Hello. I am designing a protection device for precise 1.8V circuitry. Requirements:
- Input voltage: 1.5...10V
- Output voltage: 1.8V ± 0,1V
- Output current: 0,05...10mA
- No-load consumption: max 50µA
- Noise immunity: anything between 1,8V and 10V
- Dropout: 0,15V max
- Protections: not required
- BOM list as short as possible
- BOM price no more 1$ @10kpcs
So far I managed to do this:
When not loaded, it is working fine. But when loaded with 75 Ohm resistor, it oscillates badly. The higher the load, the bigger oscillation.
I've added 1µF input and 10µF output ceramics - helped a but, but not by much.
Any thoughts?
BTW, I have also evaluated MIC5235: working good, when noise is under 2Vpp. If over, output voltage drops to below 1V. Schematic just like in datasheet, with caps, etc. How to prevent this behavior?
- Input voltage: 1.5...10V
- Output voltage: 1.8V ± 0,1V
- Output current: 0,05...10mA
- No-load consumption: max 50µA
- Noise immunity: anything between 1,8V and 10V
- Dropout: 0,15V max
- Protections: not required
- BOM list as short as possible
- BOM price no more 1$ @10kpcs
So far I managed to do this:
When not loaded, it is working fine. But when loaded with 75 Ohm resistor, it oscillates badly. The higher the load, the bigger oscillation.
I've added 1µF input and 10µF output ceramics - helped a but, but not by much.
Any thoughts?
BTW, I have also evaluated MIC5235: working good, when noise is under 2Vpp. If over, output voltage drops to below 1V. Schematic just like in datasheet, with caps, etc. How to prevent this behavior?
Last edited:
With some effort, your circuit could be frequency-compensated.
But that's only a minor problem. The real issues will be initial accuracy ( dependent on the NMOS threshold), and the temperature drift, which will be horrendous, since there is no compensation whatsoever.
So, why would you want to do that?
Among the thousands µpower LDO's available, there must be at least one or two that fit your requirements.
First example that Google provides is the 78LC series: dropout=50mV Iq=1.1µA price=0.36$.
Plus, they are compensated, stable and have 2.5% initial accuracy.
http://www.onsemi.com/pub_link/Collateral/MC78LC00-D.PDF
And with some research, you can certainly find even better on all counts.
But that's only a minor problem. The real issues will be initial accuracy ( dependent on the NMOS threshold), and the temperature drift, which will be horrendous, since there is no compensation whatsoever.
So, why would you want to do that?
Among the thousands µpower LDO's available, there must be at least one or two that fit your requirements.
First example that Google provides is the 78LC series: dropout=50mV Iq=1.1µA price=0.36$.
Plus, they are compensated, stable and have 2.5% initial accuracy.
http://www.onsemi.com/pub_link/Collateral/MC78LC00-D.PDF
And with some research, you can certainly find even better on all counts.
Elvee, thanks for this chip. I was looking at main manufacturer (TI, analog, national, linear) part lists and somehow missed this one. It looks promising, will get a sample and evaluate it.
The MIC5235 seems superior on paper, though. Input voltage up to 30V and reverse polarity protection of -20V! Sweet, will not need to implement it separately. Or so it looks on paper.
I am currently evaluating MIC5235 and having major stability issues. Here is my setup:
The schematic doesn't look like much, but it, or I think so, is properly setup:
Input Voltage is 4Vdc, output is set to 2.055V.
And here is what I get with 1.5Vpp noise (upper waveform represents AC-coupled input, lower - output):
And when noise is 2Vpp:
Instability is both amplitude and frequency sensitive. For example, With 1.4Vpp noise, regulator is stable up to 16kHz and goes crazy over that. Ant there are some lower frequencies, such as 8kHz and 4 kHz.
Any thoughts??
The MIC5235 seems superior on paper, though. Input voltage up to 30V and reverse polarity protection of -20V! Sweet, will not need to implement it separately. Or so it looks on paper.
I am currently evaluating MIC5235 and having major stability issues. Here is my setup:
The schematic doesn't look like much, but it, or I think so, is properly setup:
Input Voltage is 4Vdc, output is set to 2.055V.
And here is what I get with 1.5Vpp noise (upper waveform represents AC-coupled input, lower - output):
And when noise is 2Vpp:
Instability is both amplitude and frequency sensitive. For example, With 1.4Vpp noise, regulator is stable up to 16kHz and goes crazy over that. Ant there are some lower frequencies, such as 8kHz and 4 kHz.
Any thoughts??
You'll have to test it: nobody will tell you their circuit misbehaves badly when things get a bit tough.
Test the real regulator in your real conditions. No datasheet (or simulation) will ever replace an actual physical test, as you have already discovered.
I am testing it on my bench with close-to-real conditions. All waveforms are taken on real device. I can't feed it with complex noises until it passes these basic ones.
Here you go. this controller meets all of your design specs, except perhaps the BOM price. I doubt you be able to achieve $1 BOM with any solution.
http://cds.linear.com/docs/Datasheet/16151fas.pdf
Operates with VIN as Low as 1V
No Load: 20uA in Active Mode, <1uA in Shutdown Mode
You would need to use the SEPIC Mode for your output requirement of 1.8V with a Vin from 1.5V to 10V.
If you can do away with the 1.5 to 1.8V step up, a simple low cost buck regulator would work.
http://cds.linear.com/docs/Datasheet/16151fas.pdf
Operates with VIN as Low as 1V
No Load: 20uA in Active Mode, <1uA in Shutdown Mode
You would need to use the SEPIC Mode for your output requirement of 1.8V with a Vin from 1.5V to 10V.
If you can do away with the 1.5 to 1.8V step up, a simple low cost buck regulator would work.
- Status
- Not open for further replies.