Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Help! Fixed voltage linear regulators running way over voltage!

Status
Not open for further replies.

moonie223

Newbie level 4
Joined
Aug 10, 2018
Messages
6
Helped
0
Reputation
0
Reaction score
0
Trophy points
1
Activity points
94
Hello!

I am building a electronic module for use in a car, the main use is data transfer over canbus. The input power is ignition switched 12V, and I have two linear voltage regulators for 5V and 3.3V.

I have assembled my PCB successfully but for some reason I have linear regulators running well outside spec. I am using a NCV5501 for 5V, and a TLV73333 for 3.3V. The NCV5501 is powered by 12V, the TLV7333 is powered by the NCV5501. I also have a separate 20mA 3.3V linear regulator powered by constant 12v battery voltage, only for running "battery" backup for a GPS module.

I am measuring 5.4V from the NCV5501 and 4.0V from the TLV73333. I measure ~250mA of total input power Neither linear regulator gets too hot to touch although the 5V NCV5501 does get warm as expected. I only expected it to use ~2W in normal operation so nothing unmanageable. I measure 3.3V from the 20mA linear regulator as expected!

The GPS module and VR conditioner circuit appear to work properly. The Thermocouple, RS232 and ADC module are untested, I would like to figure out the 3.3V overvoltage before I hook up a microcontroller! I am without ideas for diagnosing why these regulators are running so far outside spec! I have read both datasheets from cover to cover twice. Is it possible I fried them both in reflow? I have a controlled oven and I had ~1:30 in preheat and soak at 150, ~20 in ramp, and ~40 seconds in 180-210C. I can't find specs in the TLV7333 datasheet, but this should be under max conditions according to the NCV5501. I also doubt heat is an issue as the smaller 20mA linear regulator is working fine!

Here is a picture of my schematic and PCB. Please excuse any messiness, my first time doing this!

https://i.imgur.com/v2PZ1BF.png

https://i.imgur.com/EREJTDs.jpg

Thank you in advance for any advise!
 

I am measuring 5.4V from the NCV5501

Some IC's can tolerate +-10 percent deviation from spec.

If you wish to try a simple homebrew voltage regulator, this type is common. The zener diode can also contain diodes and led's.

In some cases you can fine-tune the regulation by:
* adding an inline resistor to increase voltage,
or
* put a low-ohm potentiometer across a diode to obtain lower voltage.

voltage regulation 12V zener NPN to 5V 100mA.png
 

Hi,

I don't think that it's a heat problem.
I assume there is some other problem.

The quality of the schematic is too bad to recognize something.
I recommend to upload pictures using the "insert image" button at the top of the editor window.
Or use the "Go advanced" button to add attachments via the "manage attachments" button.

****
Try to measure the voltages with and without a 200 ohms load.
Give us the values.

Please show us the PCB layout so that we can see each routing layer, or give the EAGLE files.

Use a scope to verify that the voltages are clean without oscillations.

Klaus
 

Also be aware that although your board looks OK, there can be considerable voltage drops around a vehicle wiring from other loads drawing current. You have to be very careful where you measure voltage across. Measuring from chassis will almost certainly give false results, you should measure across the supply pins of each section individually.

Brian.
 

I didn't see in the original post just how far out the 3.3V
regulator is, but many LDOs have no sink capability and
any leakage current in the pass FET/BJT is going to push
the load until the load takes up more than the regulator
wants to leak.

Try adding a shunt load (in fact do a poor man's I-V with
a few load values and see if there's any range where the
regulator "gets right" (any circuit will have one point, but
regulators should stabilize across a useful range - the
lower end of the 3.3V regulator's current range should
be specified somewhere.
 

BradTheRad - The datasheet says the deviation from this NCV5501 shouldn't be more than 3%.

KlausST -

power circuits.png

Here is an annotated view of my power circuits, I tried adding a 1K load to the power rails but saw no difference in the output.

I have a DS1054z, my output voltages are pretty rock solid. It is also how I am measuring these voltages. My multimeter reads a little bit lower, 5.2V and 3.9V.

Here is an image of the board layout, I had to use imgur I think this is too large for the forum.

https://i.imgur.com/O9qlBsw.png

betwixit - I am using a bechtop power supply for testing at the moment. Part of this circuit is a RS232 decoder to read the digital value of the wideband O2 controller directly instead of reading it's DAC with another ADC. Ground offsets always happen and skew the readings at random so you can't compensate for them!

dick_freebird - The 3.3V rail is measuring 4.02V, and the datasheet for this regulator says output from 0mA to 300mA so I didn't think a shunt load should be necessary. Also the GPS module is working and obtains a lock in ~60 seconds. It has to be using at least 30mA between the antenna and receiver itself.

I will try adding a larger load to the 3.3V rail and see if it behaves.

Thanks everyone!
 

Hi,

I personally don´t like the idea of big capacitors at the output.
Doth datasheets taky about values in the range go 0.1 ... 10uF, your values are much higher.
And both talk about fast ceramics capacitors - I assume you did not use ceramics.
On the other hand I didn´t find an information that this will harm stability - as long as you are in the low ESR are like shown in the "stability" charts.

I´ll come back later....

Klaus
 

I have cut all but the 12V input capacitor off the PCB, they are not installed any longer. When I first noticed voltage problems that was my first thought, so they have not been installed for all of this testing.

I used 0.1uF ceramic capacitors on input and output on the 3.3V regulator, although I used y5v instead of e5r/e7r. Could this cause this much of a problem? I can not find a requirement for ESR on the 3.3V regulator datasheet.

I installed 50mA loads on both the 5V and 3.3V outputs, the 5V output shows no change and the 3.3V output drops to 1.9V. I did see an increase of 100mA on the total power draw.

I am starting to think I have an out of spec 5V regulator and a broken 3.3V regulator...

Thank you!!
 

The type of capacitor shouldn't cause a problem but as pointed out, a much larger value than recommended can cause instability. Think of the feedback path from the output pin to the error amplifier having it's signal delayed by the time constant of output impedance feeding the capacitor and you can it is possible to form an oscillator quite easily.

I would suggest an experiment using the same regulator types but away from your present board to see what happens. Is it possible the voltage are creeping in through some other path from a higher potential? For example through pull-up resistors on a 3.3V powered device to the 5V rail.

Brian.
 

I do have a 5V device with outputs pulled up with 3.3V, that should not divert 5V to the 3.3V bus, right? The MCP3008 is also using 5V for Vref and 3.3V for VDD, would this cause a problem?

I had a TVS diode on the canbus lines, I removed that to see if it was for some reason sinking current and saw no change. Other than that I'm not sure what would be trying to sink current.

The GPS module is feeding 3.3V-0.1V on VCC_RF through a 10 ohm resistor into a 27uH inductor for powering the antenna. I selected an inductor with a resonant frequency of 1600mHz, which is close to the RF frequency of GPS. Is it possible I have a defective GPS antenna circuit that is generating voltage somehow? I pretty much directly lifted the antenna design from the datasheets on antenna design.

For clarification, I calculated my 27uH inductor to have an impedance of 270 ohms at GPS frequency, with a bias resistor of 10 ohms I am left with 280 ohms load at 3.2V or ~12 mA to the powered antenna. Is this a correct calculation?

Looking at the schematic for the MCP3008 now, I'm wondering if using 5V on Vref and 3.3V on Vdd falls outside the "All Inputs and Outputs w.r.t. VSS............– 0.6V to VDD +0.6V" If so, I wonder if this is my problem...
 

That was my problem, the MCP3008 was sinking current to VDD from Vref. I had VDD as 3.3V and Vref as 5.0V.

Now I need to find a new way to do my ADC...
 

There are probably protection diodes from input pins (including Vref) to VDD that went into conduction and lifted supply higher.

Your solution is to lower Vref (maybe tie it to 3.3V) then scale the voltage you are measuring to keep it below 3.3V using a potential divider. Be careful to select low enough values if you are rapidly switching analog inputs so you don't limit the input charge time too much.

Brian.
 

I have a slight redesign of this same board I was working on before I found this current problem, I think to fix the ADC I will use voltage dividers and diode clamps to keep them within range of the ADC.

This little mess has lead me to realize input protection using diodes that clamp to the power rail of a linear regulator is probably not a good idea. Can I just add a zener diode on the power rail? Here is a schematic of my proposed revisions.

https://i.imgur.com/oQ6szLs.png

Most schematics I see for zener protection have a resistor between the power rail and the zener diode, is this necessary or will the LDO behave as a limiting resistor in this situation?

https://i.stack.imgur.com/mqxsA.png

Would it be better to source a regulator capable of sinking current instead?

Thank you!
 

Hi,

This little mess has lead me to realize input protection using diodes that clamp to the power rail of a linear regulator is probably not a good idea.
Not for higher current continous overvoltage signals.
But the diodes work good for ESD signals.

Can I just add a zener diode on the power rail?
For sure you can. But be aware of some limits:
* choose a zener voltage that is wihtin all tolerance (production, time, temperature...) to be below the power supply with all it´s tolerance. Mind that the dides already become conductive below their specified voltage... and the "high current" limit is much higher than the specified voltage.
I don´t think a 12V zener at a 12V input makes sense...

I don´t like zeners directely connected to analog inputs.

****
Hint on schematic:
The schematics are somehow hard to read. Please in future follow some rules.
* Signal flow (even power): from left to right.
* Higher voltages on top, lower voltages on bottom
* don´t keep the pin order (of the package) in the symbol of the schematic.

*****
Most schematics I see for zener protection have a resistor between the power rail and the zener diode, is this necessary or will the LDO behave as a limiting resistor in this situation?
--> show schematics you refer to.... (I can´t remember circuits, where protection-zeners are used with a resistor to the power supply)
The given circuit rather seems to be a shunt voltage regulator than a protection circuit. There is no 5V zener (they are 5.1V) but there are 5V shunt reference diodes.
Read the context of the schematics.

*****
Would it be better to source a regulator capable of sinking current instead?
I usually don´t do this. An exception is a power supply for an active line terminiation.

**
Onboard signals usually don´t need to be protected. A proper design usually is sufficient.
External signals usually need ESD protection and EMI/EMC filters.

Klaus
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top