If you had been following this thread from the start you would realise that in all probability the dacs are NOT balanced.
GerryM said:
So as not to be dogmatic about this bal./unbal. thing, this might interest you, if you haven't already seen it.
http://www.raleighaudio.com/passive_output.htm
Transformer iv doesn't come much simpler than this.
Interesting - I have seen this circuit before.
I still wonder - what is the benefit of having one large I/V resistor on the output side, as opposed to two I/V resistors to ground on the input side, with the primary taken off at the DAC outputs?
This cannot be the series/parallel comparison as mentioned on the linked web page as the primary side of the transformer is still effectively connected in parallel...
I guess you trade off two low value resistors for one higher value... no idea which would be more detrimental to the sound quality.
By the way there was no measurable dc offset at the DAC outputs.
I still wonder - what is the benefit of having one large I/V resistor on the output side, as opposed to two I/V resistors to ground on the input side, with the primary taken off at the DAC outputs?
This cannot be the series/parallel comparison as mentioned on the linked web page as the primary side of the transformer is still effectively connected in parallel...
I guess you trade off two low value resistors for one higher value... no idea which would be more detrimental to the sound quality.
By the way there was no measurable dc offset at the DAC outputs.
rfbrw said:
I went on this:
So, finally I have some data! I disconnected the Iout pin from the two PCM1702 DACs in one channel. Each Iout was connected thru a dedicated resistor to signal ground ... Then I connected up a dual trace scope to Both the Iouts and looked at the P-P voltages while playing a maximum amplitude test CD.
For a start the DAC outputs were identical but inverted.
I call that balanced. Interesting link btw.
guido said:
Mmm, after some more thoughts...
Looking at the evidence (logic before the dacs, shift printed on the pcb), Ray must be right.
Gerry, are you shure the dac outputs are also identical (but inverted) in the time-domain?? Guess your 'background noise' is due to the fact that the dac outputs are not happening at the same time, but one is half a sample later than the other. Can you see this on your scope?
I am using a dual trace scope, one trace for each DAC.
The scope traces of each DAC output (combined sine wave test signals) overlay perfectly when one input is inverted. (ie the input signals are perfectly out of phase and identical in shape).
When I add the traces on the scope (with one input inverted) I get one double as high output single trace.
When I add the traces with the inputs not inverted I get nothing but background very low level noise, nothing significant at all.
I cannot detect any delay / shift / stagger and to be honest I don't care - the DAC outputs on the original circuit (see post 1) are going into one differential op amp and out to one single ended output so thats good enough for me.
As long as I get and audio signal out that avoids all the active analog components I'll be happy.. what happened in the digital domain first is probably way beyond my capacity to do anything about.
The scope traces of each DAC output (combined sine wave test signals) overlay perfectly when one input is inverted. (ie the input signals are perfectly out of phase and identical in shape).
When I add the traces on the scope (with one input inverted) I get one double as high output single trace.
When I add the traces with the inputs not inverted I get nothing but background very low level noise, nothing significant at all.
I cannot detect any delay / shift / stagger and to be honest I don't care - the DAC outputs on the original circuit (see post 1) are going into one differential op amp and out to one single ended output so thats good enough for me.
As long as I get and audio signal out that avoids all the active analog components I'll be happy.. what happened in the digital domain first is probably way beyond my capacity to do anything about.
On the subject of transformers and loads... this article from Shure was an eye opener..
I had never heard of Load Loss before.
Look at the example in the appendix right at the bottom:
a +17.4 db transformer between a mike mixer unit gives only +3.7 db gain because of impedance mis-matches!!
Hmmmm....
http://www.shure.com/support/technotes/app-transform.html
I had never heard of Load Loss before.
Look at the example in the appendix right at the bottom:
a +17.4 db transformer between a mike mixer unit gives only +3.7 db gain because of impedance mis-matches!!
Hmmmm....
http://www.shure.com/support/technotes/app-transform.html
Found this:
The VRDS-8's control system is elegant, but may look unfamiliar at first. The usual track-skip keys are ditched in favour of a rotary, and an output volume control (it defaults to full output at power up), plus a switch to control the oversampling.
In essence this is eight times the sampling frequency, but a divider circuit gives an effect similar to 16-times and 24-times (labelled 'Sin') oversampling). There's even a digital filter 'mode' switch which allows the user to select a sharp attenuation above 20kHz, or a gentler rolloff above 19kHz (on paper, and in practice the gap between these two is not huge).
So in what setting are you measuring. 8x perhaps? Wonder if things (sines) start moving on your scope (timingwize), when you go to 16 or 24x. 🙂
So that's why there is all that stuff before the dacs😀
The VRDS-8's control system is elegant, but may look unfamiliar at first. The usual track-skip keys are ditched in favour of a rotary, and an output volume control (it defaults to full output at power up), plus a switch to control the oversampling.
In essence this is eight times the sampling frequency, but a divider circuit gives an effect similar to 16-times and 24-times (labelled 'Sin') oversampling). There's even a digital filter 'mode' switch which allows the user to select a sharp attenuation above 20kHz, or a gentler rolloff above 19kHz (on paper, and in practice the gap between these two is not huge).
So in what setting are you measuring. 8x perhaps? Wonder if things (sines) start moving on your scope (timingwize), when you go to 16 or 24x. 🙂
So that's why there is all that stuff before the dacs😀
Guido,
You mentioned you used I/V resistors with a transformer connected across a differential output (I will avoid the the word 'balanced'..) Could you be more specific on the setup?
- Which type of DAC?
- Max DAC current output?
- Resistor value/type?
- Transformer specs
(Make/Model nr/Ratio/input/output impedances...etc)
- What does the transformer feed into? Amp/Preamp type, input impedance, balanced (oops!)/single ended..
- Interconnects used..
I have seen a few posts with half details, 'this or that works great for me', but without the complete description of the system.. we really need a database of proven solutions for various DAC chips...
Thanks
You mentioned you used I/V resistors with a transformer connected across a differential output (I will avoid the the word 'balanced'..) Could you be more specific on the setup?
- Which type of DAC?
- Max DAC current output?
- Resistor value/type?
- Transformer specs
(Make/Model nr/Ratio/input/output impedances...etc)
- What does the transformer feed into? Amp/Preamp type, input impedance, balanced (oops!)/single ended..
- Interconnects used..
I have seen a few posts with half details, 'this or that works great for me', but without the complete description of the system.. we really need a database of proven solutions for various DAC chips...
Thanks
GerryM said:
- Which type of DAC? >> 2 tda1541A S1, one for each channel
- Max DAC current output? >> -4mA: output is 0 to -4mA
- Resistor value/type? 50 ohms, bit high for this dac
- Transformer specs
(Make/Model nr/Ratio/input/output impedances...etc)
amplimo TM3, 1:15, specs are given for the connected stuff:
input about 200 ohm (so 2*50 is a bit low, but ok), output > 45k ohm. using a multimeter for measurements kills them btw.
- What does the transformer feed into? Amp/Preamp type, input impedance, balanced (oops!)/single ended..
>> crap! Just a transistor buffer for the moment for testing.
- Interconnects used.. home made from telco cabling, preamp is tube stuff.
As you can see it is not finished.. Too much time here and too little time building. But the principle works. I'm looking at a new version, with cpld for all the digital stuff instead of logic+gal. And i have to use all the info from HtP on this particular dac which means my current design is
Things like +5 supply has audio currents through it, DEM reclocking and other stuff which is only relevant for 1541. Makes me wonder why i don't drop them for something modern. Connecting the transformer over the outputs of the dac means i get rid of the DC component of the output. Which is a healthy thing for those TM3's. I'm also not using fully balanced configuration (as i found out later
). One channel gets current sample, other channel gets inverse of previous sample.Crude way of 2x sampling i've been told. Good thing is i dont have the high freq rolloff from a 1x dac. I can really hear and measure the difference.
So one day i get myself a X95108 and continue, all works in simulation😉 Anyway, holiday first😎
- Status
- Not open for further replies.