TINKERERS POWER SUPPLY

[Alexismex]

it will be fantastic if the output for 18V = 300Ma

[Mustafa]

The shutdown pin is a good idea, I wonder how much time the power supply takes to build up again?

Good question. There is no mention of it in datasheet.

Some quotes from forum archieves:

you can still use a charge-pump DC/DC converter for the analog part. This should be controllable, which means, it should be deactivated by the micro during critical sampling processes. This can be done via PWM output disabling to avoid clocking of the DC/DC converter. Never use a DC/DC converter you can not give the clock timings. The DC/DC converter must be synchronized to the micros clock.

Certainly if you use a switching reg, it should be synchronous. But it can still cause problems because, as the load current varies, the impact of the switching reg will vary.

BTW I think Bugwhiskers is using SMPS in his QED design. Did he explained about it in GoldProspectinginOz? A universal power supply module for metal detectors would be very nice.

[Tinkerer]

PI POWER CONSUMPTION

it will be fantastic if the output for 18V = 300Ma

PI power consumption:
One of the problems of PI detectors in the past, has been the high power consumption, that demanded big batteries.
A PI does not have to consume a lot of power, but rather it is the capability to make a very powerful detector that gives it some advantage over a VLF.

But what is a reasonable power consumption?

The latest Li-Ion batteries have a very good volume/weight/power relation. 10 Amp hours at 12 Volt is still reasonable in weight. That means we could get a full day's detecting between charges, consuming 1000mA.

Now how do we distribute this power? It would seem that the TX should be having the highest priority.
The audio will consume quite a bit if a loudspeaker is used. Earphones will use less. Preferences???
Audio comfort? Piezo elements use less power per decibel, but sound horrible.

I was surprised when I switched off the TX and found that it amounted to less than 30% of the power consumption. So where does the rest of the power go?
Consider a power supply with 70% efficiency. That is 30% loss.
So lets say:
Total power ---------------1000mA
Power supply loss -----------300mA
TX--------------------------300mA
Audio-----------------------100mA
MCU and signal processing---300mA

Hmmm, no LCD display yet, how much will that consume?

OK, I should convert the numbers into Watts to call it power consumption, but anyway, I think you get the idea. I just looked at the mA meter on the bench power supply.

Now, I would prefer to boost the TX and use less for the rest.
Where could I save power?
The power supply is the most obvious place.
Then, the signal processing. Using low power devices could save quite a few mA.

I hope this is a starter. I would much appreciate input and ideas.

Tinkerer

[Tinkerer]

Good question. There is no mention of it in datasheet.

Some quotes from forum archieves:




BTW I think Bugwhiskers is using SMPS in his QED design. Did he explained about it in GoldProspectinginOz? A universal power supply module for metal detectors would be very nice.

Mustafa,

thanks for the input.

The first TINKERERS POWER SUPPLY uses 2 capacitor charge pumps that are both synchronized with the TX and RX. It gives 2 clear spaces of 100uS for the sampling.
This works fine, but the efficiency of the power supply is rather low.
Porkluvr is working on a SMPS with much better efficiency and good filtering. We will have to build it and test it to see the results.

The power supply is really one of the most important parts of the design. It is worth spending time and effort on it. There are many ways to achieve good results.
What if we made more use of the MCU for the power supply?

Tinkerer

[porkluvr]

unrush current limiter needed

Hello all -
I might have weighed in on your discussion sooner but I received no email notification of any activity in this thread.
I have not had time to read your contributions, but let me make this post, take a break, and then I'll see what I've been missing.
&&&&&&&&&&&&&&&&&&

My power supply would suffer from a large inrush current and resulting wear and tear on the input capacitors. I suppose that is a common problem - but I had been ignoring it and I can't do that.

This potential problem came to my attention as I was working to create a new design using an LT1372 SMPS controller. SMD tantalum capacitors that I thought about using would probably not have a high degree of reliability, to say the least.

The 99 amps input current shown here is only a simulated approximation, but if C1 would experience anything near to that amount of current maybe something should be done about it.

Here is an idea for an inrush current limiter explained in Motorola AN1542:
http://www.onsemi.com/pub_link/Collateral/AN1542-D.PDF
After playing with this for a few days I think I have some pretty good values worked out for the circuit shown in figure 5A. We can knock the input current down to a less scary number.

I think that my through hole power supply needs this 'IRL' incorporated. (I get tired of typing "inrush current limiter".) That will be a few days in the making, but I will show the upgraded schematic and simulation results here later this weekend, probably later today...

Maybe it should be abbreviated as "ICL" instead of IRL?
Now time to read.

[porkluvr]

Hi,
What about LM2575 or LM2675.
It has wide input voltage range: 8V to 40V.
External shutdown pin allows deactivation during critical sampling processes.
Below schematic is from http://www.mino-elektronik.de/12VDC/...h_aus12vdc.htm "Multiple power supply from 12V-DC"
+18V@50mA may be a problem for high TX pulse width.

LM2575-5 is a step-down regulator designed for a 5V output. There is a 15V version, but it must be powered from a voltage that is higher than 15V.
In your circuit, +18V is generated is through voltage multiplication. I think that voltage multage multiplication is not as efficient as using a boost regulator (like the LT1172) unless you only need a few tens of milliamps. If you try to draw more current, say 300mA, the voltage loss in the diode string will sap your efficiency.
If you want to draw the most power from the 18V output, that should be the primary power output from whatever regulator you use.
Otherwise what might happen is that as the load requirement on your 5V output changes then the pump action to the diodes would vary in like fashion and your 18V might sag. Similarly, if your 18V load changed it would not be compensated for until it caused a change in the 5V output.

You are deriving 18V from a secondary action but here we need to draw more power from the 18V source than from the others*. That is indirect.

If you want to use something from National you should look at either the LM2585-adj or LM2586-adj boost converters, assuming that you want to use 8V~12V for you input.

*If for some reason this project is going to need more current from +/-5V sources than from +15V circuit, I QUIT.

Maybe I'm just wet behind the ears but I think that 300mA for signal processing and 100mA for audio is ABSURD. Hear me now.

I'll send you my rough drafts and you can carry on without me.

[porkluvr]

Here is how to add an inrush current limiter (hereafter referred to as ICL) to my through hole power supply. I tried, but was unable to combine the ICL and RBP (reverse battery protection) into a single FET circuit.

The two FETs at the power input could be one of the dual FETS available from Vishay or Alpha-Omega, in the SO08 package. AO442 or Si4952, for instance. This is a through-hole project, using only a single sided board so SO08 would need to be mounted on the underside. Using another TO-220 device might be the easiest way to go.


I looked for a way to use an extra winding on L1 that would provide a negative voltage and eliminate the need fot the negative voltage converter IC.
I found that I could have a negative voltage, but if I drew more than only a few milliamps the voltage sagged dramatically. That was not what I was hoping for. I tried some different configurations but none of them made me very happy. That's strange because I simulated a negative converter with the LT1172 and added another winding, and was able to draw 50mA or so with no problem.

You can see turn on at 100uS in the picture.

edit:
I forgot to mention -
AO4468 is a single FET in the SO08. The RC component values I used in the simulation are tailored to that particular FET. If you use a different FET you might find that some of the RC values can be changed for better results, although, what I have here should work alright.

An IRF640 would work, but with two FETs being in the ground path, I would prefer to use devices with lower Rds.

I used the IRF640 in the RBP circuit here because I had trouble getting simulations to run with some of the other models I have.

[porkluvr]

Tinkerer,

I hear you saying that your MPU is really power hungry. In that case, I have been overlooking the most obvious solution, to simply break out separate circuit(s) directly from the battery.

I had been going on the assumption of your signal processing circuits not needing more than about 50mA, and using the LT1965 (or LT1963, or TL1963) as a pregulator before final regulation with LM7805.

For your audio circuit maybe you could use a 16 or 32 ohm speaker to reduce power there. Another possibility, and I wonder why I have not seen this in any of the many detector circuits I have come across, is to use an audio transformer to drive the speaker. It seems to me like it would be a viable solution to take a small (about 1/2") high-permeability toroid core and wind a transformer to drive a speaker. There is a concern that radiation from a transformer could cause interference in low-level amplifier circuits, but toroids are self-shielding to a large degree so that risk is lessened.

It may also be possible to reduce the duty cycle of the speaker drive signal to further reduce power requirements. Instead of having a symmetrical square wave, there are ways to reduce the duty cycle.

That's a general principal and I would need to know more about your circuitry before I could provide a more specific solution.

[Tinkerer]

Tinkerer,

I hear you saying that your MPU is really power hungry. In that case, I have been overlooking the most obvious solution, to simply break out separate circuit(s) directly from the battery.

I had been going on the assumption of your signal processing circuits not needing more than about 50mA, and using the LT1965 (or LT1963, or TL1963) as a pregulator before final regulation with LM7805.

For your audio circuit maybe you could use a 16 or 32 ohm speaker to reduce power there. Another possibility, and I wonder why I have not seen this in any of the many detector circuits I have come across, is to use an audio transformer to drive the speaker. It seems to me like it would be a viable solution to take a small (about 1/2") high-permeability toroid core and wind a transformer to drive a speaker. There is a concern that radiation from a transformer could cause interference in low-level amplifier circuits, but toroids are self-shielding to a large degree so that risk is lessened.

It may also be possible to reduce the duty cycle of the speaker drive signal to further reduce power requirements. Instead of having a symmetrical square wave, there are ways to reduce the duty cycle.

That's a general principal and I would need to know more about your circuitry before I could provide a more specific solution.

Porkluvr,

I am sorry to have confused the issue.
The TINKERERS V1 uses about 300mA TOTAL power at present.
Your SMPS is more that adequate for it.

The confusion comes fro Alexismex's comment of a 18V 300mA TX, which indeed would be a very powerful PI.

My unlucky power consumption guesstimate was merely a reminder why the power supply design is so important. It is very easy to waste 50% of the power on the supply.

Actually I have been very surprised myself how easy it is to waste power. I really started noticing it when I started running the power through an Amp meter.
So, to extract the morale of the story, IT IS IMPORTANT TO LOOK AT POWER SAVINGS FROM BEGINNING TO END.

Your power saving ideas are very welcome.

I will start a new thread for the audio. My knowledge about audio is close to zero. I can really use all the help I can get.

My own hearing is not all that good anymore, so I may not be the best judge of what is a good audio or not, but on the forums one can often see comments about not so perfect audio on all kinds of metal detectors.

Tinkerer

[Mustafa]

Hi porkluvr,
Thanks for the explanations. Please keep up the good work. It is appreciated.

[porkluvr]

Hi porkluvr,
Thanks for the explanations. Please keep up the good work. It is appreciated.

Thanks, Mustafa. I'm trying. I can see where my explanation could have been more informative but I don't know everything and I'm having to learn this stuff as we go along.

I might check out National's "Easy Switcher" software, but there are only so many hours in a day. I've been around switching regulators a lot in the past, but I was only the poor schlub who had to try and fix them when they broke, not somebody who designed them.

There is nothing to be gained by sugar coating the situation. My previous circuit should work, but as we see that more current will be be consumed by the signal processing and audio circuits than I what I had originally estimated, it becomes obvious that there are probably some much more efficient means for providing the 5V supplies. Something should be done.

Several options are available. The simplest would be to tap the 5V supplies right off of the battery. If you want something a little more efficient, adding a second switcher for the 5V is another way to go. In that case, the 15V circuit would be powered by a step-up SMPS and the 5V circuits would be powered by a step-down SMPS. The negative 5V will also need to be worked in. I tried adding another winding to the L1 inductor in my previous circuit with the goal of eliminating the need for the LT1044. It did not work very well, not even for a very small current load.

I could easily produce an upgrade to my previous through hole design, and I may do that later, but I think it makes more sense to work on an all new design. I have some ideas about using LT1372. It switches at 500kHz so it can use a substantially smaller inductor than what the LT1172 uses.

I'll try and post what I've been working on soon, but I need to make some revisions and right now there is nothing to show.

[Tinkerer]

porkluvr,

At present I am using a LM386N1 audio amplifier. I think it could run on unregulated battery power and probably save some that way.

For future designs I will certainly look for low power parts, the savings are quite impressive.

Tinkerer

[Alexismex]

Hello forum,
For audio a max of 25 Ma is just enough to have a very good sound with discretes parts (for ex: with final A13 darlington Npn transistor) but for switching pulse a minimum of 200 Ma and also with a tank of minimum 4700uF...(ideal is 10000 uF,you can see in oscillo a very clean horizontal line after swithing) 200Ma is enough to have a good penetration pulse in medium mineralized ground ...
I speaking of pulse detector for good deep receive target not Vlf for small target.
And you are alwright to say the power supply is the most important part , for me also it is the most important part of the detector...
it is like for a very good audio amplifier it is also the most important part.

Also years ago i make real test to check difference between pulse detector with power to 9v then 12V,then 15V,then 18V and really a 18V pulse supply is much better than 12V... yes I know the magnetic law power ...but in the real life with 18V i have a more strong response of a target deep to 3 meters much much better than if i use 12V....at the same current eX 200Ma.
The big problem in the power supply is the noise transient switching generated by the principal component itself .
Alexis

[Tinkerer]

Hello forum,
For audio a max of 25 Ma is just enough to have a very good sound with discretes parts (for ex: with final A13 darlington Npn transistor) but for switching pulse a minimum of 200 Ma and also with a tank of minimum 4700uF...(ideal is 10000 uF,you can see in oscillo a very clean horizontal line after swithing) 200Ma is enough to have a good penetration pulse in medium mineralized ground ...
I speaking of pulse detector for good deep receive target not Vlf for small target.
And you are alwright to say the power supply is the most important part , for me also it is the most important part of the detector...
it is like for a very good audio amplifier it is also the most important part.

Also years ago i make real test to check difference between pulse detector with power to 9v then 12V,then 15V,then 18V and really a 18V pulse supply is much better than 12V... yes I know the magnetic law power ...but in the real life with 18V i have a more strong response of a target deep to 3 meters much much better than if i use 12V....at the same current eX 200Ma.
The big problem in the power supply is the noise transient switching generated by the principal component itself .
Alexis

Hi Alexis,

thanks for the input.
could you post a circuit of the audio?

About the TX pulse. At 15V the TINKERERS V1 TX pulse is about 3 Amps. I consider it to be a medium power PI.
I see quite often posts from people who want high power PI's. Maybe we should look at the possibility of designing a high power "TINKERERS_HP" version.

I would appreciate some input on this idea.

Alexis, since you have some experience in this field, could you tell us about your experiments and recommend some ideas to try?


Tinkerer

[porkluvr]

porkluvr,

At present I am using a LM386N1 audio amplifier. I think it could run on unregulated battery power and probably save some that way.

For future designs I will certainly look for low power parts, the savings are quite impressive.

Tinkerer

I hope you will open up an audio thread soon. You are using LM386N1, but what about your VCO? The shape of the wave going to your audio amplifier affects how much power is required by the amplifier. If you can change the wave shape from a square wave to a 10% Duty Cycle pulse , that should use about 1/5 the power. You probably could not see it from where you are, but smoke came off the top of my head when I read that your audio might consume 100mA.

One thing to watch out for to make sure that your 10% DC signal does not go through an inverting stage and come out as 90% DC when it drives the speaker. You lose in that scenario.