RELIC HAWK

[porkluvr]

That's not good, and I don't have any good news.

What do your + and - supplies measure? If there are no regulators, then the capacitive voltage divider would split the battery roughly in two. It would not be a nice and neat regulated +/-5V that is easier to work with. But that is not important issue... .

Ignore what I said about a 2.5V output from U5B. I don't see that today. I don't know why I had it in the first place, and I may or may not figure it out. But I think it has to do with the fact that there is no negative feedback around U5B (U10B in Max's drawing).

Looking again at the circuit; IF U5B were a perfect opamp, having exactly equal gain on both the + and - inputs, and IF you had a perfect -2.5 input on both + and - inputs (cough, cough), then you have 0V output. If. But that is not going to happen.

U5B is not a perfect opamp. Without NFB a real world opamp's output is not predictable unless you predict it to be either HIGH or LOW. Any in-between condition will be accidental and will change with temperature, resistor tolerance, loading, power supply voltage, and maybe even the phase of the moon! Very tempermental, and practically useless - except as a comparator.

So, I don't know what I did to have a -2.5V output before, but it may have been an accident, or my mistake. I am no closer to figuring out what purpose that opamp sub-circuit serves.

I cannot hope re-engineer this circuit because all I have to work with is second hand information, and no physical lab.

TL272 and LF442 are not same as TL27M2. They're all amplifiers, but have different characterics. But I don't think this is your problem, there is probably something else.

Without an oscilloscope you have very slim chance of figuring out where errors are because you seem to be only person building this detector. You probably need more help than I can give you.

[ivconic]

Tend to agree 100% with you Porkluvr.
More help will come soon here... i am just waiting original device to come.
Week or two more...
Until than

[robopatt]

Ok.
We will wait for help to came.
Until then just some ideas.On max schematic there are 3 x 4066.
When I think one is for switching allm/disc other two to determinate s/m/l.
Am I right for this?
So you see my point.See U9 on max scheme.You have used only U3.

And the use of 339.This must be the fast shutdown.

[Max]

I agree with you.But they can't produce 10000 units with trail and error?Just we need to find what is still missing.And oscillator will work from the first switch on.
I have some ideas on this unknown TR now.But problem is elsewhere.Because every TR can be changed with similar.
I will draw my conclusions and will post them so I can hear your opinion.

Hi,
sorry for latency... too busy these weeks.
No you misunderstud.... I mean trial and error for us, homebrewers that don't know exact recipe... e.g. tolerances required and selected gain they use, nor we know the exact e.g. transistor they use there etc.

Sure once you have the exact recipe and quality checks, and also the selected gain well you can make even 1 billion of them all the same. Now... ML has military grade parts, checks , instruments , plants etc.... no homemade md will match ever the ones produced in that industrial heaven!
That's a fact we must face everytime we feel furstrated by some lazy homemade oscillator.

Other stuff.... is by the book... quite easy, like digital parts nothing special.

Kind regards,
Max

[Max]

the R28 mistery....

Here's something I don't understand. It looks like all that U5B does is output -2.5V and then draw 25mA through R28.

WHY?? What possible useful function does this perform?

Ummh, besides being a 62.5mW space heater?

Maybe it is supposed to stabilize ground. If so then it seems to me like an inefficient way to do so. You won't find me copying that scheme. It's wierd. WIERD. (Arrgh.)

Hi,
it's tricky , I see. And ugly also...at first. Seems it increase the universe's entropy stuff... and nothing more....

Uhm... let make it clear, if we can.

The U5b it's a not-inverting op-amp there.... so... let consider negative reference the -5v and we see the reference is at -2.5 (so 2.5v up the -5v).

Now look at the gain of a non inverting op amp: gain is = 1+Rx/Ry... in our case 1+(100k/100k) = 2

Then if we put +2.5v (think respect the -5V, absolute negative voltage there) you get vout = +2.5*2 = +5V

In other words the vout of U5b is exactly +5V above the absolute negative voltage (-5v , aka -Vee).

Now... that's actually the GND level in the circuit, that's cause we indicate that for gnd and the two opposite -2.5V and +2.5V respect to that GND, used for e.g. references in trimpot section.

Why that ? Simple... RH needs that, needs a floating GND level with two simm up and down voltages, very stable (lm336 is temperature compensated) probably cause it's required for correct alignment by trimpots: remember that whole gain of chain if fairly high there in RH, so voltages must be VERY stable respect e.g. to thermal variations.

Now... you probably were misleaded by all that voltages... you suppose there were 2.5v drop at R28 with big waste of power....+62mW... resistor will be hot! actually isn't so hot, though the resistor is there for a reason.

The reason have to do (probably) with impedance required. You know ... at gain=2 the op amp has very low closed-loop apparent output impedance (that's a good case for us in general) but what about that ?

The current the op amp must provide to the rest of circuit is a factor, the noise propagation is another factor and so on: the resistor is there cause THEY WANNA MAKE an higher impedance separation between the op amp output stage and GND rail.

Why that so ? I think good explaination is about noise propagation : at some frequencies the noise can play a role in op amp gain variations , with potential effects on stability. Think about e.g. some digital noise that will enter inductively the GND tracks... it will be picked-up by op-amp closed loop too and will affect stage gain a bit (small variations but relevant if md whole gain is high such in RH). A solution I saw in e.g. audio hi-fi amplifiers is separate output of op amps from next loads by increasing impedance between stages : that keep noise back propagation harder and let amplifier chains gains be fairly stable vs noise, also at higher gains.

So the answer is maybe a single word.... stability.

Kind regards,
Max

[robopatt]

Hi,
Max.
I am very glad that you are finaly here.
Have you checked this last schematic?Why there are 3 X 393?In yours there is only one 393?
Now I have some "idea" about that TR.Could it be this?

[Max]

Hi,
Max.
I am very glad that you are finaly here.
Have you checked this last schematic?Why there are 3 X 393?In yours there is only one 393?
Now I have some "idea" about that TR.Could it be this?

Hi,
really not... cause had no time to read here in past weeks... not that now I have much more time.... just today is special.

Anyway, in which page is that schematic ? I'd like to see it.

Kind regards,
Max

[robopatt]

Schematic post #748
Sprintlayout post #754

[porkluvr]

ground reference

Yes, the circuit is ugly but that does not mean that it will not work.
Now I've got it. My test setup was faulty and I was acting like a dog chasing his tail.

One bad thing about spice is that it is hard to simulate circuits that generate ground because spice likes for ground to be the starting point, not an end. And, if it's ground then it's always going to be zero volts. (duh) You can't measure it, it's just THERE. Believe it. It's true.

Here I have hopefully corrected my error. It would take a lengthy explanation or else a picture. I'll draw a picture because I don't feel smart enough to make a lengthy explanation.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%

Here I have called battery(-) "ground" and ground is "0V". LTspice will not allow me to draw the circuit as it is would be in real-life, but I must transpose. In my picture we can see that if you have a 10V battery, the "0V" will be almost exactly half-way between B+ and B-. V+ and V- would get out of symmetry when the battery voltage is NOT 10V, so I don't understand why Minelab would do this. But, if it works, who cares?

Whew, I hate these crazy power supplies, but so what? Max, I thought your drawing was wrong but it was me that was wrong. Sorry.

[Unregistered]

Hello porkluvr,
For this circuit , it is not a temperature compensated for the Zener diode???just a idea ,because i do not know for sure !!!!

[porkluvr]

... For this circuit , it is not a temperature compensated for the Zener diode???

If I understand you correctly you have noticed that I have substituted LM385-2.5 for LM336-2.5? That is purely a convenience for me. I do have LM336-2.5 model (thanks to Aziz) but I have not modified it to conform to the "standard" of other parts in my collection. Out of habit (and laziness) whenever I need a 2.5V reference I use the LM385.

LM385 and LM336 do have substantial differences, but for the purpose of supplying 2.5V in a simulation, I am not real picky. Sorry if I have caused (even MORE) confusion.

I think the LM336 in the RH circuit has much better guaranteed temperature stability than LM385. LM385 will operate over a wider range of current than LM336. Those are the main differences.

I have to make substitutions all the time because sometimes the "correct" parts just don't seem to want to get along with each other. (Notice the OP15C that I used; it's more "friendly" than the TL07x. They are not internally the same, but "sort of" close in performance.)

If that is not what you were asking please try and rephrase your question and I'll try again to answer.
*******************
(Referring to Max's original drawing): Here I would like to explain that when I had question before about the U10B circuit I was being stupid because I was operating U10 with +/-5V supplies. Well... If we had nice +/-5V supplies, then we wouldn't need U10B then, would we??
And, so long as the + and - supplies don't change too quickly, they don't really need to be symmetrical.

(It is more complicated and expensive but I am in love with the idea of using a low power switching power supply and having +/- 5V to operate the machinery. But that's probably just because I'm old, and my brain inflexible.)

edit: I'll edit my LM336 symbol in a day or two. I think the symbol/model works, but I need to make it easier to invoke.

[porkluvr]

Hello porkluvr,
For this circuit , it is not a temperature compensated for the Zener diode???just a idea ,because i do not know for sure !!!!

I just re-read your question and put into context with Max's last post.
Yes, the voltage reference should be LM336-2.5.

Every time I post a virtual circuit there is a likelyhood that I have taken shortcuts that, although my changes may be OK, probably would need some extensive testing in real-life before you or anybody would want to rely if you were three-hundred miles from civilization and a soldering iron!

[porkluvr]

Could somebody tell me, does Relic Hawk use a rechargable battery pack, or 8 AA alkaline batteries, or what? Reason I'm asking is because I'm wondering about the range of supply voltage change from new battery to dead battery.

Thank you

[robopatt]

Hi,
Porkluvr.
RH use 12V seled lead-acid battery.Max have excplained all.Just read.This diode that we tweeked at TX stage have to be 3V3.All there excpained.
Now when I switched on this PCB refuce to work like when I worked early.
Need time and then start oscilate.
I will recheck all.
Can you fins similar TR like this that I showed.Can this be the mistery SMD TX?

[porkluvr]

I appologize for not taking time to read. I would always get mad when somebody asks questions that have already been answered in a thread,
and here I am doing the same thing.

Yes, the oscillator startup can be a problem if everything is not "just right" and I haven't figured out exactly what is "just right". Not yet (although I might not EVER). I must give my 5kc RH oscillator virtual circuit a little "kick" or else it won't start. A little injection of coil current does the trick.

Just on a hunch I just now fired up my 20kc oscillator after removing the "initial conditions" (.ic ...) statement that I had been using. It did nothing for about 4ms and then took only 2ms to reach full output. It's like comparing apples and oranges, but maybe a clue as to what is happening.

If I Remove .ic from the 5kc oscillator it just sits there and does nothing. Forever. (Well, maybe not forever, but you get the picture.)

I think that the low-Q tank is a likely source of slow start-up, but then you also must have hi GBW transistor at Q2 and probably a low loss tank capacitor (polypropylene!!).

Increasing my 20kc coil resistance from 2.5 to 5.5 caused startup time to increase fom 6ms to 22ms, but it did start.

I lowered the 5kc oscillator coil resistance from 3.5 to 1.5, and it still does not seem to want to start without .ic.

Back to the salt mine.

edit: Oh yeah, using a gel-cell battery would definately make the floating ground to be a good cheap, effective power supply solution. What, it ranges from about 12.6V to 10V and then it's dead? (just a ballpark guess, not reasearched)