TINKERERS POWER SUPPLY

[Tinkerer]

Here is my idea about power and the power supply. Let's discuss that and make the changes that we all can agree on.

An important difference between the PI and the VLF technology, is the power.
Ampere/Turns produce the magnetic field that excites the eddy currents in the target. More field strength = more response.
We will build a fairly powerful IB-PI, pulsing the TX coil with up to 3 Amps current.
A standard Li-Ion polymer battery of 11V and about 5AH should be OK.
We boost the voltage with a capacitor charge pump and regulate it from 8 to 12V
Attached is a schematic. Quite a few improvements can be made. Let's work on it.

Tinkerer

[porkluvr]

... A standard Li-Ion polymer battery of 11V and about 5AH should be OK.
We boost the voltage with a capacitor charge pump and regulate it from 8 to 12V
Tinkerer

My initial impression was that we would be better off having a negative voltage driving the FET because the output (and received input) can be more easily referenced to ground. However, with an IB coil and a separate coil for receive the same constraints as when using a mono coil don't exist. (Duh). So, a positive voltage supply should work just fine . *

But, I think the charge pump is not such a great idea. So, what about I try and devise a switcher? I'm kind of partial to Linear Technology's line. I know they aren't the only players in the field but their software and available models is very convenient. And, LTspice will do efficiency calculations if you know how to ask it.

An LC pi filter would eliminate power loss associated with a linear regulator. If pi filter is unsatisfactory, THEN we can simply add a 7812, or whatever.

If anybody can point me to switchers that have spice models available from other manufacturers, I'll have a look.

*Which reminds me: I have 2/3 devised a PI where I will be using a DD coil and I could probably scrap the negative power supply voltage in favor of positive. (Or else I'll save that for NEXT time.)

[porkluvr]

An LC pi filter would eliminate power loss associated with a linear regulator. If pi filter is unsatisfactory, THEN we can simply add a 7812, or whatever.

Wrongo. There are plenty of LDO regulators to choose from. No need to skimp on having a stiff power supply.

[Tinkerer]

Wrongo. There are plenty of LDO regulators to choose from. No need to skimp on having a stiff power supply.

Thanks for the feedback. The positive Mosfet drive has been working fine on the bread board. It is less common, but the separate RX are less common too.
I feel that my power supply is not very efficient. A well designed boost switcher would probably waste a lot less energy.
At present, the TX power is around 2 to 3 Watt.
If I could apply the lost energy instead to the TX pulse power, it would increase S/N and depth.

It is important though that whatever noise there is, is synchronized with the TX and RX sampling windows, so that the switching noise does not fall on a target or ground sample.

With this first PS version, I synchronize the switching of the ICL7662 and the Darlington's with the TX and RX, as shown in the timing schedule.

I would like to keep the variable voltage regulation. The regulation is very important in this design. However, to switch from low power to high power use we still have the possibility to change the TX pulse length. (which means that we would have to change the timing software.)

Hey, what is a LDO regulator?

Tinkerer

[porkluvr]

LDO = Low Drop Out voltage regulator. Specially designed to operate with a few tenths of a volt overhead. More efficient (and of course more expensive) than LM7812 etc.
LM2931 is an example of a low power LDO regulator. For some higher power versions look here (i love Linear Technologies' power products):
http://www.linear.com/pc/viewCategor...03,C1040,C1055

A regulator circuit like I have in mind can't be synchronized with your clock. Using a filter before the LDO, and proper grounding should prevent noise problems from the power supply. ( - And treating it like a high frequency circuit also helps.)

It might be a good idea for me to make a PCB layout for a separate power board for prototyping. Switchers cannot usually be breadboarded succesfully, not if you want low noise.

I am not in a good place to easily make PCBs but I can create the pattern using Eagle Lite.

[Tinkerer]

LDO = Low Drop Out voltage regulator. Specially designed to operate with a few tenths of a volt overhead. More efficient (and of course more expensive) than LM7812 etc.
LM2931 is an example of a low power LDO regulator. For some higher power versions look here (i love Linear Technologies' power products):
http://www.linear.com/pc/viewCategor...03,C1040,C1055

A regulator circuit like I have in mind can't be synchronized with your clock. Using a filter before the LDO, and proper grounding should prevent noise problems from the power supply. ( - And treating it like a high frequency circuit also helps.)

It might be a good idea for me to make a PCB layout for a separate power board for prototyping. Switchers cannot usually be breadboarded succesfully, not if you want low noise.

I am not in a good place to easily make PCBs but I can create the pattern using Eagle Lite.

Thanks for your patience. Low Dropout Regulator, of course, sometime I am just slow and it does not click.

Yes, a switcher, filter and LDO would probably improve the power efficiency considerably. It would be beneficial to boost the Voltage higher too for the TINKERERS.
The TINKERERS also works with DD coils by the way.

Tinkerer

[porkluvr]

entry number one

Lots to choose from in the way of switch mode controllers. LTC makes it easy to look at their products by way of readily available models for most of their products, and the venerable LTspice - a free and fully functional spice simulator that beats the pants off the competition imho.

I'm just getting started but before I forget let me make a plug for an old favorite, the LT1172. It almost looks like it is custom made for a PI detector because, from the simulation results, it can easily produce 12.4V at 225mA or more at efficiency approaching 90%. The input voltage and inductor selection have a lot of affect on conversion efficiency, so it is not a one-size-fits-all solution. (But what is?)

The main problem with this entry is that it need a pretty large inductor, and in the topology shown here, dual windings. And, it "only" switches at 100kHz so the filter cap(s) also tend to be biggish. Coming in a TO220-5 package it would be hard to destroy and it has two bigger brothers you don't want to meet.

For my personal detector I might go with a 10.5V output and pi filters (and no post regulator) but for a community project maybe 12.4V output is more appropriate, to be followed by a 12V LDO (if anybody knows of one) or an adjustable such as LM2941.

Does anybody know who makes a 12V LDO regulator to handle about 300mA?

[Aziz]

Hi porkluvr and others,

Does anybody know who makes a 12V LDO regulator to handle about 300mA?

It is quite difficult to find such an IC for the required specification. A discrete solution with P-mosfet, op-amp, bandgap reference, some resistors and capacitors would allow a high current (>1 A), adjustable regulator with very low drop out (<200 mV).

If there is an interest for this LDO, we can develop one by ourself.

Aziz

[B^C]

Hi Guys,

Is this the sort of thing your after : http://www.reuk.co.uk/LM2940-12V-1A-...-Regulator.htm

[Tinkerer]

Hi Guys,

Is this the sort of thing your after : http://www.reuk.co.uk/LM2940-12V-1A-...-Regulator.htm

Aziz and B^C,

thanks for the help.
Yes, LDO's will definitely help making a more efficient power supply.
Lets look at the whole power supply and look for other improvements.

For starters we want to add a Mosfet reverse Voltage protection on the Ground.

I am not so sure if it is the best way to take the input of the ICL 7662 for the negative -5V from the regulated TX Voltage. Besides that, I think the ICL7662 is obsolete, I just happen to have a few in my parts box.

General musings:

The present choice of 600 pulse per second PPS seems to be good. It synchronizes with the 60Hz power supply as well as the 50 Hz power supply.
With the pulse length of 100uS the current can rise to 2.5 to 3 Amps with 12 to 15V and 3 to 4 Ohm total coil resistance.

The present Power supply has a large 6800uF capacitor, this seems to work fine too.
100uS out of 1666uS is only about 6% duty cycle, so the actual amps delivered by the power supply are not that high.
There is the question of the step function, when the TX starts. I have not looked at the voltage drop on the capacitor yet, but locking at the energy stored in the capacitor, the whole 100 uS TX only draws a fraction of the energy stored in the capacitor.

So, probably a power supply with about 300mA and 4.5W capacity should be OK.

Surface mount technology is so much better than the old way, but for the prototyping, the old way has its advantages. Specially for me that has great difficulties to produce a PCB. So I would say, let's use the TO220 for the prototyping phase but think of surface mount for a production run.

Tinkerer

[Tinkerer]

Lots to choose from in the way of switch mode controllers. LTC makes it easy to look at their products by way of readily available models for most of their products, and the venerable LTspice - a free and fully functional spice simulator that beats the pants off the competition imho.

I'm just getting started but before I forget let me make a plug for an old favorite, the LT1172. It almost looks like it is custom made for a PI detector because, from the simulation results, it can easily produce 12.4V at 225mA or more at efficiency approaching 90%. The input voltage and inductor selection have a lot of affect on conversion efficiency, so it is not a one-size-fits-all solution. (But what is?)

The main problem with this entry is that it need a pretty large inductor, and in the topology shown here, dual windings. And, it "only" switches at 100kHz so the filter cap(s) also tend to be biggish. Coming in a TO220-5 package it would be hard to destroy and it has two bigger brothers you don't want to meet.

For my personal detector I might go with a 10.5V output and pi filters (and no post regulator) but for a community project maybe 12.4V output is more appropriate, to be followed by a 12V LDO (if anybody knows of one) or an adjustable such as LM2941.

Does anybody know who makes a 12V LDO regulator to handle about 300mA?

porkluvr,

thanks for your help.

Maybe I started this discussion on the power supply the wrong way. I should have started with the battery.
Is my choice the best choice? Are there better alternatives?
I was thinking of a LI-ION polymer battery of a nominal 11.1V and about 5 to 7AH.
The fully charged battery will have 12.75 V and the low discharge limit is 8.55V
We need a regulated Voltage for the TX.
For the IB-PI sampling during the TX, it is an advantage to have a high TX voltage, since it allows to push more Amps into the coil in less time, that is better di/dt and more target response.
These are the reason why I have used a voltage doubler.
Modern switchers are so much more efficient and use a low foot print.
An important matter is the noise.

Tinkerer

[porkluvr]

Thanks to everybody who replied, BC, Aziz, Tinkerer.
I don't use why we can't use LM2940. Sure, there are some newer, higher performance regulators available, but we're not sending a probe
to Mars. I like things simple. Does anybody else? I just wish there were spice models available because I'm having to fake
simulation with other devices. If anybody has a better LDO please present it. (I will do some looking, myself.)

Tinkerer, the Li-PO batterys are an awsome development, lightweight and powerful. Two bad marks against them are:

(1) Li-PO batteries tend to go nuclear if you accidently short their terminals when they are fully charged,
(2) Li-PO batteries are usually not inexpensive.

(Gee, things are tough all over.)

When you want a lightweight, powerful battery (and can afford the best) they are they way to go. I Keep looking for LiPO batteries to
show up on my favorite surplus parts websites, but no such luck.

Who has a circuit for reverse polarity protection?

My tests yesterday showed me that I have been a complete blockhead in wanting to power my own project with 10.5V, and wanting to
reject anything but 12V for the Tinkerer project and was worried that it would be difficult to achieve good efficiency over a
sufficient range of output powers. I could see no need for having a power supply capable of 4.5W, judging from the data I've seen
on the projected coil, coil current, and pulse train waveform.

Efficiency tests results with the LT1172 at 15V are good, almost great. For various inputs between 8-14.4V and output between 150-300ma.
Better than 90 percent efficiency is possible for all combinations of input output. The single 58uH inductor circuit with 0.1Ω series
resistance shouldn't be too tough.

LTspice can make some wacked out calculations sometimes but when you have 20 in a row maybe there's a pattern.

I realized that LT1172 is best when used as a step-up regulator. The circuit I posted the other day is called SEPIC which allows a very
wide range of input voltages, but power efficiency suffers. So long as the step-up mode is used the LT1171 (big brother to LT1172) can
be used and with a smaller inductor. LT1170 can also be used, with an even smaller inductor, all with very good efficiency.

I will look at some more possibilities.
edit: If we use 15V as TX voltage then we can use step-up switchers in the step-up mode even with 14.4V battery input.

[Tinkerer]

porkluvr,

thanks for doing a great job. I think it is a good idea to set the TX voltage to 15V. The LT 1171 looks good, simple and efficient.

Here is a page about batteries of all sorts, including Li-Ion.

http://www.batteryspace.com/index.as...S&Category=897

Tinkerer

[porkluvr]

low noise post regulator

LT1171 can pump harder and thus use a smaller inductor to get the same job done, and also do bigger jobs than the LT1172. Either would probably work OK but 1171 is choice.

There could be several power supply versions. An economy version might use the LM2940 as a post regulator. I have no simulation model for the LM2940-15 so I experimented with some LTC adjustable regulators. I found one I like very much. There may be others but I made some mistakes. Comparison testing did not go well. I may try again, but I don’t want to spend a lot of time.

For a low noise supply I think the LT1963 (adjustable) would be a good choice. LTC provides a model for this device. In simulation I am able to get about 3mV ripple at the output of the post regulator using a two small pi filters, one at the switcher output and another at the post regulator output. I was pushing 4 amps (peak) through the coil at 600pps. With a different pi filter, ripple could be even less.

The LT1963 is $5.50 from Digikey. It's not inexpensive, but probably hard to beat for this purpose.
http://www.linear.com/pc/productDetail.jsp?navId=H0,C1,C1003,C1040,C1055,P2 222

The LT1963 datasheet lists thermal resistance junction to ambient ӨJA = 30ºC /W. By limiting the differential between input and output voltages to <0.5V, and even with 500mA load, the junction temperature should rise only 12.5º C. No heatsink required. LT1963 maximum dropout is 0.35V at 500ma (from -40 º to +125 º) so we’re in good shape.

The LT1171 is rated ӨJA = 75 degºC/W and should dissipate about 0.3W or less throughout the battery's discharge (from 14.4V~8.5V). Temperature rise could be about 25degºC as the battery nears its discharge voltage, when device dissipation would be highest. LT1171 has Tjmax of 100deg º C, so it would not hurt to have a small heatsink although it’s not absolutely necessary.

We need a negative voltage converter for the -5V supply. ICL7662 are still available, but if a resistive drop were between the converter IC and 15V supply, we could safely use a wide range of devices having maximum input voltage ratings well below 15V.

I'll get started on PCB layout. (We do want a separate power board, right? I need one, anyway.)

[Stefan]

LT1172 for SD2000 supply

Hi,

I used a LT1172 for a replacement for an SD 2000 and used initially an LT1963 LDO and a LT1964 LDO to get the positive and negative supply. I changed the LT1963 to a LT1763 to give me 500 mA. I used the LT1172 to generate a + 5.8 vDC from a - 6VDc to -8.2 vDC supply. I also used ferrite bead pi network filtering post the regulator. I built the supply on a prototype PCB board and the results were beyond my expectations. Surpisingly very little noise. I noticed that the pi filter knocked out a lot of the noise from the LT1172 output. The noise input into the LDO was about 30 mV peak to peak. I was not able to measure the output noise from the LDO. At a guess it was less than 2 mV peak to peak.

So the LT1172 is a great chip, easy to use and very robust.

Regards,

Stefan