Deep Technical musings on the Superdac 2496

Well, I've been looking more closely into the specifics of the SuperDAC 2496, and have come across a few interesting things...

The first time I opened it up, I couldn't find a local oscillator. I assumed I wasn't looking hard enough. Today I opened it up again, and still can't. So, I did a little research...

The CS8414 digital interface decoder chip that is used in the superdac is the answer. This chip regenerates the clock from the incoming datastream. It has a PLL on board to lock to that. There is also a voltage controlled oscillator within the chip (presumably for when there is no input signal). The 8414 then gives this clock as an output for the DAC (the rather unassuming AKM AK4393).

So what? Why do we care? Simple... it means you can't reclock the superDAC. It relies on the accuracy of the clock in the transport. Thus you will see significant gains in the performance of the superdac if you reclock the transport to which it is connected. This would also explain why WM has had much success with it, with his highly modified, clocked transports.

This also suggests that the superDAC will be more responsive to transport differences than, say, the DAC64 is, with its FIFO input buffering and local reclocking.

A further, slightly less technical musing is that I have found the superDAC to respond a lot to being attached to the trichord powerblock with a eupen. The on-board rectification of the 9V (external transformer) AC input is done by 1N4001 diodes (the cheapest of cheap diodes... the sort that you give to school kids). With the level of difference I found with the change in power source, this suggests that an improvement (for example to a 30p 1N5821 schottky) in the rectification could have some surprisingly large effects (especially for the meagre sum of £1.20). Of course, some improvements to the caps on board would also not go amiss, as the present electrolytics are VLSOM 20% (basic, cheapy ones). A few black gates, OSCONs or whatever fancy caps are your fancy could do quite a bit.

Of course, WM already mentioned changing the op-amps from the sub 50p NE5534s. Again, this may come down to personal preference (AD825s, OPA604s, etc). Also, WM mentioned biasing the outputs to class A.

With all of these modifications, you will have changed a significant proportion of the components in the device. Beyond that, there is stuff like improving shielding (perhaps some ERS, or otherwise the traditional copper) and of course the improvement of the external psu (I started with this because my superdac came without the original psu, so I had to build or buy one anyway).

All in all, by putting a little money into modding and providing it with clean power, I think the superdac is a seriously capable bit of kit. It needs good ancillaries though!

 

intresting, I'm still gonna get one of these soon. with a shopping list to go with it

 

I would like to point out here, for the guys at home, the transport we use it on in test room, isn't clocked or touched in way at all (YET), I also recommend (a job for issac here), subsituting the dac chip for a 24 bit 48 KHZ A/D version (higer s to n ratio) again fairly cheap.
I've ordered all the bits Issac has outlined above, no oscons and B/G are out due to the total cost, plus very few oscons are made that will replace the caps in there 35V 470Uf, 16 V 470 Uf, 50 V 10 Uf, etc, and the B/G's are serious money and it's a budget job, ok I got all the uprated caps and Diodes for £16 today, so we'll see how well it does on thursday. Wm
Ps can Isaac sus, the other fairy cheap mods to the dac, that will reap good rewards so far grand total including purchasing of said dac and mods so far are £158+£16 I spent today

 

Keep an eye on the voltage with the new diodes, it may creap up a bit so a resistor may be needed to bring it back down again.

 

With regard to caps....

Probably a good vfm (as well as available) choice for the 470uF 35V is the Starget (RS 215-5720). This will also work for the lower voltage 470uF, although it is somewhat larger... There is a 10uF 25V oscon... one would have to look carefully at the voltage across the 50V 10uFs, because I suspect that it is 15V (from the regulators, the caps being local smoothing for each individual op amp). However, the 10uF 25V oscon is 85p, and the 30V version is £1.93, and there's a 16V at 56p (RS176-8274, 188-3137 and 188-2998 respectively... any of those part numbers will take you to the page with them all on). If they are, as I suspect, between 15V and 0V then the 16V version will suffice, but with 12 appearing in the DAC, it's expensive whichever way you go. Cheaper alternatives are the rubycons (rubycon make black gates... they also have cheaper lines of caps).

The voltage output of the rectifier will indeed likely creep up somewhat with new diodes as they have a lower forward voltage than the old ones. However, since the rectifier output feeds into 7815 and 7915 voltage regulators, it's not an enormous issue, as these will still be within spec for their input voltage.

 

It may also explain why, the little monarchy upsampler produced a good hike inperformance, when even in stock form
IT's reclocking ability & SRC helping the little super dac to achieve bigger and better things

 

Another capacitor option is the new Sanyo SH solid aluminium, which seems to be an oscon of slightly lower spec and cheaper. farnell 431-1206 will get you to the page.