Hi
I am planning to drive high quality o/p transformers from a differential o/p DAC. Most of the current o/p DACs seem to have a fixed current draw on their o/p and perhaps the same is true for voltage offsets on voltage o/p DACs.
Audio transformers will want to avoid almost completely any DC current through their primary. Assuming a primary with a centre tap or two separate primaries that can be joined, there is an option to connect the centre tap to ground or not. I am told that some current o/p DACs will not work without this ground connection (e.g. PCM 1794) whereas others will (e.g. ES 90x8). I don't believe there is enough information in the datasheets to know why this is. In the latter case it seems that no DC current should flow to ground.
As an example, in my old Marantz CD94 using the TDA1541 each o/p apparently draws ≈2mA. So the designers have organised a 4k7 and 2k7 resistor in series from +14V to each o/p (with a 10µF cap to ground between the resistors). The behaviour of this is a bit mysterious to someone used to dealing with voltage-based circuitry but clearly 2mA across 7k4 ≈14.8V meaning that the o/p should be sitting somewhere near 0V and it measures at -0.4mV at the virtual earth. However, what happens when an AC current signal starts to appear into the virtual ground of the chip that carries out the I/V conversion? Is this resistor from +V a recommended method or just a crude get-by? I have experimented with a much larger cap here for better power supply decoupling and I got gross distortion - so I think I must be misunderstanding something about this circuit element in terms of it's AC behaviour.
Anyway that's a side issue, what I'm really interested in is if there are any recommended strategies for dealing with/removing/cancelling DC voltage offsets or current flow in the transformer connected to the DAC o/p's. I'm considering a PCM1794 but as I say above I believe it will need to have a ground connection from its outputs through the primary and they draw more than 4mA with no signal.
TIA
M
I am planning to drive high quality o/p transformers from a differential o/p DAC. Most of the current o/p DACs seem to have a fixed current draw on their o/p and perhaps the same is true for voltage offsets on voltage o/p DACs.
Audio transformers will want to avoid almost completely any DC current through their primary. Assuming a primary with a centre tap or two separate primaries that can be joined, there is an option to connect the centre tap to ground or not. I am told that some current o/p DACs will not work without this ground connection (e.g. PCM 1794) whereas others will (e.g. ES 90x8). I don't believe there is enough information in the datasheets to know why this is. In the latter case it seems that no DC current should flow to ground.
As an example, in my old Marantz CD94 using the TDA1541 each o/p apparently draws ≈2mA. So the designers have organised a 4k7 and 2k7 resistor in series from +14V to each o/p (with a 10µF cap to ground between the resistors). The behaviour of this is a bit mysterious to someone used to dealing with voltage-based circuitry but clearly 2mA across 7k4 ≈14.8V meaning that the o/p should be sitting somewhere near 0V and it measures at -0.4mV at the virtual earth. However, what happens when an AC current signal starts to appear into the virtual ground of the chip that carries out the I/V conversion? Is this resistor from +V a recommended method or just a crude get-by? I have experimented with a much larger cap here for better power supply decoupling and I got gross distortion - so I think I must be misunderstanding something about this circuit element in terms of it's AC behaviour.
Anyway that's a side issue, what I'm really interested in is if there are any recommended strategies for dealing with/removing/cancelling DC voltage offsets or current flow in the transformer connected to the DAC o/p's. I'm considering a PCM1794 but as I say above I believe it will need to have a ground connection from its outputs through the primary and they draw more than 4mA with no signal.
TIA
M
In the vendor sub-forum there is a guy that makes output transformers for dacs, including for ESS dacs. Some people claim they sound better than an opamp I/V stage. Anyway, so far I know there is no need to ground a center tap if the dac has differential outputs.
Hi @Markw4
Yes I have swapped a couple of messages with him. I think he might be the person who suggested that for the PCM 1794 you don't get any output unless you ground the centre tap.
Yes I have swapped a couple of messages with him. I think he might be the person who suggested that for the PCM 1794 you don't get any output unless you ground the centre tap.
Okay. A transformer can be designed to work with a DC bias current. Also, the bias currents from a balanced output dac flowing through an output transformer center tap should tend to produce flux that cancels out. To the extent the outputs may not be perfectly balanced at DC and or to the extent the center tap isn't exactly at the effective center for flux cancellation, the core could sized to handle it okay.
Is there any circuitry that can be put around the current DAC o/p and primary to cancel any small DC currents that might not be cancelled due to small imbalances in the DAC o/p?
How about for DACs with voltage outputs?
How about for DACs with voltage outputs?
Sure, capacitors or opamps (maybe not caps if you need a DC current path to ground). If you're going to use those things though then you probably don't need a transformer.
Why worry about small current offsets though if you are buying transformers from a guy that knows how to make them for that type of dac?
Last edited:
@Markw4 To answer your Q, there are no transformers that enjoy DC currents flowing through them, some will just tolerate small ones but with reduced bass response and increased distortion. I'm trying to find a way of circumventing the problem in a way that avoids active devices if possible and doesn't mess up sound quality. It may be that in some cases the DC current between differential outputs is so small that it's not worth worrying about but data sheets are pretty vague about this kind of thing.
@rfbrw thanks for the link but I know that page already and have already had some discussion with Brian Sowter. The issue in my mind is not so much about the transformers which I understand pretty well, more about the DACs, their output offsets and potential circuit ideas to null them.
@rfbrw thanks for the link but I know that page already and have already had some discussion with Brian Sowter. The issue in my mind is not so much about the transformers which I understand pretty well, more about the DACs, their output offsets and potential circuit ideas to null them.
Transformers can be designed for DC currents. Running DC through one not designed for it is where problems occur. This is stuff any EE should be able to understand.
Do you mean DC-to-DC converters for power supplies? I'm struggling to think of an audio transformer that works with DC and Faraday's law.
I just found this actually:
Might even be directly relevant to DAC I/V?
Might even be directly relevant to DAC I/V?
How about output transformers used for tube amps? They have tube DC plate current flowing through them.
Transformer cores have a B-H curve. One designs a transformer (among other things) to stay out of excess saturation. The full scale operating region of a DC biased transformer may stay on one side of the horizontal axis, but it doesn't have to saturate at LF. Entirely up to the designer.
The main consequence for audio might be that the harmonic distortion changes from mostly 3rd order to somewhat more 2nd order.
Transformer cores have a B-H curve. One designs a transformer (among other things) to stay out of excess saturation. The full scale operating region of a DC biased transformer may stay on one side of the horizontal axis, but it doesn't have to saturate at LF. Entirely up to the designer.
The main consequence for audio might be that the harmonic distortion changes from mostly 3rd order to somewhat more 2nd order.
Last edited:
Yes of course output valves have a DC current through them and so the transformers connected to them must too.
It might be nice to have 2nd harmonic rather than 3rd harmonic. Would the level of distortion tend to go up? It feels like you're only using one side of a symmetric mechanism unless you balance it somehow.
Maybe I am worrying about it too much. I just know that some core materials are very sensitive to it in audio transformers. I have never measured this myself and so am just taking the words of others about it and avoided it.
The level of harmonic distortion depends on transformer design. Hysteresis distortion is another matter, but it also depends.