Seeking advice on a voltage converter circuit, 3,7V(or ±3,7V) into ±5V.

[David_]

Hello.

I'm sorry for the weird title for this post but it was the best I could do.

Anyway I have begun a new project for a LCR meter and the first part of it all is power supply, I need a symmetrical power supply(Both positive and negative voltages) derived from one of two possible options:
1, A single battery of +3,7V.
2, Two batteries connected to give ±3,7V.

In order to convey the requirements for the power supply let me just tell you about how the device is thought to look like in the end.
This LCR meter will consist of a microcontroller(µC), a couple of data converters, some amplifiers for the front end to buffer the excitation signal and amplify the resulting voltage/current signals. ADUM4160 will be used to isolate a USB bus that will connect the device to a PC running Matlab where I will plot Impedance vs frequency and other stuff. But there will also be a LCD display so it can be used without a PC, oh and while it is connected to a PC I want to charge the batterie(s) with the USB 5V rail.
Oh I almost forgot, I will also design for a DC-bias voltage but that will have to wait to later in the design, but I don't want to use those IC's that convert a positive voltage to a higher positive voltage and a negative voltage because they only enable ≈20mA and I will want to have more current than that.

The meter will use the IV measurement technique to acquire the complex voltage and the complex current and then calculate the phase difference as well as all other parameters wanted.

So that is what will need to be powered and ether I'll use a single 3,7V battery to be converted into ±5V or I'll use two batteries and then convert a +3,7V rail into a +5V rail and a -3,7V rail into a -5V rail.

converting +3,7V into +5V appears to require a Boost DC-DC converter but what about -3,7V? do Boost converters handle both positive and negative voltages? or is there any topology that can produce ±5V from a single 3,7V source?

Then there are the question of how to isolate the USB 5V rail, does anyone have any topology to recommend?

One important requirement, I want a frequency for the converters that does not(nor its harmonics) happen to fall on the same frequency of the excitation signal which I hope will be from 100Hz to 1MHz. But it don't know if it is at all possible for me to design converters running at >1MHz but maybe I'll should give it a go.
The other option is to tune the converters to some weird frequency such as 112kHz and then mark that and its harmonics in software so they are not ever used. But can you really tune converter frequency's with such accuracy to enable say 112kHz? or do I calculate components for 112kHz and then it will run a somewhere near that...

Here it is anyway, I appreciate all comments or suggestions.

Regards
David_

 

[sreepss]

you can use the USB Self powered topology (Use D+, D- and Ground, don't use the USB ports +5V to your system)

 

[KlausST]

Hi,

A lot of text...
But you missed to specify a current for +5V and the current for -5V.

In short: you need to generate +5V and -5V.
There are cheap ready to buy converters with +/-5V output. Sadly they are often not regulated, so the voltage may drift with load current and input voltage. Additionally the output voltage may be not as clean as you need it. (You didn't specify this)

One approach is to use a step up = boost converter. It is regulated output voltage. If you need the voltage to be more clean you could filter it. There are buck converters that can be synchronized to an external frequency.

There converters for negative voltage.

I recommend to look at converter IC manufacturers internet pages. They usually have interactive selection guides to find the best converter.

One example could be LT3471....But search other manufacturers, too.

***
Frequency causing errors.
Specify the max allowed ripple voltage in the supply rails. Then test if a RC or LC filter is sufficient.
If not you may generate higher voltages and use linear LDOs to get clean supply voltages.

But the very important will be the pcb layout. If this is not good, then no filter will help you.

(I'very used a converter 10 mm away from a 16 bit ADC and can't find additional error. The ADC gives stable output signals down to one LSB, even when I change the load of the SMPS)

***
Instead avoiding the same frequencies in SMPS and measurement, you could go the other way: synchronize the SMPS to overtones of your measurement frequency. Control both with the same clock source.
This sounds crazy? Not at all.. All overtones of the SMPS may cause errors in your circuit, but because the are synchronous to your sample frequency the will generate only a small amount of DC offset.
And this DC offset often can be easily cancelled out with software.

With ADCs all frequencies above nyquist frequency will be mirrored to lower frequencies, and because the ADC sampling frequency is synchronous to the SMPS in the end they will be mirrored to 0 Hz = DC.


Klaus

 

[David_]

you can use the USB Self powered topology (Use D+, D- and Ground, don't use the USB ports +5V to your system)

Really? That I would never had thought would be sufficient to charge batteries. I'm not saying that I think you are wrong but I question the constancy of such a supply. I will try it out though since it would be very much easier than using a converter.

I tried to paint up the scenario as best I can and since most components are not jet determined I can't be sure about the current need. But I can with a good probability say that the circuits will need >=100mA but not more than 200mA but that is a calculation done excluding any current draw by the DUT.
How much current can be needed for a versatile LCR meter do you think? I mean regarding the excitation signal and resulting current?

I will replay to all other raised points very soon, by the way if I would change my batteries so as to function as one(3,7V * 2) or change them for one 7,4V battery I have found a ready made reference design from LT using LT8471, it has a reference design designated DC1854A:
http://cds.linear.com/docs/en/demo-board-manual/DC1854AF.PDF
here is LT8471's datasheet if anyone is interested:
http://cds.linear.com/docs/en/datasheet/8471fd.pdf

DC1854 does show a design using one LT8471 + 18 components if I didn't misread it to create a converter that accepts 6-32V on its input and delivers +5V@1,5A and -5V@550mA, that should do it, but I am still looking to see if there are any simpler designs available.
I will have quite a hard time developing the software for this so I am willing to cheapskate my way through the hardware design, I mean a fully functional and independent DC-DC module could be an option if the price allows for it.

- - - Updated - - -

One example could be LT3471....But search other manufacturers, too.

The LT3471 circuit look a lot smaller than LT8471, I will continue searching though and I did not think that there would be so many IC's for this purpose but there seem to exist in abundance.

About that SMPS used 10mm from a 16-bit ADC, what kind of inductor was in use?
I'm not after a very detailed answer to this but I am often concerned with the magnetic field radiating out from inductors in SMPS and I have jet to build a magnetic probe to find out, I have for some time intended to start a new thread about hall-effect sensors since they are the best option for me to build a magnetic sensor though there are confusingly many different kinds of such circuits.

When you say "synchronize the SMPS to overtones of your measurement frequency" do you mean that if I am running a measurement at 10kHz then the converter would be set to perhaps 20kHz or any other harmonic frequency of the measurement frequency?

The only thing that feels strange about such a arrangement is the fact that I intend to use this device to characterize Impedance over frequency over a range of 100Hz to 1MHz and could it not cause problems for the converters if they would have to respond to those wide frequency sweeps?
Obviously they would not be made to follow further down than 10-20kHz if they/it can even get that low as most dc-dc converters seems to be set to >=100kHz.

Regards

 

[D.A.(Tony)Stewart]

Best to read what TI says about Port Hub power and pay attention to the ESR*C values used on +5. Provisions to prevent excessive hub surge and also low ESR < 1 Ohm >>> * C value <100us

https://www.ti.com/general/docs/lit/getliterature.tsp?baseLiteratureNumber=slva049&fileType=pdf

 

[BradtheRad]

I am often concerned with the magnetic field radiating out from inductors in SMPS

A simple voltage doubler can be driven by a 555. You can operate it below 100 Hz. Capacitors do not produce magnetic interference.



This illustrates a concept. It is not necessarily a finished design.

The 555 does not output more than 200 mA. You'll probably need greater current, therefore add a half-bridge between the 555 and the voltage doubler.

The capacitor at the power supply has an effect of reducing reverse current back up through the supply.

- - - Updated - - -

------------------------------------

It is possible to convert positive DC pulses into a negative supply. However this may be impractical because diode drops take a major bite from the output voltage. Therefore consider using diodes with low fwd V.

Or as an alternate, start with -3.7 V (which is one of your options). This will make it easier to get -5V. Use a similar schematic as above, but reverse the orientation of all diodes. You'll need a driver circuit that applies negative pulses to the negative voltage doubler.

 

[KlausST]

Hi,

Inductor: it is a shielded choke from wuerth. PD-S? Don't remember the partname.

Synchronized can also be a 1MHz smps and a 100 Hz ADC. It is important that the switching edge of the SMPS has fixed relation to the ADC hold edge.

Klaus