Experience with this DIY DAC ?

[legarem]

Originally posted by fordgtlover
Sorry for the silly questions.

In legarem's diagram in post #418, the black wire is connected to the ground connection. Is that the analog ground of the DAC?

http://www.diyaudio.com/forums/attac...amp=1241223954

Also, on the same diagram, is the type of capacitor across the transformer primaries important (0.0047uF)?

Yes, it is connected on the analog ground

In my diagram, I forgot a 0 on the cap value. The good value is: .00047 uF not .0047 uF. Sorry.

If you keep the ,0047 uF, you can also change the 1K for 100 ohms and you'll get a lot more gain with the right cut off frequency.

The capacitor can be polypropylene, polystyrene, mica.

I had emali exchange with Bill Whitlock from Jensen transformers. Here is some interesting things related to the schematic included in this message:

1. The low-pass filters are used to supplement the characteristics of the transformer itself. Since the Faraday shield in the JT-11SSP-8MA (or any transformer with such a shield) functions well only below a few MHz, the two 220 pF capacitors and 249 Ω resistors function as a 3 MHz low-pass filter for common-mode to insure that RF energy does not interfere or cause strange intermodulation distortions in downstream equipment. The differential filter, consisting of the 2.2 nF capacitor and 249 Ω resistors, is set to be -3 dB at 160 kHz. This frequency is sufficiently high to add negligible phase distortion (that's deviation from linear phase, not phase shift) while low enough to control ultrasonic artifacts that can also cause strange distortions in downstream equipment (see the 1988 AES paper by Deane Jensen and Gary Sokolich for details on "Spectral Contamination").

2. The differential voltage across the transformer primary must be kept to an absolute minimum (<1 mV) to prevent a shift in the magnetic operating point of the core material. Any significant dc flow in the transformer will alter its distortion characteristics. Normal (for Jensen-selected core materials) transformer distortions consist almost entirely of 3rd harmonic, but adding a dc component creates a series of even-order distortion products (2nd, 4th, 6th, etc.).

[udovdh]

About the voltage across the primary: how?
The DAC chip has a certain offset.
Or will it be bear-zero? (no opamps)

[legarem]

Originally posted by udovdh
About the voltage across the primary: how?
The DAC chip has a certain offset.
Or will it be bear-zero? (no opamps)

As stated in a previous message, after modifying 6 of these dacs, there is always 0mv between the + and - of each output. You don't have to bother with any voltage across the primary of the transformers with the CS4398 and CS4397.

Use matched resistors at the primary of the transformers.

[Bill Fuss]

All the ones measured so far have had less than 1mv but thats with DMMs. I dont think anyone has measured with a voltmeter sensitive enough to really see anything. I have assumed that the DC level is not high enough to be a concern, but I may be wrong, as I often am.

[fordgtlover]

Originally posted by Bill Fuss
Those aren't silly questions.

The caps should be good quality polypropylene film.

Bill

Thanks

Originally posted by legarem


Yes, it is connected on the analog ground

In my diagram, I forgot a 0 on the cap value. The good value is: .00047 uF not .0047 uF. Sorry.

If you keep the ,0047 uF, you can also change the 1K for 100 ohms and you'll get a lot more gain with the right cut off frequency.

The capacitor can be polypropylene, polystyrene, mica.

Thanks for the response. No harm done, I haven't ordered the parts yet.

Just to clarify my understanding of this. The Resistor (1K) and Capacitor (0.00047uF) are acting as an RC filter.

My calculations, using the formula F = 1/(2 * pi * C * R), have the cut off frequency at 338691.43 Hz (338 KHz). Is this correct? If it is correct, why does it need to be so high?

Using the value of 0.0047uF for C, gives a cutoff frequency of 33KHz. Is this not high enough?

[piero7]

Originally posted by Franz Gysi
All Digitec's are tested and ready for shipment:


@Xaudiox: I miss your mail, still two pieces reserved for you!

Kind regards
Franz

Hi franz, I'm interested in your transformer, due I'm a new user can't still email to you. Please contact me

tnx

[legarem]

Originally posted by fordgtlover


Thanks



Thanks for the response. No harm done, I haven't ordered the parts yet.

Just to clarify my understanding of this. The Resistor (1K) and Capacitor (0.00047uF) are acting as an RC filter.

My calculations, using the formula F = 1/(2 * pi * C * R), have the cut off frequency at 338691.43 Hz (338 KHz). Is this correct? If it is correct, why does it need to be so high?

Using the value of 0.0047uF for C, gives a cutoff frequency of 33KHz. Is this not high enough?

You have to do the sum of the two resistors so 338 khz become 169 khz

If you don't have a preamplifier, use 2 X 100 ohms instead of 2 X 1K .
The caps value become 680 pF

33Khz is too low. Remember that the cut off frequency is at -3db

[Franz Gysi]

Hi franz, I'm interested in your transformer, due I'm a new user can't still email to you. Please contact me

You have good luck and I sent you a mail

Why?

All Digitec's are gone, just the pair I sent to Legarem for comparison will come back.

His opinion is, they play on the bright side. Soon we will read (hopefully) about Legarem's comparison and I am looking forward for that.

I still like the Digitec's, and I recommend to play with the surrounding parts (i.e. passive filtering at the primary side).

They are very linear over the whole frequency range and Legarem and I could not measure any overringing.

For some sound I still prefer the Digitec's over the Lundahls, for some not.

My theory: because of the bigger core of the Digitecs, they play "cleaner" bass than smaller trannies, wich means lower distortion in the bass area.

Franz

[hoverdonkey]

Originally posted by legarem


If you don't have a preamplifier, use 2 X 100 ohms instead of 2 X 1K .
The caps value become 680 pF

Legarem, I am new to this and so would be grateful if you could explain the reason for raising the cutoff if not using a preamp (I used the formula for 2x100 Ohm with 680pF, giving 1170kHz).

I use an integrated amp (Primaluna ProLogue 2) which I guess, by definition, has preamp built in. I am therefore planning to try your original filter (2x1K res, 470pF cap). Am I heading down the right path?

Thanks

[johnm]

Franz are you still using 10R resistors (and nothing else) between the DAC and primaries of your Digitecs?

Cheers,

- John