I was thinking the same thing. Markings all look genuine but same for purses.
Been reading with interest on your mods and performance changes, I haven't done chip stuff for quite a few years, thanks for the tip on the LT3042 regs.
I'd recommend you have a listen to shunt regs for the I/V and output, they just seem to sound right.
Then see if you can swap the I/V for NP's D1 while your building your valve version.
I'd recommend you have a listen to shunt regs for the I/V and output, they just seem to sound right.
Then see if you can swap the I/V for NP's D1 while your building your valve version.
Thanks for the tips.
Well I had to take out the 797 whether fake or real . But in my parts bin I eyed a couple of AD844 of 20 year vintage. Swapped them in, I was having fun anyways so why not right?
These were removed from my old vintage Assemblage DAC using Burr Brown 20 bit ladders I think. Anyways the Assemblage/Sonic Frontiers DAC2. Popped them in and powered them up, listened to Steely Dan Gaucho. Wow! the power of the choral background was all there. That was one track that killed turntables and wimpy amps. The dynamics in Babylon Sisters. Then the cymbals....the detail. Switched over to Diana Krall, Live in Paris and then the space of the live performance was delivered. Now Sergelisses was talking power in the bass. This op amp allowed the sub amp to grab hold of the cones and not let go. Control. Uncanny. Like getting a closed box sub that has a Q of 0.5. My subs are 4th order with passive radiators.
What is more interesting is that it actually neutralized a lot of the high end distortion of the 4392SRC I have on loan from Markw4. Before this I was starting to tire of the 4392. Why? it had a dual nature. It tended to deliver more of what was on the CD, opened the soundstage BUT it also introduced some HF distortion. It was a question of what you want and could tolerate. The AD844 seems to have neutralized it.
But like any system and change, the key is long term listening. Does it fit you. This one was like trying a shoe that fits upon first fit. It might get too loose or it might get better. Time will tell.
http://www.analog.com/media/en/technical-documentation/data-sheets/AD844.pdf
Well I had to take out the 797 whether fake or real . But in my parts bin I eyed a couple of AD844 of 20 year vintage. Swapped them in, I was having fun anyways so why not right?
These were removed from my old vintage Assemblage DAC using Burr Brown 20 bit ladders I think. Anyways the Assemblage/Sonic Frontiers DAC2. Popped them in and powered them up, listened to Steely Dan Gaucho. Wow! the power of the choral background was all there. That was one track that killed turntables and wimpy amps. The dynamics in Babylon Sisters. Then the cymbals....the detail. Switched over to Diana Krall, Live in Paris and then the space of the live performance was delivered. Now Sergelisses was talking power in the bass. This op amp allowed the sub amp to grab hold of the cones and not let go. Control. Uncanny. Like getting a closed box sub that has a Q of 0.5. My subs are 4th order with passive radiators.
What is more interesting is that it actually neutralized a lot of the high end distortion of the 4392SRC I have on loan from Markw4. Before this I was starting to tire of the 4392. Why? it had a dual nature. It tended to deliver more of what was on the CD, opened the soundstage BUT it also introduced some HF distortion. It was a question of what you want and could tolerate. The AD844 seems to have neutralized it.
But like any system and change, the key is long term listening. Does it fit you. This one was like trying a shoe that fits upon first fit. It might get too loose or it might get better. Time will tell.
http://www.analog.com/media/en/technical-documentation/data-sheets/AD844.pdf
Last edited:
Mike, You could try connecting the SRC4392 I2S connector pins to the I2S input pins on your dac board. Although one of the dip switches on the SRC board can configure the I2S port to be used as an input, it may very well be configured as an output by default. Maybe the distortion will be less that way, I don't know. But, DAC-3 does have an SRC4392 in it and the audio output of the dac doesn't sound like that SRC4392 distortion we hear with this dac. I don't know why that is, but trying I2S instead of SPDIF seems like it might be worth trying.
I ordered some 2x single to dual adapters to try the dual 844s on the first stage of the IV.
However since then I have received my OPA1611 and OPA1612 to install and I have listened.
I have done away for the time being with the SRC4392. Too much HF edge. I am now using Async USB via XMOS with Sitime oscillators. I will eventually powe these from a 5V linear supply.
Nice surprise actually. I agree with Sergelisses opinion of these op amps. On this particular board that we are using it has come out to be the best combination so far. It is not that any prior op amps were "bad" but it is the small details in refinement.
To give a gauge of where these op amps are. Moving from the 5532/34 series to the LME497xx series was a step up. How large? Suppose we rate that at one. The step up from the LME497xx series to the OPA161x series looks to be at least a 2 or maybe a three.
Why? do they sound better. In my system, just as Sergelisses pointed out, they have the power and grunt on the bottom octave that the AD844s had....without sounding bloated and they retained the balance of midbass to ultra low. Thus not making them sound boomy at all. No. They are well balanced in the bottom. Now at the high frequencies, they are so much better than the LME497xx series. Absolutely no hint of any splashiness no matter the track. None. Yet they will trace the high frequencies with utter ease. Now you can really hear the inner harmonics of the piano strikes. Change the piano used...in a session and you will likely pick that up. Yet absolutely no hint of harshness....like a good vinyl record. The next part is the imaging. On my system they give good image height and width and depth. Remember I commented that the 797 sounded closed in. This opens back up wide and gets the height correct. This aspect will vary by system.
Definitely a recommended pair for this board. If ESS is now using these in their evaluation boards, I can see why. The interesting thing is that specwise, they don't appear that much better than the LME497xx series.....but we all know specs don't tell everything.
The 797 sounds dated, yeah it's smooth and relaxed but maybe too much so.
I also received my Crystek, and the other parts to the Super Reg and my power supply box....all towards finalizing the project and when everything is placed then I will cut out my power lines from flat sheet copper foil and cut them exactly where they need to go. I am still waiting for aircraft connectors that will take power from the box to the dac.
I am also checking out some OPA1611s with current boosted transistors on an adapter. This in combination with an LTC6655 might end up as my op amp AVCC. We'll see. If I go this route, I will transform my AVCC from one line into a L&R arrangement.
However since then I have received my OPA1611 and OPA1612 to install and I have listened.
I have done away for the time being with the SRC4392. Too much HF edge. I am now using Async USB via XMOS with Sitime oscillators. I will eventually powe these from a 5V linear supply.
Nice surprise actually. I agree with Sergelisses opinion of these op amps. On this particular board that we are using it has come out to be the best combination so far. It is not that any prior op amps were "bad" but it is the small details in refinement.
To give a gauge of where these op amps are. Moving from the 5532/34 series to the LME497xx series was a step up. How large? Suppose we rate that at one. The step up from the LME497xx series to the OPA161x series looks to be at least a 2 or maybe a three.
Why? do they sound better. In my system, just as Sergelisses pointed out, they have the power and grunt on the bottom octave that the AD844s had....without sounding bloated and they retained the balance of midbass to ultra low. Thus not making them sound boomy at all. No. They are well balanced in the bottom. Now at the high frequencies, they are so much better than the LME497xx series. Absolutely no hint of any splashiness no matter the track. None. Yet they will trace the high frequencies with utter ease. Now you can really hear the inner harmonics of the piano strikes. Change the piano used...in a session and you will likely pick that up. Yet absolutely no hint of harshness....like a good vinyl record. The next part is the imaging. On my system they give good image height and width and depth. Remember I commented that the 797 sounded closed in. This opens back up wide and gets the height correct. This aspect will vary by system.
Definitely a recommended pair for this board. If ESS is now using these in their evaluation boards, I can see why. The interesting thing is that specwise, they don't appear that much better than the LME497xx series.....but we all know specs don't tell everything.
The 797 sounds dated, yeah it's smooth and relaxed but maybe too much so.
I also received my Crystek, and the other parts to the Super Reg and my power supply box....all towards finalizing the project and when everything is placed then I will cut out my power lines from flat sheet copper foil and cut them exactly where they need to go. I am still waiting for aircraft connectors that will take power from the box to the dac.
I am also checking out some OPA1611s with current boosted transistors on an adapter. This in combination with an LTC6655 might end up as my op amp AVCC. We'll see. If I go this route, I will transform my AVCC from one line into a L&R arrangement.
Last edited:
Just found an interesting OP AMP for AVCC. Looks like it can get the job done for the PRO chips.
http://www.ti.com/lit/ds/symlink/opa1622.pdf
140MA output at 3.3V!
http://www.ti.com/lit/ds/symlink/opa1622.pdf
140MA output at 3.3V!
I pointed that part out months ago to you Mark. The thing is that where do you get one. Second thing, can someone like me fab a board for it. Third thing is...specwise it was ho hum relative to what other LDO makers were doing. I think, they were behind on this. Next ESS Pro chip family...watch for some integration of some type into the chip.
There is a 9038Q2M DAC made in Europe or somewhere over the pond that uses that chip and performance was so so. When I read that, I totally dropped that chip off the radar.
Here's where it gets really interesting for the people like me unable to fab boards or just plain too lazy to do so.
OPA1622 | eBay
Dual OP Amp Preamp DC Amplification Board PCB for NE5532 OPA2134 OPA2604 AD826 699951710450 | eBay
Change around some parts, decouple the board nicely and bam you've got AVCC OP amp for L&R. I am going to try this and see what happens. 3.3V reference feed will be from an LT3042.
There is a 9038Q2M DAC made in Europe or somewhere over the pond that uses that chip and performance was so so. When I read that, I totally dropped that chip off the radar.
Here's where it gets really interesting for the people like me unable to fab boards or just plain too lazy to do so.
OPA1622 | eBay
Dual OP Amp Preamp DC Amplification Board PCB for NE5532 OPA2134 OPA2604 AD826 699951710450 | eBay
Change around some parts, decouple the board nicely and bam you've got AVCC OP amp for L&R. I am going to try this and see what happens. 3.3V reference feed will be from an LT3042.
Regarding a design that didn't do well using the ESS AVCC part, there are so many ways to get a dac wrong, using a possibly good part cannot be expected to save an otherwise problematic design.
Thing about the ESS part is it is both a voltage source and a shunt sink, so they were apparently aiming for what some people say they like in a regulator. An opamp in similar in that it can both source and sink current. If one is certain there is no reason to sink, such as maybe to dampen some overshoot, then a source-only solution should be fine. i.e. maybe just use an LT3042 or LT3045 as a buffer for an LTC6655. That will give the lowest noise at low audio frequencies, I am thinking.
And, as it happens, I just used an LTC6655 as a reference for my improved opamp buffer for AVCC power. Between that and dedicated regulation for VCCA, it moves sound quality another notch towards DAC-3.
The other thing is what Allo says it is doing with Katana output stage filtering. I have not posted any recommendation yet, but I am convinced I can make some improvement there too. Won't say how I know though.
Thing about the ESS part is it is both a voltage source and a shunt sink, so they were apparently aiming for what some people say they like in a regulator. An opamp in similar in that it can both source and sink current. If one is certain there is no reason to sink, such as maybe to dampen some overshoot, then a source-only solution should be fine. i.e. maybe just use an LT3042 or LT3045 as a buffer for an LTC6655. That will give the lowest noise at low audio frequencies, I am thinking.
And, as it happens, I just used an LTC6655 as a reference for my improved opamp buffer for AVCC power. Between that and dedicated regulation for VCCA, it moves sound quality another notch towards DAC-3.
The other thing is what Allo says it is doing with Katana output stage filtering. I have not posted any recommendation yet, but I am convinced I can make some improvement there too. Won't say how I know though.
Last edited:
If you can't buy then build, if you can solder then you can 'Veroboard' it.
I'm going to give this a go.
If you know the AVCC current sink/source requirements, then you can 'Class A' bias the output with a resistor to the appropriately supply rail.
Attachments
Let us know how it turns out. I was referring to the ESS9311 voltage reg not the 1622.
Listened some more and the OPA1611/12 gets it right to me. Wonder if I still want to build that hybrid mosfet/6922 buffer now. The tubes are on their way so it has to be a go. When I want that romantic sound, I can patch it thru the buffer. Yeah, it adds something but food tastes better with spices. Right. Same for music?
Even food is usually best without the same spice added to every meal. Sometimes the natural flavors of good basic ingredients can be just the thing.
Beyond that, I sometimes have this suspicion opamp rolling could be more analogous to the use of spice on food way back before refrigeration was invented.
Beyond that, I sometimes have this suspicion opamp rolling could be more analogous to the use of spice on food way back before refrigeration was invented.
You mentioned the opamp earlier, I just threw something around it; I'll post my results.
Yes,
that's why I said build instead, and it's fun.
OK, I was looking at this circuit and applying some of its ideas here.
http://www.kelm.ftn.uns.ac.rs/literatura/mpi/pdf/Op Amp Applications Handbook.pdf
Page 232. It makes a suggestion about leakage current on electrolytics which might be of use in your circuit.
Why use 1n4002s to protect the inputs. Would you be better off with some IN4148 or 1N914s to clamp down quickly? The idea of protecting the outputs of the op amp is nice. I never thought of that before.
Now in my application I will continue to use an LT3042 at 3.3V to source the voltage input. That makes bootstrapping the input caps inconsequential as there is plenty of current from the 3042 reg. But I imagine using a TL431 or even an LTC6655, it might be of benefit because you will always need an electrolytic on the input side.
You realize there is a need for such a board in the mod arena if you're successful. Also what is nice if it is possible is that the opa1622 is a dual op amp but we need only one voltage reference for the input. Just hard to imagine that the tiny opa1622 can put through nearly 300ma in total. Is it possible? I imagine some heatsinking might be necessary.
http://www.kelm.ftn.uns.ac.rs/literatura/mpi/pdf/Op Amp Applications Handbook.pdf
Page 232. It makes a suggestion about leakage current on electrolytics which might be of use in your circuit.
Why use 1n4002s to protect the inputs. Would you be better off with some IN4148 or 1N914s to clamp down quickly? The idea of protecting the outputs of the op amp is nice. I never thought of that before.
Now in my application I will continue to use an LT3042 at 3.3V to source the voltage input. That makes bootstrapping the input caps inconsequential as there is plenty of current from the 3042 reg. But I imagine using a TL431 or even an LTC6655, it might be of benefit because you will always need an electrolytic on the input side.
You realize there is a need for such a board in the mod arena if you're successful. Also what is nice if it is possible is that the opa1622 is a dual op amp but we need only one voltage reference for the input. Just hard to imagine that the tiny opa1622 can put through nearly 300ma in total. Is it possible? I imagine some heatsinking might be necessary.
Last edited:
I have been very busy with other tasks. In the meantime I have come up with the following that will be done over the next several weeks as time allows.
1. Install the Crystek crystal. Feed that directly from its own LT3042.
2. Assemble my Super Regulator board and configure it for 17V DC rails in an effort to maximize headroom for the op amps.
3. Assemble the hybrid op amp power supply as shown in attachment using an LTC6655 precision reference to generate a stable voltage. The assembly is essentially the same as shown in the LME49600 datasheet on page 31.
This assembly will take the place of the LT1963 mounted on the DAC motherboard. The footprint will allow direct mounting to minimize leads and also save space and allow for power dissipation as well.
I've also received my external power supply enclosure and connectors and that will slowly be assembled together as well.
Perfect? No. Good enough yes. Since the DAC is already at a good level, I'm taking my time since it is working well so far.
1. Install the Crystek crystal. Feed that directly from its own LT3042.
2. Assemble my Super Regulator board and configure it for 17V DC rails in an effort to maximize headroom for the op amps.
3. Assemble the hybrid op amp power supply as shown in attachment using an LTC6655 precision reference to generate a stable voltage. The assembly is essentially the same as shown in the LME49600 datasheet on page 31.
This assembly will take the place of the LT1963 mounted on the DAC motherboard. The footprint will allow direct mounting to minimize leads and also save space and allow for power dissipation as well.
I've also received my external power supply enclosure and connectors and that will slowly be assembled together as well.
Perfect? No. Good enough yes. Since the DAC is already at a good level, I'm taking my time since it is working well so far.
Attachments
Last edited:
Diagram should show 2 decoupling caps per rail. Just sayin' -- I know you know.
Something I have been wondering about is whether LF noise from LT3042 will affect oscillator phase noise at 10Hz. Unfortunately, Crystek doesn't say much about PSRR that I have seen. Sure their clocks must have some, just don't know how much. Anyway, these days I feel comfortable using LT3042 as a single 5v supply for everything since total load is less than 1 amp. Then I use local 3.3v regulators, sometimes dedicated ones for things that might be especially sensitive to LF noise or to other noise that might be encountered on a shared rail. Of course, this may be overkill. Not having measured, I don't know for sure. Still, I am okay with a little overkill for a one-off project. Better than under-kill.
Be very interesting to see how it turns out, Mike. Please post some pics along the way, if that would be okay, since we have a few rather inexperienced builders interested in modding these dacs.
Something I have been wondering about is whether LF noise from LT3042 will affect oscillator phase noise at 10Hz. Unfortunately, Crystek doesn't say much about PSRR that I have seen. Sure their clocks must have some, just don't know how much. Anyway, these days I feel comfortable using LT3042 as a single 5v supply for everything since total load is less than 1 amp. Then I use local 3.3v regulators, sometimes dedicated ones for things that might be especially sensitive to LF noise or to other noise that might be encountered on a shared rail. Of course, this may be overkill. Not having measured, I don't know for sure. Still, I am okay with a little overkill for a one-off project. Better than under-kill.
Be very interesting to see how it turns out, Mike. Please post some pics along the way, if that would be okay, since we have a few rather inexperienced builders interested in modding these dacs.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.