Induction Balance Stuff - Single/Multi Frequency Response, GB, Disc, Measurements, Ideas, Fun, etc.

[Aziz]

UAC2 userspace driver implementation estimation (see PDF).
You see, that there is much more work. I have to bypass all the Android Audio Subsystem.

Why was Quirkdroid developped? Ok, for developping Apps. Not custom drivers!
Google, go f$k yourself!

[Aziz]

Hi all,

finally, the AI says,
that implementing the metal detecting software in Android systems with exclusive access to the external USB sound card won't work very very likely on the budget consumer android tablets ( < 150 EUR). These tablets didn't implement the MMAP (Memory Mapping / NOIRQ-Mode). This is optional for vendors to implement/support it. Why should do they?

The AAUDIO_SHARING_MODE_EXCLUSIVE mode will fail and I would get the AAUDIO_SHARING_MODE_SHARED mode instead. The AudioFlinger will not allow the features I need.

I didn't find a solution to overcome these issues yet. No, I'm not going to implement my own UAC2 driver (would solve all the issues however).

Quirkdroid has big errors by design.
Aziz

[Aziz]

Hi all,

pity, the low budget consumer android tablets won't support the features I really need. Not yet.
I would require Android version 14+.
And Oboe-C++-wrapper API isn't required at all as there is a native AAudio API.
But the OboeTester App can show me, whether an android tablet with the connected USB sound card supports the features I need.

I don't think, that many vendors have enabled the features I need in the OS.
Only premium budget android tablets will have it enabled, maybe. This is not sure. And may be even restricted to few features only.

No, I'm not going to do the job of the vendors: Writing my own USB Audio Driver (UAC2).

Anyway, I have to buy a modern low budget consumer android tablet to see it with OboeTester App.

Android is a bunch of quirk development. Every new version brings new quirks.
Quirkdroid!

Aziz

[Aziz]

Hi all,

bad news on Android OS.

It seems, that Android is broken by design.
What the hell is you doing Google?

Bit-perfect mode above Android 14+ is supported in theory but it is optional for the vendors to implement it.
You can bet, that no one is going to support it.
Only expensive Google Tablets support the Hi-res Audio, which is bypassing the system mixer.

Mixer no good.
Direct mode to HAL-layer is required.

[Aziz]

Hi all,

there are some interesting & cheap Android tablets available.
The most cheap variants are the kids tablets with reduced horse power (2 GB RAM, 4 core). They are even quite small size: 6 to 7 inch.

What I need is at least:
- Android version 14+
- free USB-C port (USB 2.0 minimum)
- Wi-fi debugging feature (USB port must be free for the external USB sound card connection)
- 24 bit / 96/192 kHz support under Oboe-API
- support for running the App with switched off display (saving battery power)

The App does not really require too much processing power.
In the X-mas time, these toys get even cheaper.
Aziz

[Aziz]

BTW,

I have deciced to buy a cheap Android Tablet. Android Version 14+ is required according to my AI prompt and requirement analysis.
Minimum 4 MB physical RAM, 4 or 8 core ARM Cortex CPU tablet.

Just to play with Android Apps. Using the Wi-fi debugging feature (Wireless Debugging).
Using Oboe API.

https://github.com/google/oboe

Good bye M$, good bye Intel.
Go f$ck yourself now!

Programming language Kotlin/Java is required (a few % only). Rest of the app is in native C/C++. DSP part is C only.


I am somewhat pi$$ed off of M$.
PC is kaput! Totally kaput!
I am going to f&ck them now.

[Aziz]

Hi all,

I have been very very lazy recently.
I didn't fire up my soldering iron.
I didn't finish the ultra f%kn ultra-low-noise amplifier.
I didn't order even new parts.
I have to make an inventory of my parts yet.
There are some missing parts in my parts box.

And I am thinking of to pi$$ off to Microsoft and Intel.
I am thinking of making Android Apps on cheap Android Tablets.
Enough is enough.

Aziz

[Aziz]

Hi all,

the promissed amplifier schematics:
- 9 V Version with NE5532
- 4.5 V - 6 V Version with OPA1612

See the LTspice simulation files please.
Cheers,
Aziz

[Aziz]

Hi all,

I have seen, that the op-amp OPA1612 can be operated down to 4.5 V supply voltage (single). I am going to use it as it gives very very good performance.
So the amplifier will be optimized for 4.5 V - 6 V supply voltage. It will draw less power too.

Unfortunately, op-amp has 2 V common mode input voltage limit to the power rails (like the NE5532) but it has a rail-to-rail output. The whole transistor stage operating point will be so set, that the common mode input voltage is around mid-level of the supply voltage.

I will add a 9 V supply voltage with NE5532 to the simulation files too.
I am not finished yet.

PS: Increasing the input impedance won't work well. We need the frequency compensation feature for stable operation. It will be around 7 kOhm however.

[Aziz]

Hi all,

well ok, I will increase the dynamic Rin of the amplifier a bit (10 kOhm min). The frequency compensation feedback capacitor Cf must be reduced or totally eliminated in this case.
The bootstrapping method works then too. But this isn't required I think.

PS: Thanks for clarification Carl.
Aziz

[Carl-NC]

Here is a direct comparison between a Lee-type CB preamp and a traditional opamp preamp:


The two methods are adjusted to have the same overall output swing, indicating an overall gain equality:

The blue is the std preamp and the green is the Lee/CB preamp. The frequency is 10kHz. The Lee has an offset because N4 hasn't fully settled, that takes ~3-4s and I'm not gonna wait that long on the sim. No big deal. Notice the 90° phase difference; that's because the std method (voltage mode) has a -90° shift due to the induction derivative and the Lee method (current mode) does not. If you have ITMD3, it's all explained in Appendix A.

Here is the output noise for each:

At 10kHz the output-referred noise is about the same. Std = 28nV/rtHz and Lee = 23.2nV/rtHz. Equivalent input-referred noise is 1/20th of these numbers, a little over 1nV/rtHz.

The reason the Lee method isn't much better is because the low input impedance forces the coil into current mode, and the function of the CB amp is to first convert the current mode back to voltage mode. This is something you get for free with the voltage mode connection. If you were to strictly consider the voltage gains of the two methods, the Lee amp has ~30X more voltage gain and would seem to be the hands-down winner, but here it's being used as a transimpedance amp, not a voltage amp.

So, yes, it works, and with some tinkering might do a little better. But I agree with you Aziz, it ain't the way to go.

[Aziz]

Missing the LTspice zip file...

PS: Why it is missing? I have uploaded the file.
Or is it owed to the fact of enjoying a Johnnie Walker (20+ years old)?

[Aziz]

Hi all,

where to get the OPA1612?

Ali Baba and the Forty Thieves?

https://www.alibaba.com/countrysearch/CN/opa1612.html

This is for sure a fake op-amp.
;-)

Aliexpress?
https://de.aliexpress.com/w/wholesale-opa1612.html
They cost much more. I'm sure, they are all still fake.
;-)

They all will probably underperform NE5532.
;-)

Be careful. NE5532 is still a good choice.
Aziz

[Aziz]

Hi pito,

perfect ferrite balance, the first it is a ferrite, the second any metal, the blue it is a Uy, orangre Ux

nice. very nice.
I should look at the ferrite induction balancing. Maybe I can leave the IB-nulling part and drive the TX with differential input (more TX power).
Aziz

[Aziz]

Hi all,

I have made a comparison circuit simulation using the data in the recent AI paper.
4 independent configurations with ideal op-amps (Level 1). The low input impedance amplifier is made in inverting ideal op-amp amplifier configuration.
The high input impedance amplifier in standard non-inverting configuration with a load resistance at the input simulating the input impedance.

As the operating frequency of 45 kHz is very high, the reactance of the RX coil dominates heavy. It is acting like a choke in very low input impedance amplifiers.
This is really a revealing evidence.

Look at the spice simulation file in the attachment.
Cheers