And now, you swear this will solve the problem, right?
I'll modify the circuit right now.

Can't swear to that, just going by the recommendations of the manufacturers, as you should. But the AMS1117 may well have been oscillating on its own without the correct capacitors connected. See you later.

 

I built the circuit. Nothing drastic happened. No visible improvement. Still beeping.
The capacitor on the ADJ was making the device working worse. So I remove it. Basically now is like you recommend it.
The C5 cap is an electrolitic one, but the C1 22uF is a tantalum fat SMD with 2 new legs soldered.
Still beeping...hmmm...
This is the updated version:

 

Why does it beep? Do you suppose the voltage is dipping too low when the meter takes readings. It may have a deep dip that your voltmeter cannot see because of its limited response time.

A quick and dirty solution might be to add a large capacitor to the input or output of the AMSM1117. That, or you can work on further lowering the output impedance of your DC-to-DC converter. Or perhaps better yet is to rip the speaker out of the voltmeter (just kidding).

 

A quick and dirty solution might be to add a large capacitor to the input or output of the AMSM1117

I dont think this will be wise, because the voltage may be very high from the capacitor, potentially frying the AMS1117 if placed in it's input. Remember he can work with up to 15V max.
And in it's output, the high voltage accumulation in the capacitor, will definitely fry the poor 3V device under test.
I can understand you are putting ideas on the table.

 

I built the circuit. Nothing drastic happened. No visible improvement. Still beeping.
The capacitor on the ADJ was making the device working worse. So I remove it. Basically now is like you recommend it.
The C5 cap is an electrolitic one, but the C1 22uF is a tantalum fat SMD with 2 new legs soldered.
Still beeping...hmmm...
This is the updated version:
View attachment 280901View attachment 280905

I can't suggest any more with this circuit, except take the ammeter out from between the bridge and point A. You don't need to measure the current there anymore. By the way, your bridge rectifier is connected wrong in the diagram, needs rotating 90 degrees anticlockwise!
Maybe it's time to try MrDanko's post #57 circuit, and re-wind the transformer with the windings going right around the toroid, and secondary over top of primary, where they will be more efficient?
Just a suggestion.

 

I made another test today.
I used a bigger core, same exact diameter and thickness only it is taller. I put them side by side to see the difference. It's almost double size. I really hoped it will solve the current problem. It didn't. Aaaah.
In the very beginning it created the usual short of 350mA in the PRIM. That was because too many turns in the SEC. And I took out some of it's coils from it's SEC. From 50 down to 30. It is still not calibrated since the mA got still quite high in the PRIM. Up to 130mA, frying and cooking the Tr's if left too long. But I could take the readings and not kill anything.
I am actually very impressed how close the values are between the 2 quite different cores. I was expecting bigger and stronger mA but is really quite equal with the other. I am very surprised about this ! You can see it in the 1.31mA for NoLoad, and 31.5mA when Loaded with the 100R. WOW. The voltages are quite low but is because I didnt fully calibrated/optimize. And I dont plan to do it. This test told me everything I wanted to know. Which is, no significant amperage increase. I also tested the [device] and still beeping. Aaaah. It should have run ok especially now when all this new setup, theoretically should have been more stronger. And is the same. WOW. Right? I start to believe.... maybe some other components around this coil might affect the final result. Not only the transformer alone. Which I concentrated fully only on him, 100% as you already know.

Maybe it's time to try mister @Danko #57 circuit,

That is plan B. Haha. And I probably will. Since I am left without any ideas myself also.

 

By the way, your bridge rectifier is connected wrong in the diagram, needs rotating 90 degrees anticlockwise!

I totally didnt see it. Somehow I managed to rotate it and leave it like that. Thanks for seeing it. I didnt try anything 'new' with that configuration, hahaha.

 

I dont think this will be wise, because the voltage may be very high from the capacitor, potentially frying the AMS1117 if placed in it's input. Remember he can work with up to 15V max.
And in it's output, the high voltage accumulation in the capacitor, will definitely fry the poor 3V device under test.
I can understand you are putting ideas on the table.

Have you looked at these points with your scope? Are there any large spikes on the waveforms? Larger capacitors won't make the voltage any higher, it'll just hold up what is there better. Put a 1N4000 series rectifier diode across the AMS1117, cathode to input, anode to output, for protection.

 

Ive corrected the BR in the diagram. I forget it rotated like that.
I never had any problems with the VReg so far. Is true I didnt scope it. I should, right? I have no clues if spikes are present in this circuit. If they were, something would have died from how many tests I did so far. I will take a scope reading on AB and CD points. If no spikes are visible, then no need for extra protection over the VReg.

 

Ok, I just did a scope reading and everything is very boring on the scope. In the sense that no spikes whatsoever. And the voltage rises accordingly to the point is measured and remains there. Since is a DC Voltage. It behaves normally and in parameters. Nothing out of the ordinary. If its 10V on that point, is showing 10V. And thats it.
This is on the smaller core. The taller core I test it today indeed, but put back the smaller working one, after I finalized the tests with it.

 

I dont think this will be wise, because the voltage may be very high from the capacitor, potentially frying the AMS1117 if placed in it's input. Remember he can work with up to 15V max.
And in it's output, the high voltage accumulation in the capacitor, will definitely fry the poor 3V device under test.
I can understand you are putting ideas on the table.

There is no reason for the output to attain a high voltage if the AMS1117 is working.

The reason the ideas were offered are self-evident.

I agree with @sarahMCML 's idea that using @Danko 's circuit has a good chance of solving your problem in a more elegant way that by slapping large capacitors into the circuit. When Danko suggests something you are receiving a gift of very high value.

 

I made another test today.
I used a bigger core, same exact diameter and thickness only it is taller. I put them side by side to see the difference. It's almost double size. I really hoped it will solve the current problem. It didn't. Aaaah.
In the very beginning it created the usual short of 350mA in the PRIM. That was because too many turns in the SEC. And I took out some of it's coils from it's SEC. From 50 down to 30. It is still not calibrated since the mA got still quite high in the PRIM. Up to 130mA, frying and cooking the Tr's if left too long. But I could take the readings and not kill anything.
I am actually very impressed how close the values are between the 2 quite different cores. I was expecting bigger and stronger mA but is really quite equal with the other. I am very surprised about this ! You can see it in the 1.31mA for NoLoad, and 31.5mA when Loaded with the 100R. WOW. The voltages are quite low but is because I didnt fully calibrated/optimize. And I dont plan to do it. This test told me everything I wanted to know. Which is, no significant amperage increase. I also tested the [device] and still beeping. Aaaah. It should have run ok especially now when all this new setup, theoretically should have been more stronger. And is the same. WOW. Right? I start to believe.... maybe some other components around this coil might affect the final result. Not only the transformer alone. Which I concentrated fully only on him, 100% as you already know.
View attachment 280909View attachment 280910

That is plan B. Haha. And I probably will. Since I am left without any ideas myself also.

I think you will find an enormous difference with MrDanko's circuit. For instance, my second version takes 52mA with no load, at an output voltage of 6.1V. This drops to 3.82V out and 82mA primary with a 100 Ohm load. That's with 22 turns for each primary, and 22 for the secondary. That's with 5V input. It gets better with more input voltage.
Your new transformer looks a lot better wound, by the way!

 

Maybe it's time to try MrDanko's post #57 circuit, and re-wind the transformer with the windings going right around the toroid, and secondary over top of primary, where they will be more efficient?

That is plan B. Haha. And I probably will. Since I am left without any ideas myself also

I finished experiment with converter 5 V/3 V, using this circuit:

L1-L4 are windings of special DIY transformer.
Information about transformer I will place in next post.
D3-D5 are limiter for spikes at level 16V on coil (16+5=21 V relative
to circuit's ground).
Because of excellent coupling between coils L1, L2, instead HV spikes
on collectors of transistors appear only fast decaying oscillations
at self-resonance frequency of coils, which do not need to be suppressed.
Converter, w/o load, takes 26.4 mA from 5 V PS, with V_out = 6.5 V.
On 100 Ω load voltage decreases to 4.14 V, consuming 60.8 mA from PS.
So, efficiency is 56 %.
See oscillogram of collectors voltages w/o load:


And below is oscillogram with load 100 Ω:

 

Here is my next best idea, desperate one in fact. I really cant think on anything else after this.
Im also blown away by the results. I am missing something very essential and basic. For sure.

Explanation:
So... what I did in this experiment: I put 60 turns in both PRIM coils. and leave 50 turns as it was in SEC.
We are looking at line 19 now.
The PSU reading for [No Load] was 100mA in PRIM with 8.7V in AB points and 3V in CD points at 1.3mA in SEC.
After inserting the [100R LOAD], PSU rises to 120mA in PRIM. With 21.5mA in SEC and 3.17V in AB and 2.1V in CD points.
Good... All this with 1k resistors in the bases of the PRIM Tr's.
Now, the next line (20).
What I did different at this line, I changed into 10k those base Tr's in the PRIM.
And read the values yourself.
The conclusion, was that everything gets Down (or UP) depending on these resistors values. They modify the entire circuit behaviour indeed. But in the way that ---everything--- rises (or fall), not only parts of the circuit as I expected. All at once is modified. Is kind of a tuning element for the rest of the cct. Which is cool... but really didnt improved the mA in the SEC as I targeted. Another failure, another lesson.
I also had some new complications, after waiting a little bit, literally about ~30sec, the Tr's got warm and heat up. I could see the PSU mA rising as well, and then the entire cct in the SEC shut down. In the sense I could not get any kind of reading from my DMM voltmeter and Ampmeter !!! Very strange, right? This was some kind of Cut-Off element of the circuit. OAU. Why im so sure it was a cut-off and not a wire getting loose? because I close the power, I blow cold air over the tr's for 5sec, and then started the Power again, and everything was reading as before, as in my list there. This cut-off it also may have coincide, with me putting the probes on that electrolitic capacitor where the AB points are actually, to get the voltage. But... after a certain time, after the tr's got a bit hot. In the begining, when everything is cool down, I could read repeatedly this points with no cut-off. THats the complication.
It doesn't matter now. I explain it anyway because I like to be thorough.
I will make the second circuit of mister @Danko design. I hope it will work as advertised, haha.

Information about transformer I will place in next post.

Well? the information? Give it. Haha.

 

Well? the information? Give it. Haha.

All dimentions are in millimeters!

https://www.digikey.com/en/products/detail/laird-signal-integrity-products/LFB090050-000/571942

 

So 10 turns for both bifilar coils. Ok.

 

So 10 turns for both bifilar coils. Ok.

Both winding are bifilar, each winding contains 10 turns.

 

I am starting your circuit mister @Danko
One thing I learn in my past, is that Zenner diodes are extremely sensitive to currents. Pretty much like LED's.
I have added a minimum protection of 100R on that zener. Actually 100R will be like nothing is there, at least 500R in my experience, but I will close an eye on this and try it riskier with 100R. Which is foolish of me. Also is a bit high value of 16V. I have a good range of zenners. We should include 12V or a bit less to be in specs with the VReg I have. If you dont know already, this AMS1117 that I use, is accepting voltages up to 15V. I usually put it in circuits with 12V, never to its maximum of 15.V.
Make these 2 modifications in your simulator and see if the results are still ok.
Thank you.

 

Mister @Danko
I made your circuit !
Good news: -Your circuit worked from the first try. Which is impressive. I didnt thought it will. Very lucky guy!
Bad news: - My device still beeping, in sigh of not enough amperage.
My modifications:
I put 11turns instead of 10 for both coils. I had a bit more wire left and I didnt want to cut it.
As I mentioned already, I put 100R in series with the 16V zenner.
I included the VReg set on 3V after your final output of 5.3V (+) and (-).

Converter, w/o load, takes 26.4 mA from 5 V PS, with V_out = 6.5 V.
On 100 Ω load voltage decreases to 4.14 V, consuming 60.8 mA from PS.

My PSU mA is showing me: 30mA with [No Load], ~50mA with [100R Load] and ~40mA with [device under test].
With [100R Load], it drops to 3.5V at AB points and at CD points(over the 100R Load) is 2.3V
Very close to your results I may add. You had luck this time ! You are a lucky guy.
Everything else is fix as in your circuit.
Also I suggest to include in your next simulation test, this VReg circuit as well. Because it is part of the circuit after all.
- I wait your next update. Make it good !

 

You had luck this time ! You are a lucky guy.

With engineers like Danko luck has nothing to do with it. It is all skill and knowledge.

Congratulations to you both.