They can't even 0.1% , did you guys measure your whole system THD at 2meters? at 100 db? it is probably very high in high order THD of all kinds....
and it doesn't stops there, phase delay, IMD, and of course resonances and standing waves + room modes, floor bounce, nuls, etc.
This doesn't makes finding a good opamp futile, actually, how I found the ad8597 is I was using a V-out dacs and wanted the best dac output opamp, so I had a few to test and used the ad8620, this is a solid contender in the opamp list. However I liked it more with a single opamp and the ad8620 was OK as driving a power amp, but this is where I found the ad8599 or 8597 which was able to do both.
LM4562 oscillating, unstable, are you joking????
Please use the recommended circuit, quality feedback and bias components, please bias class A. Use a regulated dual +- regulated to at least 80 PSRR and balance both + and - sides with trimmers.
If anyone is interested I sell a modified prodigy hd2 for PC, it runs LM4562 in differential mode and I bypassed the output buffer opamp and send the signal to the RCA L and R outputs, you just need a RCA LR to headphone adapter and connect your headphones, it is a PCI card, my new computer has no PCI slots.
AUDIOTRAK PRODIGY HD2 BLACK EDITION PCI Sound Card 2ch Digital Audio 190755000025 | eBay Note my version is the whide with the Oscon cap, and transistor upgrade, a rare model with caps upgrades etc, it is better than the deluxe version.
I sell it with the 2 LMs and the mod, which is reversible and you can plug the buffer opamp, but i don't recommend that.
Last edited:
I like the idea of using a cohort of non-harmonically related tones for an IMD measurement - a dozen tones will create 100's or 1000's of IMD products and their total power is surely a better guide to the degradation of a real music signal than any single tone or two tone measurement. For instance crossover distortion will create very many more IM products than 2nd harmonic distortion.
It has lower common-mode input distortion, so as a follower for one. Also, it measures better as a transimpedance amplifier it would seem based on various DAC app notes.
The AD812 is a dual version (sort of) of the AD811 current feedback opamp. I've used about 1000 AD811 as video amps in commercial video gear. While they are really good as video amps, if you haven't used current feedback amps before, they have some quirks like the opamp bandwidth is set by the feedback resistor which means you cannot set gain by altering that resistor. If you want to run a unity gain buffer you MUST have the feedback resistor unless you WANT an oscillator. Never tie the '-' input directly to the output. Using them as part of a composite amp like Walt Jung did successfully is another thing.
G²
And video amps in general don't have great specs for audio, they have great specs for video.
The PSRR, CMRR, distortion, bias current, offsets will all be pretty lacklustre as the amp is designed to slew ultra-fast above all else. Current noise performance is pretty much irrelevant too for video as the impedance is 75 ohms everywhere. Input current distortion may bite you too as the bias currents are large - basically use such devices at low impedance only and don't expect great things other than bandwidth. PCB layout will be critical for the feedback network too.
The PSRR, CMRR, distortion, bias current, offsets will all be pretty lacklustre as the amp is designed to slew ultra-fast above all else. Current noise performance is pretty much irrelevant too for video as the impedance is 75 ohms everywhere. Input current distortion may bite you too as the bias currents are large - basically use such devices at low impedance only and don't expect great things other than bandwidth. PCB layout will be critical for the feedback network too.
But how will you separate that from actual noise? The nice thing about specific harmonics that you know where they should be and where they should not be is that you can use synchronous averaging and tweezer them accurately away from the system noise. With a whole pile of harmonics, you're limited by the noise floor, which is not a limitation for our ears either.
There's a trick with using CFB opamps which is set out in a TI application note, might be interesting for some designers. I've used a ferrite bead to good effect with AD8017 and AD815 : http://www.ti.com/lit/an/slyt081/slyt081.pdf
Sum all the power in all the peaks that aren't the test signal, gives the total signal + distortion power. Same as usual, but more peaks. Yes the higher order products will drop down into the noise, but contribute so little to the total distortion power you can ignore. Longer FFT will pick up more peaks in the noise if needed. You'll need a large FFT anyway to distinguish the many peaks.
What's the best dual I/V opamp these days: AD8066, LM6172, AD812 ???
Hello All,
I'm reactivating this thread...
I've spend quite a lot of time trying to read different post to make my mind on that matter, in my application.
So first, the context.
DAC Chip is TDA1545A with Vref 1V56 => Iout Max 2mA
Data are upsampled to 176.4KHz.
The circuit is single supply with a "Virtual Ground" Vbias of 3V3
I use low feedback R of 560 Ohm.
Output capacitor is Elna Silmic.
I have build a circuit with the AD8066, but I find that the result is a little on the bright side. Bass is good, sound stage also, good positionning, slam, but a little too bright.
I'm using a butterworth Low pass filter with a OPA1612, however the cutoff frequency is too high (around 50KHz), is it the cause of the bright sound ?
None of the opamp seem to oscillate although I have no way to certify it, other than to say that they are running cool.
I'm using separate fast regulator LP298x serie for each chip close to them.
I will first change the filter to a Bessel, fc 23KHz, do you think it will suffice to warm up the sound (low medium) ?
Capacitor types are SMD PPS.
Second, I want to try another IV chip
The two other candidates are LM6172 and AD812.
On sim, the AD812 is the best of all.
But I wonder if it will not be even brighter ???
Your advices are welcome.
Jean
Hello All,
I'm reactivating this thread...
I've spend quite a lot of time trying to read different post to make my mind on that matter, in my application.
So first, the context.
DAC Chip is TDA1545A with Vref 1V56 => Iout Max 2mA
Data are upsampled to 176.4KHz.
The circuit is single supply with a "Virtual Ground" Vbias of 3V3
I use low feedback R of 560 Ohm.
Output capacitor is Elna Silmic.
I have build a circuit with the AD8066, but I find that the result is a little on the bright side. Bass is good, sound stage also, good positionning, slam, but a little too bright.
I'm using a butterworth Low pass filter with a OPA1612, however the cutoff frequency is too high (around 50KHz), is it the cause of the bright sound ?
None of the opamp seem to oscillate although I have no way to certify it, other than to say that they are running cool.
I'm using separate fast regulator LP298x serie for each chip close to them.
An externally hosted image should be here but it was not working when we last tested it.
I will first change the filter to a Bessel, fc 23KHz, do you think it will suffice to warm up the sound (low medium) ?
Capacitor types are SMD PPS.
Second, I want to try another IV chip
The two other candidates are LM6172 and AD812.
On sim, the AD812 is the best of all.
But I wonder if it will not be even brighter ???
Your advices are welcome.
Jean
If you're going to use lm6172(or anything else) try this Technics(SP-L1000) based variation and you won't be dissapointed, as for the thin sound when it comes to proper bass reproduction , i suspect that more chips than just lm6172 need mild bass boost which is easy to do and you can see it approached there too.This circuit simulated less than -150db in odd harmonics...and less than -120db in even harmonics...
https://www.diyaudio.com/forums/digital-source/338152-nec-cd-810-revival-mods.html#post5809151
https://www.diyaudio.com/forums/digital-source/338152-nec-cd-810-revival-mods.html#post5809151
Your set-up is quite similar to how mine was when I started playing with multibit DACs. I'm still using a very close relative of TDA1545A, TDA1387. I ended up ditching TDA1545 as it gave some weird effects on piano music, especially at higher bias currents. You might try switching to TDA1387 even though it can only go up to 1.2mA (at 6V).
I also found brightness when connecting an opamp direct to the output of a CMOS multibit DAC. I suspected it was caused by glitches upsetting the opamp. I was able to tame it at first with a long string of ferrite beads between DAC and opamp. Eventually I designed a 'proper' CLC filter placed between DAC and opamp - you can find an example in this schematic : lingDAC - cost effective RBCD multibit DAC design
Without the filter I found I/V opamps vary in how much brightness they generate. AD811 was the best but note its CFB. I doubt brightness is connected to your downstream filter, the damage is done at the I/V stage.
looks like ad8066 having very low capacitance driving capabilities...so it's not oscillating, but it's settling too hard...thus it might be useful to use the ad797 datasheet for inspiration(fig 40...43) :
https://www.analog.com/media/en/technical-documentation/data-sheets/AD797.pdf
ad8066 also use the same input Butler architecture of OP275 which was found to be a bit messy on harmonics ...not sure if it's greatly improved in the ad8066.
Lots of designers ignore the 51...100 ohms series resistor with the integrating capacitor...You can also add a 100...330pf capacitor at the input.Technics variation is naturally solving this problem.
https://www.analog.com/media/en/technical-documentation/data-sheets/AD797.pdf
ad8066 also use the same input Butler architecture of OP275 which was found to be a bit messy on harmonics ...not sure if it's greatly improved in the ad8066.
Lots of designers ignore the 51...100 ohms series resistor with the integrating capacitor...You can also add a 100...330pf capacitor at the input.Technics variation is naturally solving this problem.
Last edited:
Thank you to you 2
Dreamth, I like your approach, however, I want to keep my design as simple and cheap and small (footprint) as possible.
It's a Raspberry Pi Hat that includes clock and full galvanic isolation with a virtual battery for the Pi.
Also, my board is already designed and your suggestion means too many modifications to my layout.
However, I have to try this one day...
EDIT to your AD797 suggestion:
I had thought about it although, finally didn't made it, I should try it as per fig 40.
Right now my C feedback is SMD PPS 100pF, i could just add a 100 Ohm resistance, it is simple enough
"You can also add a 100...330pf capacitor at the input."
Do you mean, like Cs//Rs in fig 40 of AD797 ?
Abraxalito, from what you say, I shoud try the AD812 (cheap dual CFB similar to AD811).
I initialy designed the board with AD8066 because of the jFET input stage said to be a better protection against RFI.
Moving to the AD812 from the AD8066 is easy enough to try.
I will definitly try to use a ferrite as suggested.
I may also consider to replace the OPA1612 in the Sallen Key filter by an OTA buffer for better IMD although more noisy and with higher distorsion.
What do you think ?
Jean
Dreamth, I like your approach, however, I want to keep my design as simple and cheap and small (footprint) as possible.
It's a Raspberry Pi Hat that includes clock and full galvanic isolation with a virtual battery for the Pi.
Also, my board is already designed and your suggestion means too many modifications to my layout.
However, I have to try this one day...
EDIT to your AD797 suggestion:
I had thought about it although, finally didn't made it, I should try it as per fig 40.
Right now my C feedback is SMD PPS 100pF, i could just add a 100 Ohm resistance, it is simple enough
"You can also add a 100...330pf capacitor at the input."
Do you mean, like Cs//Rs in fig 40 of AD797 ?
Abraxalito, from what you say, I shoud try the AD812 (cheap dual CFB similar to AD811).
I initialy designed the board with AD8066 because of the jFET input stage said to be a better protection against RFI.
Moving to the AD812 from the AD8066 is easy enough to try.
I will definitly try to use a ferrite as suggested.
I may also consider to replace the OPA1612 in the Sallen Key filter by an OTA buffer for better IMD although more noisy and with higher distorsion.
What do you think ?
Jean
Last edited:
ad812 not great at driving capacitance either... its slew rate goes down by 10x with very small capacitance so you need Rs fig 41...43:
https://www.analog.com/media/en/technical-documentation/data-sheets/AD812.pdf
https://www.analog.com/media/en/technical-documentation/data-sheets/AD812.pdf