For more than 2 years I have been working on and off (mostly off) designing a high-end Audio Analyzer. Parts of the design have been implemented as separate units. I have finally put the pieces together on one PCB (+ some small existing interface PCB's).
Some of the design goals were:
1. The Audio Analyzer should work as a sound card, enabling use with a wide selection of Audio Analyzer SW packages.
2. The performance should be somewhere between a good high-end sound card and an AP, R&S, SR etc.
3. The performance should approach the limits possible with the ADC and DAC used.
4. It should be possible to use it with any relatively modern PC. This is the reason for choosing a USB interface to the PC. SPDIF/Toslink was considered, but this makes it difficult to use portable PC's.
5. The PC should be electrically isolated from the Analyzer to avoid ground loops and noise.
6. No fan for cooling!
I have selected to use high-end converters from AKM. I have previously made a design based on AK5394A and AK4399. This design shows a very good performance, but it was not ideal in terms of flexibility, due to limitations in the input and output stages. The input stage was a relatively straight-forward design based on OPA1612 and OPA1632. The performance is fairly good, but there is no attenuator, so the input range is limited.
The output stage has a limited drive capability, since it is based on the OPA1632.
For this design I originally chose the newer AKM converters AK5397 for the ADC and AK4490 for the DAC. After testing the AK5397 I decided to go back to the AK5394A, since the distortion is much lower.
The input stages of the new design are based on a JFET/op-amp combo, with LSK389 and LME49990. This part of the design has been verified on a separate PCB, with attenuator and gain options.
The output stages are based on LME49990 buffered by LME49600.
All these parts of the design could be found in kynix.
1. AKM AK5394A 24-bit Super High Performance Stereo ADC, 192 kHz sampling, 123 dB SNR (changed from AK5397)
2. AKM AK4490 32-bit Premium Stereo DAC, 192 kHz sampling, 123 dB SNR
3. Dual fully differential inputs and outputs on XLR connectors
4. Dual single ended outputs on BNC
5. AC and DC input switchable
6. 44.1, 48, 88.2, 96, 176.4 and 192 kHz sampling, selected by the Analyzer SW. Potentially 384 and 768 kHz sampling later on (TBD).
7. Full scale inputs from 100 mV (-20 dBV) to 100 Vrms (+40 dBV) in 10 dB steps.
8. Overvoltage protection on the inputs.
9. Full scale outputs up to 10 Vrms differentially (5 Vrms single ended). Can be attenuated to 1 Vrms and 100 mVrms full scale differential output.
10. Single ended output impedance of 50 ohm (changed from 25 ohm)
11. Single ended input impedance of 100 kohm
12. On-board low jitter crystal oscillators (NDK NZ2520SD).
13. Parallel control lines for relays, controlling the attenuators, gains and AC/DC settings.
14. Power supply is integrated on the board (except for the transformer).
15. USB interface is implemented using a completely isolated XMOS based design (changed from USBStreamer).
16. Expansion connector for future digital I/O
17. 20x25 cm PCB
The parallel control lines may be controlled from either manual switches, e.g. rotary switches for the level settings. Other possibilities are to control it with a microcontroller or from the PC via a USB to serial converter. Initially I plan to use manual switches. Settings can be controlled from the PC via the USB interface.
For the power supply a 2 x 15 V and 1 x 9 V transformer is needed. I plan to use an R-core transformer with 2x15 V and 2x9 V, 30 VA, which I have also used for the AK5394A/AK4399 design.
The transformer may be changed to a 2x15V or 2x18V ring core transformer with a slightly modified power supply design. Update 161222: A ring core transformer is used. Primary voltages selectable between 100 to 120V and 220V to 240V.
The isolation between the Audio Analyzer and PC is done with a Master Clock and Isolator for the MiniDSP USBStreamer. The isolation is now part of the XMOS based USB interface.
I have just received the PCB yesterday. It will take some time before I will be able to start testing, since I plan to mount the PCB by hand. And I still need to get some of the parts. There are around 750 parts in the design.