Hi BIH, Markw4,
Can you please consider for the pcb design to leave lots of space for experimenting with the resistors? It really does matter soundwise, for instance, to use non-inductive types. I suggest both an option for through hole, and smt (preferably also an arrangement to try a quad smt arrangement - series/parallel, as discussed elsewhere on the diyaudio-forum).
Can you please consider for the pcb design to leave lots of space for experimenting with the resistors? It really does matter soundwise, for instance, to use non-inductive types. I suggest both an option for through hole, and smt (preferably also an arrangement to try a quad smt arrangement - series/parallel, as discussed elsewhere on the diyaudio-forum).
@av-trouvaille,
All resistors are to be thin film SMD, or if not that then metal film.
If thin film SMD, I would recommend Vishay PLT types.
For the through hole component board I have specified 0.1% tolerance metal film.
The resistor types are recommended by ESS and are used because they are the lowest distortion types available.
None of the recommended resistors are inductive types, so please take a look at the BOM paper I posted, then go to Mouser and look up the part numbers, and read the specifications on the data sheets for the resistors. If you can find some that are lower distortion, we would be interested to hear about them.
EDIT: The reason we are using through hole resistors for this board is because it was requested to make a project beginners could do. If you are a very experienced builder and are able to build with SMD parts, I would recommend you take a look at the pictures of my SMD layouts that I used for my first dac board. The pictures are high resolution so you can see how they were built very clearly. But you do need to be aware that when you click on a high res pic it does not show in full resolution at first. To see it at full size, first you click on the image to open it, then while the image is open you hover the mouse over the lower left corner of where a white X should appear. If you click on the white X the picture will expand to full size. Some pictures of SMD layouts to mod this dac board can be found here: ES9038Q2M Board - Page 38 - diyAudio
The bottom of the board contains the AVCC and I/V circuits, and the differential summing stage is on the top of the board.
EDIT 2: Regarding quad SMD, I don't think that is necessary for this dac board. Voltages are pretty low and good quality SMD resistors should be linear enough. The dac chip itself is only capable of getting down to -120dB distortion if you do a nearly perfect job of building all the mods. If the dac chip could get down another 10dB or 20dB lower distortion, then yes, we might be looking at ways to further linearize resistors. As it is you will have to work hard and do all the mods including clock and Arduino control of I2C bus to take control of dac chip registers. You will also need an AK4137 board. I can show you how to get down around the -120dB ESS says is possible and probably as good as any ES9038Q2M dac you could buy for any price. We will know more about that last point when the Allo Katana 1.2 dac comes in for review in another week or two.
All resistors are to be thin film SMD, or if not that then metal film.
If thin film SMD, I would recommend Vishay PLT types.
For the through hole component board I have specified 0.1% tolerance metal film.
The resistor types are recommended by ESS and are used because they are the lowest distortion types available.
None of the recommended resistors are inductive types, so please take a look at the BOM paper I posted, then go to Mouser and look up the part numbers, and read the specifications on the data sheets for the resistors. If you can find some that are lower distortion, we would be interested to hear about them.
EDIT: The reason we are using through hole resistors for this board is because it was requested to make a project beginners could do. If you are a very experienced builder and are able to build with SMD parts, I would recommend you take a look at the pictures of my SMD layouts that I used for my first dac board. The pictures are high resolution so you can see how they were built very clearly. But you do need to be aware that when you click on a high res pic it does not show in full resolution at first. To see it at full size, first you click on the image to open it, then while the image is open you hover the mouse over the lower left corner of where a white X should appear. If you click on the white X the picture will expand to full size. Some pictures of SMD layouts to mod this dac board can be found here: ES9038Q2M Board - Page 38 - diyAudio
The bottom of the board contains the AVCC and I/V circuits, and the differential summing stage is on the top of the board.
EDIT 2: Regarding quad SMD, I don't think that is necessary for this dac board. Voltages are pretty low and good quality SMD resistors should be linear enough. The dac chip itself is only capable of getting down to -120dB distortion if you do a nearly perfect job of building all the mods. If the dac chip could get down another 10dB or 20dB lower distortion, then yes, we might be looking at ways to further linearize resistors. As it is you will have to work hard and do all the mods including clock and Arduino control of I2C bus to take control of dac chip registers. You will also need an AK4137 board. I can show you how to get down around the -120dB ESS says is possible and probably as good as any ES9038Q2M dac you could buy for any price. We will know more about that last point when the Allo Katana 1.2 dac comes in for review in another week or two.
Last edited:
Right. The LTC6655 gets input power from the +5v supply, and the dual opamp runs on +-15v rails. Also, it turns out the opamp +-15v power needs a very low impedance path back to the big 1,000uf filter caps or to the power supply output. The last change I made to my existing SMD AVCC circuit can be found here: ES9038Q2M Board - Page 229 - diyAudio
Last edited:
Ok, and existing 2xcapacitors 47uf directly placed onto DAC board (near to chip ) for avcc R and avcc L are still there (or those ones 47uF on avcc buffer 3.3V out should be only used and then no extra capacitor near to the chip...) ,i see that only different type are used on your board... in past you mentioned that increasing of these AVCC capacitor has affect on SQ (more bass but less resolution ?)
Blackfear,
If using the opamp circuit, increasing the 47uf caps should not be necessary or desirable so long as the opamps have a low impedance path back to the +-15 supply or to big 15v rail filter caps.
If people don't use an opamp AVCC supply, sometimes they try putting really big caps there for AVCC. In that case bigger caps do help bass, sometimes enough to get close to opamp-AVCC-supply-quality bass, but overall SQ is never as good as with opamp regulation.
Regarding what type of 47uf caps to use, I used some organic polymer caps there because they have low ESR. Good quality electrolytics probably work too, but I haven't done an A/B comparison. The main downside to organic polymer caps is that they are relatively leaky, so more leakage current has to be supplied by the opamps than if using standard aluminum electrolytic caps.
If using the opamp circuit, increasing the 47uf caps should not be necessary or desirable so long as the opamps have a low impedance path back to the +-15 supply or to big 15v rail filter caps.
If people don't use an opamp AVCC supply, sometimes they try putting really big caps there for AVCC. In that case bigger caps do help bass, sometimes enough to get close to opamp-AVCC-supply-quality bass, but overall SQ is never as good as with opamp regulation.
Regarding what type of 47uf caps to use, I used some organic polymer caps there because they have low ESR. Good quality electrolytics probably work too, but I haven't done an A/B comparison. The main downside to organic polymer caps is that they are relatively leaky, so more leakage current has to be supplied by the opamps than if using standard aluminum electrolytic caps.
Last edited:
Oki thanks, this is exactly what i want to hear, using opamp buffer for avcc to increase SQ, now less work only cut feeding traces to AVCC L and R
Corrected through hole component output stage layout, with differential stage non-inverting input caps added, is attached below.
If additional space is desired to make fitting components easier, opamps sockets could be moved one hole farther apart vertically and or one hole horizontally.
They could be moved more apart than that if desired, but I think the circuit can be built without needing to move them apart at all.
If additional space is desired to make fitting components easier, opamps sockets could be moved one hole farther apart vertically and or one hole horizontally.
They could be moved more apart than that if desired, but I think the circuit can be built without needing to move them apart at all.
Last edited:
Well, a quick and easy full AVCC schematic could be like the one attached below. However, some of the lines are very thin so it may display most clearly using Adobe Reader.
Someone would certainly be welcome to redraw it in an easier to read format.
Last edited:
Well, a quick and easy full AVCC schematic could be like the one attached below. However, some of the lines are very thin so it may display most clearly using Adobe Reader.
Someone would certainly be welcome to redraw it in an easier to read format.
Thanks Mark
Concerning the sound results , do you think that it is better than a LT3042 ?
Or it's just for the new output stage .
Serge
Yes. LTC6655 can only be better as a voltage reference than LT3042, and the LTC6655 and opamp solution will absolutely have lower noise at lower audio frequencies.
Some of the first people that started getting excited about using LT3042 for AVCC have abandoned them and say they don't sound as good as other solutions.
Last edited:
Hi Mark!
I'm a bit confused because on the ESS AVCC schematic I see single +12V power for opamp. (?)
Hi Mark, as you wish
I'm waiting for LTC6655 till end of october because not in stock. This is a AVCC board with 3042 ref. On the AVCC schematic i didn't see decoupling cap for the opa power rail. It is omissible? But I'm soldered in 10uF X7R. Thx. D'Attachments
Hi Mark, as you wish
Hi Deenoo,
Thank you for the picture. Just checked and it looks like Digikey has LTC6655 in stock: LTC6655BHMS8-3.3#PBF Linear Technology/Analog Devices | Integrated Circuits (ICs) | DigiKey
Also, for decoupling opamps, unless the data sheet for a particular opamp says otherwise, we always use 10uf tantalum in parallel with .1uf X7R ceramic directly from each opamp power pin to ground. Best if there is a ground plane, of course.
However, if the effective leads are very short 10uf ceramic SMD might be okay. A reason for using two caps is because one cap can become self-resonant at some frequency. If using a large cap like 10uf, a small cap like .1uf in parallel will help prevent self resonance of 10uf from becoming a problem. Also, lead length plays a big part in self resonance. If no ground plane then power wiring can provide the necessary inductance in series with a cap to create resonant conditions.
Last edited:
Most modern opamps can run from a range of supply voltages. The opamps we typically use such as LME49720 could run on 12v or +-15v, and also some other voltages. So long as the supply voltage is okay for the opamp then it should not be a problem. If unsure, please check the data sheet for an opamp you may be considering for use to see what supply voltages it is rated to work with.
True, for AVCC, 5v and ground should work too, at least in theory. However, very low impedance power rails back to the supply matters and is important for best sound quality it seems. At lower voltages it might matter even more, don't know haven't tried it.
EDIT: I should add as people seem to be getting more interested in moving ahead on AVCC that the AVCC 3.3v power ground return back to the power supply needs to be very short, direct, and low impedance. It should use a ground plane connected to the dac board ground plane just the same as we are doing with the output stage we are working on. To get the best possible sound quality out of a dac like this, I feel there is a need to keep reminding people to pay attention to what ESS says about layout of things like AVCC and output stages. Current always flows in loops and a high impedance ground wire from AVCC to dac and back is just as bad as high impedance 3.3v wire. Always use ground planes for critical circuitry like we are doing here. (for some other circuits, maybe a ground plane isn't the best approach but it is for what we are doing here.)
EDIT: I should add as people seem to be getting more interested in moving ahead on AVCC that the AVCC 3.3v power ground return back to the power supply needs to be very short, direct, and low impedance. It should use a ground plane connected to the dac board ground plane just the same as we are doing with the output stage we are working on. To get the best possible sound quality out of a dac like this, I feel there is a need to keep reminding people to pay attention to what ESS says about layout of things like AVCC and output stages. Current always flows in loops and a high impedance ground wire from AVCC to dac and back is just as bad as high impedance 3.3v wire. Always use ground planes for critical circuitry like we are doing here. (for some other circuits, maybe a ground plane isn't the best approach but it is for what we are doing here.)
Last edited:
In further preparation for the through hole component output stage, did some work to get the dac and output stage boards ready. Two pictures below show the modifications so far.
First, all unneeded components from old dac output stage were removed from the top side of the board, as was the +15v power input jack.
Holes were drilled in the output board to line up with existing holes in the dac board. The +15v trace that ran over to the jack was cut so that the output stage board could not short out the power. The cut point is outlined in a red box.
Also outlined in red boxes are 4 holes to bring down the dac output signals to the bottom side of the ground plane so they can be run down to the output stage board from there.
With the two boards securely attached together and the +15v trace cut, it should now be safe to scrape of some of the solder mask on the bottom of the dac board in preparation for soldering the output stage ground plane to the dac ground plane.
Next steps will probably involve positioning the opamp sockets, then laying down some strips of ground plane leaving openings for tantalum caps and power connections, and for any other circuit nodes I may decide to attach to the board.
Also, since I am using some of the dac board mounting holes to attach the output stage board, I may work out some way they can also support standoffs, or I may decide to use some holes on the far edge of the output stage board as mounting points to help support the entire assembly.
First, all unneeded components from old dac output stage were removed from the top side of the board, as was the +15v power input jack.
Holes were drilled in the output board to line up with existing holes in the dac board. The +15v trace that ran over to the jack was cut so that the output stage board could not short out the power. The cut point is outlined in a red box.
Also outlined in red boxes are 4 holes to bring down the dac output signals to the bottom side of the ground plane so they can be run down to the output stage board from there.
With the two boards securely attached together and the +15v trace cut, it should now be safe to scrape of some of the solder mask on the bottom of the dac board in preparation for soldering the output stage ground plane to the dac ground plane.
Next steps will probably involve positioning the opamp sockets, then laying down some strips of ground plane leaving openings for tantalum caps and power connections, and for any other circuit nodes I may decide to attach to the board.
Also, since I am using some of the dac board mounting holes to attach the output stage board, I may work out some way they can also support standoffs, or I may decide to use some holes on the far edge of the output stage board as mounting points to help support the entire assembly.
Attachments
Last edited:
Looking at where to two boards meet a little more closely, decided to trim back some of the output stage board, so maybe I can try to attach the ground planes together right where the dac ground plane ends next to the power trace running across through there, and just before I start getting into an area where there are a lot of holes to work around. Time will have to tell how that ends up working out.
Also, there is a little horizontal trace on the dac board right in there where I would like to solder the output stage ground plane down. I may cut the traces on top leading to that short trace on the bottom, and solder jumper wire on top to restore the circuit connectivity. That way I don't have to worry about shorting out the little bottom trace, as it will at that point no longer be in use.
Also, there is a little horizontal trace on the dac board right in there where I would like to solder the output stage ground plane down. I may cut the traces on top leading to that short trace on the bottom, and solder jumper wire on top to restore the circuit connectivity. That way I don't have to worry about shorting out the little bottom trace, as it will at that point no longer be in use.
Attachments
Last edited:
Hi DRONE7,
Sorry, I think we lost track of plans for you to try working on an output layout for one hour then stop. May I ask if that happened, and if so how did it go?
Also, no one else committed to trying for some kind of layout drawing, but if anyone did give it a try at all, may I ask how it went?
No one has to show what they came up with if they don't wish to, but it would be helpful to know if anyone even tried at all to start working through it on paper?
Or, maybe if anyone just tried to mentally visualize some layout parts without drawing anything?
If nobody tried at all, why not? I know it can't be all because of no time, not if there is time to do a little fun stuff instead... or to build a board..
Sorry, I think we lost track of plans for you to try working on an output layout for one hour then stop. May I ask if that happened, and if so how did it go?
Also, no one else committed to trying for some kind of layout drawing, but if anyone did give it a try at all, may I ask how it went?
No one has to show what they came up with if they don't wish to, but it would be helpful to know if anyone even tried at all to start working through it on paper?
Or, maybe if anyone just tried to mentally visualize some layout parts without drawing anything?
If nobody tried at all, why not? I know it can't be all because of no time, not if there is time to do a little fun stuff instead... or to build a board..