DAC 2496 (AK4393) DAC KIT With CS8416+AK4393+5532

Testing to find best output opamp

I tried LME49720 biased in class a, AD8066 and LT1364. LME49720, although it had great timbre when biased in class a, was outperformed by the other two opamps. Ad8066 and lt1364 where both really good in terms of resolution and seperation. Ad8066 has a huge soundstage, but LT1364 is musically more engaging and is currently my favorite. I intend also to try LT1028 and OPA627

When I complete my DAC, I’ll post a comparison of Dario’s BOM opamp (OPA827 aka OPA827AID) to one my brother recommended that I try, ADA4075-2. I don’t know if it’s a familiar one to anyone else posting on this forum. Does require an adaptor. (My brother designs and builds audio mixers for a living). Should be interesting.

I plan to apply Dario's shunt design, assuming it tests out well as the discussion moves forward over the coming weeks. I've aready set out to build the PSU on a separate board, so incorporating some inexpensive upgrades to the default DAC design looks very doable. (Thanks Dario and all for working on this option)!
 
its also a bit pointless with modern opamps, which mostly have output stages that already operate in Class A, so just doing a blind mod without a scope to see if you are actually doing harm not good is not such a great idea IMO, I say just stick a good chip in there and leave it.
 
As others reported the most rewarding swap is for opamp rails and analog 5V (U8, U9 and U10)

But you still need U9 and U6 for the analog 3.3V (CS8416)

hi Dario,
could I have you advice once more. Do i really ned U9 if I have 3 separate salas regs attached to green points (att)?
So I built 80mA 5v version for analogue part to completely remove U8.
Another salas +-12v 200mA is built for two opamps - DAC's output and preamp (LM49720HA) that I'll put in the same box. I guess opams need no more than 50mA, salas requires some 30mA, so I still have about 50mA left to pass to U6 for A+3.3...:confused:
 

Attachments

  • s.png
    s.png
    22.8 KB · Views: 743
could I have you advice once more. Do i really ned U9 if I have 3 separate salas regs attached to green points (att)?
...
I guess opams need no more than 50mA, salas requires some 30mA, so I still have about 50mA left to pass to U6 for A+3.3...:confused:

According CS8416 datasheet (Va3.3 is for it) it wouldn't need more than 10mA so it should be fine.

But, in a state of the art mod, I wouldn't mix digital PS with Analog.
 
DAC lives!

On my DAC kit the PCB traces were shorted between pins as follows:
(this is correct according to the kit's schematics and my DAC certainly works – very well)

On the AK4393/6:
12 shorted to 13
15 shorted to 16
17 shorted to 18

27 and 28 are shown as shorted to GND on the schematic.

On the CS8416,
10 shorted to 11 (and GND)
20 shorted to 21

You should be able to see, between the pins, where the PCB traces have joined/shorted them.

Well when I finally got around to checking this advice in detail, checking the pin numbers against data sheets etc, I found that pins 4 and 5 on my AK 4396 had a solder bridge. I cut that and everything sings!
Burning in out on the bench....

I'm very happy I didn't fry anything.:)

Cheers,
Bill.
 
Problem fixed....

As usual it was some overlooked detail.
Thanks to Jnewbold and the datasheets for the key, and thanks to everyone for kind and helpful advice.
My transformer wiring was wrong too, that was a bit thick!
Multiple errors are always a wind-up...
DAC is sounding very good already! Great news considering I was so close to giving up.

Cheers (again)
Bill
 
btw, it's easy to understand from CS8416 datasheet current consumption (7th page), but for CS4398 and 74HCT04 I was puzzled. 17th page of CS4398 shows 25 mA for digital analogue parts, but there are also other parameters like interface current and PS current in power down mode. 74HCT04 manual doesn't show any current at all or maybe it's "additional supply current per input" on 9th page?
 
74HCT04 is a switching device who's current is composed of (1) static and listed in the data sheet, and (2) dynamic which is dependent on switching speed of the gate. This current is what charges and discharges the internal capacitance as the device switches. There should be a plot of current vs speed. That said, it is most likely below 2mA unless you are using it to buffer a high speed clock or such.

Data sheet shows quiescent (1) current as less than 40uA over temp range.

what does it drive?

The additional supply current per input is specified at Vi = VCC-2.1V (NXP data sheet). This is defining the current for a gate strapped in the linear (pseudo) region such as one would see with a feedback resistor used as an R-C oscillator.
 
Last edited:
The CS2496 typical current demand is:

25mA at 5V on the Va supply which is for the analog section.
1.5mA at 5V for the Vref section
25mA at 5V for the Vd digital section if powered from 5V.
18mA at 3.3V for Vd digital section if you are running it on 3.3V instead of 5V

I just looked at the schematic of the CS4398 board I have on the way and it uses three sections of the 74HC04 to drive LEDs from the 3.3V supply. A fourth LED is directly powered to indicate power present. With no values given I have to guess at the power but it would be reasonable to sink 2-4mA per gate so figure a total of 16mA for the LEDs.

The CS8416 and CS4398 are powered from V3.3, so use the 3.3V 18mA consumption.
 
Last edited:
Interesting. On Dario's upgrade BOM, I see a heat sink for the positive 7812 regulator. It is mentioned as MANDATORY.

I ordered it (60 cents), but is it really necessary? If I need to install a heatsink the 7812 should be providing at least 250 mA.... My transformer can supply 500 mA, so that will not be the problem..... but why did Dario put a heatsink in the list?
 
To know how much heat is being dissipated on the regulator, measure the current that goes through it. Then multiply it by the in/out voltage difference.

That should give you a watt number, which should be less than that specified on the data sheet for no-heatsink operation.

But as Dario says, if the 7812 gets hot you will need a heatsink. Have you put your finger to it on several occasions to see how hot it is?

That will also depend on how many volts are going in on your specific place and on Dario's, which may vary from country to country or city to city.