Soekris' DAC implementations

Du you plan to inject these voltages in J2? Is J2 "south" of any onboard regulation i.e. sees the loads?

//

I have no idea - I had read you could bypass the onboard rectifier and regulators. Any clarification about this is welcomed!

I see the place on the PCB (drawing) and wondered if that was it. Even more importantly I wonder if I can leave out the +-12 volts supply since I will not be using the output stage.

To the fellow who thinks batteries are dumb, I agreed until the A123s 18650 were brought to my attention. Look at the specs and tell me they can be bettered with any AC supply. They may be overkill, I do not know, but they have an extremely low ESR which extends into the high frequencies that cannot be matched by any regulator.

I am actually not using a reg after the paid of 3.3s but will charge them only to 5 volts with a very high quality regulator. The regulator becomes transparent.
 
I can't find the spec of the bridge : MCC M8TSS ? Do I read well ?

How is working the second bridges when 2 are in serie (in the case we are feeding the DAC with a 10 to 12 VDC) ? Does it just drop the voltage of one diode of a half bridge and adding its noise !

Is it a liminting factor if it is less faster than the first one can add with a DC supply (like Schottky diodes bridge e.g.) ?


What could be the interest of a super reg with a low impedance (shunt reg ?) through the first stage of embeded filtering caps (anyone understood the repartition of the 6 caps ? 4 for the negative rails and two for the positive ?????)?

A usefull information for a simple DC could be to know how much uF are present on each rail ! Don't want to use a first cap (for the pulse current) on my DC PS (a simle Pi filter) whith a value superior to the one present of the dac board !

Any advice between the "two schools" : 30 VA R-core VS 5VA torroid ?

Are there really low impedance cells like Alkaline are but whichh can be reloaded ????
 
Last edited:
3x 820 uF on positive rail, 1x 820 uF on negative rail, then RC filter with 12R and 820 uF for each of the output buffer rails. The positive rail draw about 3 times as much power as the negative as all digital logic runs on the positive rail.

It's designed to be powered directly by 2x 7-8V AC voltage from a transformer, all the internal power supplies are regulated, filtered and decoupled as needed, there are no advantages of external regulators, batteries or snake oil.

You are of course welcome to experiment as you want, as long as the absolute input voltages don't get higher than +- 15V....
 
3x 820 uF on positive rail, 1x 820 uF on negative rail, then RC filter with 12R and 820 uF for each of the output buffer rails. The positive rail draw about 3 times as much power as the negative as all digital logic runs on the positive rail.

It's designed to be powered directly by 2x 7-8V AC voltage from a transformer, all the internal power supplies are regulated, filtered and decoupled as needed, there are no advantages of external regulators, batteries or snake oil.

You are of course welcome to experiment as you want, as long as the absolute input voltages don't get higher than +- 15V....

Thanks Søren,

I've been watching the thread(s) from the sidelines and I'm guessing this must be quite a challenging time for you with so many questions being raised and having no control over what people are actually doing with your boards. I hope the effort of supporting the DAC isn't taking you away from earning your daily crust.

The HiFiDUINO post looks to be a useful wiki type resource so let's hope it gets referenced.

https://hifiduino.wordpress.com/2015/01/30/building-soekris-r-2r-dac/

A couple of questions of my own if I may;

referring to you post above, I'm planning to use a custom wound toroid with secondaries for the DAC boards, the required 3.3V isolator supply and a network renderer. I think I'll reflect the asymmetric load represented by the DAC board and specify different secondary currents for the positive and negative DAC supplies?

How much current can the raw and buffered outputs source? I may need to run several metres of interconnect to my amplifiers.

Anyway, thanks again for all your hard work, it is really appreciated.

Ray
 
Last edited:
Has anyone tried using DC power?
My board is on the way and it seems you just "plug" into the proper spots?
I am going to use the A123 batteries - one for 3.3 and two (regulated down) or 5.Will not be using the buffer so can I ignore the +/- 12 volts? Sure hope so.

Why in the world would one want to do that?? 😕

As we know the board is designed to run off AC or DC, and I guess the implementation of the supplies is as well thought out as the rest of the board.

So where is the need to fiddle with injection in places that were not meant to be used? This board is as easy as it gets in setup and ready to provide
world class performance (judging by the findings of the first users) but it seems that some are only happy when they can throw as many uber-regs or batteries at something as they possibly can.
Next will be to add a tube output stage 😉
 
A couple of questions of my own if I may;

referring to you post above, I'm planning to use a custom wound toroid with secondaries for the DAC boards, the required 3.3V isolator supply and a network renderer. I think I'll reflect the asymmetric load represented by the DAC board and specify different secondary currents for the positive and negative DAC supplies?

How much current can the raw and buffered outputs source? I may need to run several metres of interconnect to my amplifiers.

Anyway, thanks again for all your hard work, it is really appreciated.

Ray

The power input has a full bridge rectifier, it use both windings for both supplies so they should be symmetric.

The raw output has a zout of 625 ohm, the buffered one has zout of 10 ohm. You can use reasonable long cables for both as long as they're shielded, several metres is not a problem.
 
3x 820 uF on positive rail, 1x 820 uF on negative rail, then RC filter with 12R and 820 uF for each of the output buffer rails. The positive rail draw about 3 times as much power as the negative as all digital logic runs on the positive rail.

It's designed to be powered directly by 2x 7-8V AC voltage from a transformer, all the internal power supplies are regulated, filtered and decoupled as needed, there are no advantages of external regulators, batteries or snake oil.

You are of course welcome to experiment as you want, as long as the absolute input voltages don't get higher than +- 15V....

Thanks Soren,

It's not really I want to experiment (no electric eel planned nore snake oils), it is just I have a 9V/30 VA r-core on hands.

So as already said I planned to reduce the voltage in the 10-12 VDC you advised. I was just asking for the rails caps repartition because I believed in the case of a Pi (which a DC before your caps looks like !), the first cap in front of the rectifier bridge (the one I try to do for my DC) does not have to be superior in capacitance in relation to the cap(s) after (those last can be equal or superior but NEVER below the first cap of a Pi DC Pi filter... at least it is what I believe to understand as I'm not technician, nore Nobel price and precisely want to avoid snake oils 😀)

So because the standalone negative 820 uF, my two cents question is : Should I stay for my own DC under or equal to 820 uF for the both rails (and then after it : the resistor value needed not to be above the 12 VDC max you advised & of course not above the 15V toasted grill for eels, salmon, smoked herrings with caps smoke or whatever...).

Sorry if my questions are too simple and are an insult of the great intelligence of some others members🙄.... it's just to paste to what is needed with parts on hands and NOT to try to do better (btw do you tried yourself and listen to difference between AC and feed by laboratory DC unit ?)

At least, if a united gain buffer stage with a traffo or a simple traffo at the output can tailor the sound, I really don't see why one have to deprive oneself or be ashamed to try it if the subjective result is improved !

For the moment I just plan to feed the DAM1201 without toasted it and deprove the result with a bad DC PS !

thanks for your help to a modest diyer (with a simple goal to understand more and avoid fatal break !) !🙂
 
Last edited:
Why in the world would one want to do that?? 😕

As we know the board is designed to run off AC or DC, and I guess the implementation of the supplies is as well thought out as the rest of the board.

Why in the world wouldn't you want the possibility of it beeing there?
No matter WHAT it is designed for, different supplies give different results.
This is tested plenty of times, and is no snake oil.

Myself i would love the possibility to experiment with different supplies, this is part of what diyaudio is about for many people.
 
am I right in thinking we still need the negative rail if we are to experiment with bypassing on board regs
we still need +4, -4 , + 3.3 and 1.2v
1.2 v will be difficult for lifepo4 fans might have to use something else but I doubt it will matter for this application
 
am I right in thinking we still need the negative rail if we are to experiment with bypassing on board regs
we still need +4, -4 , + 3.3 and 1.2v
1.2 v will be difficult for lifepo4 fans might have to use something else but I doubt it will matter for this application

Yes, the R-2R network itself need both positive and negative supply.

And I DON'T recommend supplying power to the J2 connector or other places beside J1 , the change of burning the board is big....
 
Brought my DAC up. In an effort to use what I had available, it's more complicated than it needs to be for Soren's fantastic board. I'm using a 12V S03 for the DAC, with 2 Reflektor-D's (5V, 3.3V) to power the Sonore USB->I2S board, and supply the ISO 3.3 for the R-2R. The 2 Reflektor-D's are sharing a transformer, as they're powering the clean side of the Sonore, and the dirty side of the R-2R Iso. Unnecessary double rechecking going on, plus the top of the line oscillators on the Sonore are of little benefit...

Anyway - at first impression it sounds very good. Just settling in for some listening.

An externally hosted image should be here but it was not working when we last tested it.
 
Why in the world wouldn't you want the possibility of it beeing there?
No matter WHAT it is designed for, different supplies give different results.
This is tested plenty of times, and is no snake oil.

Myself i would love the possibility to experiment with different supplies, this is part of what diyaudio is about for many people.

Søren has paid close attention to the PSU for the DAM1021 - as should be clear from the information collated over at https://hifiduino.wordpress.com/2014/10/12/r2r-for-the-rest-of-us/

The description of the supplies for the resistor ladder, in response to a concern about the matching of positive and negative supplies, sounds suspiciously like an onboard super reg:

I have a -4V reference that also sent though an inverter with 0.01% resistors generating the +4 reference, both are then filtered and buffered for each rail and channel, all with precision opamps.

Have a look at part 1 of Walt Jung's Regs for High Performance Audio, and you'll see a filtered reference buffered by a precision opamp is used in the both Sulzer and Jung "super regs". Same principle is used in the circuit described in Walt Jung's article on ultra low noise voltage references. What improvement do you think a tacked on supply will bring over what is already on-board?

The other area that is critical is the XO supply. Perhaps if the DAM1021 was using a fixed frequency XO with 0dB power supply rejection, this would be an issue but one of the features of the is that it Si514 has an on-board LDO. The specs on the Si514 data sheet indicate that with a 100mV sine signal overlaid on the power rails clock jitter increases by around 3-3.5 ps.

This is the SiLabs spiel:
All Si51x devices include on-chip voltage regulation, minimizing the impact of system-level power supply noise on clock jitter and keeping jitter below 1 ps rms jitter even in noisy system environments. This noise immunity is advantageous in FPGA-based systems that rely on tightly regulated switched mode power supplies. By filtering power supply noise inside the device, Si51x XO/VCXOs can be mounted next to FPGAs without requiring external low drop-out regulators for power supply filtering.
Silicon Labs Oscillators Minimize Jitter and Cost in Communications and Embedded Computing Systems | News and Press Releases | Silicon Labs

Anyway I'm sure you get the picture. Using super regulators isn't going to hurt performance, but don't expect it to make much improvement.

The board sounds very nice with a 7VAC 5W transformer hooked up, and I highly recommend that you give this setup a try before assuming that it cannot possibly sound good with a trafo directly connected.
 
Last edited:
I highly recommend that you give this setup a try before assuming that it cannot possibly sound good with a trafo directly connected.

Which is the reason i earlier asked a question if anybody comparied both. And Yes i totally agree, that if the R-2R rail voltages are really important, then it's best to use whats on board - but it still would be great if it was possible to bypass the bridge/caps, and the higher voltage for the buffers (since some people won't use them).
What i said in my last post, is that i don't see how anybody can think its stupid to make it POSSIBLE - its not like i said it was bad using the onboard regulators (since i know nothing about them) - but it doesn't hurt giving the possibility.
 
Which is the reason i earlier asked a question if anybody comparied both. And Yes i totally agree, that if the R-2R rail voltages are really important, then it's best to use whats on board - but it still would be great if it was possible to bypass the bridge/caps, and the higher voltage for the buffers (since some people won't use them).
What i said in my last post, is that i don't see how anybody can think its stupid to make it POSSIBLE - its not like i said it was bad using the onboard regulators (since i know nothing about them) - but it doesn't hurt giving the possibility.

The buffer op-amps are powered from an RC filter. It's not too difficult to locate the two 120R resistors located ahead of the final 820uF caps, so if you were bent on disabling the power to the buffers you could just remove these 2 resistors. If you use something like Chip Quik ( https://www.youtube.com/watch?v=7kyaz4Zrd78 ) you are extremely unlikely to damage the board in the process, and it's easy enough to replace the resistors.

Same goes for the main caps. If you wanted to completely bypass, remove the last cap in the main smoothing cap section, and use the now vacant pads to insert your supply.

I'm not advocating you do this but if you really want to mess with this stuff it's not really beyond the scope of DIYing. But if you can't work this out for yourself I wonder if you are really equipped to be doing these kind of mods.
 
Last edited: