Varying voltage in a circuit

[Kisen]

Hi,

I suspect that what i am about to ask will be OK, but wanted to get opinions.

I have a design i am working on. It is a low power design running from a coin cell. The coin cell voltage is boosted to 5v and stored in a supercapacitor.
The supercap is then feeding a variable regulator, in that i can switch it to 1 of 2 preset voltages. In my case 2.0v and 3.3v.

Typically i intend for the regulator to be in 2.0v and not doing anything particularly exciting. an MCU wakes up periodically and reads some data from sensors and eeprom etc. These devices are all within spec running at 2.0v all capable of upto 3.6v.

There are however 2 occasions where i need more voltage, the 3.3v supply. For transmitting RF and for an LED controller.

I plan to switch the regulator to 3.3v, which will also increase the supply voltage to the afore mentioned devices, periodically. The voltage change has slew control so its not "violent" and i can account for the voltage change in software, i.e waiting for the voltage to transition.

I have never done this before, my supply voltages are normally stable 3.3v etc. So my question is...

Does a device usually experience a negative effect of such voltage fluctuations when done so in a controlled manner?

Thanks

 

[Nigel Goodwin]

Why not just run at 3.3V all the time?, changing supply voltages is quite likely going to cause intermittent problems.

I'm also dubious about the point of the supercap?, generally people seem to try and use them for purposes they aren't suitable for, and of course a coin cell has very little power capacity to start with.

As in all of these types of thread, you'd do a lot better to tell us EXACTLY what you're trying to do, and why this way.

 

[Kisen]

The purpose of the supercap is to store energy that can be discharged quickly. The coincell has about 600mA capacity, but discharging it above a few mA will shorten its overall life. So i am feeding the energy into a supercap with a current limit of 1mA. This takes time to charge the supercap. As such the energy within it should be used correctly.

For a 1F Supercap, the useable energy between 3.3v (Circuit) and 5.0v (Charged) is ~7Joules, whereas the energy between 2.0v and 5.0v is ~10 Joules. Using 3.3v clearly gives me less useable energy if i use it 100% of the time. So the idea i have is to use 3.3v for demanding jobs like LEDs, RF transmission. and then once they are done with fall back to 2.0v where there is plenty of useable energy available while the supercap charge recovers.


Nigel Goodwin You mention intermittent problems... This is what i would like to get to the bottom of.
So to ask the question again... Does a device suffer in any way when it voltage fluctuates between 2 fixed points within its acceptable voltage range?

For example... an eeprom. 1.6v to 3.6v range. If i design it to work on 1.8v it happily will. If i design it to work on 3.3v it happily will. But what happens if it was working on 1.8v does some work, and then moves to 3.3v, does some work, then back to 1.8v again?

 

[MrAl]

The purpose of the supercap is to store energy that can be discharged quickly. The coincell has about 600mA capacity, but discharging it above a few mA will shorten its overall life. So i am feeding the energy into a supercap with a current limit of 1mA. This takes time to charge the supercap. As such the energy within it should be used correctly.

For a 1F Supercap, the useable energy between 3.3v (Circuit) and 5.0v (Charged) is ~7Joules, whereas the energy between 2.0v and 5.0v is ~10 Joules. Using 3.3v clearly gives me less useable energy if i use it 100% of the time. So the idea i have is to use 3.3v for demanding jobs like LEDs, RF transmission. and then once they are done with fall back to 2.0v where there is plenty of useable energy available while the supercap charge recovers.


Nigel Goodwin You mention intermittent problems... This is what i would like to get to the bottom of.
So to ask the question again... Does a device suffer in any way when it voltage fluctuates between 2 fixed points within its acceptable voltage range?

For example... an eeprom. 1.6v to 3.6v range. If i design it to work on 1.8v it happily will. If i design it to work on 3.3v it happily will. But what happens if it was working on 1.8v does some work, and then moves to 3.3v, does some work, then back to 1.8v again?

One idea would be to keep it regulated at 2v while the power hungry part gets the 3.3v.

Another idea would be to change the voltage gradually over time. Even a second is probably enough.

 

[crutschow]

How are you limiting the current from the battery?
A resistor for that purpose will waste significant energy from the battery.

 

[Nigel Goodwin]

The purpose of the supercap is to store energy that can be discharged quickly. The coincell has about 600mA capacity, but discharging it above a few mA will shorten its overall life. So i am feeding the energy into a supercap with a current limit of 1mA. This takes time to charge the supercap. As such the energy within it should be used correctly.

For a 1F Supercap, the useable energy between 3.3v (Circuit) and 5.0v (Charged) is ~7Joules, whereas the energy between 2.0v and 5.0v is ~10 Joules. Using 3.3v clearly gives me less useable energy if i use it 100% of the time. So the idea i have is to use 3.3v for demanding jobs like LEDs, RF transmission. and then once they are done with fall back to 2.0v where there is plenty of useable energy available while the supercap charge recovers.


Nigel Goodwin You mention intermittent problems... This is what i would like to get to the bottom of.
So to ask the question again... Does a device suffer in any way when it voltage fluctuates between 2 fixed points within its acceptable voltage range?

For example... an eeprom. 1.6v to 3.6v range. If i design it to work on 1.8v it happily will. If i design it to work on 3.3v it happily will. But what happens if it was working on 1.8v does some work, and then moves to 3.3v, does some work, then back to 1.8v again?

I would VERY strongly advise that you keep the electronics (processor, memory etc.) running from a consistent and constant voltage - preferably 3.3V - you don't want massively varying voltages.

A common use for super capacitors is to provide high current pulses for an SMS/GPRS modem which required high current pulses, which long life batteries aren't capable of providing.

I would suggest using a switch-mode converter to charge the super capacitor to 5V, then a 3.3V low drop out regulator to give a stable 3.3V supply for the electronics. But a coincell sounds a poor choice for what you seem to be trying to do?.

 

[danadak]

Affected datasheets, some, will have allowable ramp rates (min) for their
operation. Slow ramp rates can cause excessive current draw in their
supply leads, affecting V as well. That leads one to consider lowering
C, but then that bites you in the rearend trying to suppress transients.

You might even aggravate the situation to get a feel for how tolerant
your design is. You can use a f() generator to supply ramps to specific
part Vdd and look at its current consumption.

When you switch the regulator you are under its control loop, so I would use
a scope looking to trigger at out of spec V. To insure you dont get any triggers.
Single shot trigger, slope polarity try both under and over spec for a given
ramp.

 

[ZipZapOuch]

For example... an eeprom. 1.6v to 3.6v range. If i design it to work on 1.8v it happily will. If i design it to work on 3.3v it happily will. But what happens if it was working on 1.8v does some work, and then moves to 3.3v, does some work, then back to 1.8v again?

It is possible that an I/O pin will be triggered as the supply voltage jumps or drops, ADC data collected during the transition may be inaccurate (depending on a lot of things), timers & counter date may be an issue, crystal oscillators can be an issue depending on how accurate the time has to be in your application. The question is a good one, but I've never seen it addressed. Your fear that a battery drains faster when it is run at higher currents must be balanced with, a system that converts from one voltage to another is inherently in efficient so, which causes more power loss?

Can you use sleep functions / shutdown features to save power between events instead of try to change voltage?

 

[Kisen]

How are you limiting the current from the battery?
A resistor for that purpose will waste significant energy from the battery.

This is one of the built in functions of the boost regulator. It can limit from 1mA to 100mA
Its the TPS63900 for anyone who wants to take a look.

 

[Kisen]

It is possible that an I/O pin will be triggered as the supply voltage jumps or drops, ADC data collected during the transition may be inaccurate (depending on a lot of things), timers & counter date may be an issue, crystal oscillators can be an issue depending on how accurate the time has to be in your application. The question is a good one, but I've never seen it addressed. Your fear that a battery drains faster when it is run at higher currents must be balanced with, a system that converts from one voltage to another is inherently in efficient so, which causes more power loss?

Can you use sleep functions / shutdown features to save power between events instead of try to change voltage?

OK, So these are some of the things i hadnt actually considered. I/O triggering and such.

I am going to keep the voltage at 3.3v as it seems that there is too many unknowns, as has been previously mentioned.
I had planned to use sleep/shutdown functions and in some cases use load control switches if Iq was too high for devices in shutdown.

I would VERY strongly advise that you keep the electronics (processor, memory etc.) running from a consistent and constant voltage - preferably 3.3V - you don't want massively varying voltages.

A common use for super capacitors is to provide high current pulses for an SMS/GPRS modem which required high current pulses, which long life batteries aren't capable of providing.

I would suggest using a switch-mode converter to charge the super capacitor to 5V, then a 3.3V low drop out regulator to give a stable 3.3V supply for the electronics. But a coincell sounds a poor choice for what you seem to be trying to do?.

Advise taken!