Dealing with reference voltages and Vdroop

[Moffitt]

My friend and I are building a telemetry system for a baja car to measure transmission temperature, fuel level, RPM, etc. that would include a wireless module for realtime monitoring as well as an SD card for backup, all powered from (probably) three 1.2V NiCd AA batteries in series with a voltage booster to get 5V.

From past experience building a circuit to boost the voltage from AAs to 5V, I found that when load is applied to the circuit, the voltage will droop below 5v.

Because we are going to be using ADC converters in our setup that use a reference voltage of 5V, will the droop in voltage effect the results of the data? If so, what can we do to combat this?

 

[berkeman]

My friend and I are building a telemetry system for a baja car to measure transmission temperature, fuel level, RPM, etc. that would include a wireless module for realtime monitoring as well as an SD card for backup, all powered from (probably) three 1.2V NiCd AA batteries in series with a voltage booster to get 5V.

From past experience building a circuit to boost the voltage from AAs to 5V, I found that when load is applied to the circuit, the voltage will droop below 5v.

Because we are going to be using ADC converters in our setup that use a reference voltage of 5V, will the droop in voltage effect the results of the data? If so, what can we do to combat this?

Welcome to the PF. What circuit did you use for your boost DC-DC? It should have pretty good load regulation, say on the order of +/-5%. Can you post the schematic?

 

[Moffitt]

Schematic and parts http://www.ladyada.net/make/mintyboost/parts.html" . I found that this circuit would drop as far as .4v under load, and while that might be fine for charging a USB device, I'm worried about how that will effect our measurements.

I was thinking aboot this last night and it occurred to me that if the pic we are using is receiving less than 5V (4.6V for example), and it is providing the same voltage to a sensor, then the voltage that the pic gets back from the sensor would be a proportional factor lower than had it gotten the full 5V.
Is this correct?

For reference, we are using PIC18F45K20, and considering the TPS61202 boost converter.

TPS61202 schematic is attached
Mouser page for data sheet etc. http://www.mouser.com/ProductDetail/Texas-Instruments/TPS61202DRCTG4/?qs=sGAEpiMZZMuTgk%252bQPI7Id/48AfwjCIF/"


Thanks!

 

[berkeman]

Schematic and parts http://www.ladyada.net/make/mintyboost/parts.html" . I found that this circuit would drop as far as .4v under load, and while that might be fine for charging a USB device, I'm worried about how that will effect our measurements.

I was thinking aboot this last night and it occurred to me that if the pic we are using is receiving less than 5V (4.6V for example), and it is providing the same voltage to a sensor, then the voltage that the pic gets back from the sensor would be a proportional factor lower than had it gotten the full 5V.
Is this correct?

For reference, we are using PIC18F45K20, and considering the TPS61202 boost converter.

Thanks!

Hmm. Not sure why you were getting such poor load regulation. What was your full load current?

Also, since you are wanting to power analog measurement circuitry, doing so directly off the output of a DC-DC converter is generally a bad idea. Even if the average value of the 5V rail has good load regulation, you are going to be getting significant ripple at the switching frequency, which can play havoc with the accuracy of your analog measurements.

A better topology for mixed signal circuits is to boost to something like 6V, and use a Low-Dropout (LDO) linear 5V post-regulator. The dropout voltage of the LDO (and the ripple on the boost output rail) will determine the minimum voltage that you need to boost up to.

You might also consider using two 5V LDOs, with one powering the analog circuitry, and the other powering everything else. This will further reduce the noise in your analog measurements. Be sure to use a star ground layout floorplan, with the power supply circuitry in the middle, the analog circuitry with its regulator on one end (say the left side), and the other circuitry on the other end (say the right side) of the PCB. This star ground floorplan minimizes the shared ground return impedance between the analog circuitry and the rest of the circuitry.

 

[Moffitt]

Hmm. Not sure why you were getting such poor load regulation. What was your full load current?

Full load was around 500mA.

A better topology for mixed signal circuits is to boost to something like 6V, and use a Low-Dropout (LDO) linear 5V post-regulator. The dropout voltage of the LDO (and the ripple on the boost output rail) will determine the minimum voltage that you need to boost up to.

Would this switching frequency be considered noise?

So the LDO removes the 'noise' cause by the switching frequency of the IC? Couldn't this also be done using capacitors?

Thanks for your help.

 

[berkeman]

The ripple comes from the discontinuous nature of the charging of the 5V caps. Part of the cycle there is no charging current (when the bottom of the inductor is grounded), and part of the cycle there is charging current (when the bottom of the inductor is released, and it flys back (up in voltage) and is caught by the output rectifying circuit.

So the 5V on the output caps is rising while being charged, and falling when there is no charging current from the boost. This creates voltage ripple on the output caps at the switching frequency, the amplitude of which depends on the current and the ESR of the output caps. It depends on the value of the caps somewhat too, but that value is generally constrained by feedback considerations of the DC-DC and its stability.

The LDO removes this ripple because it is a linear regulator, and passes continuous current to the load. As long as the input voltage dips due to ripple do not get below the minimum input voltage to maintain the output at 5V, you will have a smooth 5V rail.