E1DA Cosmos ADC

[David_M]

Maybe this has been answered before but does @AVX plan to make an industrial version of this ADC/DAC/Filter system to present a genuine rival to the AP $50K plus system?

 

[Salt]

IIRC @IVX said NO.

 

[David_M]

IIRC @IVX said NO.

That's very unfortunate. Thanks!

 

[KSTR]

Maybe this has been answered before but does @AVX plan to make an industrial version of this ADC/DAC/Filter system to present a genuine rival to the AP $50K plus system?

You understand the difference between a simple ADC and a full blown test station? The ADC comprises maybe like 1% of the cost and effort of an APx machine (or any other full-featured audio analyzer).
It's like the difference between, say, an injection system and the complete F1 car.

 

[BKDad]

You understand the difference between a simple ADC and a full blown test station? The ADC comprises maybe like 1% of the cost and effort of an APx machine (or any other full-featured audio analyzer).
It's like the difference between, say, an injection system and the complete F1 car.

Well, yes and no.

Imagine a Cosmos ADCiso, Scaler, and 9039S all in the same box. Use larger connectors (maybe), add a couple relays to the Scaler so that it would attenuate in addition to gain to facilitate power amplifier measurements, add build out resistors to the 9039S outputs to reduce the possibility of "oopses", and use an external power supply. This would then be similar to the basic AP product. No tracking filter, of course, but fixed filters like the Cosmos APU could be added either internally or externally if desired. It wouldn't have digital outputs and inputs, if those are needed, and wouldn't have the very high sample rates of the AP, but it would be quite serviceable and useful.

A big thing would be to rely on externally build software, like REW and Multitone, rather than building his own. The software is a giant part of what makes the AP unit the price it is.

BUT... Ivan has mentioned here or elsewhere that the APU, which is purely for making high quality measurements, hasn't sold well at all. So, that could be a signal that the market isn't there for a modestly priced AP type unit for hobbyists and repair shops. He certainly knows his business better than anybody - WAY better than I could imagine - so I understand why there isn't a Cosmos AP in his web store. All we can do is ask and hope.

 

[David_M]

You understand the difference between a simple ADC and a full blown test station? The ADC comprises maybe like 1% of the cost and effort of an APx machine (or any other full-featured audio analyzer).
It's like the difference between, say, an injection system and the complete F1 car.

Wow...yes, I know the difference between the two but thank you nonetheless! A superb ADC/DAC system is the heart of a precision audio test system. Add some powerful hardware DSP and / with sophisticated software on a powerful PC platform and you have a system that's almost ready to go. Add extras H/W and S/W features to distinguish yourself from the competition and you are ready to go.

I designed Automated Test Systems (ATE) for years so yes, I do know what goes into a full-blown test station, but thanks for reminding me anyways. Peace!

 

[David_M]

Well, yes and no.

Imagine a Cosmos ADCiso, Scaler, and 9039S all in the same box. Use larger connectors (maybe), add a couple relays to the Scaler so that it would attenuate in addition to gain to facilitate power amplifier measurements, add build out resistors to the 9039S outputs to reduce the possibility of "oopses", and use an external power supply. This would then be similar to the basic AP product. No tracking filter, of course, but fixed filters like the Cosmos APU could be added either internally or externally if desired. It wouldn't have digital outputs and inputs, if those are needed, and wouldn't have the very high sample rates of the AP, but it would be quite serviceable and useful.

A big thing would be to rely on externally build software, like REW and Multitone, rather than building his own. The software is a giant part of what makes the AP unit the price it is.

BUT... Ivan has mentioned here or elsewhere that the APU, which is purely for making high quality measurements, hasn't sold well at all. So, that could be a signal that the market isn't there for a modestly priced AP type unit for hobbyists and repair shops. He certainly knows his business better than anybody - WAY better than I could imagine - so I understand why there isn't a Cosmos AP in his web store. All we can do is ask and hope.

A 3rd party could use his modules as the foundation of a full-blown test system, and add more HW and SW features (if AVX is willing to license his code or simply write their own) and you'll have a test system with a BOM cost well under $10K...then some serious 3X profit and you could have a rival system for under $30K.

 

[Sokel]

A QuantAsylum on steroids is not the worst idea...
At about $2k would be a poor man's AP.No software,no nothing,only decent drivers and functional.
What's the alternatives in the middle?

 

[KSTR]

Imagine a Cosmos ADCiso, Scaler, and 9039S all in the same box. Use larger connectors (maybe), add a couple relays to the Scaler so that it would attenuate in addition to gain to facilitate power amplifier measurements, add build out resistors to the 9039S outputs to reduce the possibility of "oopses", and use an external power supply. This would then be similar to the basic AP product. No tracking filter, of course, but fixed filters like the Cosmos APU could be added either internally or externally if desired. It wouldn't have digital outputs and inputs, if those are needed, and wouldn't have the very high sample rates of the AP, but it would be quite serviceable and useful.

A big thing would be to rely on externally build software, like REW and Multitone, rather than building his own. The software is a giant part of what makes the AP unit the price it is.

Well, anyone interested in audio-testing and having the skills *) can do this already (and many people already do it, me included), use the Cosmos building blocks and add your own hardware (safety features, connectors, routing, helper circuits etc) in a box around those blocks. Then use REW as the software (or Multitone).

*) If you don't, I see no reason why lay-people should deal with sophisticated audio testing anyway, you have to fully understand what you are doing, why you are doing it and in general, what's the background of all this. If you don't understand the background, leave it to the pro's.

 

[Bamboszek]

Actually there are many options already.

Apart from obvious Audio Precision stuff:
- RTX6001 which started as similar DIY project
- QuantAsylum QA403
- Stanford Research SR1
- Spectral dScope Series, previously known as Prism Sound
- Keysight U8903B
- Rohde & Schwarz UPV
- NTI FX100
- Audiomatica CLIO
- Shibasoku AD725 with impressive THD capabilities to this day
- Panasonic VP-7721A
- Chinese AP System Two clones
- PicoScope oscilloscopes with 16 bit mode
and probably many more.

Lots of positions at this list is already EOL. Guess why?

There is no market for high performance audio analyzers anymore.
No big companies are interested in R&D of audio products. Nothing better than 0.01% or even 0.1% THD is usually considered during designing of current products.
Nowadays HIFI is niche market, and objective audio performance focused customer is niche of a niche.
So every audio analyzer better than -100 dB is an overkill.

For hobbyists 150 USD audio interface is usually more than sufficient. Small businesses will buy slightly more expensive interface or QuantAsylum. Medium companies can lease Audio Precision or send devices for external testing. Handful of big corporations still somehow involved in audio can afford industry standard AP and don't care about everything else.

For manufacturing end of line testing you don't need that kind of performance, easy automation for simple OK/NOK is most important aspect.

Unfortunately, relatively ultra small community focused on reaching astronomically low THD numbers is not an sustainable target for business.

For me it's an engineering exercise and hobby to design circuits and measure such low level distortions.
From customer point of view, transparency is reached much, much earlier than you want to measure with more advanced analyzers.
As @KSTR pointed, I don't see anyone else from non-pro community who should be interested in such kind of product.

 

[MaxwellsEq]

Lindos LA100

 

[BKDad]

Well, anyone interested in audio-testing and having the skills *) can do this already (and many people already do it, me included), use the Cosmos building blocks and add your own hardware (safety features, connectors, routing, helper circuits etc) in a box around those blocks. Then use REW as the software (or Multitone).

*) If you don't, I see no reason why lay-people should deal with sophisticated audio testing anyway, you have to fully understand what you are doing, why you are doing it and in general, what's the background of all this. If you don't understand the background, leave it to the pro's.

Well, I've done all that, too. But, having a bunch of boxes spread across the workbench isn't good test procedure because there's variability every time you reconnect things.

There are hobbyists out there (like me) who actually do care about getting really exact measurements of the other gear they build. I fully admit that the market for such test equipment is small. Probably QuantAsylum has that market all to itself. The rest of the gear you mention is already obsolete or not reasonable for purchase by most hobbyists. Obsolete isn't always viable because of the difficulty in maintaining the equipment. That can become a hobby or chore unto itself.

I guess I should just shut up, design a box, have the box fabricated, remove the various building blocks from their cases, and be done with it. Use the time I've spent reading forums to build my own.

 

[KSTR]

Well, I've done all that, too. But, having a bunch of boxes spread across the workbench isn't good test procedure because there's variability every time you reconnect things.

I fear these kind of issues are still present even when you have sophisticated pro equipment. I'm a happy owner of an AP SYS2322 and found you always have to implement baseline/sanity checks and best practice is to measure some known DUT to see whether the setup is still consistent and stable.

But for 99% of my typical measurement tasks on line-level stuff and the occasional power amp an ADI-2 Pro with some accessories (USB isolator, low leakage power supply, attenuator etc) does the job.
Recently I've added the Cosmos ADC and APU which are a nice addition but not essential.

 

[BKDad]

No AP here.

Most measurements here are made with a Cosmos ADCiso, Topping E50, Cosmos APU as needed, Cosmos Scaler (for preamps) and Didden Autoranger (for power amps). Plus a similar array of USB battery supplies, USB isolators, and so on. I also have a QA401 and QA480 for sanity checks. They cover half the workbench when they're not in the closet. In all, they probably average about 40 hours of use each year. I'm ok with that level of investment and space for a hobby. Not so for an AP system. The annual AP software fee is more than I have invested.

 

[HaveMeterWillTravel]

@IVX kindly sent me a sample of the new Cosmos ADC to try out. Here are my impressions of it so far.


The Cosmos ADC does have one drawback in that the input impedance is relatively low, ranging from 640 Ω to 3.5 kΩ (measured at 1 kHz) depending on the the sensitivity setting. This may be too low for some sources.

That's all I have for now. I'll update if I think of any more relevant tests to perform. Suggestions are of course welcome.

I know this is an old thread and I did browse through it but I may have missed a discussion on it's input impedance.

How well established (e.g. measurement verified) that there is no appreciable change in performance (of the device being tested, not the E1DA) between a 640 Ω load impedance and a more typical load of say 10k to 100k Ω.

As an example, if this ADC is being used to measure an RCA output of a DAC or pre-amp. Typically it would be loaded by the next device's input impedance, say an amplifier.

How do we know that the performance of the DAC or pre-amp does not change and thus, the measurement isn't representative of the actual performance driving the amp without the E1DA connected? Has it been quantified on at least a few devices? After all, we know amplifiers typically will measure different with a 4 or 8 Ω load. That is only a doubling (2x) of load resistance. 640 to 100k Ω is ~156x.

It seems like one should always use their Cosmos Scaler except for when measuring equipment which will be driving a low impedance device.

In any event, the performance out of such a small and inexpensive device would be almost unimaginable just a short time ago.

HMWT

 

[restorer-john]

Many devices will not like a 640R load and will produce more distortion than they would when unloaded. A typical input impedance for an analyzer is 100k/200k.

That said, there are plenty of devices that will be quite happy into 640R and not suffer. But you will have to take into account the voltage drop the Cosmos causes as it becomes a series resistance with the output impedance of the device.

Measure a device for THD with and without the scaler. Obviously, noise will vary (be worse with the scaler). Pick something with a highish output impedance and see what the effects are. I find for the small penalty in absolute noise, it's better to have a consistent input impedance across all ranges than constantly needing to make calculations. That's where a dedicated analyzer is much better.

 

[jkim]

I've been playing with the ADC and scaler for a couple of weeks now. Thoroughly impressed.

I see a couple of posts above about the input impedance of the ADC. Interestingly I was going to comment on that topic, too.

Although I see why one wants to have high input impedance on testing equipment---i.e., to cover all kinds of DUT by providing essentially no load---, I do notice a disadvantage. Such high input impedance causes interconnect cables to be vulnerable to noise! Of course, in most cases the noise is way below audible levels. But these days we hobbyists measure these things not just from audibility concerns. We measure devices whose noise floors are extremely low.

For example, today I found that my very short (2 ft) Canare L-4E6S star quad balanced cables, used to connect a DAC to the Cosmos Scaler, was very finicky with their position (literally posture) relative to other cables. The effects were clearly shown on measurements---not a beautiful picture I'd like to see! Took some time to troubleshoot and finally find the culprit: the scaler's 200 kOhm input impedance. That made sense.

For testing modern devices, 10k - 20k Ohm should be high enough. I wish the scaler had selectable input impedance

 

[oal]

It seems like one should always use their Cosmos Scaler except for when measuring equipment which will be driving a low impedance device.

I think so. As restorer-john said, many devices are fine with an impedance of 640 ohms. However, some devices are not. The Scaler or other input buffer is required to guarantee measurement results.

Let me show you an example. This is a measurement of an ES9023-based I2S DAC that I bought for about $10, with the Scaler set to 0dB between the DAC and the Cosmos ADC. The input range of the Cosmos ADC was set to 4.5 Vrms (1.66k ohms). The vertical axis is voltage. The signal is 1.8Vrms, THD is -95.3dB, SINAD is 92.7dB. They are good for $10, and will sound fine when connected to an amplifier.

However, if you remove the Scaler, you get the following results. The voltage drops to 1.5Vrms, but not only that, harmonic distortion increases: THD is -51.5dB and SINAD is 51.5dB. If the input range of the Cosmos ADC is set to 2.7Vrms (1k ohms) or 1.7Vrms (640 ohms), the distortion will increase further. The input impedance of my amplifier is 47k ohms, so this level of distortion does not occur.

 

[johny_2000]

W

Let me show you an example. This is a measurement of an ES9023-based I2S DAC that I bought for about $10, with the Scaler set to 0dB between the DAC and the Cosmos ADC. The input range of the Cosmos ADC was set to 4.5 Vrms (1.66k ohms). The vertical axis is voltage. The signal is 1.8Vrms, THD is -95.3dB, SINAD is 92.7dB. They are good for $10, and will sound fine when connected to an amplifier.

Was this an unbalanced connection?
If so, based on the Scaler's 200/100kOhm (bal/unbal) input impedance specs, your DAC's output was loaded with 100kOhm.

 

[oal]

Was this an unbalanced connection?
If so, based on the Scaler's 200/100kOhm (bal/unbal) input impedance specs, your DAC's output was loaded with 100kOhm.

Yes, it was an unbalanced connection, so the input impedance of the Scaler to the DAC was 100k ohms as you pointed out. The output impedance of the Scaler was 10 ohms and the input impedance of the ADC was 1.66k ohms.