Originally Posted by aringgo
Here is one way to turn a 4520 into something that sounds comparable to the good and powerful CLASS-A systems costing $hundredsK.
Very few good Class-A amps exceed 50 watts/channel - so a really good one that you can buy if you don't build it yourself would literally cost you your house.
Here is a photo (if I can learn how to upload pictures again) history of how I made mine (in the sequence I planned because I started knowing I'd do all the necessary modifications spending only reasonable amount of cash that the wifey will not notice)
Part 1: The DACs - Why don't they use the best parts all throughout? They put in BB1795 but channelled its output to cheap I/Vs and buffers. It's like pouring distilled water thru a dirty pipe.
The reason I found out is that the japs will not buy the BB chips if TI does not show the jap chips in their documentation of example applications for the top-of-the-line BB chip".
TI is a business entity, so...
Engineers play safe and fast normally - so they design according to the catalogue, so...
So, there you have it.
A top-of-the-line BB DAC chip whose outputs are channelled to cheap generic jap I/Vs and buffer chips. In fact the DAC circuit of the 4520 is exaclty as what you'll find in the BB1795's datasheet with only the old jap chips replaced with other generic jap chips and the resistors adjusted so as not to overload the newer generic jap chips.
Tough luck everyone - at the price we peasants, (tongue-in-cheek), are willing to pay - that's all we can buy.
But using science, (and some heart), the BRAVE ones can right it - so that the top-of-the-line BB DAC chips work in perfect harmony with the top-of-the-line BB I/V and buffer chips together with scientifically verified (and classified as) non-distorting types of filter capacitors.
Cost: $300 for 6 x OPA627 buffers, $70 for 22 x OPA2107 I/Vs, Polystyrenes filters @$2.70/piece.
Why 6 OPA627 and not 5 or the whole 11? Not 11 because I am poor, not 5 because the damn circuit layout is for dual-amp chips so I must have 6 to fix the five most important channels.
Here is the Denon AVR4520's DAC section after replacement of the filter capacitors and I/V converters. The DSP board containing the 3 processors is detached.
Here is the same section, with the DSP board attached, after putting in the expensive OPA627's. The OPA627's are mounted under the adapters (upper mid of picture) connected by magnet wires to the lower mid of the picture. The adapters are inverted - pins upwards - so that I can connect them with ease of course.
The blue and red wires are my power harnesses for the +-15volts to power the DAC section. The DAC section was originally powered with +-7volts only but the replacement chips require min of +-10volts - hence the reason for their existence. IT IS SCIENCE.
The ADC: The ADC requires 10ufd coupling cap per channel to faithfully encode signals down to 1HZ. The unit uses polar electrolytics. That's not good. But again, that's what they put in so they can make more money with what we are willing to pay.
Orginal polar electrolytics cost less than $0.15 each. The only viable alterntive is to parrallel 4.7ufd polypropylenes at the cost of $5.97 each. I'd be happy if if can encode only down to 5HZ as long it does it cleanly without introducing any sound that was not present in the input.
Also, why not use a better BB chip specifically manufactured for buffering that signal - cost of replacement = $1.12
Here is the Denon AVR4520's ADC section showing the paralled polypropylenes fixes. It is in the underside of the DAC section. I used double-sided sticky-tapes to attach the caps to the board. You see larger resistors too. They look ugly but that's what I had to correspondingly adjust the time constant when replacing 10ufd electros with 9.4ufd polyprops.
[Next] Part 2: The analogue signal paths