The DAC can work with the bit clock as DAC clock, but possibly also with the master clock. The bit clock will generally have a lower frequency (except for DSD512) and result in better settling, but the master clock may be cleaner than the bit clock. I don't know what will overall work best, although my guess would be the bit clock for DSD rates up to and including DSD256 and the master clock for DSD512.
Some time ago you mentioned a USB or reclocking board having master clocks of 45.1584 MHz and 49.152 MHz rather than 22.5792 MHz and 24.576 MHz. As these frequencies are too high for the DAC core, I included a frequency divider to divide them off to 22.5792 MHz and 24.576 MHz. An advantage over using 22.5792 MHz and 24.576 MHz crystal oscillators is the well-defined duty cycle you get when you divide off these higher clocks.
P8, P9 and P10 are meant to choose between normal mode and divide-by-two mode by placing jumpers:
Normal mode: P8 shorted, pin 1 and pin 2 of P9 connected, pin 1 and pin 2 of P10 connected
Divide-by-two mode: P8 open, pin 3 and pin 2 of P9 connected, pin 3 and pin 2 of P10 connected
Always use normal mode when you use the bit clock. When using the master clock, you have to use divide-by-two mode when the master clock is above 27 MHz and you can use the divide-by-two mode whenever the divided frequency is greater than or equal to the DSD bit rate. That is, with 22.5792 MHz and 24.576 MHz master clocks, the divide-by-two mode can work with DSD256 or less, but not with DSD512.
By the way, when using the Amanero-compatible connector, you have to place the 0 ohm resistors R48, R57 and either R58 or R59 depending on what clock you want to use. When using the U.FL connectors, it is best not to place any of the 0 ohm resistors R48, R57, R58 and R59.
Last edited:
So the build is officially underway!
Picked up the smd soldered PCB from my chum today - he's done a really good job of hand-soldering all those tiny parts.
I placed an order this morning for lots of through hole parts so I should get some more work done at the weekend.
Looks good!
One comment, though: you have to remove either R58 or R59, as explained in my previous post, otherwise you short two clock lines of your Amanero or reclocker board. I forgot to mention this in the document I wrote on 19 July, so my apologies for the inconvenience. Removing SMD resistors is fairly easy with two soldering irons, one for each side, although it can also be done with one.
Edit: I wrote this before reading post 303.
Last edited:
Thanks. I'm lucky to have a friend with those skills!
No need to apologise marcel, I'm just extremely grateful for your responsiveness to my questions.
I'll return the PCB later in the week so he can add the diode I forgot to order and I'll get him to remove R59 at the same time.
I was being a bit tongue in cheek when I said about using LEDs to illuminate the valves from below but I'm now thinking that it might look quite cool. 3mm LEDs can be easily glued into the centre holes of the valve sockets, however, I already have enough 5mm diameter LEDs in a colour I like but, of course, they're too big for the holes in the sockets. I don't really want to start messing with trying to open up the holes in the sockets, as I can foresee some breakages, and thought I might just drill out the holes in the PCB and glue the LEDs to the board. Drilling out the board will remove the through hole plating that would have connected the tube centre pins to the -300V plane but as I'm not using that I don't see any problems with opening up the PCB holes?
Last edited:
For the E88CCs, there is about 1.5 mm of clearance between the centre hole and the 0 V plane on layer 3. When you increase the hole diameter from 3.5 mm to 5 mm and everything is perfectly centred, it gets 0.75 mm bigger on all sides, so the clearance to the ground plane on layer 3 drops to 0.75 mm. For an inner layer, that should still be enough, besides you get more vertical distance to the -300 V because the hole plating is removed.
The outer layer tracks nearest to the centre hole are the cathode tracks at about 1.9 mm distance. Drilling out the hole should reduce that to 1.15 mm. On top of that, you get 0.36 mm of vertical distance to the -300 V plane on layer 2 when the hole plating is removed. There can be about 170 V over this 1.15 mm + 0.36 mm distance. According to the KiCAD calculator, 1.25 mm of distance on an outer layer should suffice.
All in all, I wouldn't dare doing it, but I can't think of anything that goes wrong as long as the bigger holes are properly centred.
The outer layer tracks nearest to the centre hole are the cathode tracks at about 1.9 mm distance. Drilling out the hole should reduce that to 1.15 mm. On top of that, you get 0.36 mm of vertical distance to the -300 V plane on layer 2 when the hole plating is removed. There can be about 170 V over this 1.15 mm + 0.36 mm distance. According to the KiCAD calculator, 1.25 mm of distance on an outer layer should suffice.
All in all, I wouldn't dare doing it, but I can't think of anything that goes wrong as long as the bigger holes are properly centred.
It would be very ironic to destroy an expensive PCB even before it gets up and running for future cosmetic reasons.
Or use a 3mm LED
Jan
I have zero intention of doing that, which is why I checked with Marcel. If he wouldn't dare do it then nor will I, especially just to save the cost of buying a handful of LEDs.
Anyway, moving on from frivolous stuff like LEDs, as the build is now inderway I'm starting to think about the output filter boards, however, I'm a bit reluctant to spend several hundred pounds on the parts for the filters I would ideally build (with the DON Audio filters and Lundahl transformers) without getting a feel for what the Valve DAC can do so I'm looking at some cheaper interim arrangements. If all works well I can replace them with the optimised filter boards later.
I was initially thinking about using the JLSounds filter boards but on reflection I think it is probably better to build some filters that are closer to those Marcel has designed.To that end I want to assemble an interim filter based on Marcel's third order schematic.
For the output transformers these look to fit the bill; they're copies of Lundahl LL1527s which themselves are essentially smaller, cheaper variants of the LL1684s;
R24 imitation Swedish LUNDAHL1527 Lunda cattle 1+1:1+1 audio isolation transformer microphone line input-in Transformers from Home Improvement on Aliexpress.com | Alibaba Group
It's a bit more difficult when it comes to the inductors though as, obviously, there are no standard products with precisely the correct value, however, these seem to be fairly close;
RFS1317-185KL Coilcraft | Mouser United Kingdom
and I'm wondering if they would be too much of a compromise or if they could be accomodated by adjusting the filter?
Thoughts?
I was initially thinking about using the JLSounds filter boards but on reflection I think it is probably better to build some filters that are closer to those Marcel has designed.To that end I want to assemble an interim filter based on Marcel's third order schematic.
For the output transformers these look to fit the bill; they're copies of Lundahl LL1527s which themselves are essentially smaller, cheaper variants of the LL1684s;
R24 imitation Swedish LUNDAHL1527 Lunda cattle 1+1:1+1 audio isolation transformer microphone line input-in Transformers from Home Improvement on Aliexpress.com | Alibaba Group
It's a bit more difficult when it comes to the inductors though as, obviously, there are no standard products with precisely the correct value, however, these seem to be fairly close;
RFS1317-185KL Coilcraft | Mouser United Kingdom
and I'm wondering if they would be too much of a compromise or if they could be accomodated by adjusting the filter?
Thoughts?
Last edited:
With Lundahl LL1527 transformers the series branch R3, C9 should be changed into just a 4.3 kohm resistor and no series capacitor. I hope the same holds for the imitation transformers. The resistors on the primary side have to change from 402 ohm to 475 ohm each.
At low audio frequencies, the distortion of an LL1527 may not be entirely negligible compared to the rest of the DAC, but it is not very large either; again I hope the same holds for the imitations.
The coils should work fine; they are not designed for low distortion, but the signal levels are very far below the point where they start to saturate.
At low audio frequencies, the distortion of an LL1527 may not be entirely negligible compared to the rest of the DAC, but it is not very large either; again I hope the same holds for the imitations.
The coils should work fine; they are not designed for low distortion, but the signal levels are very far below the point where they start to saturate.
Last edited:
I'm working on my reconstruction filter PCBs - any recommendations for capacitor types - I'm thinking smd components to keep everything compact but can go for through hole if they'll be less of a compromise. I'm thinking thin film smd resistors too. smd parts need to be not too small if I'm to solder them myself!
Thanks
Ray
Thanks
Ray
Polystyrene, polypropylene or ceramic class 1 (for example NP0, also known as C0G) with a tolerance of 5 % or less (that is, at least twice as accurate as the inductor, so capacitance inaccuracies won't dominate). Don't use class 2 ceramic capacitors (X7R, X5R, X8R, Y5V, Z5U...), as they are inaccurate and nonlinear.
NP0 ceramic multilayer capacitors are readily available as SMDs. A working voltage of 25 V is more than enough.
For the resistors, metal film through-hole or thin film normal SMD or thin film/metal film MELF. Manually soldering MELF resistors is slightly more difficult than manually soldering normal SMD resistors because they tend to roll off the PCB.
NP0 ceramic multilayer capacitors are readily available as SMDs. A working voltage of 25 V is more than enough.
For the resistors, metal film through-hole or thin film normal SMD or thin film/metal film MELF. Manually soldering MELF resistors is slightly more difficult than manually soldering normal SMD resistors because they tend to roll off the PCB.
Last edited: