LP2950 Indecision - How would you do it?

[Old Steve]

I have a need for a <100mA low-dropout regulator with an output voltage of 3.45V to 3.5V

It has to feed two modules, one with a minimum supply voltage of 3.4V, and the other with a maximum supply voltage of 3.6V. (Between a rock and a hard place.)
I'll be feeding it with a minimum of 4.2V or thereabouts. (3 x AA cells in series, which could drop to ~4.2V when close to flat.)
I don't want to specially order a low-power adjustable LDO regulator like an LP2951. I'll never need one again and this is a one-off. Delivery is free, by courier, (RS), but I wouldn't feel right buying just one, and don't need anything else right now.

So I've decided to use an LP2950-3.3 and jack it up by 0.15V. No problem, but I can't make up my mind which way to connect the output capacitor - between the output pin and the ground 'pin', or between the output and ground.
Either would work, but which is preferable? (Keeping in mind that an LP2950 is very unstable without a decent, low-ESR output cap.) >=2.2uF is recommended for the LP2950-3.3

Any input is most welcome. (Please pick A or B below.)



Edit: I modified the circuit - that current should read 3.3mA now

 

[AnalogKid]

The output cap is there to improve system transient response. As such, its return current should *not* be involved with the feedback voltage divider. The 10 uF in the middle of the divider is there to keep impressed output noise out of the control loop. Picture B.

The 10 uF cap across the 47 ohm resistor might cause oscillation; won't know until you try. There are other LDO's that have a fully adjustable version, so you wouldn't be modifying an internal control loop with an external control loop.

ak

 

[Old Steve]

The output cap is there to improve system transient response. As such, its return current should *not* be involved with the feedback voltage divider. The 10 uF in the middle of the divider is there to keep impressed output noise out of the control loop. Picture B.

That makes good sense. It's the way I was leaning, ak, but I wasn't too sure.

The 10 uF cap across the 47 ohm resistor might cause oscillation; won't know until you try.

I thought it was a good idea for the same reason as it's recommended for an LM317 or similar - to stabilise the voltage across the resistor. The load will include a serial to RF board, which will draw 30 mA in very sort bursts of about 50mS per second when transmitting. I thought this cap would stop the output voltage varying too much during these short periods. I'll see what happens when I try it, in that regard, I guess

There are other LDO's that have a fully adjustable version, so you wouldn't be modifying an internal control loop with an external control loop.
ak

True, but I don't want to order just one or two of these since RS Components are being good enough to ship all online orders for free by courier, don't need other parts right now and also don't want to spend extra money buying other parts that I don't need.

As I mentioned in my original post, I could buy an LP2951 otherwise. This is not a project for someone else and is just a series of tests for now. I plan to buy an LP2951 when I do order parts next and replace this temporary regulator.

I know this will do the job in the meantime, once the best layout is determined.

 

[Arouse1973]

I have a need for a <100mA low-dropout regulator with an output voltage of 3.45V to 3.5V

It has to feed two modules, one with a minimum supply voltage of 3.4V, and the other with a maximum supply voltage of 3.6V. (Between a rock and a hard place.)
I'll be feeding it with a minimum of 4.2V or thereabouts. (3 x AA cells in series, which could drop to ~4.2V when close to flat.)
I don't want to specially order a low-power adjustable LDO regulator like an LP2951. I'll never need one again and this is a one-off. Delivery is free, by courier, (RS), but I wouldn't feel right buying just one, and don't need anything else right now.

So I've decided to use an LP2950-3.3 and jack it up by 0.15V. No problem, but I can't make up my mind which way to connect the output capacitor - between the output pin and the ground 'pin', or between the output and ground.
Either would work, but which is preferable? (Keeping in mind that an LP2950 is very unstable without a decent, low-ESR output cap.) >=2.2uF is recommended for the LP2950-3.3

Any input is most welcome. (Please pick A or B below.)

View attachment 22001

Edit: I modified the circuit - that current should read 3.3mA now

I would choose "B" personally. As @AnalogKid said It has better ripple rejection and transient response than "A". Also something to consider is a diode across the device, some devices can be damaged if the output is shorted or just when powering down. However you only have a 10uF on the output so as long as you don't add too much more you should be fine. But for the cost of a diode it might be better to be safe.
Adam

 

[Old Steve]

I would choose "B" personally. As @AnalogKid said It has better ripple rejection and transient response than "A". Also something to consider is a diode across the device, some devices can be damaged if the output is shorted or just when powering down. However you only have a 10uF on the output so as long as you don't add too much more you should be fine. But for the cost of a diode it might be better to be safe.
Adam

Yep, I usually do put a diode backwards across the regulator, Adam, I just didn't worry on that diagram.
After AK's response, I pretty much made up my mind to go with B, but I still haven't had a chance to put it together and test it with and without the capacitor across the 47 ohm resistor. Personally, I think that cap is a good idea, but I'll soon see tomorrow when I test it on a breadboard.

It's for my Mattel Mindflex headset hack. I want to connect a 433MHz RF module directly to the headset's TGAM1 chip to transmit EEG info, but the RF module's datasheet, (APC220), says that it's Vcc min is 3.4V. In reality, 3.3V might be fine, but 3.4 or a fraction higher will be better. I want it to be as reliable as possible. (The headset's TGAM1 chip has a Vcc max of 3.6V, and I especially don't want to blow that one - $$.)

 

[Arouse1973]

Yep, I usually do put a diode backwards across the regulator, Adam, I just didn't worry on that diagram.
After AK's response, I pretty much made up my mind to go with B, but I still haven't had a chance to put it together and test it with and without the capacitor across the 47 ohm resistor. Personally, I think that cap is a good idea, but I'll soon see tomorrow when I test it on a breadboard.

It's for my Mattel Mindflex headset hack. I want to connect a 433MHz RF module directly to the headset's TGAM1 chip to transmit EEG info, but the RF module's datasheet, (APC220), says that it's Vcc min is 3.4V. In reality, 3.3V might be fine, but 3.4 or a fraction higher will be better. I want it to be as reliable as possible. (The headset's TGAM1 chip has a Vcc max of 3.6V, and I especially don't want to blow that one - $$.)

Cool let us know how you get on.
Cheers
Adam

 

[Old Steve]

Cool let us know how you get on.
Cheers
Adam

I will. My headset was due today, but didn't turn up, so tomorrow I guess. Most of the similar hacks that I've seen have used a micro in the headset, but I don't see the need, since the TGAM1 outputs serial data at 9600 baud anyway. I already have everything else ready to go. I want to do motor control etc, but also wrote a 'Processing' sketch so I can view the data in bar-graph form, for diagnostics and practice in concentrating & relaxing to hopefully eventually control 2 channels with it, motor speed and steering, or similar things. All good fun.....
I was inspired by Spike124's Mind-Controlled Robot Car, in the 'Project Logs' section.

 

[Old Steve]

I tested this circuit this afternoon. I only tested the 'B' version and didn't bother with the 'A'.
I had a 100uF electro cap on the input, and 10uF tant caps across the 47 ohm resistor and from output to ground.

From an oscillations point-of-view, it's fine, with or without a cap across the 47ohm resistor. I'll still use one in the final circuit, it will help stop the output voltage jumping up during the short term load increases while the RF module is actually transmitting, I reckon.
Load regulation suffers a little with the resistor added from the ground pin to ground, with the output voltage increasing slightly with load due to the increase in the regulator's ground pin current. I tested with five 330 ohm resistors, adding one at a time:-
1 x 330 ohms: V_out = 3.46V, I_load = 10.5mA
2 x 330 ohms, R = 165 ohms: V_out = 3.46V, I_load = 21mA
3 x 330 ohms: R = 110 ohms: V_out = 3.46V, I_load = 31.5mA
4 x 330 ohms: R = 82.5 ohms: V_out = 3.47V, I_load = 42mA
5 x 330 ohms: R = 66 ohms: V_out = 3.47V, I_load = 53mA

Only 10mV variation with loads from 10mA to 53mA. That's acceptable to me. I'll still buy a couple of LP2951s and do it properly when I place my next RS order, but will stick with the LP2950 for the moment.

Edit: I just tested it running the RF module and a PIC 'F84A and it works fine. The RF module was the part that bothered me, and it's fine on 3.46V.

 

[GPG]

I have a need for a <100mA low-dropout regulator with an output voltage of 3.45V to 3.5V

It has to feed two modules, one with a minimum supply voltage of 3.4V, and the other with a maximum supply voltage of 3.6V. (Between a rock and a hard place.).

http://www.dfrobot.com/wiki/index.php/APC220_Radio_Data_Module(SKU:TEL0005)
Says 3.3V is ok

 

[Old Steve]

http://www.dfrobot.com/wiki/index.php/APC220_Radio_Data_Module(SKU:TEL0005)
Says 3.3V is ok

That's a typo/error. Look under "Specification" in the same Wiki, it says "3.5V to 5.5V".
I have the datasheet for this version, which says 3.5V. (Attached)

I also have the datasheet for the identical Dorji DRF7020-D23 module, which says 3.4V. Although I said APC220 because they're more common, I also have a few of the Dorji modules, and will probably use one of those. They're identical, though, and I tested with an APC220 the other day at 3.46V.

In the end, I needed more parts anyway, so ordered a few LP2951s. I've made a little PCB that will hold the LP2951, 2 resistors and 3 caps, so it's really a moot point now. I'm good to go.

The Mindflex headset finally arrived today, too, so I can get started tomorrow.