Great - looks the same as my plot!
I designed this filter to do NOS droop correction - so it could be retro-fitted to any NOS DAC that's using active I/V (opamp or discrete) but I've since thought it could be a more generally applicable tweak to any DAC with current output where external opamps are in use (think AD1955 and PCM1794). For those applications the NOS boost won't be desirable - if anyone's interested I can tweak the filter to give a flat response.
I designed this filter to do NOS droop correction - so it could be retro-fitted to any NOS DAC that's using active I/V (opamp or discrete) but I've since thought it could be a more generally applicable tweak to any DAC with current output where external opamps are in use (think AD1955 and PCM1794). For those applications the NOS boost won't be desirable - if anyone's interested I can tweak the filter to give a flat response.
Last edited:
Random, possibly silly, thought: what if the NOS rolloff is actually what makes NOS appealing to (some) people? And if that is true, could that be approximated by creating a filter that deliberately starts rolling off the high frequencies around 16khz like NOS, but applied to sigma delta DACs (I.e. the exact opposite of what abraxalito's droop correction filter does)? Only suggesting this for curiosity's sake...
hi abraxalito,
My FS is always fixed at 192Khz. I need a (nearly) flat response until 96Khz, and follow by a sharp decline. Could you tweak the filter? Thanks.
PS. My IV is OPA861 which has 10 ohm input resistance.
Poting
Hi Poting
I can't recommend a filter which stays flat all the way to 96kHz given that lower bandwidth is going to reduce noise and is known to sound better. Here's a version which is flat to 20kHz (about 0.3dB down). Note there's 23R for the termination resistor - for 10ohm input impedance you'll need 13R between the filter and your I/V.
I can't recommend a filter which stays flat all the way to 96kHz given that lower bandwidth is going to reduce noise and is known to sound better. Here's a version which is flat to 20kHz (about 0.3dB down). Note there's 23R for the termination resistor - for 10ohm input impedance you'll need 13R between the filter and your I/V.
Attachments
Does the SIM account for the R (damping) of the inductor?
BTW is it air core or not, and if not, what is the suggested core? sendust?
I assume this is done on a PCB so there's no significant parasitics etc?
I would very interested in seeing how the circuit performs in a real application vs the SIM prediction. Are you planning to try?
Yes, I will try that. I am using certain ferrite bead at this moment. I wish to filter out more HF grabage before IV stage.
Poting
abraxalito,
Thank you for the help. I will build it and post my result here.
Poting
Yes, but it turns out not to be particularly critical to this.
Air core for 150uH is going to be fairly bulky but might work. It'll be sensitive to picking up all kinds of stray noise though so I'd not recommend it unless you're ready to shield it externally. The suggested core is gapped ferrite (as used by the SLF7045 range I've already mentioned).
Are you kidding me? Sendust is great for power supply chokes but too lossy here.
Parasitics will depend on layout, however its a very low impedance circuit so a few pF here or there won't be a concern.
I've been using the NOS droop version myself for over a week. What would you like to see? I haven't plotted its FR but its sounding great.
@canvas - I look forward to your listening feedback
Well, the caps don't sound too bad with their bypasses. I removed them and they test okay. Value measures as 20uF (marked 22u) with 0.3 ESR and 1.8% voltage loss. Rather than replace them, I shorted the holes, fed a little DC into the buffer op amp, and trimmed the DC output to zero. No cap sounds better
I have heard only three implementations of TDA1387, one of which is the filterless Pi hat you designed, Matt, and then Richard's first LingDAC 1.1 (which I believe uses the same filter being discussed) and lastly the Audiophonics board, which has a less precise NOS droop filter that can be seen here.
They all sound good to me, but I actually DO prefer the NOS rolloff for headphone listening, where I find it prevents fatigue. Over speakers, I think both correction filters sound more realistic than none. Of course there are other differences in the implementations to consider, but I have at least A/B'd LingDAC with and without the filter engaged, and found this to be true. Somebody with very bright speakers may disagree.
I really enjoyed the sound from the balanced AD8067 I/V followed by AD8065 3rd order filtering. It made music more exciting. But only at first, after a few days it gave me a slight sense of unease. My gf wasn't particularly impressed she said the HF wasn't as clean as from lingDAC (her hearing goes up higher than mine). I figured the difference might well be in the steepness of the filtering - lingDAC uses a much steeper filter than the single-stage MFB circuit I've posted up here.
So I decided to build an active version of lingDAC's steeper filter, using 12 * AD8065s (4 channels, 3 opamps per channel). This took rather longer than I imagined because I couldn't get it to stop being a 50MHz oscillator. Perhaps no surprise when using 100MHz+ opamps on prototyping board, no groundplane. In the end I managed to tame it with a little bit of help from some ferrite beads.
I've been listening for a couple of days now and while it still sounds slightly forward, its definitely more subdued. To my ears the HF is silky-smooth even on the most demanding material.
I changed my mind and ordered the AD1865 from Taobao a few days ago.
It arrived this afternoon and I hooked it up. It took an hour to settle in and then sounded very good - better than the other DAC from the same seller. It had similar "Ch" "K" but was more natural on "S" "Sh" and "th" female vocals. I left if to burn in... and during one of the periodic checks, I found the left channel had stopped working and the buffer op amp was extremely hot.
I will have to return it I guess. :-(
It arrived this afternoon and I hooked it up. It took an hour to settle in and then sounded very good - better than the other DAC from the same seller. It had similar "Ch" "K" but was more natural on "S" "Sh" and "th" female vocals. I left if to burn in... and during one of the periodic checks, I found the left channel had stopped working and the buffer op amp was extremely hot.
I will have to return it I guess. :-(
It will take a long time to reach the seller by China post (the 'China' side), unless you can ship it with SF Express. I'd just replace the faulty op amp. I'm assuming it's NE5534?
Poting
Hi,
I don't think it is the buffer op amp (OP42 - out of production). I let it cool and then switched on again - it works - and in fact it is the I/V op amp that heats up first. The AD1865 also gets quite warm. BTW, if anyone is curious, the logic gate in between the AK receiver and the AD DAC is 74HC7266A and I think the branding is the Fairchild "F".
It will be hassle to return and I will lose money but I'm not going to waste time fixing badly made gear from China. So, SF Express it is - thanks ;-) I wasn't that impressed with the other DAC from this seller, although that could be the nature of the IC itself with it's rather large amount of ultrasonic noise... and that's still a work in progress...it's good enough and quite satisfying to listen to but not dynamic or finely detailed or any of that. I have no experience with TDA1387 other than this one though so maybe it can be better. Keep trying...
Besides all that, for the money, I can get a decent ES9018k2m, which I prefer $ for $ to these NOS dacs. Still, I will try to find a better AD186x DAC because I really want to see what a good R2R DAC can do.
I don't think it is the buffer op amp (OP42 - out of production). I let it cool and then switched on again - it works - and in fact it is the I/V op amp that heats up first. The AD1865 also gets quite warm. BTW, if anyone is curious, the logic gate in between the AK receiver and the AD DAC is 74HC7266A and I think the branding is the Fairchild "F".
It will be hassle to return and I will lose money but I'm not going to waste time fixing badly made gear from China. So, SF Express it is - thanks ;-) I wasn't that impressed with the other DAC from this seller, although that could be the nature of the IC itself with it's rather large amount of ultrasonic noise... and that's still a work in progress...it's good enough and quite satisfying to listen to but not dynamic or finely detailed or any of that. I have no experience with TDA1387 other than this one though so maybe it can be better. Keep trying...
Besides all that, for the money, I can get a decent ES9018k2m, which I prefer $ for $ to these NOS dacs. Still, I will try to find a better AD186x DAC because I really want to see what a good R2R DAC can do.
Last edited:
Bummer! This DAC looked promising, and the eBay seller's item description makes it seem like the sound would be quite nice, in spite of the funny translation. Too bad you got a defective board. What are the coupling caps used? And, could this DAC be modded to bypass SPDIF input, and instead accept I2S signal direct to the chip, like what Matt has considered doing with the Little Giant board?
Coupling caps are marked Rubycon but electros get faked a lot in China so I wouldn't place any faith in what the skin says. The same ones on the other DAC tested okay though.
Yes, it can be hacked although it's not one of the easiest I've seen for this. It's really a question of how good your skills are.
- Status
- Not open for further replies.