If I'm don't remember to wrong, you also added a decoupling capacitor across the shunt.... Forgot that?? I didn't 😀
Anyway... I have some years of experience using OrCAD, and know how to set up the simulations. And we get the results shown in this thread. Funny thing is, that when we measure on the PCB, we also get this result. So I guess it's about time you put your faulty simulations some place where the sun don't shine 😱 And move on....
You guys are about as subtle as a knee in the crutch! You are antagonistic and cannot take any criticism and you infer that TPA are dishonest! That to me says it all. You have been given advice by many and all you do is throw it back in the donor's faces.
Bad form!🙄
I have nothing to prove to you. We have a successful commercial product, and we would like to protect our design for a little longer. As you claim to know, the key is in the details. I just wish you would stop claiming speculation as fact.
I don't forget that easily. OK, I'll repeat once more: I have recreated your circuit exactly as shown in your schematic.
Listen carefully: your exact circuit I simulated with Orcad. Different results than what you claimed. FYI, that capacitor in or out does not change absolutely anything substantial. This claim should also be embarrassing.
You've measured your real circuit at such output impedance at 10MHz? You should definitely check your instruments. This is not science, this is circus.
Edit: and because I'm a nice guy, here's what you can do to improve your regulator. Deal with the self noise; you're an engineer, it should be obvious how. Your circuit as implemented should show upper range audio frequency noise of around 500-800uV. To lower and extend the bandwidth of your output impedance, use active loads on the two driver bjts. Even then it will have the performance you claim your original one has, but it will be much closer to that.
Listen carefully: your exact circuit I simulated with Orcad. Different results than what you claimed. FYI, that capacitor in or out does not change absolutely anything substantial. This claim should also be embarrassing.
You've measured your real circuit at such output impedance at 10MHz? You should definitely check your instruments. This is not science, this is circus.
Edit: and because I'm a nice guy, here's what you can do to improve your regulator. Deal with the self noise; you're an engineer, it should be obvious how. Your circuit as implemented should show upper range audio frequency noise of around 500-800uV. To lower and extend the bandwidth of your output impedance, use active loads on the two driver bjts. Even then it will have the performance you claim your original one has, but it will be much closer to that.
Last edited:
This debate is kind of silly because the TPA DAC is readily available whereas the other one is made of unobtanium.
Let´s get back on the ground!
It takes much much longer time and effort to design NFB discrete circuits, than just using op-amps. Op-amps are fixed circuits with certain requirements to function as intended. You can influince them by changing the properties of their feedback loop, their supply voltages and their loads. Normally the vendor knows excactly how this influinces the op-amp performance, as this is the most important data, any customer might need. So you just go on from there.
At this time, we decided to do a second DAC, which will be a lot more simple to assemble, and it will be featuring op-amps, and 32 bit upsampling as well as 32 bit conversion. Just we do not bother listen a lot to it, simply because we cannot make it right r at least influence it a lot anyway. All important decissions were already taken by the vendor of the op-amp. But so be it.
Though you will have the opportunity to choose between any single op-amp you´d like to use, and the AD844 NFB op-amp.
But anyway, this project will be exciting in only one way, namely its 32 bits. The rest is just as boring as the Buffalo, except the supllies are a bit more ambitious, and we´ll try to settle for just one opamp pr. ch.
Btw. it will eventually be featuring CS8416, CS8421 and AK4399 @ a price a bit more reasonable than most kits.
It takes much much longer time and effort to design NFB discrete circuits, than just using op-amps. Op-amps are fixed circuits with certain requirements to function as intended. You can influince them by changing the properties of their feedback loop, their supply voltages and their loads. Normally the vendor knows excactly how this influinces the op-amp performance, as this is the most important data, any customer might need. So you just go on from there.
At this time, we decided to do a second DAC, which will be a lot more simple to assemble, and it will be featuring op-amps, and 32 bit upsampling as well as 32 bit conversion. Just we do not bother listen a lot to it, simply because we cannot make it right r at least influence it a lot anyway. All important decissions were already taken by the vendor of the op-amp. But so be it.
Though you will have the opportunity to choose between any single op-amp you´d like to use, and the AD844 NFB op-amp.
But anyway, this project will be exciting in only one way, namely its 32 bits. The rest is just as boring as the Buffalo, except the supllies are a bit more ambitious, and we´ll try to settle for just one opamp pr. ch.
Btw. it will eventually be featuring CS8416, CS8421 and AK4399 @ a price a bit more reasonable than most kits.
Last edited:
Thank you very much for your advice.
But I think we´ll stick with our own design.
It both simulates, works and measures perfectly, even on different computers and oscilloscopes. The components you´ve added unfortunately makes our design oscillate.
Of course, easily fixed, but since you're not interested, we'll leave it at that.
It would be interesting to expose a naked DAC directly to vibrations or to
massive sound waves and see if anything pops up anywhere in the circuit.
I somehow doubt that, except maybe for tubes or very large caps.
Besides all the delta sigma and switched cap filter stuff inside the CS4398,
quote from datasheet page 23, 4.7.5 Interpolation Filter:
These filters have been designed to accommodate a variety of musical tastes and styles.
In other words, the DAC Chip itself fails to reproduce music correctly 🙂
So much for the ultimate DAC...
quote from datasheet page 23, 4.7.5 Interpolation Filter:
These filters have been designed to accommodate a variety of musical tastes and styles.
In other words, the DAC Chip itself fails to reproduce music correctly 🙂
So much for the ultimate DAC...
Last edited:
I for one do not care a bit about the BOM price for the Buffalo. IMO the rapid implementation of the ESS chip by Russ and Brian, the excellent sonics of the result, and the great support provided is well worth whatever they might earn.
Hardcore diy'ers, or diy'ers on a budget, may rightfully think different but then again one cannot say than Hurtig and KvK's DAC provides any kind of viable alternative for those.
Nic
Ahh!
The variable coifficients are in your mind.
Every DAC available by now includes these variable coifficient filters invented originally by Sony arround 1997 I think.
So what is the news actually?
We chose the sharp roll of type of anti aliasing filter @ 24 bits and 96 KHz, which means anti aliasing above 96 KHz.
A lot of other opertunities are available in the Crystal chip, as they also are in every other chip available incl. BB, AKM, WM, ESS and AD.
So what are you actually trying to incinuate?
The CS4398 is in fact a state of the art chip unsurpassed by any other U out chip so far.
Well spoken mr. NicMac!
But I and others still consider the Buffalo a left hand implementaion of whatever chip you might choose.
It is a standard by the book implementation of a DAC chip with the usual op-amps to follow.
The DAC chip I do not know, just rumours say that the data are achieved by very heavy oversampling and paralleling, but the op-amps I do know, and they are mediocre all together.
Btw. Soon you will se how the Buffalo´s schematics are done, and then you can see for youself, if it is hot or not.
Until then this thread is about our DAC based on CS8416, AD1896/SRC4192 and CS4398 followed by a beautyfull analog stage way out of reach of any op-amp in this world.
Last edited:
umm I dont get what you mean? BTW , the switch cap filter is designed such way that the phase shift caused is corrected beforehand with the digital filter, hehe. These chips are just too widespread to ignore. That cheap e-mu 1212m soundcard comparison in the other thread Is missing the point, as the 4398 also found its way into professional interfaces like pro tools digidesign192, along with some DDS clock generation. Thats the stuff for those studios who cant afford EMM Labs or Lavry,there are many hehe.
Last edited:
let me quote Lavry:
"Multibit sigma delta design was a problem a few years ago. Now it is a good thing. PCM has a lot to offer, it can beat sigma delta if one is willing to deal with a lot of difficulties in the design, and a much higher cost of manufacturing including a lot of "by hand" calibration... "
___________________________
even, 1bit has a lot to offer..........
"Multibit sigma delta design was a problem a few years ago. Now it is a good thing. PCM has a lot to offer, it can beat sigma delta if one is willing to deal with a lot of difficulties in the design, and a much higher cost of manufacturing including a lot of "by hand" calibration... "
___________________________
even, 1bit has a lot to offer..........
Mr KvK,
unlike you, I don't rely on my eyes to judge a DAC - I rely on my ears🙂
I will stop here as I am in the risk zone of becoming just as bad an advocate as you for an otherwise open-minded and civilized country like Denmark.
Enjoy your DAC. As they say in Naples: "Ogni scarrafone è bello a mamma soia", meaning something like everything is beautiful to its creator.
Nic
We will both enjoy our DAC and we will publish the schemitics of the Buffalo on this site when back engineered, I don´t get the "The Buffalo DAC must remain indisclosed" idea, I was told that this is a DIY forum, where people share their designs.
But It is kind of weird, when we did not want to publish unfinished schematics for our own DAC, all hell broke loose, the same seems to be the case, if the Buffalo will be disclosed.
@Tritosine!
As you surely know, one cannot judge any chip on where you find it. All of them can be bought at low prices if you buy huge quantities.
The WM8741 has been seen @ less than 7$ sold through a dealer 10K.
The BB1794A is about 10$ 1K
AD1955 is about 7$ 0,5K
CS4398 is about 3,43£ = 5$ 0,5K
BB1795 brand new 32 bit DAC is just 4$
Non of these are very expensive chips, and all of them might end up anywhere.
The only chips that really are expensive and probably will remain to be so, are the R2R ladder DACs such as 1704K, which will cost more than 10 times as much as a 1794A, incl. a digital filter, that makes around 100$ for one DF1706 and 4 pcs. of 1704K if you buy 1K units - new world record, I think.
As you surely know, one cannot judge any chip on where you find it. All of them can be bought at low prices if you buy huge quantities.
The WM8741 has been seen @ less than 7$ sold through a dealer 10K.
The BB1794A is about 10$ 1K
AD1955 is about 7$ 0,5K
CS4398 is about 3,43£ = 5$ 0,5K
BB1795 brand new 32 bit DAC is just 4$
Non of these are very expensive chips, and all of them might end up anywhere.
The only chips that really are expensive and probably will remain to be so, are the R2R ladder DACs such as 1704K, which will cost more than 10 times as much as a 1794A, incl. a digital filter, that makes around 100$ for one DF1706 and 4 pcs. of 1704K if you buy 1K units - new world record, I think.
Last edited:
- Status
- Not open for further replies.