Custom transformer for Discrete DAC

Some 3 years ago I built the famous “Starlight” discrete DAC and CD transport, the DIY version from Sonic Illusions.  This is, in my opinion, one of the best DACs I’ve every listened to.

One of the disadvantages of the DAC output is the high (10kΩ) resistance and AC coupling requirement (unless you are prepared to balance DC, etc.) The traditional output configuration is an output capacitor feeding a step-down transformer to provide lower output impedance. The DAC can do only 1Vrms maximum due to the TTL output logic. With a 4:1 output I had to date, I wasn’t able to drive properly my VT25/10Y preamp into the 4P1L PSE amplifier

I decided to try a 1:1 output transformer but I had a clear challenge. The big primary inductance (Lp) needed demands a careful design and construction of the transformer to minimise the leakage capacitance and inductance which will impact the HF response. As I’m driving a long cable into the DHT preamp, this presents a demanding load of 47kΩ||400-600pF.  

I contacted DvB transformers to order a custom transformer to fit my needs:

With this configuration I could get nearly 2dB of loss and a good response above 20kHz. I asked Dorin for a minimum Lp of 200H, and Dorin sent me a first prototype of 307H.  

DvB used for this transformer the highest grade mumetal-permalloy available aiming for lowest THD under 200Hz/2Vrms and finest magnet wire produced by Elektrisola. Computerised non-bifilar hybrid winding technique, involving high interleaving rate and organic materials for isolation, in order to achieve the lowest leakage impedances I needed for this transformer.

So I submitted the transformer to some tests before replacing them on the DAC:

The initial results were very encouraging. Without the capacitance load this transformer can do 150kHz but when loaded properly, the response is like this:

The LF pole is about 2Hz due to the high inductance of the primary. The HF rolls off about 26.5kHz. Ideally should aim closer to 30kHz, which is something we will look doing by reducing the primary inductance and therefore reducing the leakage impedances from both primary and secondary windings. 

How does it sound?

I have these transformers running for a couple of weeks now. The sound of the DAC is very clear and detailed. I have now the gain needed in my system, so I can drive effectively the amp. These are a pair of amazing handcrafted transformers. I will review further prototypes as DvB have them available.

3 Replies to “Custom transformer for Discrete DAC”

  1. Nice transformer. Just coincidental i am buzzy with a interstage se wich measures >300H without the gap. I tested it as in your set-up and i get almost the same results, -3dB 3,5hz-35kHz. The big differance is that i use a “big” nanocrystaline c-core and that the design is intended for se (with a gap in the final configuration).
    I guess with a little different winding technique and the same H it is possible to get bandwith a bit better for the high frequencies.

    Martin Mug

    1. Hi Martin, great hearing from you. I’d be nice to have more details about your transformer.
      The load capacitance is a swinging factor here. Are you testing it with a similar load? If I reduce it slightly I can easily improve FR. As per my results, without capacitive load, it can do over 100kHz. Of course is not reflecting my real life setup.
      I will ask Dorin for more information as I know he’s working on other prototypes as he also found that FR could be improved.
      For this particular DAC, it’d be ideal to have 30kHz (-3dB) response, not more.

Comments are closed.