@Serge, the 9028pro board has LT1963-3.3 reg for AVCC. apparently not good enough to handle specific current demands in this case. one may add an op amp buffer as discussed above and ECC recommends. or build own regulator something like this:File:Voltage stabiliser OA, IEC symbols.svg - Wikipedia
Last edited:
The examples given int eh document here are for driving AC into varying loads. Distortion will increase because of increased crossover distortion at higher currents. When using an opamp as a buffer for DC as in the case of an Avcc regulator, crossover distortion won't be an issue.
If wanting a separate I/V and bal-se converter, the IVY by Twisted Pair will work very well. You'll have to adjust the I/V resistor to suit needs.
The IVY Balanced Line Stage
The IVY Balanced Line Stage
Maybe. Of course it could be done but I don't think we are quite ready yet. For one thing I think a board should have space for some kind of AVCC circuit, at least if it would be intended to work with a Q2M DAC. In that regard, so far nobody has measured AVCC current at the very highest sample rates any of us can do. Maybe I could get to it eventually, but I'm working on other stuff now.
Also, I don't want to make an IV board for this DAC until I understand what is going on with my too-warm op amps.
This morning I have been cleaning up my DAC board, restoring the original wiring for the 1.2v regulator. Think it should be fine on the digital 3.3v bus, or maybe fine to leave it disconnected entirely.
I am also adding a little more filter capacitance to the 5v and 3v rails to see what that does. I am using an LT1084 for the 5v regulator (its what I had in stock) and the 10uf tantalum output cap I have on it is very minimal. I want to try a little more while I'm just cleaning up a few odds and ends that are not big problems, but they maybe aren't as good as I think good practice would suggest.
Also, setting up to do a test separating 5v and 3v power from +-15 rails. Just starting to peel back the layers of possible coupling between RF and analog. Could also be through the DAC outputs, probably some HF noise there, but not sure that is all there is to look for. Another possibility would be ground currents and resulting ground plane voltage gradients introducing RF noise into analog circuitry. A different kind of possibility would be radiated RF noise, which I know LME49720 can be sensitive to (like from DECT wireless phone base stations).
Why all this? Because there is no way a differential amplifier with 10k resistors around it and a few k-ohms output loading, and with maybe 4 Vp_p maximum peak output level audio should be very warm to touch. Especially even when it is hardly outputting any voltage. It doesn't seem credible to me that should be normal for this application. Problem is I can't see any reason for it on the scope. So, I will have to try some other things to narrow down the investigation.
Given all the above, laying out a board is not my top priority right now. If somebody else wants to start working on it that would be great. Download the free Eagle PCB layout software and read the manual. (its not really free, you pay to have them make boards, or pay them to convert the layout files to Gerber format so you can have someone else make the boards). Or, use the app of your choice if that sounds better.
Last edited:
I will try a NewClassD (UWB) Regulator 3.3v. i have one ....
wait and see .
The muse and AD927 are not so good
Serge
The muse and AD927 are not so good
Serge
I do not think there is a need for muses, even though AD797 is a very good op amp for the filter. of course all discrete op-amps sound "better", but mostly because producing extra harmonics. actually pretty appropriate might be OPA16XX members, or one could try to increase current handling performance of older ones by adding a buffer.
Attachments
Last edited:
Current supply for the LT1963 is fine,, but it's noise and PSRR performance is much worse than state of the art regulators like the LT3045 or ADM7150.
When I used the LT3042 with the current boost FET, I found that it was displaying some sensitivity to implementation at lower frequencies even when the 22uF bypass cap was implemented. This convinced me that despite the lovely specs on paper, it did not perform as expected and I was hoping that my transition to the OP AMP AVCC regulator would be an improvement. We need nothing but the best for AVCC as a base when modding these ESS DACs. Each change shows up in the sound.
I am hoping to get around this with the 9028pro board I ordered. In viewing the board ( and the same board is sold by many vendors on Ebay and AliExpress and TaoBao) It looks like the board has a totally separate power channel for AVCC. It starts off with a dedicated 6V ac input that feeds a bridge arrangement into some filter caps and then to two separate LT1963s. My plan is to completely not power on this channel at all. Thereby having my own line/s for AVCC and powering up AVCC via OP AMPs by tapping onto the decoupling caps or soldering onto the LT1963 output pins. This gives complete control of AVCC. The next part is that the analog side is driven by a simple 317/337. Again my plan is to remove these and the two resistor dividers to set voltage and then interject either a Super Regulator or from my parts boards a Sulzer regulator as a low noise low Z supply by taking the output of the regs into where the 317/337 are soldered on making this a "clean" and easy mod. I would reuse the initial rectifying circuits to feed the added regulators. These same added regs could feed the op amps for AVCC. I have found a $2 board that is useable for creating AVCC reg as well. No fabrication of boards is necessary. I am also currently studying a dual NE5534 board that might be useable for creating a dual AVCC supply using AD797s as well. For me fabrication of boards is a problem. I'd rather get something off Ebay and desolder the parts required and re purpose the board.
What is even more interesting is the 3 channel power supply that was also mentioned. That would possibly be an easy to implement mod to this board. The 3rd channel might be good enough for AVCC directly or at worst be the power supply for AVCC op amps.
So my plans create a high quality DAC is likely possibly by purchasing all prefabbed boards from Ebay. The intial -93db reading without the mods sure seem like a significantly better place to start. I can't wait for my board to arrive.
Last edited:
Can you say more about the sensitivity to implementation, such as what aspect of the implementation do you think was causing the issue? What did it sound like?
BY changing the quality of the Cset caps on the LT3042, you could hear differences in the character of the sound primarily in the lower frequencies. My final arrangement was to use paralleled film caps to correct the sound. This is something I had not expected but there is possibly more about these regs that we have not uncovered from an audio perspective. I bumped into the LT3042 after Didden had mentioned the TPA47 something and I was shocked to see the extremely low noise rating for these regs. Orders of magnitudes better than that and quite amazing. So I was expecting this to be superior to any kind of op amp supply and the intitial use of it to AVCC was eye opening to me and then I started to listen more carefully. I feel it is possibly better suited as a voltage reference for AVCC and I plan on using it for that as I have it. But I am not 100% on using it to drive AVCC itself. I think eziitis also commented on this but it seemed he was even less positive about it.
Guys, it is great to see you working on or thinking about making an AVCC supply design for 9028PRO. But, please, try to think like engineers I know you can do it.
Here is the thing: you cannot design an AVCC supply simply by talking it over and weighing each other's thoughts and arguments. Engineers are not taught to work that way and for good reason. You must work with engineering models, math, numbers, and measurements. It isn't so much a social consensus type of activity.
If you want to design an AVCC supply for Q2M or for 9028PRO, you need to come up with a model, and some numbers that will be your design goals within that model. First thing to do is find out what ESS says is needed for AVCC. I have no idea what 9028PRO needs according to them. I will tell you they give a number for AVCC current for a Q2M in what they consider to be a more or less standard operating mode when used in something like a cell phone. However, for cell phone use ordinarily the DAC is run at less than full thottle, pedal-to-the-metal mode. They don't give a number for that mode. So, the options would then be, (1) ask them, or, (2) measure it. That would give an engineering number for current draw to work with.
For AVCC voltage supply stability, there is no spec. There is an example circuit that you have all seen and warnings about op amp and passive choice component selection for AVCC and IV stage use. That's all we have to work with and as engineers we would need to somehow decide on how good we need. That gets us into tricky territory because there is only limited research on what people can hear, and engineers have widely different views on the quality or reliability of that research. So, it is up to the designer(s) to deal with voltage stability question on their own given the information we have and in deciding if it is good enough to work with, or what. AVCC current though, is a given, it is what it is, you just have to measure it.
Proceeding then with the voltage stability question, an unknown we need to wrestle with. Once we know the maximum current draw for the way we will be using the DAC we will need to look at the best known voltage regulator typologies that could be used to meet our current needs. First thing we would look at would probably be whether op amp designs like ESS recommends can meet our current requirements. We can't do that yet, because we don't know the current. Once we do know the current we can tell for sure if AD727 will work. If so, great, we are almost done. Otherwise, we would probably look at more discrete level super regulator circuits and do some analysis to see if any can meet our needs or if not how close we can come to meeting our needs. We could also look at other types of circuits to see how they compare. Once we have out best candidates, the only way to know how they will work in our poorly defined application would be build them and test them. Since we don't have test equipment and numbers for voltage stability we will need to conduct listening tests using humans to get our results. This is another tricky area. Humans can be used to get good answers, but the hearing of individual humans can vary a lot, so we need to be very mindful of that at all times.
Personally, I would still build one or more test circuits that seem likely to provide great, not good, performance. If trying something relatively new like LDO RF regulators, i would be wary that there is not a lot of experience in the engineering world about using them for what we want to do. On the other hand, there is lots of existing knowledge and experience on making good op amp regulators similar to superregulators.
I mentioned before there is a service manual for an older version of m-dac out there. Anybody google and look at the AVCC supply? If no, why not? You lazy? Better do it, you need more information about what is likely to work. Or at least what one rather obvious type of candidate circuit would probably look like. An immediate question would then be how much complexity are we willing to consider? I don't know, but it doesn't really matter because I have no idea at this point what problem I am even trying to solve. First, I need, and you need, to know how much current you are going to need to need your circuit to supply. Are you starting to get the idea that somebody, anybody, please measure the freaking AVCC current so we can have something factual to actually talk about. Sorry, if I am am being unpleasant about this, but if you guys were my students or my junior engineers working on this I wouldn't let you do another thing until you come up with a prospective way to measure AVCC current and try it and come up with some numbers for us, or you, to proceed with. Then, you will be learning how to do something more like engineering and we will all be better off, and actually things will then get simpler. Alright, I will shut up and wait to see what you all want to do. Thank you for listening.
Here is the thing: you cannot design an AVCC supply simply by talking it over and weighing each other's thoughts and arguments. Engineers are not taught to work that way and for good reason. You must work with engineering models, math, numbers, and measurements. It isn't so much a social consensus type of activity.
If you want to design an AVCC supply for Q2M or for 9028PRO, you need to come up with a model, and some numbers that will be your design goals within that model. First thing to do is find out what ESS says is needed for AVCC. I have no idea what 9028PRO needs according to them. I will tell you they give a number for AVCC current for a Q2M in what they consider to be a more or less standard operating mode when used in something like a cell phone. However, for cell phone use ordinarily the DAC is run at less than full thottle, pedal-to-the-metal mode. They don't give a number for that mode. So, the options would then be, (1) ask them, or, (2) measure it. That would give an engineering number for current draw to work with.
For AVCC voltage supply stability, there is no spec. There is an example circuit that you have all seen and warnings about op amp and passive choice component selection for AVCC and IV stage use. That's all we have to work with and as engineers we would need to somehow decide on how good we need. That gets us into tricky territory because there is only limited research on what people can hear, and engineers have widely different views on the quality or reliability of that research. So, it is up to the designer(s) to deal with voltage stability question on their own given the information we have and in deciding if it is good enough to work with, or what. AVCC current though, is a given, it is what it is, you just have to measure it.
Proceeding then with the voltage stability question, an unknown we need to wrestle with. Once we know the maximum current draw for the way we will be using the DAC we will need to look at the best known voltage regulator typologies that could be used to meet our current needs. First thing we would look at would probably be whether op amp designs like ESS recommends can meet our current requirements. We can't do that yet, because we don't know the current. Once we do know the current we can tell for sure if AD727 will work. If so, great, we are almost done. Otherwise, we would probably look at more discrete level super regulator circuits and do some analysis to see if any can meet our needs or if not how close we can come to meeting our needs. We could also look at other types of circuits to see how they compare. Once we have out best candidates, the only way to know how they will work in our poorly defined application would be build them and test them. Since we don't have test equipment and numbers for voltage stability we will need to conduct listening tests using humans to get our results. This is another tricky area. Humans can be used to get good answers, but the hearing of individual humans can vary a lot, so we need to be very mindful of that at all times.
Personally, I would still build one or more test circuits that seem likely to provide great, not good, performance. If trying something relatively new like LDO RF regulators, i would be wary that there is not a lot of experience in the engineering world about using them for what we want to do. On the other hand, there is lots of existing knowledge and experience on making good op amp regulators similar to superregulators.
I mentioned before there is a service manual for an older version of m-dac out there. Anybody google and look at the AVCC supply? If no, why not? You lazy? Better do it, you need more information about what is likely to work. Or at least what one rather obvious type of candidate circuit would probably look like. An immediate question would then be how much complexity are we willing to consider? I don't know, but it doesn't really matter because I have no idea at this point what problem I am even trying to solve. First, I need, and you need, to know how much current you are going to need to need your circuit to supply. Are you starting to get the idea that somebody, anybody, please measure the freaking AVCC current so we can have something factual to actually talk about. Sorry, if I am am being unpleasant about this, but if you guys were my students or my junior engineers working on this I wouldn't let you do another thing until you come up with a prospective way to measure AVCC current and try it and come up with some numbers for us, or you, to proceed with. Then, you will be learning how to do something more like engineering and we will all be better off, and actually things will then get simpler. Alright, I will shut up and wait to see what you all want to do. Thank you for listening.
Last edited:
Mikett, I agree with you that it is very unlikely to see an I/V interface board at reasonable price. However, not all are lost. There are a few "Balanced to Unbalanced XLR-RCA Audio Converter Preamplifier PCB" for $3.66 on ebay that could be modified to do the job.
Low Distortion Balanced to Unbalanced XLR-RCA Audio Converter Preamplifier PCB 699937072503 | eBay
From what I heard from Markw4, the first dual IC should be laser trimmed SSM2141 instead of the NE5532. Other parts may need to be hand selected too.
A dual channel board using SSM2141 can be useful too. But the seller did not provide any schematic, which should not be too difficult to trace out.
SSM2141 Balanced input and unbalanced output Dual Channel Preamplifier Board 4857664135595 | eBay
I may take on this little project in the next few week. But I have to wait for my ES9038Q2M box to arrive first.
Last edited:
Rather than SSM2141, That THAT 1200 Series would probably quite it bit closer to the quality the DAC can provide.
http://www.thatcorp.com/datashts/dn133.pdf
There are other ones too like THAT 1240 Series that potentially could be used to make a differential stage. Does't provide the filtering the ESS recommended circuit does though.
SSM2141 is good in terms of CMRR but distortion is between .01% and .001% depending on output loading. That means good CMRR but worse distortion.
But while a 5532 is a better audio amp, it can't make a differential amp with high CMRR without precision resistors, presumably, thin film.
Best circuit hands down for a high quality DAC is the ESS circuit. Sorry.
http://www.thatcorp.com/datashts/dn133.pdf
There are other ones too like THAT 1240 Series that potentially could be used to make a differential stage. Does't provide the filtering the ESS recommended circuit does though.
SSM2141 is good in terms of CMRR but distortion is between .01% and .001% depending on output loading. That means good CMRR but worse distortion.
But while a 5532 is a better audio amp, it can't make a differential amp with high CMRR without precision resistors, presumably, thin film.
Best circuit hands down for a high quality DAC is the ESS circuit. Sorry.
Last edited:
Markw4, you are a difficult guy to keep up with. Fortunately in this case, the "THAT 1200P08-U" is available from Mouser for $6.10 in single quantity.
SSM2141PZ is $4.72 in single quantity also from Mouser. THAT 1200P08-U is not too big a jump in cost.
The big question is whether either of these balance line differential receiver can be used as is for the I/V interface input stage? If not, how can one modify them?
Last edited:
Sorry, I should take a more thorough look at the datasheets for every idea that people run by. Sometimes I feel like it is better to say something rather than nothing, but without checking everything. Please though, I would like to encourage everyone to look at the datasheets carefully for parts you are considering. Not just the first page, but learn how to read the graphs and don't skip a careful read of the application notes section. I know it is a lot to do, it is for me too, but its part of being a competent designer. Over time and if you ask questions you can learn how to understand everything in a datasheet. Rarely would it require knowing any calculus or high-level physics. It is something good technicians learn how to do. If you want to design its just part of the gig, and it can get you surprising far actually.
"This is the PCB of a very low distortion preamplifier that can accept balanced signal source. Within the board the balanced signal is converted to unbalanced output.
The circuit make uses of a famous balanced line receiver SSM2141. The chip can offer a distortion as low as 0.002% across the whole audio frequency band. This is followed by another opamp stage.The gain is also adjustable so as to complete a preamplifier function."
From the LME 49720 datasheet:
Key Specifications
– Power Supply Voltage Range: ±2.5 to ±17V
– THD+N (AV = 1, VOUT = 3VRMS, fIN = 1kHz):
– RL = 2kΩ: 0.00003% (typ)
– RL = 600Ω: 0.00003% (typ)
– Input Noise Density: 2.7nV/√Hz (typ)
– Slew Rate: ±20V/μs (typ)
– Gain Bandwidth Product: 55MHz (typ)
– Open Loop Gain (RL = 600Ω): 140dB (typ)
– Input Bias Current: 10nA (typ)
– Input Offset Voltage: 0.1mV (typ)
– DC Gain Linearity Error: 0.000009%
Why on earth would you use an ES9038Q2M with LME49720 IV stages and then run it though something with "as low as 0.002%" distortion when in fact you will probably not get all the way to the 'lowest' in your application?