Ideas wanted -- USB DAC product

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
FW/TBT require more than a few weeks or months of work. FW is dead, and nobody uses that anymore besides the nostalgic guys. TBT is essentially PCIe, so I need to develop a PCIe enabled gadget. PCIe enabled FPGAs are expensive, and I need to pay $4-digit amount of licensing fee to use the PCIe controller IPs, plus equally expensive membership fee to use TBT interface (TBT has a different protocol that encapsulates PCIe, so I need to use both technologies).

USB is royalty free with all documents available for free with no non-waived patents. The only fee to use USB is when you need a proper USB vendor ID, you buy that for $3500, but there are cheap sources VID/PID pairs banned by USB consortium, but the IDs still work and are still unique. Also, if I want to put USB compliant logo, I need USB membership, which is $2000 per year. In my case, I don't care about a USB logo, so I don't need to join USB-IF.

Overall, there's no fee for me to use USB besides the time invested on writing my own USB audio class 2.0 driver stack.
I've read they will open Thunderbolt 3 soon -in 2018 for more companies and with no royalties and it is the protocol of the future that will replace all other ports -but of course I understand it is your design / business choice taking into account several factors.


FWIW, my design uses a reversible USB type C, so even someone wants to plug it the wrong way around, it't not possible.
Finally! It took them 18 years to think of a reversible connector (1996-2014).
So in another 2 decades they will make it completely orientation free!
An externally hosted image should be here but it was not working when we last tested it.

Talk about dinosaur progress. They should have asked a public school class to help them.

560px-USB_Type-C_Receptacle_Pinout.svg.png



If anyone can't plug the USB plug on the computer side in the right orientation, then they are not my customer. This is a prosumer grade gear, not for some 16 year old kids wanting a boombox connected to their phones.
Well, first, for a universal standard designed to be used by anyone from 16y olds to grandmas in everyday life that should have been able to plug it blindly, they have failed in an epic way.

Second, I'm not 16 and not clueless either, and I bet all members here, you included have attempted to plug a USB only to find out you need to turn it to the other side to your frustration. Now expand that number to the billions of USB consumers in the whole world x the number of failed attempts since 1996 and compare that to 10 minutes of thinking for a single developer required to figure out the proper connector design. That was my point.



Yes, but there are technical complications. The DAC has a master digital volume control, so the signal applied to integrated HPA is already attenuated. If I want to implement independent volume control between master out and HPA, I have to add a PGA chip, and even that requires master output can't be totally muted. Also, PGA chips are noisier and less linear than my digital volume control, so anyway the headphone output will be crappy. If I want to have truly independent, lossless volume control, I need a second set of physical DAC, which almost doubles the cost.
No need for overkill measures. You could use the master digital volume with two memory settings: one for the main DAC output and one for the headphone amp, then switch to the proper setting when the user enters the headphones mode, with the push of a button and/or by inserting a headphone plug, and provide an elegant touch volume control eg with light feedback and you're done, with the minimum hassle.


I am also working on my own open source audio analyzer that focuses on detecting extremely low residual harmonics for THD and IMD characterization. DR (basically noise) is relatively easy to measure. The design will be in an open source license, so everyone can use it.
Interesting. I will need an audio analyzer for my electrostatic headphones amp project -either I'll build one, or use one, but my measuring requirements are beyond 1ppm (<120db).


However, I'm not going to compare measured data with competitors' products. It's considered very offensive and uneducated. Let's the reviewers do the comparison.
You don't need to do that in an offensive way. My point was to design a scientific (read: repeatable) test that will prove the highest quality distinguishable from a significantly lower one, in an official, public test. If one achieved this, he would contributed to the audio community with that test mehod (would break a record actually). Everything else beyond an independently verified scientific test, is details. There should be ways to avoid offending and damaging the competitive companies.
 
I mean, a PWM generator generating 10000000 doesn't have exactly 1/2 of output energy than the same PWM generator generating 11000000 pattern.

Therefore, you need to insert a zero between every symbols, basically to get 11000000, you need to generate 1010000000000000 to make sure each symbol returns to zero.

I doubt that, because the relation between the number of ones and the total charge or flux is still affine with a constant offset (linear except for a constant offset).

With patterns like 10000000, 11000000, 11100000, 11110000, 11111000, 11111100, 11111110, each first 1 has a slightly different weight than the other 1s due to the rising edge and each last 1 has a slightly different weight than the other 1s due to the falling edge, but as there is exactly one rising and one falling edge per eight-bit pattern, this only causes a constant offset.

A more problematic issue is the fact that the centre of the pulse is not always at the same place, that is, there is an unintended phase modulation. You can reduce the impact by letting the loop filter run on the PWM clock and feeding back each bit individually (so you don't feed back 5 into a loop filter running at fPWM/8, but 11111000 into a loop filter running at fPWM).

The following patent is remarkably readable and deals exactly with issues like this. It is 21 years old, so it must be about to expire. Besides, it describes many useful techniques that are not claimed (such as the one I just mentioned):

Peter G. Craven, Analogue and digital converters using pulse edge modulators with non-linearity error correction, US patent 5548286, 20 August 1996
 
Last edited:
Interesting. I will need an audio analyzer for my electrostatic headphones amp project -either I'll build one, or use one, but my measuring requirements are beyond 1ppm (<120db).

As I mentioned in a previous post, synchronous averaging will get down well below -120dB even with a computer sound card. Read this down near the end of it to see how (image below from page 10 of the document): http://www2.electron.frba.utn.edu.a...Documentos/Fundamentals_Modern_Audio_Meas.pdf
 

Attachments

  • DM.jpg
    DM.jpg
    63 KB · Views: 133
Last edited:
I don't understand, but did you mean he has an ADC to sample the output of DAC back and use that to correct his filter's tap coeffs?

No, I didn't mean that - that sounds like an adaptive filter. But his filter I understand has fixed coefficients.

I mean, a PWM generator generating 10000000 doesn't have exactly 1/2 of output energy than the same PWM generator generating 11000000 pattern.

Right but I would guess the very lowest and highest symbols won't get used, just those in the middle.

Therefore, you need to insert a zero between every symbols, basically to get 11000000, you need to generate 1010000000000000 to make sure each symbol returns to zero.

But for PWM the symbol is representative of not just single bits, its a multi-bit symbol. So you need zeros (bits) between symbols but not a 'zero symbol' between symbols. For say 4-bit PWM perhaps only 14 out of 16 possible symbols get used and maybe just two bit-times are used to minimize ISI.

So 000111111111111110 would be the most positive symbol and 000000000110000000 the most negative one. Here I'm using two bit times for symbol separation (18 bit times = 180nS symbol rate).


Can you drop me a link to his blog? I think I can learn a lot from his posts.

Mola Mola
 
So 000111111111111110 would be the most positive symbol and 000000000110000000 the most negative one. Here I'm using two bit times for symbol separation (18 bit times = 180nS symbol rate).

Those patterns are also mentioned but not claimed in the Peter Craven patent (which really looks more like a tutorial than like a patent). They much reduce distortion by keeping the centre of the pulse at a fixed place.
 
I doubt that, because the relation between the number of ones and the total charge or flux is still affine with a constant offset (linear except for a constant offset).<snip>

That's a good paper to read. Huge thanks. If my understanding is correct, it basically inserts at least 2 ones even for the lowest output code ISI isolation barrier whose transition energy becomes DC bias which cancels since it is a fully differential DAC). Then, valid data bits are embedded inside the 2 ISI isolation walls. Brilliant.
 
No, I didn't mean that - that sounds like an adaptive filter. But his filter I understand has fixed coefficients.

After seeing their picture I finally understood what did you mean by analog FIR. It's a low jitter 32-bit "SPU receiver" thing with analog coefficient programmed resistor per tap. I never thought of that -- that's indeed a good way of getting rid of the first stage of LPF opamp and to squeeze some extra THD reduction, besides to get rid of high bandwidth requirement of IVC opamp. As a side benefit, extra output levels grant higher DAC element SNR. That's a brilliant design with one stone hitting three birds.
 
I've read they will open Thunderbolt 3 soon -in 2018 for more companies and with no royalties and it is the protocol of the future that will replace all other ports -but of course I understand it is your design / business choice taking into account several factors.

Intel said to open the standard in 2012 since the first debut of TBT1. It's always the next year.

Finally! It took them 18 years to think of a reversible connector (1996-2014).
So in another 2 decades they will make it completely orientation free!
Talk about dinosaur progress.They should have asked a public school class to help them.

What do you mean by "orientation free"? Like a ball in a socket that can move and make contact at any degree? What's the point?

Well, first, for a universal standard designed to be used by anyone from 16y olds to grandmas in everyday life that should have been able to plug it blindly, they have failed in an epic way.

Second, I'm not 16 and not clueless either, and I bet all members here, you included have attempted to plug a USB only to find out you need to turn it to the other side to your frustration. Now expand that number to the billions of USB consumers in the whole world x the number of failed attempts since 1996 and compare that to 10 minutes of thinking for a single developer required to figure out the proper connector design. That was my point.

I've plugged USB-A ports blindly to Ethernet jacks. And I believe I'm not the only one. But it doesn't hurt anything as Ethernet port is not supposed to be energized (no, POE won't deliver power besides a weak, current limited 5V initial voltage before negotiation).

Just flip it if you tried to plug it the wrong way around and it will be all fine.

No need for overkill measures. You could use the master digital volume with two memory settings: one for the main DAC output and one for the headphone amp, then switch to the proper setting when the user enters the headphones mode, with the push of a button and/or by inserting a headphone plug, and provide an elegant touch volume control eg with light feedback and you're done, with the minimum hassle.

Will consider. I need some time working my spec sheet.

Interesting. I will need an audio analyzer for my electrostatic headphones amp project -either I'll build one, or use one, but my measuring requirements are beyond 1ppm (<120db).

Will update. Stay tuned!

You don't need to do that in an offensive way. My point was to design a scientific (read: repeatable) test that will prove the highest quality distinguishable from a significantly lower one, in an official, public test. If one achieved this, he would contributed to the audio community with that test mehod (would break a record actually). Everything else beyond an independently verified scientific test, is details. There should be ways to avoid offending and damaging the competitive companies.

Good point.
 
After seeing their picture I finally understood what did you mean by analog FIR. It's a low jitter 32-bit "SPU receiver" thing with analog coefficient programmed resistor per tap. I never thought of that -- that's indeed a good way of getting rid of the first stage of LPF opamp and to squeeze some extra THD reduction, besides to get rid of high bandwidth requirement of IVC opamp. As a side benefit, extra output levels grant higher DAC element SNR. That's a brilliant design with one stone hitting three birds.

I gather you are writing about a FIRDAC: digital delay line driving a bunch of DACs of which the outputs are added (with or without weighting). The main advantage of FIRDACs is reduced sensitivity to jitter. The patent on it has passed its due-by date, so you see FIRDACs all over the place nowadays.

Heinrich Pfeifer, Werner Reich and Ulrich Theus, Circuit arrangements for averaging signals during pulse-density D/A or A/D conversion, US patent 4947171, 7 August 1990
 
What do you mean by "orientation free"? Like a ball in a socket that can move and make contact at any degree? What's the point?
I meant cylindrical, with ring-like contacts, with a protocol that prevents short-circuit while inserting or pulling out by enabling the outputs only when the plug is fully inserted.

And you can make it completely zero-force, contact-free, (and water and dust/dirt proof) as it should be in our era, by replacing the ring contacts with coils that will transmit the data inductively, and provide a magnet at the tip that will hold the plug inside -which can hold or release the plug at will (even being pushed out) by a host tiny electromagnet.
We are already 6 decades ahead at the year 2077 -in dinosaur company years.:p.

That is, if we'll wait for those dinosaurs, instead of making an open, free to license protocol with intelligently made connectors and make all other rubbish "inventions" worth only to through in a trash bin. Because the technology to do this, exists today and also existed many years back.
 
Last edited:
The best way to lower the cost is to make the system modular...

This means that provide interfaces but attach cards!

Cards that provide additional functionalities like Ethernet, Bluetooth, headphone amplifier, DAC upgrade and others!

Think about this as a computer where I can add a graphics card, ram, Ethernet card, WIFI card, sound card!
 
Sorry for the loooong delay. I was working on the FIRDAC concept and I am still working on it. So far it seems very promising.

As for the connector based expansion card concept, I don't think this is going to work. USB interface is provided by DSP, and it is anyway needed for the normal operation of the system, as the DSP serves as interpolation engine. So the delta cost is zero.

Same for Ethernet -- anyway for device firmware update capability, I need an external CPU+ROM to load the DSP firmware and FPGA bitstream. A CPU+FLASH combination will be ~$2 minimum, and an ESP32 system on module (CPU+WiFi+Bluetooth+FLASH) is only $4.

And the same can be extended to WiFi and Ethernet -- they are provided on chip of ESP32, and the additional cost is only antenna cost.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.