EMU0204 mods for FFT measurements

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
EMU 0204 right channel mod

I have arrived at R values to scale 0dB to the FS of the AK4385B.

The ADC FS is 2.9Vpp differential.

I give some details of performing the mod in another post.

Here are some schematics and results.

The first sch is the original. The second sch is the AKM recommended and the third sch is the mod.
The rest is self explanatory.

Cheers,
 

Attachments

  • Emu orig sch.GIF
    Emu orig sch.GIF
    15.4 KB · Views: 1,071
  • 0dB mod 96k ns.GIF
    0dB mod 96k ns.GIF
    77.4 KB · Views: 473
  • 0dB mod 96k.GIF
    0dB mod 96k.GIF
    82 KB · Views: 477
  • 0db FS mod 48k.GIF
    0db FS mod 48k.GIF
    76.6 KB · Views: 492
  • 0db FS mod 44 1k ns.GIF
    0db FS mod 44 1k ns.GIF
    76.3 KB · Views: 919
  • 0db FS mod 44 1k.GIF
    0db FS mod 44 1k.GIF
    77.3 KB · Views: 961
  • Emu MOD sch.GIF
    Emu MOD sch.GIF
    19.1 KB · Views: 1,050
  • AKM input.GIF
    AKM input.GIF
    26.1 KB · Views: 1,032
As an alternative you can input 1Vrms and bring the gain control up to clipping.
Reduce the input level to-10dB and calibrate ARTA with 316mV.

But the differential voltages to the input won't be balanced.

This mod is a bit cleaner though.

This is the alternative.
 

Attachments

  • Altern.GIF
    Altern.GIF
    77.3 KB · Views: 422
Last edited:
Member
Joined 2004
Paid Member
I was confused at first but I believe what you are doing is making sure the differential inputs are equal at the DAC. this would give the largest dynamic rage on it. The lowest distortion is usually less than full input, typically -10 dB or so.

The AKD5394A uses almost the same input circuit and the docs are more complete (see below). Its not necessarily the best but it will give perfect balance at the chip.

For general purpose audio I find 1V is not enough. A preamp needs to be able to drive 3V or more to get full output from most power amps. The interface board project is intended to address this limit.

I have an EMU Tracker Pre, 0202, 0404 and Juli@. The Juli@ has a much different circuit so scaling is a different task, especially since it has different circuits for balanced and unbalanced inputs and JRC4580 opamps in a unique footprint. I'll open the other boxes up and post closeups of the circuits. I got the 0404 because it uses the AK5394A, but its not optimized for performance it seems. I am having a terrible time getting the EMU stuff to work in Win 7 64 bit. Its just funky. . .
 

Attachments

  • AKM Input circuit.JPG
    AKM Input circuit.JPG
    71.8 KB · Views: 843
I was confused at first but I believe what you are doing is making sure the differential inputs are equal at the DAC. this would give the largest dynamic rage on it. The lowest distortion is usually less than full input, typically -10 dB or so.

The AKD5394A uses almost the same input circuit and the docs are more complete (see below). Its not necessarily the best but it will give perfect balance at the chip.

For general purpose audio I find 1V is not enough. A preamp needs to be able to drive 3V or more to get full output from most power amps. The interface board project is intended to address this limit.

I have an EMU Tracker Pre, 0202, 0404 and Juli@. The Juli@ has a much different circuit so scaling is a different task, especially since it has different circuits for balanced and unbalanced inputs and JRC4580 opamps in a unique footprint. I'll open the other boxes up and post closeups of the circuits. I got the 0404 because it uses the AK5394A, but its not optimized for performance it seems. I am having a terrible time getting the EMU stuff to work in Win 7 64 bit. Its just funky. . .

Hi Demian,

This same circuit is on Ak5385B data sheet. I didn't show it because I only addressed the right unbalanced channel.

The 1V is just our instrument maximum input level. What difference does it make if the input has to be attenuated to fit the range. It's a relative measure and quite proportional.
Having to attenuate gives a better SNR.

We do need what you have offered for a front end to handle the scaling for higher voltage inputs.

The magic is not in the mod it's the principle of scaling the entire ADC range for the measurement range and calibrating for that. It drives the relative noise floor down.
Then we can afford to add a front end system which will surely drive the noise floor up.

The maximum input level into the input amp on the 339a is 1Vrms that's full scale.
 
Does this mod invert absolite phase for this channel? It looks like it would.

It does invert the signal. This shouldn't be a problem unless the two channel are summed physically. The emu would be useless for music applications at this point.
An alternative is to leave the input op amp in the non inverting mode, ground the added resistor and adjust the R value for this mode.

I prefer the inverting amp.

Cheers,
 
Member
Joined 2004
Paid Member
I got the EMU 0404 working. It starts with a reformatting of the hard disk and complete re-install of Win 7 with no Dell drivers, and ended with the laptop's battery self destructing. . . I'll grab a quick shot of its performance in the morning as is. I think the gain structure is limiting the noise floor. I'll know more soon. It seems the residual distortions are in the -115 dB to -120 dB in loopback. If the software works then its worth making an effort to get the analog performance optimized.
 
I haven't looked at the left channel. It resembles what the 0404 looks like from the outside.
I bet they're the same input circuit. Whatever EMU used for a balance input.

EMU likes to direct couple each stage of there inputs. This make it difficult to minimize the number of op amps required. The driver bias voltage appears at the non inverting input.
This forces the preceding stage to operate a bit in class A. The output bias voltage is dependent on having the source resistor of the driver tied to a low Z ground reference source. Other wise we would not get the 2.5V bias required at the inputs of the ADC.
 
Member
Joined 2004
Paid Member
The datasheet for the AKD5384a has most everything you need on it. The bigger challenge is programming the settings for the chip for different sample rates. It can be done with a multipole switch or a micro if you have the skills. Or you can use the hardware here: http://www.diyaudio.com/forums/digital-source/185761-open-source-usb-interface-audio-widget.html which already has support for the AK53954A. The ADC code is already done so its just a PCB. The analog side does not need to be complex. I could be up for contributing if someone else is ready to do the bulk of the work.
 
The datasheet for the AKD5384a has most everything you need on it. The bigger challenge is programming the settings for the chip for different sample rates. It can be done with a multipole switch or a micro if you have the skills. Or you can use the hardware here: http://www.diyaudio.com/forums/digital-source/185761-open-source-usb-interface-audio-widget.html which already has support for the AK53954A. The ADC code is already done so its just a PCB. The analog side does not need to be complex. I could be up for contributing if someone else is ready to do the bulk of the work.

Hi Demian,

Put it up I'll do it.

I've been looking into this anyway.

I do want try the parallel ADC technique. The summing can be done with serial adders and the division by a shift padded with a zero in a serial shift register. Doing this with discrete logic would be a real chore and DSP would be over kill. There must be a happy medium???
 
The datasheet for the AKD5384a has most everything you need on it. The bigger challenge is programming the settings for the chip for different sample rates. It can be done with a multipole switch or a micro if you have the skills. Or you can use the hardware here: http://www.diyaudio.com/forums/digital-source/185761-open-source-usb-interface-audio-widget.html which already has support for the AK53954A. The ADC code is already done so its just a PCB. The analog side does not need to be complex. I could be up for contributing if someone else is ready to do the bulk of the work.

I've done this with other codecs. It can be done with a header and shunt jumpers.
This way it's expandable to uP by replacing the shunts with a header connector and jumping to a different board.

The simpler the analog end is the better. I don't like long chains of op amp in the signal chain. I think the ADC should be integrated with the front end design and this is very doable.

We just need someone here who's really good at programming software. It's not my area of expertise. I can pull off the imbedded stuff okay but the windows programming not so well.

I don't know how well Visual Studio ports to Win7 when compiled in XP. The drivers are not the same.

Does anyone here know how to write drivers for Windows?
 
Last edited:
There was a great project a little while ago, here:

http://www.diyaudio.com/forums/equi...onverter-project-audio-measurements-tool.html

And, of course, for those who don't want the complications, there's the pcm4222evm, which, for $150 gives you some amazing performance; see post 4 here:

http://www.diyaudio.com/forums/equi...ject-audio-measurements-tool.html#post2072409

Thanks.

That DIY ADC does not have USB or other computer interface.

Yeah, I have the pcm4222evm, but it is a real hassle to install in a box. One set of connectors is not designed to be mounted in a box because they are recessed too far on the board. Unsoldering the connectors lead me to damage the board. So, while it may perform admirably, it's a hassle, and the board could be much smaller also. It also does not have holes in it for standoffs, but rubber feet instead. So, I either have to drill the board myself (scary) or use hot melt glue to hold it in place. I'm just not that excited about the latter and the former would make me worry that I'm further damaging the board. I don't want to rig up some kind of clamp that others have done to hold the board in place. The board has a lot of options that are probably not needed for audio ADC use (why two clocks?), and seems spread out, like it was designed by an amateur (though I doubt that). Obviously, I'm annoyed by this board....

Basically, what I'm looking for is a high performance ADC with balanced/unbalanced inputs, and a computer interface output (USB, etc.). It's not complicated, but sure is a pain to find as a DIY kit or board. I've got an I2S->USB interface board, so no problem there interfacing the TI evaluation board with a computer, but I'd obviously prefer a more integrated solution.

I don't mean to hijack this thread... sorry!
 
Last edited:
Hi Demian,

Put it up I'll do it.

I've been looking into this anyway.

I do want try the parallel ADC technique. The summing can be done with serial adders and the division by a shift padded with a zero in a serial shift register. Doing this with discrete logic would be a real chore and DSP would be over kill. There must be a happy medium???

I'm certainly willing to buy what ever board you make, as long as it has the interfaces I need. I've gotten a lot better at soldering SMD's now... they take forever though.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.