TINKERERS POWER SUPPLY

[porkluvr]

18V

Alright, this may be a little bit off topic but I'll get this off my chest now.

Alexismex, if you want to use this at 18V you might try the '(ICH) 55071-L9/1' toroid from surplussales.com. I make no guarantee but from their stock number I think there is a very good possibility that this item is a 55071 from mag-inc.com. That would be the next larger size in the CM270060 series of toroid cores, only a different manufacturer. If I had money to burn I would buy a dozen of these but I am leaning more towards developing an SMD supply. The 55071 is no baby, at 1.3" diameter. The more turns you can fit on a core for a given inductance the cooler it should run, for a given bulk. ?!

Using the formula N=sqrt(L/AL), where N=number turns, L=nH, AL=61, wind about 39T of #18AWG to obtain about 95uH. It should work fine if you use (stranded?!) hookup-up wire and can't fit the entire 39T, but you should avoid more than one layer of turns. And, do not have start and finish turns too close together; leave a small gap.

I was thinking about buying several of these to try and upgrade the previous supply I designed but Tinkerer informs me that his power requirements are more for the +/-5V supplies than I earlier suspected, so 55071 would not work any better for me.

Use one of the 55118 or A-263104 cores for L2 and put about 30T of #18 for L2. No worry if those turns make a double layer.
Remember, I have not tested this so make sure that L1 does not get hot at low AND high battery (8V~12.5V). Use good quality Low ESR cap for C2 (or else). (And remember to wear safety glasses blah blah.)

[porkluvr]

Here are some obvious improvements to my original plan. BUZ11 FET has a lower Rds for the inrush current limiter and reverse battery protection. There is less
than 100mV dropped across the two series FETs, even at low battery voltage when current draw is highest.

Another improvement is adding secondary windings to the storage inductor to develope the +5/-5 power sources. I stated in an earlier post that I had a problem
with this approach. My mistake was that I did not simulate a heavy enough load on the primary. That's explained a little better below.

My first circuit taps power from the 15V output and wastes power unnecessarily. That might be OK under some circumstances, but if the analog and digital circuits
draw in the neighborhood of 100mA the wasted power is not trivial. Tinkerer said he doesn't need it, but I want to have an option
for a separate voltage regulator for speaker power - drawing audio power directly from the battery. Efficiency would be less but driving
low impedance speakers would be less likely to cause voltage fluctuations elsewhere.

Several important issues to mention: In this power supply version, power for the +5/-5 loads is parasitically drawn from the action of developing the +15V current.
Without sufficient power drawn in the primary, you won't have solid +5 and -5V. That means that if you use short PW, you need a high PPS number.
The minimum allowable +15V and maximum +/-5V loads are inter-related and will need to be determined and will vary depending on usage. There will be a minimum
allowable transmit power which varies according to the processing overhead.

I still have not determined the total power that can safely be drawn using the L1 magnetic I selected. If my original choice is a disapointment I will need to
move to to the next larger size. The CM330060 is almost twice as expensive (and the CM270060 wasn't cheap) so I keep my fingers crossed.
Mag-inc.com has direct substitutes (550894 and 55071) if you want to read their data sheets. I cannot make sense of the Magnetics company's numbering system
but they have a whole lot of theoretical and component data available on their newly revised website. Thrilling.

The CM330' has a wider inside diameter allowing many more turns of wire than CM270'. I have already invested in some CM270' but having the CM330' might allow
interweaving primary and secondary windings, instead of having to wind the secondary on top as I will need to do.

Surprisingly, CM330060 has a lower inductance index (AL=61) compared to CM270060 (AL=75) meaning that you would need to wind more turns for a given
inductance, and that (in addition to its added bulk) will make it run cooler. There must be a trade off in the size of toroid
core used vs allowable circuit losses, and here I think there are probably clear winners in each category, but that does not necessarily mean that there is one best
choice - unless the CM270060 proves to be clearly inadequate. Clear as mud?

The efficiency calculation performed by LTspice cannot possibly predict how much power would be lost in the magnetic (unless that's what
the "parallel resistance" attribute is for). But, heck - it's free so I can't complain.
Using the largest practical MPP core should go a long way towards minimizing that loss. Cool-Mu (aka High-Flux) and Sendust cores
are second and third choices, but not necessarily in that order.

I wound an L1 inductor primary using CM270060 and #18AWG magnet wire. Now I need to probably get some #20 or #22 Teflon insulated wire for the secondaries.
Pre-wrapping the primary winding with glass tape might be a good idea but I don't have any and don't particurlarly want to lay out the cash for any.
To use the CM330' would probably allow using teflon wire in the primary and using either magnet wire or teflon insulated for the secondaries.
It is important to minimize interwinding capacitance in L1. Teflon insulation makes a nice distance cushion between windings or layers. It would be
impossible to correctly fit the necessary number of turns of #18 teflon wire on CM270' for the primary, leaving a gap between the start and finish. I just now
took a look at my wound inductor and realize that I might need to remove a turn and slide the others around, because I have no gap between the ends. oops. The datasheet says I should be able to fit about five MORE turns, and that ain't gonna' happen.

There is substantial skin effect at 100kHz so stranded wire or even Litz could be effective. I tried to get a handle on calculating the benefit of using Litz wire
and I thought I had it worked out. Then I did a second calculation with a different subject at 500kHz and I came up with a negative number.
I think that was an impossible answer so I gave up. For now.

Remember that rectifier diodes on the secondary will have a positive voltage on one end and negative on the other so using a fifty
volt device is not over-kill. 60V devices are more plentiful (and might be less expensive).

One day I will get the rest of necessary parts together to build and test this power supply. But first I'll need to finish
with a PCB layout.

I used Linear Technologies LDOs in the simulation because they are about the only comapny that I know of who bothers to provide simulation models for regulators.
At least one or two of them will make their way into the physical product because LTC makes good stuff. Hoorah.

[Aziz]

Hi porkluvr,

thanks for your deserving efforts. I was experimenting with SMPS last days and found the timer chip 555 more attractive for this purpose. Because it is easier to get and of course cheaper (but needs an external mosfet, so the cheaper argument melts down). I know, this timer chip will last for a very very long period, whereas other SMPS chips be won't.

Another bonus with 555 is the possibility to synchronize to a system clock (cycle period). The better cmos versions (TLC555, LMC555, ICM7555) could also provide a high switching frequency (rated to 2-3 MHz), more timing and frequency precision and less power consumption.

There is a good source for 555 applications in the internet (including SMPS boost converter). Just have a look.

I like your comments and observations and will follow this interesting stuff here. Keep on good work.

Aziz

[Tinkerer]

Hi porkluvr,

thanks for your deserving efforts. I was experimenting with SMPS last days and found the timer chip 555 more attractive for this purpose. Because it is easier to get and of course cheaper (but needs an external mosfet, so the cheaper argument melts down). I know, this timer chip will last for a very very long period, whereas other SMPS chips be won't.

Another bonus with 555 is the possibility to synchronize to a system clock (cycle period). The better cmos versions (TLC555, LMC555, ICM7555) could also provide a high switching frequency (rated to 2-3 MHz), more timing and frequency precision and less power consumption.

There is a good source for 555 applications in the internet (including SMPS boost converter). Just have a look.

I like your comments and observations and will follow this interesting stuff here. Keep on good work.

Aziz

Aziz,

porkluvr has been making an enormous effort to design an efficient power supply for the TINKERERS. I believe his effort will also be helpful for many other detector designs. It could be looked at as a "universal power supply"
I like the way he uses the simulation to find the most efficient way and components and specially how he spends a lot of time explaining every detail. And how he thinks of every detail, like the reverse polarity protection and the input surge current control.
A good power supply is a very important part of any design.

It would be interesting to see other, different approaches too, so please Aziz, post your SMPS here too.

I might add a word of warning about my own capacitor charge pump design at the beginning of this thread. It functions and I have been running the TINKERERS breadboard off it for some time now, BUT, the efficiency is very low.

Tinkerer

[Aziz]

Hi all,

here is a basic concept of the 555 boost converter:
http://www.elecfree.com/electronic/s...as-controller/

The voltage regulation feedback will be fed via voltage divider into the base of the bipolar transistor (reference voltage = Ube = ~0.65V).

This makes the design much more easy and flexible. The voltage regulation of the boost converter should be seen as a rough regulation. So you generally need to filter and post regulate this.

The reset pin of the 555 can be used with one capacitor, two resistors and one transistor for synchronisation input trigger. The reset pin should be set to ground for few µs with the synch transistor implemented as a mono flop.

The NE555 can be used up to 1 MHz switch clock. The better cmos types can be operated up to 2-3 MHz and some precautions to mosfet driving must be taken (the cmos types have lower drive capability).

Aziz

[Aziz]

Hi,

here is the basic NE555 boost converter. But I do not recommend it. It is just for educational purposes. But its simplicity could be used elsewhere, where the demands are not much critical.

Have fun with spice simulations.
Aziz