Is there any real advantage to using single opamp in the DAC output stage?
Most of the equipment i see uses dual opamp in the output stage, would this be just to save cost and board size?
Most of the equipment i see uses dual opamp in the output stage, would this be just to save cost and board size?
Duals come in more variety, and are cheaper per op amp and pcb.
Singles can have better channel separation and layout.
Some high performance op amps are only available as single.
Singles can have better channel separation and layout.
Some high performance op amps are only available as single.
Single opamps often have compensation pins brought out, duals can't as there aren't sufficient pins on the package. So using a single may allow the opamp compensation to be optimized to a particular application.
AD811 is an example of this - to my ears its the best opamp for I/V.
AD811 is an example of this - to my ears its the best opamp for I/V.
Sometimes PCB layout is simpler with single OP.
No other advantages seen. Channel separation is not an issue, it is more than enough with dual ones.
Alex.
According what the dac output is : current or voltage, single or dual channel, single channel (two dac chips) or differential output in the more modern ic, either single or dual OPA could be better from a layout point of view.
Also further OPA after the first one, can have a role as well : filtering, buffer, ... some OPA have buffer strong enough already, etc.
Then/so, according your needs/requirements, one OPA migth not be enough or not. The load may dictate some choice for some.
I am not sure a CFB like the AD811 is the best choice according the DAC ic. What is the need I/V or pehaps just buffering voltage output dac ic? Quite different choices for the OPA choice, imhe.
just my 2 cents
Also further OPA after the first one, can have a role as well : filtering, buffer, ... some OPA have buffer strong enough already, etc.
Then/so, according your needs/requirements, one OPA migth not be enough or not. The load may dictate some choice for some.
I am not sure a CFB like the AD811 is the best choice according the DAC ic. What is the need I/V or pehaps just buffering voltage output dac ic? Quite different choices for the OPA choice, imhe.
just my 2 cents
Last edited:
Channel separation in the LM4562 dual opamp is 118 dB (@ 1 kHz). It's better than 150 dB (1 kHz, 5 kΩ load) and better than 120 dB (1 kHz, 600 Ω load) in the OPA1612. So while you're correct in theory, channel separation is not really much of a practical concern.
I prefer the layout of the dual opamps, in particular that the output is right next to IN-, which makes the feedback network nice and tight. The single opamps allow for a nice flow left-to-right, so line the opamps up horizontally (assuming pin 1 is in the upper left). The dual opamps allow for a good layout end-to-end, so align them vertically, one above the other (again assuming pin 1 in the upper left). That dual layout also has the advantage of allowing for easy routing of the power supply, which isn't as easy with the single opamps.
I don't really have an issue with single vs dual, but in audio I'll always use the dual because I always need one per channel and single opamps provide no tangible benefit to me.
Some high-performance opamps are only available as dual, especially for audio opamps. Take the LM4562/LME49720 for example. There used to be a single version, the LME49710, but TI killed it. But, yes. There are some that are only available in single. OPA627 springs to mind.
Tom
Can you explain why this opamp is good for stage I/V?
I've seen this opamp in medical equipment, but never in audio.
One reason why its good is its a current-feedback type. Meaning that the -ve input pin is at a low impedance independent of the negative feedback loop. A typical VFB opamp will experience instantaneous overload of its input stage when the DAC's output changes, before the feedback loop gets to correct it. A second reason is its a very low noise device and further to that it also has a wide bandwidth with low THD and a very high slew rate.
If you'd like a more in-depth technical treatment, try this paper by Walt Jung : https://cfas.waltjung.org/High_Performance_Audio_Stages_Using_TransZ_Amps.pdf. I shared another relevant paper recently over here : https://www.diyaudio.com/community/...apacitor-which-frequency.414555/#post-7723805
If you'd like a more in-depth technical treatment, try this paper by Walt Jung : https://cfas.waltjung.org/High_Performance_Audio_Stages_Using_TransZ_Amps.pdf. I shared another relevant paper recently over here : https://www.diyaudio.com/community/...apacitor-which-frequency.414555/#post-7723805
... but settling time so so and a little high input current noise at the inverting input VS a fet based one...
The low input Z migth be the only real advantage but need a big resistor at input >500 ohms
There are now OPA VFB that acheives better spec but the cool low Z input.
Some too use Transconductance OPA used in transimpedance conf (OPA860/861) as well.
The low input Z migth be the only real advantage but need a big resistor at input >500 ohms
There are now OPA VFB that acheives better spec but the cool low Z input.
Some too use Transconductance OPA used in transimpedance conf (OPA860/861) as well.
What's 'so so' about 65nS to 0.01% settling time? I agree that the -ve input current noise is the biggest drawback of CFB opamps, they seem all to have rather a high number. Poor SNR is the CFB's achilles heel.
The AD811 is good for I/V and was very popular 10-20 years ago. It is fast, but since it is the so-called "TOC", there are some nuances in their use. For example, it is very problematic to put a capacitor in its feedback in order to obtain a low-pass filter.
But now we have OPA1656, OPA1612, OPA1632, and other excellent OPs for I/V.
Not so many opamps designed "for audio" are truly good. In most cases, industrial operational amplifiers are better, and many “for audio”, on the contrary, are terrible (for example, anything with “Muses” written on it, is a nightmare, IMHO, of course 🙂).
And not just operational amplifiers, this is also true for all other components (capacitors, resistors, etc.)
As my colleague said many years ago: “for audio” they write on something, that is not suitable for anything else".
That is, for something serious it cannot be used, but for audio is possible.
Last edited:
I assumed the settling time was very important for that conversion task with small signal (more than SR). I choose OPA with 2 to 4 nS at 0.1 and same at 0.01%... and SR good enough 60 to sometimes more than the AD811 (200 SR) and with lower noise. Fets input sometimes may have advantages too according the dac chip (for instance a clear on the Sun shiny side Analog Device dac chip may ptrofit VS a smoother Philips dac ic à la TDA... I don't know if those fets (mosfets as well despite noiser) play a role or not at filtering dacs output "overshhots")
But don't get me wrong I have not said it sounds bad, just it is not the alpha and omega of I/V conversation.
I surmise the good results rather depends more about the WJ regs than the OPA here. But what about a no feedback reg (aka stabiliser) with a very good PSRR and CMR) ? as far the impedance is a lowthan WJ's reg. ?
Also I wonder indeed about that resistor choice value and topology before the inverting input... should be smd, no ? MELF certainly a bad choice cause added inductance VS a flat thin film (maybe marginal) but I really wonder despite being noiser how subjectivly could give a carbon resistor here or even carbon comp. (Vishat at least has good carbon MELF, so sall form factor)
And of course the same schematic with the LME49713HA (anyone have two for me ? 🙂 ) which is said uch better than the AD811 on which it was based.
I hope it is not blah, ahahah ! (but usefull inputs for some as well for the cause) 🙂
Last edited:
Which would that be? Seems likely that such a tremendously fast opamp would have a poor noise performance in the audio band (high LF noise corner) as it'd be designed for RF.
I google to find them, there are a lot of OPAs.... For illustration (I use that one has a cheap alternative to AD797 for buffer in my dac (not for I/V btw) as I am sure I am unable to hear the difference between their noise floor : AD8597 -single channel- sota PSRR and cMR ; curious to have your input and advices after you read the datasheet btw)
https://www.diyaudio.com/community/threads/opamp-single-or-dual-dac-output-stage.414520/post-7725836 one among others or have I missread the datasheet ? And why this "us" Abraxalito ?
That link not working for me, just takes me back to this same page. I used 'us' because I assumed I wouldn't be the only one interested in such a device. If you just give the part number I'll examine the DS.
Oh did you mean to indicate AD8597 was the 2-4nS opamp you'd found? If so then yes, you did misread '2uS' as '2nS'. And the slew-rate is 14V/uS. But for sure, its lower noise than AD811.
Oh did you mean to indicate AD8597 was the 2-4nS opamp you'd found? If so then yes, you did misread '2uS' as '2nS'. And the slew-rate is 14V/uS. But for sure, its lower noise than AD811.
Last edited:
Is that one is working https://www.analog.com/en/products/ad8597.html ? If not I upload here the datasheet.
I have given the part number in the post above already. I use it as buffer as the SR is maybe a little weak but some claims SR is notthe first item on the list for I/V. There are a lot of equally good OPA but higher SR that that ones...
I have given the part number in the post above already. I use it as buffer as the SR is maybe a little weak but some claims SR is notthe first item on the list for I/V. There are a lot of equally good OPA but higher SR that that ones...
Ah I see alzheimer on me ! Indeed 2 uS = to 2000 nS ! Not the same. AD811 is 65 nS !
aH I have checked, my I/V OP is 16 nV at 0.01% with big SR and bif bandwidth so this one chose for the task should be good and the AD8597 good as unity gain buffer 🙂
for the populus : Ad
(4 layers, yes it sounds good, no sound description blah sorry) : (and I have posted gerbers of a 2 layers version of a miro1360 modified version, cause attacks of some big mouths that pointed me as "us" in a to quoque filli mi attitude) 🙂
aH I have checked, my I/V OP is 16 nV at 0.01% with big SR and bif bandwidth so this one chose for the task should be good and the AD8597 good as unity gain buffer 🙂
for the populus : Ad
(4 layers, yes it sounds good, no sound description blah sorry) : (and I have posted gerbers of a 2 layers version of a miro1360 modified version, cause attacks of some big mouths that pointed me as "us" in a to quoque filli mi attitude) 🙂Attachments
Last edited:
what about 7nsec.. I know I know it does not have the dynamic range, but otherwise no slouch.. OPA856.
It can do ~1,1Vrms.. After this truly impressive I/V converter, a 6dB/10dB output stage would beef it up to standard level. Filtering could be applied before the output stage..