Valve DAC from Linear Audio volume 13

[MarcelvdG]

Yes, just replace it with a jumper.

[SonnyMarrow]

Got the heatsink for the Maida tapped and drilled and got the standoffs figured out for the DAC and filter board.

The 9-pin connectors will be as close as possible to minimize that distance, then the XLR outputs will be right next to the filter outs.

I also added two standoffs/holes to the filter near the output transformers because they are heavy and to combat the flex. Since there is no solder mask, I put some paint and electrical tape around the holes so the standoffs don't touch the ground pour.

[Mikky1975]

Hallo,
I had very little time in the last few month but I’m now ready to (finally!) begin building the dac. I’m therefore re-reading all documentation and the thread.
Concerning noise floor, as measured by Marcel recently: how that can be improved? Low noise regulation? B+ regulation? Do good E88cc make a difference?
I’m asking just to better understand where to put my attention since I’ve not started yet.

Thanks,
Mirko

[SonnyMarrow]

I can't say that any of it is necessary, but since this is an expensive project anyway, I went all-out and tried to minimize any noise, usually overpaying for over-spec'd parts just for my own satisfaction lol. Nothing crazy, Vishay RN/CMF resistors, WIMA, Kemet, etc. Every resistor is at least 1%, then the ones that are specified in the schematic to be 1%, I used 0.1%.

I'm using external PSUs for the -300V, the 6.3V, and a 5V for my USB thing. I'm also using this reg for the on-board 5V supply.

My case layout is also really geared towards separating power things from signal, and the signal ground connects to the chassis at one point using this circuit.

Hopefully the efforts aren't wasted. I won't have any way to measure the noise, so hopefully I just don't hear it.

[MarcelvdG]

There are a couple of resistors that should be stable and shouldn't have too much 1/f noise, particularly the anode resistors and anode trimming potmeters (RV1 and RV2). I recommend using metal film for the resistors and cermet for the trimming potmeters.

E88CC valves with good matching between their halves could very well lead to a lower noise floor than E88CC valves with somewhat worse matching. If you can measure or record the noise, you could try various permutations of the E88CCs, trim the anode trimming potmeters every time and see if it makes any difference. I used old old stock E88CC's in random order, by the way.

Retrimming the anode trimming potmeters RV1 and RV2 every now and then definitely helps.

Crosstalk from the sigma-delta modulate to the voltage reference or to the clock that drives the lower E88CCs worsens the noise floor, but I don't know any way to reduce that further than it is already reduced. Reducing this crosstalk was the reason to go for a four-layer board with the analogue/mixed signal circuitry on the top, the digital stuff on the bottom and two planes in between. C5, C24, C25 and C34 are also there specifically to keep the crosstalk to a minimum.

For the raw DSD version, you could try different DSD rates and using the bit clock or the master clock.

[SonnyMarrow]

I bought the WaveIO, which has data, word clock, bit clock, master clock, Vin, and 5 GNDs. I think to connect to P9, it's:

data to pin 5
word clock to pin 11
bit clock to pin 9
master clock to nothing
Vin to pin 19 or 20
GND to pins 2, 4, 6, 8, 10?

[acg]

Originally Posted by MarcelvdG

There are a couple of resistors that should be stable and shouldn't have too much 1/f noise, particularly the anode resistors and anode trimming potmeters (RV1 and RV2). I recommend using metal film for the resistors and cermet for the trimming potmeters.

May be a waste of $$, but I have used Vishay Foil 1240W series trimpots in my build, simply because of their position at the anode and because they are less likely to wander than the cemets.

[Mikky1975]

Thanks Marcel, acg, SonnyMarrow for the hints, already processed as per your suggestions

I don't seem to find a BOM for the B3 Lundahl filrters, apart from transformers and RM8 inductors suggested by Ray. Does anybody have the BOM handy?

Many thanks in advance for the support!

Mirko

[diyAchim]

Re: trimpots - they need to be trimmable from above, otherwise this will be a PITA.

One turn trimmers for the tube balancing seem to me optimal in this position. With a multiturn trimmer it will not be easy to sweep, and sweeping gets you there!

[Mikky1975]

Originally Posted by acg

May be a waste of $$, but I have used Vishay Foil 1240W series trimpots in my build, simply because of their position at the anode and because they are less likely to wander than the cemets.

I see Vishay 1240W are 1/4W, enough for what they have to dissipate?

Ciao,
Mirko