Line In amplifier to E1DA Cosmos ADC

[sonic tac]

I ordered the E1DA Cosmos ADC thinking it was the best interface in the world with very low distortions.

But I didn't know my source has output impedance of 3.5 KiloOhm.

The following is the dip switch underneath the E1DA Cosmos ADC where you can set the input impedance based on the expected voltage.

"
1.7Vrms 640Ω,
2.7Vrms 1kΩ,
3.5Vrms 1.3kΩ,
4.5Vrms 1.66kΩ,
6.7Vrms 2.46kΩ,
7.6Vrms 2.82kΩ,
8.5Vrms 3.12kΩ,
10Vrms 3.48kΩ,
43Vrms 13.6kΩ
Note: Unbalanced impedance about 30% less."

My source signal is only around 0.5Volts. If I'll set it to 1.7Vrms 640 Ohm. And my source output impedance is 3.5 KiloOhm. The input would be attenuated by over 5 times.

What if I'll just amplify the voltage to make it reach 10Vrms and make use of the 3.48 KOhm setting? What Line In amplifier that can amplify the signal without introducing distortions or artifacts and what exactly are the artifacts etc produced?

Whatever. Any artifact can be ok as long as it won't make it worse than the Behringer UMC204HD or M-Track Solo.

Thanks.

[Hugh Robjohns]

I think you've just found the interface between geekery and practicality.

The designer eliberately omitted a proper input buffer stage to avoid any theoretical distortion it might introduce... but in doing so he's provides input conditions which are virtually unusable in the real world with normal consumer equipment.

The maximum input impedance of 3.6k Ohms (for a 10Vrms /+22dBu signal) is really too low for any real-world equipment, and while I've no idea what your source is, an output impedance of 3.5k Ohms is ridiculously high too, making things even worse.

So, yes, you need an input buffer/impedance converter that will provide an input impedance of around 50k Ohms, and an output impedance of around 100 Ohms. That has to be an active device of some sort.

If you're into DIY electronics it would be simple to build one, either from scratch or using a kit board that you'd have to box and provide a power supply. Unfortunately, I'm not aware of a suitable standalone commercial line buffer.

Monitor controllers or hifi preamps would do the job, but to match the performance of the ADC you'd be looking at serious money.

Alternatively, send the ADC back and buy an interface that is usable in the real world. Many high-end interfaces use ESS Sabre converters...

[James Perrett]

The Behringer HA400 is actually 8 buffer amps in a box. It is unbalanced though and, as a headphone amp, I feel it degrades the sound very slightly.

As Hugh says, if you aren't into electronics you would be better off sending it back and buying something more suited for the job.

[sonic tac]

I think you've just found the interface between geekery and practicality.

The designer eliberately omitted a proper input buffer stage to avoid any theoretical distortion it might introduce... but in doing so he's provides input conditions which are virtually unusable in the real world with normal consumer equipment.

The maximum input impedance of 3.6k Ohms (for a 10Vrms /+22dBu signal) is really too low for any real-world equipment, and while I've no idea what your source is, an output impedance of 3.5k Ohms is ridiculously high too, making things even worse.

So, yes, you need an input buffer/impedance converter that will provide an input impedance of around 50k Ohms, and an output impedance of around 100 Ohms. That has to be an active device of some sort.

If you're into DIY electronics it would be simple to build one, either from scratch or using a kit board that you'd have to box and provide a power supply. Unfortunately, I'm not aware of a suitable standalone commercial line buffer.

Monitor controllers or hifi preamps would do the job, but to match the performance of the ADC you'd be looking at serious money.

Alternatively, send the ADC back and buy an interface that is usable in the real world. Many high-end interfaces use ESS Sabre converters...

I may keep the E1DA Cosmos ADC for other uses like recording audio. I once had the Turtle Beach Multisound. It has flat frequency response like the E1DA. So what thing to plug in the E1DA to get Multisound like recording?

For my source. I think I can get the M-Track Solo. It has 1 MegaOhm input impedance. But I think the gain is only for the microphone. What thing like it has gain also for the Line-In? With 1 MegaOhm input impedance and line level gain. I think it should be able to amplify 0.5 volts signal at 3.6kOhm output impedance to be recognizable, isn't it?

[James Perrett]

If you are happy with the performance of the Multisound then a cheap (but not the cheapest) Behringer interface will give you better performance in most respects. OK, if you are doing seismic research then the Multisound wins because it will record down to DC, but the Behringer would be more practical for recording because you don't want something that goes down to DC for normal audio recording.

[sonic tac]

I'm studying the patterns of amplifier noises as it interacts with ambient power line frequency field. The 0.3 volt signal are noises. I consider it as line in since line is has sub 1 volt voltage while microphone is in the millivolts. And is it not a power amplifier is supposed to amplify the line in. Why can't I just use a power amplifier to amplify the 0.3 volts into 3 volts? What is usually the output impedance of power amplifier?

I bought the E1DA Cosmos ADC thinking it may be able to distinguish amplifier noises better. I can't return it anymore. I'd receive it in one week. But if the E1DA has so low distortions. Even if the input are just millivolts. Won't it be able to display it property without minimal distortions?

[James Perrett]

This sounds like an engineering problem where an oscilloscope or wide band spectrum analyser would be more appropriate. When testing amplifiers you need to consider much more than the audio range. The amplifier could be oscillating or putting out noise at a much higher frequency than your audio ADC can measure. There are a few famous cases where amplifiers were thought to sound different but, in reality, the amplifier was unstable and oscillating at high frequency.

[Hugh Robjohns]

I'm studying the patterns of amplifier noises as it interacts with ambient power line frequency field.

I appreciate there may be a language mismatch here, but I don't understand what you're trying to do. What amplifiers? What noise patterns? And how is this 'ambient power line frequency field' involved?

The 0.3 volt signal are noises.

That's a heck of a lot of noise! That's almost -10dBV, the standard operating level of semi-pro audio equipment!

And is it not a power amplifier is supposed to amplify the line in.[/quote]

Yes, if you want tens or hundreds of Watts of output! A standard preamp also accepts a line input and provides line level outputs...

Why can't I just use a power amplifier to amplify the 0.3 volts into 3 volts?

Because its not the right tool for the job. Power amps usually introduce 25-30dB of voltage gain which you don't need. A simple line preamp would be more appropriate and introduce less noise and distortion too.

Won't it be able to display it property without minimal distortions?

I'm sure you'll get something usable from it. The issue is that with such a high output impedance and ultra-low input impedance, the small signal level you're starting with will be much smaller by the time it gets into the ADC. Moreover, the output circuitry of your source msy struggle to delivery sufficient current into the low impedance, potentially causing distortion.

[sonic tac]

This sounds like an engineering problem where an oscilloscope or wide band spectrum analyser would be more appropriate. When testing amplifiers you need to consider much more than the audio range. The amplifier could be oscillating or putting out noise at a much higher frequency than your audio ADC can measure. There are a few famous cases where amplifiers were thought to sound different but, in reality, the amplifier was unstable and oscillating at high frequency.

I need a spectrum analyzer that can analyze audio frequencies to study all kind of artifacts and noises that appear like audio signal. .

Unfortunately. The main problem is most hardware oscilloscopes start at 9 kHz and up. Only very old scopes have audio frequency or below 20 kHz and these are not only very slow but also very expensive sold used at ebay.

There is almost no wide band spectrum analyzer that can cover the audio band since it would be extremely expensive costing more than the Cray supercomputer.

The cheapest solution is a sound card or audio interface with software spectrum analyzer. And the best of them so far is the E1DA Cosmos ADC.

Mr. Robjohns. A preamp can amplify Line In too? In the Behringer UMC204HD or M-Track Solo. There is a gain control for the Microphone/Line In. The gain control can work for the Line In too? Or does one need to press the microphone button even if the input is Line In to make the gain control work?
If it will work, then maybe I can connect the Behringer or M-Track Solo outputs to the input of the E1DA Cosmos ADC? It will be delivered soon and unreturnable. I'll be stuck with it forever.

[Hugh Robjohns]

The cheapest solution is a sound card or audio interface with software spectrum analyzer. And the best of them so far is the E1DA Cosmos ADC.

I can accept that a hardware audio spectrum analyser is expensive (and rare), and digitising the signal to use a software analyser is a very pragmatic option. However, there are plenty of USB audio interfaces that would suffice and be easier to use.

A preamp can amplify Line In too?

Yes, of course.

In the Behringer UMC204HD or M-Track Solo. There is a gain control for the Microphone/Line In. The gain control can work for the Line In too?

Yes.

[sonic tac]

The cheapest solution is a sound card or audio interface with software spectrum analyzer. And the best of them so far is the E1DA Cosmos ADC.

I can accept that a hardware audio spectrum analyser is expensive (and rare), and digitising the signal to use a software analyser is a very pragmatic option. However, there are plenty of USB audio interfaces that would suffice and be easier to use.

A preamp can amplify Line In too?

Yes, of course.

In the Behringer UMC204HD or M-Track Solo. There is a gain control for the Microphone/Line In. The gain control can work for the Line In too?

Yes.

Let's say I'm detecting something similar to seismic wave and it is in 0.20 Volts. I don't want any amplification which can introduce distortions (via THD). The A1DA Cosmos ADC and a spectrum analyzer software can see the 0.2 volts frequencies clearer than amplying them to Line level 1.5 Volts, right?

About a preamp. A preamp is said to amplify 0.001 volts to 0.1 volts. What if the signal is 0.25 volts. Would the preamp only amplify the 1 volt values and clip above it? If not. You mean even if the voltage is 1 Volt. The preamp can amplify it to 2 volts? What is the rule or typical?

I'm thinking if the A1DA Cosmos APU is what I need too.

[Hugh Robjohns]

I don't want any amplification which can introduce distortions (via THD).

All amplification and all converters introduce distortion. It's just a question of how much, and what level is acceptable for your project.

The A1DA Cosmos ADC and a spectrum analyzer software can see the 0.2 volts frequencies clearer than amplying them to Line level 1.5 Volts, right?

I don't know. For minimum distortion you need the signal to be close to peak level in the ADC. Is there an input level setting that provides sufficient sensitivity while still presenting a load impedance the source can handle without adding distortion of its own? That's going to depend on the source... which you haven't told us about yet.

About a preamp. A preamp is said to amplify 0.001 volts to 0.1 volts.

Who said that? What preamp?

What if the signal is 0.25 volts. Would the preamp only amplify the 1 volt values and clip above it? If not. You mean even if the voltage is 1 Volt. The preamp can amplify it to 2 volts? What is the rule or typical?

Different preamps are designed to accept different signal levels under different conditions and amplify them by certain amounts while maintaining certain headroom margins. The options are myriad!

For example, the ART Cleanbox Pro accepts unbalanced inputs up to +21dBu (8.7Vrms) via an input impedance of 100k Ohms, and boosts that signal by up to 21dB (raising 0.25Vrms to 2.8Vrms), delivering a balanced output signal from an impedance of 100 Ohms, while introducing less than 0.1% THD (meaning distortion artefacts are more than 60dB below the peak signal level).

The ART is a relatively cheap box, but there are others with lower distortion. The Sonifex RB-UL2 has 0.01% THD (-80dB), for example, but only +15dB of gain.

So it really all comes down to the dynamic range of the noise source, and what level of distortion is acceptable for your measurement purposes.

[James Perrett]

Exactly what are you trying to detect here? What is the complete signal chain (makes and model numbers please). It might help if you gave us the full story rather than just a tiny little piece.

[sonic tac]

I don't want any amplification which can introduce distortions (via THD).

All amplification and all converters introduce distortion. It's just a question of how much, and what level is acceptable for your project.

The A1DA Cosmos ADC and a spectrum analyzer software can see the 0.2 volts frequencies clearer than amplying them to Line level 1.5 Volts, right?

I don't know. For minimum distortion you need the signal to be close to peak level in the ADC. Is there an input level setting that provides sufficient sensitivity while still presenting a load impedance the source can handle without adding distortion of its own? That's going to depend on the source... which you haven't told us about yet.

About a preamp. A preamp is said to amplify 0.001 volts to 0.1 volts.

Who said that? What preamp?

What if the signal is 0.25 volts. Would the preamp only amplify the 1 volt values and clip above it? If not. You mean even if the voltage is 1 Volt. The preamp can amplify it to 2 volts? What is the rule or typical?

Different preamps are designed to accept different signal levels under different conditions and amplify them by certain amounts while maintaining certain headroom margins. The options are myriad!

For example, the ART Cleanbox Pro accepts unbalanced inputs up to +21dBu (8.7Vrms) via an input impedance of 100k Ohms, and boosts that signal by up to 21dB (raising 0.25Vrms to 2.8Vrms), delivering a balanced output signal from an impedance of 100 Ohms, while introducing less than 0.1% THD (meaning distortion artefacts are more than 60dB below the peak signal level).

The ART is a relatively cheap box, but there are others with lower distortion. The Sonifex RB-UL2 has 0.01% THD (-80dB), for example, but only +15dB of gain.

So it really all comes down to the dynamic range of the noise source, and what level of distortion is acceptable for your measurement purposes.

I read that :

"Professional line-level devices operate at an average output voltage of 1.23V. A preamp amplifies a mic-level signal (in the range of 0.001V to 0.1V) into a usable 1.23V line-level signal by replicating it and sending it through one or more amplification circuits. The resulting line-level signal is then output to the rest of your signal chain and ultimately to your audio interface, DAW, or recording device.".

So if the input is 0.3 V. Can typical Preamp amplify it to 1.23V or become higher like 1.5V?

I kept reading about balanced inputs or outputs. Does it simply mean the stereo left and right need to be balanced? What is the connection of this balanced thing to the amplifier performance? Is "balancing" related to amplifier differential signal inputs?

How do you convert between THD levels to distortion artifacts in dB below the peak signal level?

Are you familiar with the A1DA Cosmos APU? Maybe it's the one I need where the output impedance is so low enough to drive the A1DA Cosmos ADC?

Thanks Hugh.

(James, I'm working at a sensor detecting company and I'll check the non-disclosure agreement if I can even describe the product.)

[James Perrett]

(James, I'm working at a sensor detecting company and I'll check the non-disclosure agreement if I can even describe the product.)

That's what I thought. Many sensors require specialist signal conditioning before they reach the ADC. Audio ADC's are not always suitable for sensing applications. When I worked in oceanography we sometimes used familiar audio gear (like the Revox A77) to record the signals but we usually used our own electronics to allow the recording of signals that an A77 wouldn't normally record.

Edit to add: Your mention of the Turtle Beach Multisound makes me wonder whether you want to record very low frequencies and possibly down to DC? The Multisound was unusual in that it could do this, but most audio gear and audio convertors won't work down to DC.

[sonic tac]

I only used the Turtle Beach Multisound for recording music from casette tapes ages ago and also because I love audio and speakers.. It's not for any sensor use.

Ok. The sensor is just a very sensitive sound processor to be deployed at seismic disaster zone to search for survivors. Idea is to pickup very low sound inaudible to the human ears and then use sound recognition to differentiate human voice from animals or insects.

The E1DA Cosmos ADC is very appealing because of very low noise floor. So the spectrum analysis will be done on very low levels close to the noise floor (that is, to hear almost inaudible cries for help).

Also I was able to lower the output impedance of the prototype circuit to 150 Ohms, from original 3.6 Volts. And I'm seriously thinking of getting the E1DA Cosmos APU. It has input impedance of 47/23.5kΩ balanced/unbalanced. This will be very sufficient. And output impedance is 47 Ohms. I read in the http://archimago.blogspot.com/ (just google) "Of note, if you're pairing the APU with the E1DA Cosmos ADC, 47Ω output impedance should be fine even at the lowest input impedance of 400Ω unbalanced at the 1.7Vrms input setting and even better with higher input levels with higher impedances.".

But I'll use the 34 dB gain (the APU has only 34 dB and 60 dB switch with no gain adjustments in between). With 34 dB. 0.2Volts output would produce 10 Vrms just enough for the maximum 10 Vrms output and input of both the APU and ADC.

Ok. With the adjusted 150 ohm output impedance of the sound sensor/processor, and 0.2 Volt output. Do you know other alternatives that can make it work with the E1DA Cosmos ADC with input and output impedance not far from the APU? Or is there just none, and APU is the only available choice left?

I'm thinking of others so if the project fails. I still can use the preamp for ordinary audiophile leisure.

[ef37a]

If you can do a bit of soldering the LM4562 dual op amp configured as a unity gain follower will have a distortion level below the level of any THD meter except the AP gear Hugh has and even then it will be at about the level of the internal sig genny. That is driving 600 Ohms and around +12dBu.

Dave.

[James Perrett]

I don't think the noise floor of the ADC should be your primary concern. The most important noise consideration should be the noise of your input stage and the impedance of the input (assuming that you are using an analogue transducer of some kind). To optimise noise you may want to think about limiting the bandwidth.

The book "The Art Of Electronics" by Horowitz and Hill covers noise issues fairly extensively and I would suggest that you read it carefully.

[ef37a]

I thought James that the chap was more concerned with adding distortion? In any event the 4562 is very low noise, lower even than the very excellent NE5532.

Dave.

[sonic tac]

What do you think of the Topping L30 II which is one of the best headphone amps out there (google all the reviews of the performance like having THD+N of 0.000087%). It has output impedance of mere 0.1 Ohms enough for the E1DA ADC! And input impedance is 2kOhm enough to work with 150 Ohm sensor output impedance.

The maximum gain of the Topping L30 II is 16.5 dB which has gain of 6.683 times so if the input is 0.2 V then the maximum gain is 1.3366 Volts. Is 1.3366 V enough for spectrum analysis? Compared this to the E1DA Cosmos APU with gain of 34 dB that can make the 0.2 V become 10 Volts. So what do you make of 1.3366 Volts versus 10 Volts? Can you see more signal using 10 Volts when zooming in certain spectrum? This is the factor that can make me decide whether to get the Topping L30 II or APU, or other ready built preamp or interface units.

[Hugh Robjohns]

To answer all these questions we need to know the dynamic range of the source signal.

If the peak level is 0.2Vrms at what level are the components you're trying to detect in your spectral analysis.

You seem fixated on ultra-low intrinsic distortion which is forcing the use of rather quirky and difficult equipment. Are these spectral components you seek really 140dB below your peak level?

[ef37a]

I am always suspicious of THD figures in the fifth decimal place, especially for 150 quid! Try making an oscillator that good!
Then, the spec for that Topping headphone amp is a bit ambiguous? There are figures for the output section at a much more realistic 0.1%.
An output resistance of 0.1 Ohms looks great but implies some form of V/I protection. VERY hard to do that without increasing distortion just below max output.

To bang my previous drum again. LM4562 can easily be configured for variable gain, say 0dB to +20. Use a good quality 30V line lump and an NE5534 to split the rail. Beer into water for even a 6th form technology geek.
If you want a balanced input, variable gain, a bit more complicated but not really that much.

Dave.

[Wonks]

The maximum gain of the Topping L30 II is 16.5 dB which has gain of 6.683 times so if the input is 0.2 V then the maximum gain is 1.3366 Volts. Is 1.3366 V enough for spectrum analysis?

You are talking about voltages here, so the gain is 10x Log10 (V1/V2).

10^1.65 is a gain of 44.67 times, not 6.883 times, so for an input of 0.2v, a gain of 16.5dB would result in an output of 8.93v (assuming the Topping had the power rail voltages to produce such an output voltage).

If you are going on say the Audio Science Review site figures, note that the THD figures were measured with specific input and output load impedance, so may well vary a bit with different configurations.

But the output voltage vs THD graph seems to show best performance with an output of about 5v, which is achievable with the Topping with a 0.11v to 0.2v input. And 5v seems a decent signal level to analyse to me.

[Hugh Robjohns]

You are talking about voltages here, so the gain is 10x Log10 (V1/V2).

Cough, cough, splutter, splutter! Perhaps you haven't had your porridge and blueberries yet this morning...

Voltages are 20x log of the ratio. (Power is 10x log of the ratio)

The OP's sums are correct.

[Wonks]

Bugger! I wasn't 100% sure so I did check on a website and I missed the squareds on the V1 and V2 bits. I thought I saw what I expected to see so didn't look closer.

Sorry chaps, as you were.

But from the Audio Science Review graph you won't get the best noise+ THD figure at 1.33v, it would appear to be around 8-9dB worse than at 5v output. But I have no idea exactly how those tests were conducted. Obviously similar to all the other headphone amp tests conducted so good for a comparison, but it may not be 100% applicable when fed from different input impedances and into different output loads.

https://www.audiosciencereview.com/foru ... amp.36027/