What's inside a Prism Dream AD-2 converter

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
living sounds said:
"Discrete" as in resistor ladders (or other single components) - or do you mean 4 AD ICs per channel?

I'm not sure but I know there were no AD IC's that gave anywhere near that performance in the early 90's.
 
I wonder if they rolled a custom A/D converter using an FPGA? That might be what's under the heatsink....Might explain the "discrete" aspect too IE: They designed it out of discrete gates in the FPGA  ;D

 
iampoor1 said:
I wonder if they rolled a custom A/D converter using an FPGA? That might be what's under the heatsink....Might explain the "discrete" aspect too IE: They designed it out of discrete gates in the FPGA  ;D

There's not much room under these heatsinks. The ICs are really small.
 
Currently AKM's AK5578EN (8 ch A/D) for an example can be configured to parallel it's all eight converters for 130 dB SNR (112 dB THD). You will still need two chips for 2-channel conversion to get that SNR.

https://www.akm.com/akm/en/product/datasheet1/?partno=AK5578EN
 
mhelin said:
Currently AKM's AK5578EN (8 ch A/D) for an example can be configured to parallel it's all eight converters for 130 dB SNR (112 dB THD). You will still need two chips for 2-channel conversion to get that SNR.

https://www.akm.com/akm/en/product/datasheet1/?partno=AK5578EN

Yes, today you can get even better performance from a SAR chip. I'm still waiting for a commercial product for pro audio.

But the AD-2 came to market in 1998.
 
mhelin said:
Currently AKM's AK5578EN (8 ch A/D) for an example can be configured to parallel it's all eight converters for 130 dB SNR (112 dB THD). You will still need two chips for 2-channel conversion to get that SNR.

https://www.akm.com/akm/en/product/datasheet1/?partno=AK5578EN
Yup... 3dB for every doubling so 1-2 is 3dB, 2-4 is 6db, 4-8 is 9dB... not nothing but they would need to go to 16 for the next 3 dB.

JR

[edit] another strategy available to increase resolution as convertors get faster and faster is more over sampling.  If the convertor is fast enough that you can capture multiple samples within a single output sample interval, those samples can be averaged together to improve resolution following a similar rule (3 dB per doubling). I have used this inside digital meters with modest A/D resolution. Since I didn't care about aliasing (in a meter data stream) I combined together multiple samples at lower than sampling theory wants for frequency integrity to realize more meter resolution.  [/edit]
 
crap because it does Not sound better than others....

the New Prism Atlas are much better.

other interesting are the Lavry Gold, very interesting, has temperature controlled self calibrating resistor network, its a true resistor network AD converter...
but the clock is crap, has FIFO a memory buffer used as clock source, with 60ns / 80ns jitter. "too much."
also has too much latency, and warm up time. 15~30minutes.
the lavry blue bad, the black are ok.

i have apogee rosetta800 192k, the AD is very natural, dynamic, but the DA feels weird, lacks bass, and sounds less dynamic...
sometimes i want to modify it, some times i just want to sell.

the latest Focusrite RED 8Pre are very interesting they have dual parallel paths, to phase add signal, and phase cancel distortion...
because analog signal is balanced before AD converters, or After DA...

see differential signal & twisted pair technology..

The UA Apollo, classic 5532 sound, Not my thing/taste.
 
Juan,

This thread is about technology, not about what we think "is crap" or "is interesting" - that sort of discussion is much more interesting to take to gearslutz or somewhere like that, where majority of users would buy into the idea that everyone can differentiate between modern converters (in a blind test)..

Jakob E.

 
I think there is a (two or more) stage system involved. With a high frequency (18k) test tone input I can see distortion products in the midrange rise linearly while increasing the test tone amplitude. At a certain point of increase, the distortion suddenly drops. This is predictable and reproduceable.
 
gyraf said:
...where majority of users would buy into the idea that everyone can differentiate between modern converters (in a blind test)..

Jakob E.

Right?! I can't even tell if that electric bass went thru a blue stripe or if it was a blackface, in the mix. Neither spot if the vocals did or didn't go thru the pultec set on "bypass". Shitty damn ears.
 
"The mixpre3 uses the smaller 6 channel AK5576EN in a 6-to-3 mode giving it better dynamic range.  "
https://jwsoundgroup.net/index.php?/topic/33053-zoom-f6-a-32bit-recorder/&page=2
 
mhelin said:
"The mixpre3 uses the smaller 6 channel AK5576EN in a 6-to-3 mode giving it better dynamic range.  "
https://jwsoundgroup.net/index.php?/topic/33053-zoom-f6-a-32bit-recorder/&page=2

Zoom has a seemingly similar 32-bit setup in one of their new portable recorders. The Sound Devices one is patented. Wonder what's original enough to get a patent for it. And who they'll hit with it, of course.
 
cyrano said:
Zoom has a seemingly similar 32-bit setup in one of their new portable recorders. The Sound Devices one is patented. Wonder what's original enough to get a patent for it. And who they'll hit with it, of course.
Patents are published so you should be able to read what they claim...

JR
 
This kind of patent is usually heavily obfuscated, if it is already publicly available at all...
 
cyrano said:
This kind of patent is usually heavily obfuscated, if it is already publicly available at all...
By law the patent is supposed to show the "preferred embodiment" or best way. While I have never heard of this happening, obvious obfuscation could be used to deny an application from ever winning a patent.

That said companies are motivated to not give away the entire farm, but they "must" describe adequately how the invention works.

JR
 
Since I sold mine I will finally reveal what is inside: A couple (one for each channel) Cirrus Logic 24 bit delta sigma AD converter chip with a big heatsink on top bolted to the chassis top.

I made a foto but cannot find it, but it was the top of the line chip at the time (1998?). One per channel, so the gain ranging wasn't done by switching between converter chips.

The noise performance was quite spectacular, the bass response (thanks to a DC servo) very good, but the overall sound wasn't as good as I had hoped. The high end sounded kind of harsh/fizzy (for a 10k $ converter) and the dynamics were not preserved perfectly. I found it had a certain "sound" I know from many recordings that I really don't care for. Not missing it at all.

At 44.1 khz the rather steep filter cut most everything above 20khz, a little more radical than many other modern ADs.
 
Back
Top