the c58 next to it.
2. the 5v supply that the la9216 regulates for the dac feeds the 3 dac analog supply pins and half of the biasing of the internal filter network and dac output op-amp (see below). this supply is decoupled by c33, in parallel with the c58 next to it and the two c1 caps (again a left side, right side deal) closer to the chip.
3. these two supplies are clamped together by the 2 d1 diodes, exactly like it is suggested in the datasheet:
   
4. the player runs a mute line to the pin (25) on the pcm1710 and then to the mute pin of the la9216, so no pops if you decommission the latter. the 3rd c58 next to the dac chip is hooked up to this mute line. no cap upgrade needed there.
5. the la9216 is not pin compatible with a la9215.
6. the bias of the internal filter network and dac output op-amp is hooked up the hard way, forced bias:
   
   pin 12 to 5v analog and pin 11 to ground. the datasheet and the presentation where I got above image from both recommend the floating method, to avoid “the output of the op amp to rise with respect to frequency at low frequencies.”
7. the 2 r28 resistors closest to the dac chip are connected to the signal outputs.

I was not completely sure if removing the 3rd order analog filter after the dac was going to be a good idea. I had a horrible experience ages ago doing something like that to a bitstream player. so I tried this out:
output mod:

• lift the dac side of those r28 resistors;
• connect a 100 ohm resistor in series with each dac signal output;
• remove both fb1 (wire + ferrites) next to the output jacks;
• connect 2 caps (0,33 µF is enough for 100K) between the 100 ohmers and the jacks;
• pull the caps to ground at the jacks with 1 meg resistors.

cap mod:

• replaced both c9 caps with my favourite 10 µF tantalum caps;
• replaced c33 with a 120 uF cap panasonic hfq left over from a previous dac project (the Goudreau mods);
• after reading about “no low-esr caps around the la9216,” I thought: what is the point of swapping if I hardly use this chip anymore? so I did not change them.

the result? no disasters, a more direct sound, but also a bit wiry. also the next day after warming up for 16 hours or so.

undeterred, I had one more ace to play: pull the dac signal output into class A. because of the constellation of dc output voltage of the dac (2,5v) and the max signal swing (1.6v peak) the best solution is to make the pulling resistor (dac output to ground) 1/4th of the load connected to the player. as long as pull resistor paralleled with load is larger than 5k you are OK.

in my case, the 100k amplifier load (parallel with the 1 meg) lead me to use 22k pull resistors. I also took the opportunity to properly ground the the flying ends of the r28 resistors and with that the analog filter and the 3 stages of the la9216. I also put a 47 µF tantalum cap instead of a electrolytic over the clock crystal. and that was it…

so, now we have a direct, lush sound in our living room from cd. I am still startled by it.

future moves
here is something intriguing: looking at the image above of the bias of the internal filter network and dac output op-amp, I think we can tap off the signal before that op-amp: if we take the floating hook up shown above—with pin 11 via 10 µF to ground—and float pin 12, then the output signal is on that pin. attach it to a tube/fet/transistor and we have music without op-amps.

now there are some wobbles to iron out before this al works. I do not know if the dac output op-amp is normally set up for positive, unity or negative gain of the (+)-(-) signal. so what kind of signal level comes off pin 12 is unknown. floating it changed the set-up, as it also changes the filter characteristic of the internal filter. it could be worth putting a spare pcm1710 on a jig to figure that out…