• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

E1DA Cosmos ADC

it is a tech as is supposed to be, 9039Q2M may give you all harmonics under -150db@1kHz@accurate implementation. The LPF is for complete geeks to wipe harmonics even under -160db.

I guess that you're now pretty much at a noise limit. It's hard to get past the laws of thermodynamics.
 
@IVX,
Have you finalized the active low-pass filter design?
1731570237889.png
 
The second LPF version proto is tested. H2 performance(-150db is a good result for SE) and noise were improved for SE outputs. Now SE loses less than 1db of the original 9039S THD+N, -125.2db/125.8db.

9039S->LPF at the Left SE output 4.6Vrms, APU +6db, that test shows the THD+N(<-128db) performance for SE applications.

2024-11-21_11-09-54.jpg


The same test, but APU 0db, to show the THD performance of SE applications.

2024-11-21_11-14-02.jpg


9039S->SE Left 3.5V APU 0db, THD test. Ultra-high performance balance/unbalance converter, due to highly symmetrical structure and delicate PCB layout full 9039S performance translated to the unbalanced output with virtually no losses. BTW, 100mA outputs may drive headphones.


2024-11-21_11-27-25.jpg


The same test but APU 6db, THD+N performance test.


2024-11-21_11-28-13.jpg


9039S->SE Right 3.5V APU 0db, THD test.

2024-11-21_11-24-07.jpg


The same test but APU 6db, THD+N performance test.

2024-11-21_11-25-02.jpg



9039S->LPF balanced output 4.6Vrms, THD performance test, close to -155db, all harmonics about -160db or less.


2024-11-21_11-37-25.jpg



9039S->LPF balanced output 7.75Vrms, THD+N performance test, less than -131db(limited by APU).


2024-11-21_11-40-33.jpg
 
Perfect for a balanced, however, it works fine OK with unbalanced cheap DACs(I saw it for $13) too, but one channel is inverted to emulate balanced output.
Meizu Hi-Fi DAC has lots of noise and weird "harmonics" around the fundamental frequency at Fs >=96k, the best choice is 44.1k. Anyway, the THD performance lets you calibrate Cosmos ADC with $13 DAC + $50 LPF board.


2024-11-21_15-36-49.jpg



2024-11-21_15-39-48.jpg
 
Personally, I'd like this to be available in a case.

My reason is that often when testing a preamp or amp, the device under test isn't entirely enclosed in a case. There could be tools and other test gear on the workbench, too. I'd prefer not having a DMM test probe pop off and fly into the filter, damaging it. That's where the probe would fly to, of course.

This is why my original ask was for a Cosmos DAC in a box. It could be the same circuitry as a #9039S, but with just one change - a larger output resistance (more than zero ;)) like maybe 10 Ohms so that if the test cable connected to it somehow shorts, there'd be no damage to the DAC.

I understand why you may not want to offer such a product, and my plan is to buy a #9039S to go along with the filter board. Mostly, I want to protect the output of the #9039S from workbench accidents. Not that I've had any of those...

Even if an empty APU case is available to buy, it would be easy enough for me to have a new front panel made that would match the LPF board. A panel from Front Panel Express would probably cost as much as the LPF board, but better to spend the money for protection than for replacing a damaged LPF board.

The two advantages of an AP test set and even the QuantAsylum product are that they are in one box with less chance of these problems and that the ADC and DAC use a common clock, which can have some advantages sometimes. Maybe. There are probably software advantages in production use, but I don't find that to be a thing for engineering or DIY use.
 
Last edited:
BTW, I even considered taking a #9039S out of its case and mounting it on my own LPF board. Based on what you did, and the photo of the LPF board, it looks like it would fit. Yeah, I'd be reinventing the LPF and may not get the same results as you did, but it might be better on a lab bench. Less chance for an oh-shit moment.

Of course, it's probably possible to cut and paste the pcb layout for the #9039S onto the LPF board you designed and put it into one box... ;)
 
I am about to order the Cosmos ADCiso. This will replace my EMU 0404 USB which died after serving my measurement needs for 16 years. Considered repairing it if possible but decided not to bother.

I used the 0404 USB mainly for DIY speaker measurements, DAC/amp testing, and occasional stereo music recording.

I wanted to check with advanced users here before hitting the order button. Do I need both the APU and Scaler as well? For acoustic measurements I need a mic pre (w/ two channels for occasional stereo recording). But I don't think I need extremely low noise floor or low-level HD measurement ability for speaker DIY (let alone two APUs for stereo recording). The EMU (which was the SOTA at the time) was more than sufficient. Not sure I will need the APU's notch filter, either. So, I think I can pass the APU and get a more practical two-channel mic preamp.

Now, about the Scaler. I think this is worth owning because I need an input buffer---and gain settings should add flexibility---to measure some audio gear with relatively high output impedance.

Am I on the right track? Is there anything I miss here?
 
Last edited:
Do I need both the APU and Scaler as well?
The scaler is convenient for dealing with the E1DA input impedance and correcting the roll off you might get when testing something capacitor preamp. For 1 kHz SINAD, may not matter since the 10V sensitivity mode is still really clean.

The APU helps you get even cleaner measurements but it’s just a AES notch in my opinion. I didn’t buy it.

I bought mine from Audiophonics assuming that the minijack to XLR cables from Amazon would be fine, but the ports are pretty tightly spaced on the Scaler, so the factory cables are nice.

So yeah, I think you’re on the right track. I am looking forward to seeing your measurements!
 
The two advantages of an AP test set and even the QuantAsylum product are that they are in one box with less chance of these problems and that the ADC and DAC use a common clock, which can have some advantages sometimes.
Well, for the AP, DAC / Dig.Gen and ADC do not run on the same clock, the clocks are completely independent. That did cause a lot of head scratching some time ago.

As for significant advantages, IMHO it's the rugged front-end and the auto-ranging. It is almost impossible to fry the input and you'll always in a proper range even for stepped measurements that require range-switching.
And of course the core of the analyzer, the variable notch filter with automated frequency tracking, is the main asset.
 
Well, for the AP, DAC / Dig.Gen and ADC do not run on the same clock, the clocks are completely independent. That did cause a lot of head scratching some time ago.

Is that right, wow. I just assumed they would have a master clock, divided down to whatever D/A, A/D settings you selected. So you can have different clocks for the D/A and A/D?

The QA-403 has one clock, sadly it can't do 44.1/88.2 etc, but it does 48/96/192 and 384 (just for the ADC).

As a low (relatively) cost analyzer, it is truly amazing, especially with fully calibrated ranges, overload protection, proper BNCs, etc. I just bought a second one for a portable test rig.

IMG_4299.jpg
 
I am about to order the Cosmos ADCiso. This will replace my EMU 0404 USB which died after serving my measurement needs for 16 years. Considered repairing it if possible but decided not to bother.

I used the 0404 USB mainly for DIY speaker measurements, DAC/amp testing, and occasional stereo music recording.

I wanted to check with advanced users here before hitting the order button. Do I need both the APU and Scaler as well? For acoustic measurements I need a mic pre (w/ two channels for occasional stereo recording). But I don't think I need extremely low noise floor or low-level HD measurement ability for speaker DIY (let alone two APUs for stereo recording). The EMU (which was the SOTA at the time) was more than sufficient. Not sure I will need the APU's notch filter, either. So, I think I can pass the APU and get a more practical two-channel mic preamp.

Now, about the Scaler. I think this is worth owning because I need an input buffer---and gain settings should add flexibility---to measure some audio gear with relatively high output impedance.

Am I on the right track? Is there anything I miss here?

I think that the E1DA adc + scaler couple is
a good configuration for microphone recordings.

Of course you won't have 48 volts if your microphones
require it.

I made an electret microphone interface (lavalier microphone)
and it works very well. (sensitivity -36db /pascal)

Adding a scaler allows you to have an input impedance of
100Khoms in balanced mode, the ADC having an impedance too low
for microphones, you will have to provide an XLR cable -> 3.5 jack

In stereo you can have a maximum gain of 26db on each channel.

In mono you will have up to 52db of gain by putting
the 2 channels in series with a Jack3.5->jack3.5 male cable
which corresponds to a relatively honest "sound" interface.

In terms of noise you will always be better than a good
microphone, so don't worry about that.
 
Is that right, wow. I just assumed they would have a master clock, divided down to whatever D/A, A/D settings you selected. So you can have different clocks for the D/A and A/D?
Yep, the ADC for the analog analyzer is completely independent from any sample rate setting for digital outputs (internal if SPDIF/AES3/I2S etc, external for USB).
1737031058738.png
 
Is it that the effective clock sampling rates for the converters are independent or that there are separate physical clocks?

For example, the DAC might run at 48 KHz but the ADC at 96 KHz, but both might be derived from the same oscillator through the use of different dividers for each. That would keep the relative rate of any drift somewhat constant. Two different oscillators might drift differently, which could cause the FFT bins to not stay in alignment over time.

This is just academic for me, since I am never going to purchase an AP test set. It's just simply not appropriate for what I do here at home. But, it's interesting to know so that I don't make a fool of myself in the future by suggesting that the AP uses a common clock. I'll have to find a different way to look foolish. Shouldn't be hard...
 
Is it that the effective clock sampling rates for the converters are independent or that there are separate physical clocks?
I can only say that the high rates (624kHz and multiples) are not synced to 48kHz and multiples, even though the integer factor (13) would seem to suggest that.
The standard rates for the ADC may or may not be synced to the internal clock which also is the source for SPDIF etc.
 
Back
Top Bottom