wow ! And seems also abble to drive headphones.
C4 & C6 with a foilded styren cap (or a silver mica) or a less good 5 mm pitch MKP (avoiding XR7) should be usefull to preserve the clarity of the previous stage.
And for the same purpose : C4+R10 / C7+R18 could be removed if one has an input cap on the pre or amp... (but headphone?)
Btw, just has been looking at your Lingdac shematic. Is this AD8017arz a sort of AD844/OPA860(/1) transconductance IC ? Or why do you prefer it over these last ones to stay in the thread topic (whether an embeded feedbackloop in it?)
Anyway, many thanks for that.
C4 & C6 with a foilded styren cap (or a silver mica) or a less good 5 mm pitch MKP (avoiding XR7) should be usefull to preserve the clarity of the previous stage.
And for the same purpose : C4+R10 / C7+R18 could be removed if one has an input cap on the pre or amp... (but headphone?)
Btw, just has been looking at your Lingdac shematic. Is this AD8017arz a sort of AD844/OPA860(/1) transconductance IC ? Or why do you prefer it over these last ones to stay in the thread topic (whether an embeded feedbackloop in it?)
Anyway, many thanks for that.
Headphones can be driven yes - just need to check that the bias current is set (via R8 & R16) in accord with the impedance of the 'phones. Also since its classA there's a fair amount of power being dissipated in Q1/Q2/Q4/Q5 - the PCB layout (found on the lingDAC thread) has areas of copper on both sides to act as heat spreaders. One point I just noticed - the PSU is shown as +9V but in fact this buffer only needs 7V to swing 2VRMS at the output, so to keep the FETs as cool as possible its best to run only as high on the rail as needed to swing the full signal.
C4 and C6 are NP0 types, not X7R so no need to swap them out. NP0 are quite blameless as caps. Yes you could remove output coupling caps if you know the next stage has them - definitely do not remove when feeding headphones.
AD8017 is a current feedback amp with high output drive capability. I chose it mainly because its very cheap on Taobao, excellent bang-for-the-buck. AD811 performs slightly better SQ-wise but is more difficult to employ as its output doesn't swing as close to the rails, demanding more volts on the supply. It also runs very, very hot. Incidentally lingDAC only uses discretes in the OPS, its PhiDAC which uses the ADs.
C4 and C6 are NP0 types, not X7R so no need to swap them out. NP0 are quite blameless as caps. Yes you could remove output coupling caps if you know the next stage has them - definitely do not remove when feeding headphones.
AD8017 is a current feedback amp with high output drive capability. I chose it mainly because its very cheap on Taobao, excellent bang-for-the-buck. AD811 performs slightly better SQ-wise but is more difficult to employ as its output doesn't swing as close to the rails, demanding more volts on the supply. It also runs very, very hot. Incidentally lingDAC only uses discretes in the OPS, its PhiDAC which uses the ADs.
Last edited:
Hello George, could i use this schematic for the pcm1702 instead of 1704?
are all the component values ok? use 3x ad844 or only 1?
i ´m using lt1028 as i/v but i would like to try this alternative.
thks.
With good layouts I noticed NP0 in the final direct signal path gives subjectivly something not so neutral at ears (so in the listen frenquency range) : often too thin and a little brighty. Although it depends on the width of the traces on the pcb as well ! When I have choice I swap them for Panasonic PPS if SMD designs where the capicitance is notof an importance (serie signal)- Wima FKP2 (non metalized) are good too. or foilded styrens if vias (silver mica being a little too much brighty and thin as NP0 and not tin foilded styrens less good either). YMMV. I like AWG30 (or more little)width and short length output traces for some reasons before RCAs or any gears output stage (trying to have low mass serie contacts). As for the serie resistors I try to use cheap metalized Yageo which are very low noise or RD if not better RS Sussumu in the digital stage AND after.
Again YMMV, all being subjective at ears in the last details (but the Devil often in its for sure)
Again YMMV, all being subjective at ears in the last details (but the Devil often in its for sure)
Thanks for the swift responses!
There is so much info in this thread it has taken me ages to get my head round everything. I am using a TDA1541a and plan to use the Pedja Rogic DDNFS IV and probably feed that into OPA1641 op-amps. From what I could gather over the pages that unless you have OPA627 then the OPA1641 are the best for sensible amounts of money. Do I have that right and if so anybody have a rough schematic and or conditions for the 1641 buffer?
I will have to start matching transistors now!
Thanks
Ian
There is so much info in this thread it has taken me ages to get my head round everything. I am using a TDA1541a and plan to use the Pedja Rogic DDNFS IV and probably feed that into OPA1641 op-amps. From what I could gather over the pages that unless you have OPA627 then the OPA1641 are the best for sensible amounts of money. Do I have that right and if so anybody have a rough schematic and or conditions for the 1641 buffer?
I will have to start matching transistors now!
Thanks
Ian
So this DDNFS I/V needs sorting out the transistors ? What margin please ?
I have been asking about op amps for I/V task at johnc124 from TI in the opa1656 thread in the Vendors Bazar folder.
His answer : "For a DAC chip with a low output current, I would suggest op amps with FET type inputs, either JFETs or CMOS. Low output current would mean the transimpedance gain of the I/V circuit will be high, and the input current noise of the op amp will be a significant contributor. There's no shortage of candidates here, and the new OPA1656 is a great option, but you could also check out: OPA1642, OPA1678, OPA827, and the newer OPA828, or also OPA2810. If you just wanted to use a single 5V supply, the OPA1671 is good. " OPA1656: High-Performance CMOS Audio Op Amp
could be better than AD844 or OPA861 ?
I have been asking about op amps for I/V task at johnc124 from TI in the opa1656 thread in the Vendors Bazar folder.
His answer : "For a DAC chip with a low output current, I would suggest op amps with FET type inputs, either JFETs or CMOS. Low output current would mean the transimpedance gain of the I/V circuit will be high, and the input current noise of the op amp will be a significant contributor. There's no shortage of candidates here, and the new OPA1656 is a great option, but you could also check out: OPA1642, OPA1678, OPA827, and the newer OPA828, or also OPA2810. If you just wanted to use a single 5V supply, the OPA1671 is good. " OPA1656: High-Performance CMOS Audio Op Amp
could be better than AD844 or OPA861 ?
DDNF + Buffer
Hi Ian, I no longer use the OPA627 as a buffer for the DDNF. I switched to the Sparkos SS3601. I used a capacitor between the DDNF output and sized it for the lower impedance of the discrete op amp. I did have some good results with the OPA1641 and as I recall it was drop in. I remember you lose offset adjustment with the 1641. So for me it was the same schematic as the OPA627. Do check the datasheet pin outs. It was a long time since I looked at this. Transistors... There is a benefit to matching. I also bonded all of mine to a heatsink so as to eliminate excess drift. I used a heat sink adhesive for that.
Op amps for I/V
The point of the AD844 and OPA861 is to do it differently. Same with Pedja's DDNF circuit. The triple stack 844 and DDNF have capabilities beyond a fet input op amp doing I/V duties. Noise I have found has been a non-issue.
No feed back I/V
This is just an opinion.... Feedback works well with steady state signals. Test signals and the like. Music is different and it is my feeling that it creates a smear in the time domain. I have a Nelson Pass BOSOZ preamp (Balanced). What it does well is to present depth and imaging. I have the ability to buffer with an OPA1632 (balanced op amp) stage and the imaging collapses when that is applied. So I use that OPA1632 stage to drive a subwoofer amplifier. My reference is a pair of Magnepan MG2.5R Planar ribbon speaker with a pair of open baffle subs. With DSP crossovers they image surprisingly well. Transistor matching... I matched for current gain hFE. My feeling was it was better to have even current gain in the mirrors. This is just a subjective opinion.
Hi,
The input stages almost constantly driven into overload.
Either one reduces the slewrate of the current steps and/or utilizes very fast video OPAmps.
Besides that, the input impedance of a OPAmp rises from quite low frequencies on, presenting the DAC-output a rather too high impedance at the DAC´s output clocking rate (~2.8MHz for TI´s PCM DACs iIrc).
A current conveyor witout global feedback, be as simple discrete grounded-base circuit, or in integrated form like the AD844 or OP861, is ´fast´ enough and doesn´t suffer from the feedback related distortion.
Instead of current conveyor one might think of current-buffer as term for simple grounded-base circuits.
jauu
Calvin
Hawksford has shown in a meanwhile quite old paper that the steps the DAC outputs are capable of are way too fast for the typical audio-related OPAmps.
The input stages almost constantly driven into overload.
Either one reduces the slewrate of the current steps and/or utilizes very fast video OPAmps.
Besides that, the input impedance of a OPAmp rises from quite low frequencies on, presenting the DAC-output a rather too high impedance at the DAC´s output clocking rate (~2.8MHz for TI´s PCM DACs iIrc).
A current conveyor witout global feedback, be as simple discrete grounded-base circuit, or in integrated form like the AD844 or OP861, is ´fast´ enough and doesn´t suffer from the feedback related distortion.
Instead of current conveyor one might think of current-buffer as term for simple grounded-base circuits.
jauu
Calvin
I've come to the conclusion (based on listening) that its best to employ both of these measures - bandwidth limiting prior to the opamp and use a wide bandwidth video opamp. Subjectively AD811 works very well but it does run rather warm with its ~16mA quiescent current. There are other video opamps with otherwise similar specs that I'm currently exploring which operate on a much lower current, EL5165 is one that I'm looking at.
@Torchwood421 - I conjecture that the subjective problem isn't with feedback itself but other implementation aspects - in particular with opamps its the classAB output stage's tendency to dirty up the power supply lines. Scrupulously clean (by which I mean very low impedance not merely very low noise) power is mandatory for best subjective results.
I guess that you are talking about 3 x bc547c/bc557c pairs schematic
what tolerance Is allowed to match npn /pnp, are you using dc servo at output
did you add a buffer to output
would you please explain your Impression from DDNF I/V soundwise comparing with other type of I/V, for example op amp I/V (OPA861 or 3 x AD844)
Is this your current setup
what tolerance Is allowed to match npn /pnp, are you using dc servo at output
did you add a buffer to output
would you please explain your Impression from DDNF I/V soundwise comparing with other type of I/V, for example op amp I/V (OPA861 or 3 x AD844)
Is this your current setup
Hi Ian, I run at + and - 15 on my builds. Having said that it will have less current in the mirrors. As long as your OK with that it is quite OK. I would not go lower then 15 VDC. I adapted the DDNF to work with a PCM1704 based dac with great results.
Time domain smearing
Hi abraxalito, AD811... Tried it as an I/V in the past. It works although I thought it didn't sound as good as other chips. Just a personal bias maybe. I'd have to agree with all of your comments. I am forced to use my ears as I lack the test equipment to really measure what I hear. Speaking on the OPA1632, I found I preferred the direct output of the Nelson Pass BOSOZ balanced pre. It imaged better and was quite holographic. Due to excessive loading I found I had no choice but to make a buffered output to drive the subs amplifier. So for that the OPA1632 was excellent. At bass frequencies the OPA1632 does not mess with the imaging. I run the open baffles to augment the bass and run the MG 2.5R speakers full range.
Transistor matching
Hi samoloko, I used a parts analyzer one of the ones you can find on Ebay. It can test transistors, capacitors and resistors etc. I had a lot of 50 of each transistor. I tested all of them and pulled the closest into a smaller pool of parts. Retested until I found a group that more or less matched. Tried to find similar hFE match between the PNP and NPN's. You will find some temperature effects so a small fan blowing across the work area might help to get a better match. The PNP's tend to have lower hFE. So that can be a challenge. I don't have a tolerance per se. It is all based on the parts you have on hand. Try to get all the parts from similar lots if you can. It might be best to stick with one brand as I found some old Philips transistors were quite low hFE then other brands. DC servo... I found that there is some drift with temperature on the DDNF. So you could apply a servo. My solution was to thermally bond all the transistor (in a channel) together. I glued a heatsink to the transistors with a heatsink rated epoxy. In my case I just used a capacitor to block the small amount of DC offset. Using a Sparkos discrete op amp as a buffer. If I remember correctly the cap was 1 uf and a 100K resistor. There is a limit with the Sparkos as it isn't a FET input op amp. You might stick with the OPA627's. Rather expensive today. OPA1641's are great on a budget. The servo might be the path forward as some don't like caps in the signal path.