Output transformers for DACs

[bisesik]

To Zoran:

Different output currents - different i/v resistors for the same targeting Vout - simple story. Regarding 1794. Simply use the same values of both poles i/v resistors - and there will be almost no DC-offset across the Pos and neg poles (as the offsets for neg and pos poles has the same value including sign). Connect the primary coil between pos and neg then - that's it.


Regarding specs. I have some technical data, but for some particular trafos versions only. Measuring and collecting info such you described is boring process for me in most cases, so I prefer to interact with the customer on the level of the real project implementation. In such a case the knowledge of electrical parameters you have described is really fades into the background. Parameters are just parameters. They matters for sure, but in most cases within my own transformers, for my own internal needs. I don't want to fight windmills, I have no time for such.

 

[Zoran]

There are -6.2mA of DC component x 4 times, creating offset at the Riv @ direct output?

This produce -0.62Vdc off-set @ 100 ohm Riv resistor. And this -0.62V will run trough the primary of the transformer. If the transformer core is from high permeability (probably is) this can produce core saturation and magnetization.
This is specific only for this dac chip model...

 

[bisesik]

Zoran, you should understand that there is no offset between pos and neg poles. Primary coil has no deal with the gnd, so the real offset that can be found between the pos and neg poles (between Ioutp and ioutn) relates to the internal imbalance of the DAC chip between the two poles output currents. In real it is less than 5mV with the Ra/Rb iv resistors with 1% tolerances. You can make it even 0.0mV if needed by adjusting the Ra/Rb values.

 

[bisesik]

Here is an example of how to "scale" my trafos/buffers boards (using iancanada's DAC II for example, but it works excellent with any other current output type DACs) to allow DAC chip to work on to the very small i/v resistor with the high enough voltage output level after my trafos (boards). Active stage is mandatory (+/-15V) in this case.

 

[drone]

Hello,
I would like to connect Headphones (70 OHM Audeze LCD-2) direct and balanced to an Ian Canada DAC (dual mono es9038) with your transformer board. Would this work well (impedance match?) or should I go preamplifier --> Headphone Amp?

thank you
Branko

 

[bisesik]

Hello Branko.

70 ohm is too low impedance even for the active stage after my trafos (250-300R is the minimum at 1-2Vout), not talking about the direct connection after SUTs in a passive mode. So pre/amp is mandatory in your case.

BR,
Ivan.

 

[drone]

Hello Werner,
Please use 100R across the primary with ~13k across the secondary coil instead of 4k12 across the secondary. 100R resistors should be as good quality as possible. Wirewound or AN tantalum/niobium or Z-Foils, etc high quality audio resistors.
Sure you can use trafos with D3 or DDDAC.

Ivan.

Hi Ivan,

did not had time to tinker with my setting. Which is working fine btw. I use 4 Kohm on the sec. on your trafos right now (balanced to preamp).
I was planning to go with only about 16 Ohm on prim. (1/60 of 100 Ohm (or lower)).
Now I see you recommend to split load between prim. and sec. What is the best way to go? and how to split? I wish to lest my setup with "normal" resistors to get the right volume for my chain an then update to "quality" resistors.

thank you
Branko

 

[dht 4 me]

Have you ever thought about an air gapped transformer for the TDA1541a?

 

[bisesik]

Nothing to think about actually. Gapped means losing all the high-permeability (linear inductance for the whole BW) benefits. I prefer to use a very low DCR of primary coil instead.

 

[dht 4 me]

A low DCR primary coil is inherently going to have a low inductance and nickel will show hysteresis-magnetization effects negating high perm anyway.
Is that really the best way to deal with a DC current on the core?

 

[bisesik]

My cores are not nickel-based. FINMET if in two words. You can try to use a separate primary coil to compensate the DC-offset btw (gap - is really not my story with such cores). Or you can compensate the DC using extra circuit on jfet. But I prefer to avoid such extra difficulties because...
I have done some measurements couple years ago using 1:44 trafos with TDA1541 (1541 sees 10 Ohms, output impedance was ~20k).

sine 20Hz	sine 1kHz	sine 20kHz
-2dB	0dB (0,52Vrms)	-0,7dB
THD 1,42%	THD 0,0023%	THD 0,0033%

As for me results was good enough without extra coils, compensating circuits, extra PSUs, etc... 🤷‍♀️

 

[dht 4 me]

I thought I read somewhere a long while ago that there was some modulation of the DC current that would make a strip would core unhappy even with a counter coil. Maybe that is why we don't see it implemented often? I have heard about the implementation but see very little evidence it gets implemented with great success. I am not trying to stirr any hornets nest, just wondering if it would be worthwhile to have a small gapped unit for the TDA. The LF 20hz distortion in the numbers you posted certainly is higher than at 1k but if it is an ungapped core it's not as bad as I would expect and then maybe the thinnest of gap or even a polished cut may be enough to help the LF saturation.

 

[bisesik]

The LF 20hz distortion in the numbers you posted certainly is higher than at 1k but if it is an ungapped core it's not as bad as I would expect

PCM63 in the same conditions gives very similar results to the TDA1541. Diff mode for 1541 gives 0.66% at the same output level. So I can conclude that the DC-offset not affecting at the chosen configuration, or affecting slightely.

 

[Capt.Foong]

Hi Bisesik, i wan to order a fully assembled i/v transformer to hook up with an iancanada dac system, can u email me the options at capt.foong@gmail.com, thanks.

 

[drone]

Hi,
as I am in the process of buying (high quality) resistors for the primary side of my transformers I measured them today for the first time. Left channel 1-->2 -- 2.4 OHM; 9-->10 -- 110.2 OHM. Right channel 1-->2 -- 2.0 Ohm; 9-->10 -- 88.2 OHM (Fluke 177).
Do I have to change anything or could I use same resistor value across booth sides. At the moment I use 1000 OHM at the secondary and would like to go with 16 OHM on primary. Or even less (how low could/should I go?)

Thank you
Branko

 

[bisesik]

Hi Branko, 15-20R across the primary will be better than 1000R across the secondary. Try 15-20R across the primary and 10k across the secondary.

BR,
Ivan.

 

[drone]

Hello Ivan,

got my resistors today. As I have the older version of your PCB for Ian's DAC, there are only soldering points for resistors at the secondary R1/R2. So I would solder the 16R directly between the receptacles of the primary? And the 10K at R1/R2. Would that be OK?

BR
Branko

 

[Zoran]

Hi Henri,
Simply use a 3-resistor attenuation circuit instead of 10k across the secondary.
View attachment 1003439
R1,R3 = 10k
R2=150k
This will attenuate your signal by -24dB.

p.s. if you need other tahn 10k input/output impedances (they can be different as you want), tell me for recalculation.

p.s.2. transformer do not steal the power of signal for attenuation, it just transforms the impedances (power approx. remains the same), whereas resistors steals the signal as dissipate it as a heat. That is the difference.

better solution is so called bridge "T" attenuator. Passive R network.
example
https://www.electronics-tutorials.ws/attenuators/bridged-t-attenuator.html
.
(For balanced version use Bridged "H" net.)

 

[Zoran]

Hi Branko, 15-20R across the primary will be better than 1000R across the secondary. Try 15-20R across the primary and 10k across the secondary.

BR,
Ivan.

Yes because in that way the iv conversion will took a place before transformer and partially on Rdc of the copper of primmary...
.

If I may suggest:
Dont put the ractive load direct at the output of the DAC.
Transformer has complex RLC reactive impedance, and DAC will "have" it on the output...
in step-up transformers capacitance from secondary is transfered to the primary (and DAC output) by square of transformation ratio. And capacitive component become huge in that way, as the N is higher...
.
Try simple experiment:
Use passive Riv at the output of the cuurent DAC
after that for instance simple JFET buffer (lower Rdson, Ouput resistance will be app Rdson for driving OT.)
with smaller Riv voltage will be smaller amplitude and OT will be more happy with that. Also the DAC will be isolated from reactive load at the output. The load will be only pure Riv as it can be, and JFET input capacitance. Much much smaller reactive load to DAC than any transformer.
Transformer will be more happy with lower Vp-p input, buffer as much lower output impedance will easily drive the transformer reactance all of the BW.
.

 

[bisesik]

In 95% cases people don't want to deal with active stages at all.
In my particular case (Riv is 5R), suggested by you connection is far from "simple", believe me...