AMX RX

[ivconic]

There are CMS, IMS, and HMS versions, all just different temperature grades. Any will work. TRPBF means "tape & reel" and "lead free." These also don't matter.

I will also try to support the LTC family of converters listed here. They have an almost identical pinout and you can get started for as little as $20.

So now definite choice is LTC2338CMS-18, right?
No more changes?
I am asking this because I am about to place an order.
Updated schematic will not hurt here too...

[ivconic]

Folks if you want to push this project to rocket speed; vote me for boss... BUT first give me Carl's, Willy's, Tony's and all yours phone and Viber contacts!
Then all of you will regret because of that!
But the project will run at light speed!
Now as things are going on... probably my grandchildren will see the prototype... in their late ages!

[ivconic]

I tried to consult ChatGPT on this issue too...

[Willy Bayot]

So now definite choice is LTC2338CMS-18, right?
No more changes?
I am asking this because I am about to place an order.
Updated schematic will not hurt here too...

Do not rush to order as there could still be some updates after the real experiments of Carl.
Up to now, we only have seen simulation results of the XMIT part of Moodz but none for the AFE.
Give Carl a chance to verify the whole system using a simple PCB as described in this post:.

The board I'm working on will combine the TX, RX-AFE, ADC, and power supplies on one board. It will include an IO interface to easily hook it up to a micro dev board for driving the TX and reading the ADC. I will start off by building the TX & AFE only, probably using the bench supply option, to test out RX responses to ground and targets in both current mode & voltage mode. This will give me the information I need to see if any ground compensation needs to be designed in. If it does not, then I can add the ADC and start grabbing data. Even if it does, I can still add the ADC and start writing code while making design changes to the circuit.

There could be some surprise waiting around the corner for us

[ivconic]

Do not rush to order as there could still be some updates after the real experiments of Carl.
Up to now, we only have seen simulation results of the XMIT part of Moodz but none for the AFE.
Give Carl a chance to verify the whole system using a simple PCB as described in this post:.

The board I'm working on will combine the TX, RX-AFE, ADC, and power supplies on one board. It will include an IO interface to easily hook it up to a micro dev board for driving the TX and reading the ADC. I will start off by building the TX & AFE only, probably using the bench supply option, to test out RX responses to ground and targets in both current mode & voltage mode. This will give me the information I need to see if any ground compensation needs to be designed in. If it does not, then I can add the ADC and start grabbing data. Even if it does, I can still add the ADC and start writing code while making design changes to the circuit.

There could be some surprise waiting around the corner for us

It usually do, respecting the Murphy's law. Actually I should know this best; because Murphy made his law accoridng to my life!
I understand you. I am just killing time by joking. In meantime I try to find suitable substitution for the ADC.
My personal crieteria is bit different than others.
Since I am struggling with soldering small SMD components; first criteria is chip to have as few as possible pins and as large as possible packaging.
Price is next criteria. My budget is modest. But the price is not decisive factor in this case, because if it is worth; I'll gladly buy it.
My main problem is in worries how will I put it all together in my modest amateur conditions with amateur tools.
Ideal would be ADC with 8 pins, easy to solder it on pcb and to match all other criteria we mentioned before.
I failed to find such one. Obvously it not exist, with other specs that must have.
And yes, then comes exactly what you pointed on; the preliminary results and possible traps which we didn't expect in process of planning.
My older son is living in the USA, he can order directly from Mouser and then send it to me, any material.
But it takes certain ammount of time to arrive here. Not to mention shipping and other costs.
So I am constantly trying to find some better alternative.
I asked ChatGPT and it made me a list, half in the list are mistaken by ChatGPT by some spec, but I picked one which seems match.
It is ADS8881... although the price of $28 (least i could find) is not so lower than others we analyzed here.
What is your opinion about ADS8881 ?

[Olly]

The ADS8881 has quite a long sample time leaving only 290nSec for acquisition. This means that the SPI clock must be in excess of 62MHz and so becomes tricky to interface with a lot of micros.

[Olly]

The ADS8881 has quite a long sample time leaving only 290nSec for acquisition. This means that the SPI clock must be in excess of 62MHz and so becomes tricky to interface with a lot of micros.

The LT2338 on the other hand has a faster sample time and much longer acquisition time of 460nSec so needs only a 40MHz SPI clock.

[Willy Bayot]

Carl,
I would suggest you to execute a full simulation with the XMIT and AFE and applying the minimum amplification gain that you have calculated.

[ivconic]

It's frustrating when among many chips you can't find even one that fulfills "everyone's wishes"... that is, the given criteria.
Either the criteria are too high... or the criteria are set wrong...
There are plenty of detectors on the market that work just fine. How are they made?
ML is the leader in this field, can we "peek" into one of those machines and see how it's done?

[Carl-NC]

It's frustrating when among many chips you can't find even one that fulfills "everyone's wishes"... that is, the given criteria.
Either the criteria are too high... or the criteria are set wrong...
There are plenty of detectors on the market that work just fine. How are they made?
ML is the leader in this field, can we "peek" into one of those machines and see how it's done?

I don't understand... what criteria are too high or wrong?

Minelab's PIs are pretty well described in their patents. They are hard to read but lots of good info in them.

[Willy Bayot]

It's frustrating when among many chips you can't find even one that fulfills "everyone's wishes"... that is, the given criteria.
Either the criteria are too high... or the criteria are set wrong...
There are plenty of detectors on the market that work just fine. How are they made?
ML is the leader in this field, can we "peek" into one of those machines and see how it's done?

We perrfectly know how the existing traditional PI are made. (Simple pulse switching, analog integration, analog-based DSP)
We also know how the PI systems of ML are made. (Square Wave XMIT current, complex multichannel (12) analog integration, multiple multi-channel ADC's, FPGA....)

We do not want to copy any of them.

Wa want to build a basic (but sound) platform based on Square Wave XMIT current and direct sampling for further experimentation with DSP by members of this forum.
Any short turn in its design would give miserable results.

This excludes the 'fulfills "everyone's wishes"'. concept

A bit more patience, we know where we are going, just try to follow and learn!!

[ivconic]

[ivconic]

"..A bit more patience, we know where we are going, just try to follow and learn!!..."

Follow what?
Learn what?

I was under the impression that the work on this would be transparent. And that we will be able to follow the work on the project step by step.
Now it seems to me that a couple of you have gone "anonymous" and are working on it without the public.
If you succeed; at the end you will post the final outcome. It's not what I was hoping for.
Here's an indicator, I deliberately didn't want to write anything on the topic for a few days, but every day I check a couple of times if there are new posts.
And there are no new posts for several days. Topic is slowly sinking into oblivion.
Have a couple of you changed your mind about this and now want it to be a private project?
If so, ok, but at least let us others know.
Otherwise, I'm not surprised. Ideas like this never worked on the forums.
And the recipe is always the same; you start a broad debate, see what the majority thinks, pick up some interesting idea, suggestion... and then calmly retire and work on "your" project, which if it succeeds; will become commercial.
If it fails or succeeds half-heartedly; maybe someone will finally remember to give it to the people, let the people play.

[Carl-NC]

AMX isn't private. My time this week has been very limited. However, I have been doing some simulations to get a better understanding of the energy losses and how they affect the RX side. Also I'm laying out the AFE PCB, and working on the TX design, in particular the current feedback regulation. I should have something to post this week.

[ivconic]

That would be good news.
In the meantime, I reviewed the topic "My AGD analog detector work" by Auto-Mation-Assist.
And he has similar goals, to be able to detect golden flakes.
His schematics are interesting.
It may be possible to use the good things of both projects and make something third. Sublimation of both.
But also in his schematics I see some components that are not easily available in local shops.
I was thinking something... the TC for a small gold nugget was said to be 1uS.
Is that strictly so? Can certain nuggets be stretched to 2-5uS?
In other words; what to expect from a detector with the ability to set the delay to not below 5uS, or 10uS?
...
A few more interesting questions...
Whether and to what extent; good GEB and DISC depend on minimum TC?
Logic says that what is more "extracted" data from one delay period; the more reliable both GEB and DISC will be.
Or does it have nothing to do with each other?
On the other hand; GEB should not depend on it.
Unrelated to TC, the moment you put the coil on the ground; the state of the signal in coil will change due to the influence of the earth.
Does the number of samples improve GEB or does it not matter.
Further next question;
What will DISC look like?
Will Disc work on the simple principle of "rejecting" everything in the signal that is longer than 1uS, 2uS... etc.
That is; if iron is present in the first uS but still too in the next 20uS (for example) and the gold nugget is not, it is present only in the first uS:
how will the difference between iron and gold in that first uS be expressed? By amplitude?