Hello everyone,
I'm currently working on a differential DAC based around the AK4396, but I've run into a bit of a problem. Hopefully you will be able to help. 🙂
The problem is that the AK4396 have a differential voltage output centered around 2.5V. The normal practise is to feed this into a differential amplifier (most commonly an op amp), which gives a single ended output and recenters it around 0V.
Because this is for a differential DAC, the original plan was to only buffer the outputs and feed them straight to the amplifier. However doing so would effectively halve the voltage swing of the amplifier as it could never get below 0V, which is somewhat undesirable.
While I could theoretically use two op amps, with one of them being inverted, I would prefer to only use discrete components here, to have a better control the different stages. Really good op amps are also quite expensive. I've tried to use a differential amplifier made from discrete components (basically a modified long tailed pair), but without much sucess. Could anyone enlight me on the proper way to do this?
Another idea I had was to feed the AK4396 with -2.5V and +2.5V instead of 0V and 5V. This would mean that a 2.5V output from the DAC would equal 0V for the amplifier. I would prefer not to do this however, think of it more as a last resort.
Are there any other alternatives?
Any help would be most appreciated. 🙂
I'm currently working on a differential DAC based around the AK4396, but I've run into a bit of a problem. Hopefully you will be able to help. 🙂
The problem is that the AK4396 have a differential voltage output centered around 2.5V. The normal practise is to feed this into a differential amplifier (most commonly an op amp), which gives a single ended output and recenters it around 0V.
Because this is for a differential DAC, the original plan was to only buffer the outputs and feed them straight to the amplifier. However doing so would effectively halve the voltage swing of the amplifier as it could never get below 0V, which is somewhat undesirable.
While I could theoretically use two op amps, with one of them being inverted, I would prefer to only use discrete components here, to have a better control the different stages. Really good op amps are also quite expensive. I've tried to use a differential amplifier made from discrete components (basically a modified long tailed pair), but without much sucess. Could anyone enlight me on the proper way to do this?
Another idea I had was to feed the AK4396 with -2.5V and +2.5V instead of 0V and 5V. This would mean that a 2.5V output from the DAC would equal 0V for the amplifier. I would prefer not to do this however, think of it more as a last resort.
Are there any other alternatives?
Any help would be most appreciated. 🙂
I don't get your question. Are you feeding a balanced amp? Trying to avoid caps? Just getting from diff to single-ended? Why would a differential amp not work?
Just use an OPA1632 and be done with it. Read through the datasheet here:
http://direct.focus.ti.com/lit/ds/symlink/opa1632.pdf
Note:
The output common-mode voltage pin sets the DC
output voltage of the OPA1632. A voltage applied to the
VOCM pin from a low-impedance source can be used to
directly set the output common-mode voltage. For a
VOCM voltage at mid-supply, make no connection to the
VOCM pin.
http://direct.focus.ti.com/lit/ds/symlink/opa1632.pdf
Note:
The output common-mode voltage pin sets the DC
output voltage of the OPA1632. A voltage applied to the
VOCM pin from a low-impedance source can be used to
directly set the output common-mode voltage. For a
VOCM voltage at mid-supply, make no connection to the
VOCM pin.
analog_sa said:
I will be feeding a balanced amp, that's not the problem. The problem is that AK4396 outputs 0V to 5V (i.e. it's centered around 2.5V). I need it to output -2.5V to +2,5V (i.e. centered around 0V). Otherwise I halve the output swing of the amplifier.
The common practice is to use a differential to single ended converter (such as an op amp). This automatically recenters the output at 0V.
I want a balanced output, and while I could use two differential to single ended converters for this (basically creating a differential to differential converter), I want to avoid op amps all together.
I've tried to make a differential amplifier from discrete components (basically a modified long tailed pair), but so far I've only had limited success at best.
If I can't get any of this to work, powering the AK4396 with -2.5V and +2.5V seem to be my only option.
Two caps will eliminate the dc bias. And won't halve the swing 🙂
Or any differential input amp which will treat the dc as common mode.
Again, is your purpose to eliminate coupling caps?
Sadly, you can't eliminate opamps as they are internal to your dac.
A simple, ready solution is to use a Pass preamp - takes differential input and can output either balanced or se. And you can control volume at output if needed. In my setup a 1.7 Aleph works like a charm doing what you want. Of course it's full of coupling caps 🙁
Or any differential input amp which will treat the dc as common mode.
Again, is your purpose to eliminate coupling caps?
Sadly, you can't eliminate opamps as they are internal to your dac.
A simple, ready solution is to use a Pass preamp - takes differential input and can output either balanced or se. And you can control volume at output if needed. In my setup a 1.7 Aleph works like a charm doing what you want. Of course it's full of coupling caps 🙁
analog_sa said:
The problem is not the DC bias. Since this is for a balanced DAC, it's the difference between the two outputs that matter. If both are at 2.5V, the difference is still 0V. The problem is that the outputs are not centered around 0V.
For example, say we have an amplifier with a maximum voltage swing of +/- 25V. If I where to recenter the output of the DAC at 0V (i.e. it would output -2.5V to +2,5V), I could use a gain of 10 before clipping occurs (DAC outputs -2.5V to +2.5V, multiplied by 10 is -25 to +25V).
If I where to feed the signal straight from the DAC into this amplifier, the maximum gain I could use is 5, or else it will clip (DAC outputs 0V to 5V, multiplied by 5 is 0V to 25V). The voltage swing have effectively been halved.
Adding a capacitor will not change this.
analog_sa said:
I will integrate all of this into the DAC so I cannot use a ready made solution. I will take a look at the Aleph however, perhaps I can adapt it. 🙂
For some reason you are assuming that the DC levels have some kind of effect upon the AC gain of your amp. This is simply not the case. Whether your input signal has a DC component of 2.5v is completely irrelevant. What matters is the AC voltage developed between the two outputs of the dac. In case of AC coupling this is simply trivial. If you prefer DC coupling then you might need to consider getting rid of the bias voltage.
analog_sa said:
What you are saying applies only to differential amplifiers. I don't want to limit myself to using only differential amplifiers.
If I connect the positive and negative output of the DAC to two single ended amplifiers the voltage swing will be halved. The output of the DAC will never get below 0V, and as such the output of a single ended amplifier will never get below 0V.
Continuing the previous example, the amplifier can potentially swing between -25 and +25V, but it will only swing between 0V and 25V, thus the voltage swing have effectively been halved.
Cauhtemoc said:
You can use an OPA1632 and set the output common to ground, that way you get balanced>se, re-centered at ground. It's all in the data sheet. You only get about 12V peak swing but you can add your +/-25V amp after that. Question: what do you need +/-25V swing for?
Edit: forget this post, someone else already mentioned this....
Other question: would you really go to +/- 2.5V supply just to avoid an opamp? You may get more problems and worse sound than what you got rid of.
BTW If you feed the DAC outputs capacitively to your amp, you have gotten rid of the 2.5VDC without compromising swing, that will then be +/-2.5V (or whatever the DAC puts out).
Jan Didden
janneman said:
I don't. It was just meant as an example.
janneman said:
I want to avoid both capacitors and op amps. Feeding the DAC with +/- 2.5V is one way to accomlish this. I just want to know if there are other ways.
analog_sa said:
You're missing the point.
Imagine the DAC outputting a sine wave that varies between 0V and 5V (i.e. it's centered around 2.5V). Let's feed this sine wave into the +/- 25V single ended amplifier with a gain of 5 from the previous example. The output would be a sine wave that varies between 0V and 25V. Since we are already at the maximum 25V we cannot increase the gain any higher or the output will be clipped. Thus we are stuck with a voltage swing of 25V.
Now imagine the DAC outputting a sine wave that varies between -2.5V and 2.5V (i.e. it's centered around 0V). We feed this into the same amplifier as before, and the result is a sine wave that varies between -12.5V and +12.5V. This has the same voltage swing of 25V as before, but this time we can increase the gain to 10 and still be within the maximum +/- 25V. This would give us a voltage swing of 50V, compared to the previous 25V.
See what I mean?
You could use a 1:1 transformer to accomplish your goal.
No caps, no op-amp. As long as you don't mind heavy, expensive and limited bandwidth. 🙂
By the way, I have a RAKK DAC and use a 1:4 transformer and 1K resistor for V/I.
Doug
No caps, no op-amp. As long as you don't mind heavy, expensive and limited bandwidth. 🙂
By the way, I have a RAKK DAC and use a 1:4 transformer and 1K resistor for V/I.
Doug
A quality transformer is more expensive than an op-amp. Is a transformer more expensive than an active circuit and power supply, that's usually a bit more competitive. Is the transformer "better"? It can be, IMHO.
While I consider myself a transformer convert, after believing the propaganda that transformers should be a last resort, I recognise there are usually many valid ways to accomplish a design. I also recognise that there is a range of opinions on any of the technologies.
By the way. Heavy, expensive and limited bandwidth have been the complaints of transformers at least since the 1940's. For some applications, they are my first choice. Hence the smiley.
Doug
Quite in fact, there's a simple way to do that, but i wouldn't use it: Use a resistive network to substract 2.5 from each output:
DACout ----/\/\/\/\---Out---/\/\/\/\--- -2.5v
Both resistors are 1Kohm
This network will substract 2.5v from the DAC output, and also will divide the output voltage swing in half. You should notice that adding resistors in series could add noise to the signal AND you should also notice that unless the -2.5v source is perfectly filtered, you will also add the voltage source noise to the output signal. I am pretty sure the DAC has a NOMINAL 2.5 v offset, and NOT a warrantied one, so i assume it will vary with temnperature, aging, part, etc. So, instead of using a -2.5v source, i would use a servo amplifier to estimate the dc component of the output and supply a (variable) voltage that will cancel the output voltage... Is is assuming you reaaallly don't want a capacitor here (much simpler than a servo amplifier) --- Take into account you are not dealing with a high power output, it is perfectly possible to get a decent and not expensive at all non-electrolitic cap to block the DC component of the signal.
DACout ----/\/\/\/\---Out---/\/\/\/\--- -2.5v
Both resistors are 1Kohm
This network will substract 2.5v from the DAC output, and also will divide the output voltage swing in half. You should notice that adding resistors in series could add noise to the signal AND you should also notice that unless the -2.5v source is perfectly filtered, you will also add the voltage source noise to the output signal. I am pretty sure the DAC has a NOMINAL 2.5 v offset, and NOT a warrantied one, so i assume it will vary with temnperature, aging, part, etc. So, instead of using a -2.5v source, i would use a servo amplifier to estimate the dc component of the output and supply a (variable) voltage that will cancel the output voltage... Is is assuming you reaaallly don't want a capacitor here (much simpler than a servo amplifier) --- Take into account you are not dealing with a high power output, it is perfectly possible to get a decent and not expensive at all non-electrolitic cap to block the DC component of the signal.
ejtagle said:
He doesn't seem to mind a capacitor but has problems believing a capacitor can remove DC 🙂
DougL said:
I don't want to use a transformer, for reasons stated above.
ejtagle said:
I have also thought of this method, but feeding the DAC with -2.5V and +2.5V (instead of 0V and 5V) accomplishes the same thing but without cutting the output in half, and (more importantly) without the added noise of the -2.5V supply.
analog_sa said:
A capacitor will remove the DC offset, however it will still halve the voltage swing if fed to two single ended amplifiers. I also don't want to use capacitors in the signal path. For the same reasons as I don't want to use op amps or transformers.
Cauhtemoc said:
It will not halve anything. If you don't like capacitors for sonic reasons that's fine. But to imagine they will limit the swing to positive only is hilarious.
Spend five minutes thinking about it. Consider how capacitor-coupled preamps can possibly work with bipolar-supplied power amps.
- Status
- Not open for further replies.