If using opamp AVCC regulators, no need for 3300uf caps too. Somewhere around 10uf to 47uf at the output of the opamps should be enough, depending on the particular opamps. 47uf worked with all the ones tested and recommended by ESS.
MKTs? Not sure what you mean by that.
Regarding IVY output stage, don't know what might help it. Don't know that anyone has run experiments to find out if any effects from film caps on the power. Also, separably from any power rail filtering effects most likely there is going to be more distortion with IVY than if using opamp I/V simply because IVY is somewhat less of an ideal offset-virtual-ground dac load (needed for lowest distortion out of the dac chip), but don't know how much of an issue it might be for you, if any.
Very good, and maybe a good opportunity to try some pin lifting while you are working on a junk practice board (not that you need it right now, but its a good skill to have some experience with).
You are quite welcome, of course!
For those not wanting to use LTC6655, it may be fine to use LT3042 as the 3.3v voltage reference for an opamp AVCC supply. Just don't use that particular LT3042 for anything else at the same time.
Only something like a 3-opamp output stage will work. An I/V is needed as the first part of that to operate the dac in current mode. After that, differential summing removes common-mode distortion, does some more filtering, and removes any offset voltage.
Its three 1800uf electrolytics in parallel added to the existing AVCC components. That will at least lower measured distortion.
The bypassing we use for other things is intended to keep circulating currents local by keeping the impedance to ground low over the frequency range where device may draw current. For AVCC, we don't necessarily want impedance to be super low at HF/RF, since we don't want a lot of HF/RF to appear at the output of the dac and then need to be filtered out in the output stage. Electrolytics tend to become lossy at HF/RF, meaning they act more like inductors and resistors in series. Their impedance goes up, so if we wanted to keep impedance low, we might need to add some RF caps in parallel. However, that can get tricky and is often not such a good idea. It may be better not to use caps with too little ESR when we do that, but that choice might have other consequences. If you want to know more you might look for the book, Electromagnetic Compatibility Engineering, by Henry Ott. A very practical and standard reference for such things.
The thing about using LDOs for AVCC is they tend to make the sound kind of dynamically 'flat' instead of giving realistic dynamics in the reproduced sound. Also, they tend to sound more distorted than opamp regulators. Don't know why exactly but could be because the error amps inside LDOs are not necessarily low distortion amps since they aren't designed to listen to. Since AVCC has zero PSRR (power supply rejection ratio), that means any sound of the voltage regulator may come through in the audio. Good low distortion audio opamps empirically sound best for AVCC, possibly because to whatever extent they have a sound while acting as regulators they naturally tend to sound good with audio. In other words, as regulators, they have low distortion error amplifiers which may matter when PSRR is zero.
Last edited:
Thanks! I did at the time of setting up this account, how time flies!
TBH, I haven't used the optical input on the board, mainly i2s and the coax. Have you tried shorter length cables?
That's really good to know. Keeping the board changes simple and not having a big black cap near the crystal when trying to desolder it later is a good idea.
Wima MKT, polyethylene caps.
I've got a junk ES9038Q2M to practice on!!!!
I need the long cable for the Sky tv box due to physical placement of it all. It's not a bid deal, just a shame.
Oh, that is an interesting suggestion. Actually it might be viable to build the circuit now, i found some cheap (maybe fake) LME49720NA's on ebay. But actually maybe even if its fake, it still might be better than an LT3042, perhaps. However a limitation is that here, the selection of resistors and capacitors is quite poor that they dont stock any higher quality stuff. So then i would have to order that online too, and since all the resistors are 10k, and all the caps are 10 and 47uF it wouldnt be such an issue, other than waiting for everything to arrive who knows how long.
I was just wondering, wont such big caps limit the frequency range it will filter (specifically HF), or maybe this is not important?
Furthermore, maybe related question hypothetically if we used an lt3042 for vdcc and put a cap on input it would be ok, but if we put a cap on the output wont that limit bandwith? The spec sheet suggests a 4.7uF output capacitor, but i have previously thought 100uF was a good idea, maybe its not?
Another question is what about LT1963? This one has 3 separate output channels, seems like WIMA caps at the output too.
Here is the idea, using the suggestion you have given. It has 2x 3.3V output and one 5V, use the 5V output as replacement for the LTC6655 and build the AVCC circuit, use the other 3.3 for the DVCC, and the last 3.3 for eventually if i decide to build the 3 opamp output stage. In this case i would be interested in how the fast transient response of the 1963 affects opamp operation.
In theory it could be a fairly simple, "elegant" solution for power supply, that also future proofs me in case other mods need a supply, even if perhaps not the optimal. At the very least, it shouldnt matter much for DVCC or VCCA, i think LT1763 is recommended for that in this thread, and the noise is also 100uV/DIV.
The spec sheet for the LT1963 indicates a 10uF cap at the output. As opposed to the lt3042 it is optimised for fast transient response, however i cant decide if transient response is something you want or dont want on AVCC.
A concern is the noise being 20x higher at 100uV/DIV than an lt3042, so if transient response isnt important, then indeed an lt3042 might be better for building the AVCC circuit.
The PDF you have kindly provided advises low noise for the AVCC regulator.
Its dropout voltage is also only 340mV on the LT1963 and the lt3042 has 350mV, with an input of 15V and output of 3.3V that should be plenty of differential even if they are LDOs, right?
Also thanks for the book recommendation, i have read the synopsis and will go to the library to try and find it (or a similar book). I have seen the table of contents and it seems like a read and a half
I havent come around to replying to this previously, because i was still thinking and estimating it. I think this is a topping d50 output stage right?
I think if I were to build any circuit right now, it might be the the AVCC supply, because of the aformentioned reason of only generic cheap components being available to me. There's a lot of differing components and it might be troublesome to source each component, resistor and cap individually on ebay. Also im not super confident i could make it right, so i think that one remains until i prove my worth with other mods, i would first have to see how well the AVCC goes.
Regards
Do not buy on Ali, on ebay capacitors / resistors and operational amplifiers! You do not know the Chinese well !!! In my subordination they worked 2.5 years. and I know how they live and think. no. they are not bad people! but Mark is right - this smp es9038 board is cheap only because there is nothing "real" there, except for the es9038q2m chip. and it is designed simply. just to work. the rest is marketing. this makes sense to buy - we will finish it as it should. and the details you DO NOT CORRECT. take and buy these parts in 2-7 months, but in a reputable store! have respect for your work and its payment. respect yourself eventually! stop sponsoring the Chinese who bought a marriage / fake! The result is a well-built DAC on fakes / marriage
sorry. You do not know much about the Chinese traders. this does not apply to everyone! I do not want to offend the Chinese! but most of the Chinese traders cheat on you. there is a Russian proverb: "avaricious pays twice." and it means - having bought cheaply - you are doomed to buy again. and this time you’ll buy a normal product for the corresponding price. plus what you paid earlier. total - it will cost you dearly !!
Mark! Hello)) my DAC burned down again ((I don’t know why. I used avcc lm317. the voltage was 3.26 volts. the MCU worked on asm 3.3 v. the other parts were powered by TPS7A4700. lm317 for avcc sounds much better than tps7a4700! smp v1.07 will come to me tomorrow. And ak4137, which with remote control and spdif inputs (new version / blue color). I will try again. Let me remind you - I have Chinese amanero usb. With the es9038 register change, I have not progressed yet. was on a business trip. and the DAC died. replacing watches on cchd gave good results!))) I want to try in avcc either op or Nazar.
I will continue to access the DAC registers. using arduino. on avcc I will try Nazar. I do not think that op will be worse. I’ll try the output on transformers. I know that they are not without flaws. but operational amplifiers are also not without flaws. even though it is opa1612. There is far from one review of how they sound with the es9038q2m. example: ES9038Q2M dac streamertje - forum.zelfbouwaudio.nl
Mark. what are the tips when using ak4137? I also need spdif inputs. I remind you. This is the new version of ak4137. RUB 4,806.61 8% DISCOUNT | AK4137 flagship high-end DAC SRC audio 384K 32Bit DSD256 DSD PCM conversion AliExpress
Electrolytics may start becoming inductive up around 100kHz. That is more than enough frequency range for AVCC, but not enough for power pins on the opamps we use. Thus, we use different caps for opamp power bypass than we use for AVCC.
Also, one needs to be very careful about buying individual components on ebay. There have been many fakes, some of them cleverly re-marked with different part numbers. Some are manufactured in factories there, and sometimes used parts are employed, taken out of old junk circuit boards. Some are real too, hard to know sometimes. In any case, it turns out it is too hard for them to fake the dac chip, it is real. Some of the resistors and caps that are not marked with expensive brand names are just cheap parts, but not fakes.
It takes some careful reading to understand what is best for LT3042. Using a cap that is too big on the output will limit regulation bandwidth, but might allow for improved transient response as seen by the load. Probably best to start with what they recommend.
Don't know. I have used LT1963 for +15v and it worked okay. Still probably not ideal for AVCC.
I would suggest to use LT4042 module as a 3.3v reference, and put a socket for a dual opamp. For now, maybe try a plain old 5532. They are very cheap these days, so much so even the ones from China may be real. Sometime later if you can get better opamps. LM4562 are farily common and the same thing as LME49720. They can work quite well if there is no RF around or if they are put in a shielded case to keep RF away from them. Otherwise, some kinds of RF will cause some distortion in them. Some cordless phones may cause problems, for example. Best is probably OPA1612, although for AVCC AD797 can also do a great job (but the latter are singles, so two sockets would be needed).
It would probably be okay for DVCC and VCCA, but its big and keeping wires very, very short between it and the dac chip loads could be tricky. I would suggest to leave the bypass caps at the dac chip as they are for now. Also, there are VCCA pins on two sides of the dac chip. Both sides will need to be connected to the same voltage regulator.
The schematic I attached at the bottom of the post is what someone claimed to be a topping version. The general topology is a very old one, and about the same as we have always used. The I/V opamp circuit is well known, and the differential summing circuit is called an MFB (multi-feedback) differential filter. Thus, it is often called a 'standard three opamp' output stage for a current output dac. Sometimes a differential Sallen-Key or other filter circuit is used instead.
Last edited:
Link appears to be broken, maybe you can try again?
Very sorry to hear a dac was burned. Over-voltage at one or more of the pins would be the most likely cause. Arduino can do it if the design is not careful enough. A bad voltage regulator might do it. For AK4499, the I/V opamps can damage the dac chip if the dac is powered off and I/V output voltage goes too high, but ES9038Q2M may be different in that respect. It is important that the dac chip not be powered off when anything is connected to pins that can go higher than 3.3v. While the dac chip is powered up, some of the pins are 5v tolerant, as specified in the data sheet.
Regarding AK4137, for a long time I just put it between the USB board output and the dac board I2S input. AK4137 was usually set to upsample to DSD256. Sounds best with as low as DSD DPLL bandwidth can be set and remain stable. To get DPLL bandwidth down to 1, I had to divide the dac chip clock by 4 to produce a 25MHz clock signal which was used to replace the 22.5MHz clock on the AK4137 board. Best sound quality still turned out to be very sensitive to AK4137 phase relative to dac clock phase. Probably the the I2S output of AK4137 needs to be reclocked as the dac would like to see it. I need to try that at some point, but no time for it yet.
from the description it is hard to figure out what kind of power circuit you are using, but the first rule for diy ps is to hook up a scope and check out is there any overshot upon switching on/off
Thank you.
One might add that if using multiple power supplies and switching them at the same time, it is still possible that some power supplies come up to full voltage slower than others. If switched off at the same time, some can discharge the filter caps faster or slower than others. In any of the above cases, it is important that the dac chip doesn't inadvertently get too much voltage on the pins. The 5v tolerant pins may not stay tolerant long enough for any external 5v device power supplies to discharge to below 3.3v, for example. I2C and I2S pins are of the conditionally 5v tolerant type. Voltage regulators for other dac pins need to stay within what the dac chip can tolerate during power switching transients, or AC line power disturbance transients, as eziitis has thoughtfully pointed out.
Last edited:
Mark is getting so many questions he might come out of retirement and open up a consultancy firm
Interesting how many op amps are interchangeable, yeah i'll get a dip socket.
How much does it matter that the components be high quality? Im limited either buying generic ones locally or (possibly) counterfeit ones online. Considering this i dont know if this might be a counterproductive task, or still worth it.
Im looking at this schematic (because its easiest to read for me) 2x AVCC goes to one and the other output to other, ground goes to any ground. I guess i dont need to lift any pins or caps for this, just connect new ones (wouldnt the oboard 3.3v rail still be providing power if everything is just left as is on the board)?
Only thing is im not sure where Vref is on the board, i tried looking in the es9038q2m datasheet but no mention. I was looking at this posts 2nd image, i cant see where do those 2 white Vref wires connect on the board. Underneath where the capacitors GND and 3.3v pop out? But then i thought that was for the AVCC lines (where one could connect the AVCC output lines on the schematic) I will admit, i have a lot of questionmarks.. Picture. (please dont cringe)
Sorry in advance, this post might be my worst one yet
Okay, gotcha. So i just ignore everything in the AVCC diagram before the LTC6655 and build from there out of the lt3042 output?
Interesting how many op amps are interchangeable, yeah i'll get a dip socket.
How much does it matter that the components be high quality? Im limited either buying generic ones locally or (possibly) counterfeit ones online. Considering this i dont know if this might be a counterproductive task, or still worth it.
Even in the use case where its just a replacement for the LTC6655 and LME is the opamp that is feeding the AVCC?
I apologize, im not sure what this means exactly. Do you mean there is AVCC_R and AVCC_L, on pins 1 and 10 and the voltage regulator will be a single lt3042 feeding the opamp in the AVCC circuit?
Im looking at this schematic (because its easiest to read for me) 2x AVCC goes to one and the other output to other, ground goes to any ground. I guess i dont need to lift any pins or caps for this, just connect new ones (wouldnt the oboard 3.3v rail still be providing power if everything is just left as is on the board)?
Only thing is im not sure where Vref is on the board, i tried looking in the es9038q2m datasheet but no mention. I was looking at this posts 2nd image, i cant see where do those 2 white Vref wires connect on the board. Underneath where the capacitors GND and 3.3v pop out? But then i thought that was for the AVCC lines (where one could connect the AVCC output lines on the schematic) I will admit, i have a lot of questionmarks.. Picture. (please dont cringe)
This is good to know what to google to learn more about it. By the way i managed to acquire the book you recommended.
Sorry in advance, this post might be my worst one yet
Last edited:
Generic should work okay.
It okay to do the best you can afford for now, but if there is a choice then LT304x are much lower noise which makes them much better for AVCC reference use.
Referring to our AVCC schematic, just replace LT6655 with LT304x, either one at 3.3v should be fine.
Also, each existing AVCC pin supply includes an electrolytic cap, an SMD ceramic cap, and an SMD inductor. The two caps are in parallel going to ground, and the SMD inductor connects AVCC to the existing 3.3v rail. If using opamp AVCC supplies, first remove the inductor or cut the trace to disconnect the existing 3.3v. The caps might be able to stay if the electrolytic is the right value, or you could replace it with a better one if you want. The ceramic cap can stay or go. I usually leave it.
Vref goes to the standard 3-opamp output stage, if one is fitted. Otherwise, it serves no purpose.
Regarding your linked picture, if buying a pre-built LT304x module, it probably already has an output cap so no need to duplicate that (i.e. you can delete that first cap on the left in your drawing).
As for coming out of retirement, I am having fun not working
Last edited:
Once again. AK4137. 4 806,61 руб. 8%СКИДКА | AK4137 флагман высокого класса ЦАП SRC аудио 384 K 32Bit DSD256 DSD PCM преобразования
AliExpress
AliExpress