Low-distortion Audio-range Oscillator

The minidsp uses the Sigma Studio family of DSP chips. They are good and the digital pathway is 28/56 bit processing supporting double precision but the adc and dac are 100 dB class. You should not expect more for $7 or so. A good premium DAC.is better than that.

Do you really need continuous tuning at -150 dB distortion? Maybe 4 or 5 frequencies would be enough. A seperate dedicated box with source and notch connected to an analyzer would be relatively achievable and would avoid some of the difficult stuff.

Sent from my LG-H811 using Tapatalk

Indeed, spot frequencies is the way to go for these kinds of deep measurements. At minimum, 55Hz, 1kHz, 20kHz. 5kHz and 10kHz would be the next 2 I would add to get a feel for rise vs frequency. I realize this is not all of a 1-2-5 sequence. the 55Hz number is to sneak in between the harmonics of 50Hz or 60Hz. I believe it is important not to overlook LF THD.

Cheers,
Bob
 
The minidsp uses the Sigma Studio family of DSP chips. They are good and the digital pathway is 28/56 bit processing supporting double precision but the adc and dac are 100 dB class. You should not expect more for $7 or so. A good premium DAC.is better than that.

Do you really need continuous tuning at -150 dB distortion? Maybe 4 or 5 frequencies would be enough. A seperate dedicated box with source and notch connected to an analyzer would be relatively achievable and would avoid some of the difficult stuff.

Sent from my LG-H811 using Tapatalk

The comes a point where the continuous tuning is close to equal the cost of 10 frequencies.
If sota parts are used. Quality relays etc. The micro control is the least expensive part.
Software doesn't wear out or require cleaning.
 
I was thinking if the miniDSP was that good, we could write some software for the DSP to generate test signals, sending them out through that -130dB DAC and get rid of all the other boatankers. 😉
Probably wishful thinking. ...

Jan

AFAIK there are no DAC's at any price with -130dB max spurious tones or even -120db at 0dB out. The ESS DAC's claim -120dB with several ultra-low distortion op-amps external, its own voltage out only -108dB.

The miniDSP HD has -103db DNR on the data sheet.
 
Last edited:
This is the premium -HD model. Not using same as the other models.


THx-RNMarsh

All the Sigma Studio compatible parts seem to be HD compatible with 24/192 capable cores. I just went through all the associated parts here: SigmaStudio | Analog Devices none were capable of the performance you show. Can you look inside your box to see what is in it? I have used these in projects and the are quite good but not SOTA ADC's or DAC's. They have DSP only devices that can be used with external ADC's and DAC's. Maybe that's in yours.
 
All the Sigma Studio compatible parts seem to be HD compatible with 24/192 capable cores. I just went through all the associated parts here: SigmaStudio | Analog Devices none were capable of the performance you show. Can you look inside your box to see what is in it? I have used these in projects and the are quite good but not SOTA ADC's or DAC's. They have DSP only devices that can be used with external ADC's and DAC's. Maybe that's in yours.

I think it's a Crystal part but can't read any of the pics the Shark has no converters. There is a teardown of the HD version with more channels somewhere which definitely is an 8 channel Crystal chip.
 
AFAIK there are no DAC's at any price with -130dB max spurious tones or even -120db at 0dB out. The ESS DAC's claim -120dB with several ultra-low distortion op-amps external, its own voltage out only -108dB.

The miniDSP HD has -103db DNR on the data sheet.

I don't think any chip company has the chutzpah to lie to that degree (-130dB). They are too busy making other promises they can't keep. I did get the -130 dB distortion products from an AK4490 but only under ideal conditions and they would not guarantee that or anything close.
 
I think it's a Crystal part but can't read any of the pics. There is a teardown of the HD version with more channels somewhere which definitely is an 8 channel Crystal chip.

Its been a few years but I had a long talk with the principles and they told me then it was based on the ADI stuff. It may have changed since. Sigma Studio is a really good basis for a dsp system and lends itself to this stuff well. Unfortunately the tools are not openly available so commercial products need to recreate them.
 
I found a good picture of the current HD model on the minidsp site. It seems to have this chip: AK4626AVQ | Product | AKM - Asahi Kasei Microdevices Good but not SOTA and the ADC seems to be 96K max. Its a good value for a 2 in 8 out chip. Datasheet claims "The AK4626A has a dynamic range of 102dB for ADC, 106dB for DAC " "- S/(N+D): 90dB (dac)" and "- S/(N+D): 92dB (adc)" That's a long way from -130 dB.
 
I like Sigma Studio but its not for the typical hobbyist in current form. We need to get the FAE to get it running for us the first time and we just fiddle around the edges mostly. This is true for ADI, TI, AKM, Intersil-D2 etc. The China clone chips are the worst for the software. Mostly needing Windows XP to work.
 
does this concept have potential as an approach? What could you do to improve its performance?


figure_01.gif




-RNM
 
does this concept have potential as an approach? What could you to to improve its performance?

The distortion of these IC's would limit performance. I don't see any filtering approach that would work except passive L/C switched with relays. Even then the layout of the air core inductors would need to be very careful. If you want silly -150dB numbers there are no IC's that will do it.
 
We have to be clear about the output impedance requirement. Demian suggests better then 600ohms down to 50ohms. Does this mean it must operate with a 50 ohms load? We can get a virtual zero ohms from any amplifier and attenuate the input instead. Noise is higher this way but it is doable.
 
A 50R system is easier to keep quiet once the signal leaves the output jack. I imagine that it would also reinforce good practices when making connections or routing signals. Just think of all the 50R lab equipment you could use if it is characterized down to baseband, or DC. If it can be done reasonably, I think it would be beneficial.

-Chris