I guess ALL the DAC manufacturers are stupid because they ALL recomend OpAmps as I/V (with virtually zero impendance on DAC outputs). Their engineers are smart enough to design a whole DAC, but they cannot figure that a resistor is a better choice for I/V? Or is a world-wide conspiration?
If resistor was sounding better, it would be in the datasheet. I guess the "do-it-on-the-knee" companies and users on the net are the true specialists...
Last edited:
In this imperfect audio world, we have to accept compromises. Our decisions should be based on knowledge. I cannot see anything wrong with passive approach or opamp approach.
When the DAC has low dynamic range, of course passive I/V tends to give trouble, but not so if we take that into account in designing from source to speaker.
One way to improve dynamic is by using opamp instead of resistors. But opamps have their own problems. But may be not if the opamp has high slew rate and a good power supply. I prefer passive approach than using less than 10V/us opamp (It was even very difficult to beat OPA2134 (20V/us) with passive I/V). So I think I will not go the passive route as I have the money to buy parts and have sufficient knowledge, skill and ears to build a good sounding system.
So again, it is all about implementation. If money is not a problem (to buy better parts), imo it is always easier (and I think better) to go with active. And hey, you can choose from the many discrete opamp designs out there.
Successful passive implementation requires careful attention and matching from start to end of the audio chain
N.B. I wanted to edit the SNR with THD+N in previous post but for a reason I don't have an edit button so
You basically just answered your own question. Sound improvements can also be made that are unrelated to the DAC chip sets. Take the analog output stage for example. I have a Denon 2900 with a new ZAPfilter installed. It sounds way better than stock, yet the DAC chip set has not been altered.
That would make sense if the resistor I/V would be MORE EXPENSIVE than an OpAmp one. Since the cost is superior, obviously, there was no "compromise" when manufacturers selected OpAmps as the required output stage.
Good OpAmps today have more than 20V/uS and THD+N at -110dB. They are way better than the OpAmps used in '80's DAC's. What was "high-performance, audiophile quality" in 1985 (for example LM833), is "low-cost, general use" now.
Some people didn't get the memo yet.
Last edited:
A decision to use passive I/V is like a decision to use the shortest signal path, single ended design then to work out the related issues like putting massive inductors in power rail, etc. The external work may be more expensive (and time consuming) than the I/V itself.
If I/V is used but there is opamp in the signal chain, then that would be against the philosophy.
All in all, I'm against the passive I/V myself. I believe that those with good ears will know why.
Sure, even passive I/V can be followed with discrete buffer. There are many ways to make a DAC to sound good, even with a cheap DAC like TDA1453.
That's why I stated that DNR is one important parameter that defines a signature of a DAC. Because no matter what you do with the auxiliary circuits, it will never sound like what it is intrinsically incapable of. It doesn't mean that high DNR DAC design will always preferable soundwise than low DNR DAC. It depends on the design. But "dynamics" is audible, and it is a positive characteristics. Have a good listen.
Thanks Jay. I am having a very good listen. The clarity and dynamic slam that the ZAPfilter brings to the table is very addictive like a drug habit that can't be kicked. I think differences in DAC chip set are generally overrated and the analog output stage gets generally overlooked.
Yes, and, minimizing the virtual ground impedance is the primary reason they recommend an op-amp based i/v. However, the practical benefit of a near zero impedance virtual ground depends upon the specific D/A converter utilized.
Let me ask you a simple thought question. Suppose the current output of some given converter is not audibly or measurable degraded (although it may be theoretically degraded) by a non-zero virtual ground of, let's arbitrarily say, 50 ohms? Which objective technical parameters of that 50 ohm passive resistor i/v would not be superior to those of any op-amp based i/v implementation? Please list such parameters, i.e., THD, IMD, noise, slewing limit, PSRR, bandwidth, settling time, phase margin, etc.. I'll wait...
Last edited:
Problem is... there is not such thing. Current sources inside DAC are derived from switched capacitors at low voltage. They are not ideal current sources. Not at the level of 20-22 bit resolution.
Also, resistor voltage noise can be higher that the one of the OpAmp. Settling, phase, slew and bandwith (they are all the same thing) can be worse due to capacitance following it (even buffers have some parasitic capacitances). Also, due to the fact that you don't have any proper filters following that I/V (because OpAmps are devil's work) or OS (because that is also complicated and it's just another conspiration) - the resulting DAC's trully suck in the THD department.
Do you really think that the engineers are morons (from all the Europe, USA and Japan) and didn't take ALL those aspects in consideration? That any DIY without any formal training know more than them?
The manufacturers would LOVE to sell a product that is cheaper to incorporate in designs, especially if it is better that the more expensive alternative. Any of the DAC manufacturers would at lest mention that as a posibility in their data sheets if it was even remotelly fesable.
The reality is that they DID TRY it in their labs, mesure them (with tools that probably cost as much as the house you are living in), listen to them (yes they are doing that too) and the results where less than the alternative that you see in data sheets.
Resistive I/V is just kids toy. Same like when I was kid and would make "radios" from a potato (variable capacitor) and germanium diode or shaving blade. Looks cool, but still a toy.
Also, resistor voltage noise can be higher that the one of the OpAmp. Settling, phase, slew and bandwith (they are all the same thing) can be worse due to capacitance following it (even buffers have some parasitic capacitances). Also, due to the fact that you don't have any proper filters following that I/V (because OpAmps are devil's work) or OS (because that is also complicated and it's just another conspiration) - the resulting DAC's trully suck in the THD department.
Do you really think that the engineers are morons (from all the Europe, USA and Japan) and didn't take ALL those aspects in consideration? That any DIY without any formal training know more than them?
The manufacturers would LOVE to sell a product that is cheaper to incorporate in designs, especially if it is better that the more expensive alternative. Any of the DAC manufacturers would at lest mention that as a posibility in their data sheets if it was even remotelly fesable.
The reality is that they DID TRY it in their labs, mesure them (with tools that probably cost as much as the house you are living in), listen to them (yes they are doing that too) and the results where less than the alternative that you see in data sheets.
Resistive I/V is just kids toy. Same like when I was kid and would make "radios" from a potato (variable capacitor) and germanium diode or shaving blade. Looks cool, but still a toy.
Last edited:
I can tell you that at least in the Philips lab the red pencil was put to good use AFTER the engineers had done their job...
A long time ago I tried to find out some design details and talked to one of the last engineers from the CD design team and was astonished to hear that the bean counter dept cut so many good design ideas just because they would save some cents here and there.
Many design aspects were just implemented to satisfy consumer expectations, SQ was not the most important one.
Not arguing against what you are saying but what engineers know is best is def. not always what is sold in the end
A long time ago I tried to find out some design details and talked to one of the last engineers from the CD design team and was astonished to hear that the bean counter dept cut so many good design ideas just because they would save some cents here and there.
Many design aspects were just implemented to satisfy consumer expectations, SQ was not the most important one.
Not arguing against what you are saying but what engineers know is best is def. not always what is sold in the end
The op-amp i/v will never have lower noise simply becuase it has an resistor in it's feedback loop. Which means an op-amp based i/v will always have greater noise than an equivalent passive resistor i/v.
The issue, once again, isn't the technical competance of the engineers at T.I., or Analog Devices, or whomever. The issue is that all current type DAC chip outputs, theoretically perform better into a near zero impedance virtual ground. The simpliest way that I know to implement a minmal impedance virtual ground is by feedback amplificstion. The lowest cost high open loop gain feedback amplifers are op-amps. So, it's rather simple to understand why engineers of large microchip vendors recommend op-amp based i/v circuits. It seems rather clear that the primary design objectives for commodity (op-amp) based i/v circuit are a minimal virtual ground impedance, and lowest static THD.
In addition, you assume that those DAC vendor engineers have put much thought into the external i/v circuit. I can you tell you from my own 10 year experience at a major semiconductor vendor that such often wouldn't be the case. The engineers involved are mostly concerned with the performance and customer support of the device the company is trying to sell, and in getting that device out to market as quickly as possible. Those two realities very often don't lend themselves to great expenditures of thought and energy on optimizing external circuitry, particularly, over the subjective aspects of audio performance. This all has NOTHING to do with their technical competence, so, you can stop attempting to conflate technical competance with subjective listening results.
Last edited:
I know it is such an old but brilliant and short thread, but I am really curious if any of this thread participants can add anything to the facts posted ? Have they got any extra facts, details, personal experience ?
It looks like it was mainly discussion of I/V stage and not the dac chips , and specificly opamp based vs passive one.
Pardon me if I am wrong, but it is worth getting back to.
Any comments ( or all been said) ?
It looks like it was mainly discussion of I/V stage and not the dac chips , and specificly opamp based vs passive one.
Pardon me if I am wrong, but it is worth getting back to.
Any comments ( or all been said) ?
I haven't chipped in up until now, but here's my 2c.
I used to take the view, rather like SoNic that THD and SNR told us pretty much all we needed to know about a DAC's performance. If you come at DAC design with this assumption then opamps and presenting the lowest impedance to an I-out DAC are uppermost in your mind.
After I began playing with DACs and listening to how they sound I did find there was a difference between the various types. And when I played around with opamps in I/V I found there was a definite improvement (more clarity, better soundstage) in having a faster opamp and in a lower value of feedback capacitor. Neither changed the THD to a significant extent from what I recall.
On my listening journey I eventually abandoned opamps and S-D DACs altogether and nowadays I only design with passive I/V and multibit DACs as subjectively I very much prefer the sound. Use of passive I/V allows a steep passive filter to be employed between the DAC and subsequent amplifier stage. In the past few months I've been engrossed in passive filter design and inductor selection.
My hypothesis for why passive I/V and passive filtering delivers tbe best sound is that, rather than THD, its IMD we need to minimize for the best subjective effect. IMD measured with a very large number (>100) of tones, not the traditional 2 tone measures.
I used to take the view, rather like SoNic that THD and SNR told us pretty much all we needed to know about a DAC's performance. If you come at DAC design with this assumption then opamps and presenting the lowest impedance to an I-out DAC are uppermost in your mind.
After I began playing with DACs and listening to how they sound I did find there was a difference between the various types. And when I played around with opamps in I/V I found there was a definite improvement (more clarity, better soundstage) in having a faster opamp and in a lower value of feedback capacitor. Neither changed the THD to a significant extent from what I recall.
On my listening journey I eventually abandoned opamps and S-D DACs altogether and nowadays I only design with passive I/V and multibit DACs as subjectively I very much prefer the sound. Use of passive I/V allows a steep passive filter to be employed between the DAC and subsequent amplifier stage. In the past few months I've been engrossed in passive filter design and inductor selection.
My hypothesis for why passive I/V and passive filtering delivers tbe best sound is that, rather than THD, its IMD we need to minimize for the best subjective effect. IMD measured with a very large number (>100) of tones, not the traditional 2 tone measures.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.