Maker Pro
Maker Pro

Low ESR caps for LDO regulator

S

Scott Miller

Jan 1, 1970
0
I'm looking at replacing a 78L05 regulator with a TI TL750L05 low-dropout
regulator - the TI part has a lower dropout voltage, better transient and
reverse bias protection, and can provide more current (though it'd probably
toast my board and melt the enclosure in the process.) Of course, it needs
a low ESR capacitor for stable operation. I've used LDO regs before with
standard tantalums, but this data sheet has me scratching my head.

The data sheet seems to indicate that the ESR needs to be below 0.4 ohms for
best stability. When I start hunting for low ESR tantalums on Digi-Key, I
find that at 10 uF, an ESR of 1.6 ohms or so is about as low as they go. Am
I reading the regulator datasheet wrong, misinterpreting the ESR rating, or
maybe just looking in the wrong place for my parts?

Thanks,

Scott
 
L

Larry Brasfield

Jan 1, 1970
0
Scott Miller said:
I'm looking at replacing a 78L05 regulator with a TI TL750L05 low-dropout regulator - the TI part has a lower dropout voltage,
better transient and reverse bias protection, and can provide more current (though it'd probably toast my board and melt the
enclosure in the process.) Of course, it needs a low ESR capacitor for stable operation. I've used LDO regs before with standard
tantalums, but this data sheet has me scratching my head.

The data sheet seems to indicate that the ESR needs to be below 0.4 ohms for best stability. When I start hunting for low ESR
tantalums on Digi-Key, I find that at 10 uF, an ESR of 1.6 ohms or so is about as low as they go. Am I reading the regulator
datasheet wrong, misinterpreting the ESR rating, or maybe just looking in the wrong place for my parts?

Thanks,


Use a ceramic cap.

Small sintered tantalums are well known for somewhat
high ESR. If you want to spend more money, a foil
tantalum would have lower ESR. But the ceramic
will be cheaper and more stable in the long term, (at
least if you choose the right dielectric).
 
A

Andy

Jan 1, 1970
0
Scott said:
I'm looking at replacing a 78L05 regulator with a TI TL750L05 low-dropout
regulator - the TI part has a lower dropout voltage, better transient and
reverse bias protection, and can provide more current (though it'd probably
toast my board and melt the enclosure in the process.) Of course, it needs
a low ESR capacitor for stable operation. I've used LDO regs before with
standard tantalums, but this data sheet has me scratching my head.

The data sheet seems to indicate that the ESR needs to be below 0.4 ohms for
best stability. When I start hunting for low ESR tantalums on Digi-Key, I
find that at 10 uF, an ESR of 1.6 ohms or so is about as low as they go. Am
I reading the regulator datasheet wrong, misinterpreting the ESR rating, or
maybe just looking in the wrong place for my parts?

Thanks,

Scott

Look for T494C226M016AS at Digi-Key. But why use TL750L05 and not
something more standard, like LM1117?
 
S

Spehro Pefhany

Jan 1, 1970
0
Use a ceramic cap.

Small sintered tantalums are well known for somewhat
high ESR. If you want to spend more money, a foil
tantalum would have lower ESR. But the ceramic
will be cheaper and more stable in the long term, (at
least if you choose the right dielectric).

Do note that instability is possible for ESR too low as well as too
high. ;-)


Best regards,
Spehro Pefhany
 
S

Scott Miller

Jan 1, 1970
0
Look for T494C226M016AS at Digi-Key. But why use TL750L05 and not
something more standard, like LM1117?

Sorry, I should have specified through-hole. 0.1" radial lead spacing.

Aside from the TL75L05 being quite a bit cheaper, I need something in a
TO-92 package - or at least, something that'll fit in the same footprint.
This is for a kit that I've already produced boards for. I've substituted
other, more expensive LDO regs before for special applications that needed
very low dropout voltage (mostly due to very cold batteries), but the idea
here is mostly to make it more resilient to stupid mistakes (reversed
polarity, for example) and harsh conditions like automotive transients.

I've also got another board revision in the works that gives me some more
flexibility - I've added a transient suppressor, RFI protection, and might
have enough room for a TO-220 regulator. This board will be larger and more
expensive, though, and I'll probably keep both versions available.

Scott
 
A

Anders F

Jan 1, 1970
0
Yep - ceramics rule!
Do note that instability is possible for ESR too low as well as too
high. ;-)

Yep - and that s*cks big time. Seems I _need_ to put a big nasty
electrolytic cap on my LM1117 in stead of the nice 22uF ceramics I thought I
could use :-(

Cheers,
Anders
 
A

Anders F

Jan 1, 1970
0
Andy said:
Scott Miller wrote:
Look for T494C226M016AS at Digi-Key. But why use TL750L05 and not
something more standard, like LM1117?

Because the LM1117 dislikes ceramic caps? ;-)

/Anders
 
J

John Popelish

Jan 1, 1970
0
Scott said:
I'm looking at replacing a 78L05 regulator with a TI TL750L05 low-dropout
regulator - the TI part has a lower dropout voltage, better transient and
reverse bias protection, and can provide more current (though it'd probably
toast my board and melt the enclosure in the process.) Of course, it needs
a low ESR capacitor for stable operation. I've used LDO regs before with
standard tantalums, but this data sheet has me scratching my head.

The data sheet seems to indicate that the ESR needs to be below 0.4 ohms for
best stability. When I start hunting for low ESR tantalums on Digi-Key, I
find that at 10 uF, an ESR of 1.6 ohms or so is about as low as they go. Am
I reading the regulator datasheet wrong, misinterpreting the ESR rating, or
maybe just looking in the wrong place for my parts?

Thanks,

Scott

If this is the data sheet you are looking at:
http://rocky.digikey.com/WebLib/Texas Instruments/Web data/TL750L,TL751L Series.pdf

I think that figure 4 tells you that the ESR of a 10 uf output cap
should fall between .024 ohms and .4 ohms, only a 1 to 16 range.
Satisfying this in a production situation may cost a lot more than the
regulator does. They are not very helpful about how the regulator
performs with other capacitances, either, so you will have to sweat
what adding additional bypass caps to the rails will do to the
stability. I would probably try a fairly good aluminum bypass cap a
bit bigger than 10 uf, in parallel with a 1 uf stacked film Panasonic
V series) that is in series with a .1 ohm resistor.

Then I would connect the rest of my system and torment the upstream
side with pulses to see how the output was behaving. That is a long
way from guaranteeing stability with the next thousand regulators,
though. The stability of low drop out regulators is not only
dependent on the load current and capacitor characteristics, but on
the upstream voltage, also. They give me heartburn.
 
T

Terry Given

Jan 1, 1970
0
John said:
If this is the data sheet you are looking at:
http://rocky.digikey.com/WebLib/Texas Instruments/Web data/TL750L,TL751L Series.pdf

I think that figure 4 tells you that the ESR of a 10 uf output cap
should fall between .024 ohms and .4 ohms, only a 1 to 16 range.
Satisfying this in a production situation may cost a lot more than the
regulator does. They are not very helpful about how the regulator
performs with other capacitances, either, so you will have to sweat
what adding additional bypass caps to the rails will do to the
stability. I would probably try a fairly good aluminum bypass cap a
bit bigger than 10 uf, in parallel with a 1 uf stacked film Panasonic
V series) that is in series with a .1 ohm resistor.

Then I would connect the rest of my system and torment the upstream
side with pulses to see how the output was behaving. That is a long
way from guaranteeing stability with the next thousand regulators,
though. The stability of low drop out regulators is not only
dependent on the load current and capacitor characteristics, but on
the upstream voltage, also. They give me heartburn.

Yep. Very frustrating, the tight ESR range spec. An Al electro aint much
use if the temperature gets down to say -20C, as Al electrolytic ESR
then skyrockets. Sanyo OSCON caps have very low, and (cf Al) very stable
ESR. I normally use X7R caps & series R's, also ensuring the peak pulse
power rating of the resistor can hack it (IOW not an 0402 :).

Cheers
Terry
 
J

John Popelish

Jan 1, 1970
0
Terry said:
Yep. Very frustrating, the tight ESR range spec. An Al electro aint much
use if the temperature gets down to say -20C, as Al electrolytic ESR
then skyrockets. Sanyo OSCON caps have very low, and (cf Al) very stable
ESR. I normally use X7R caps & series R's, also ensuring the peak pulse
power rating of the resistor can hack it (IOW not an 0402 :).

The current lasts a very short time and is linited by the regulator,
so I doubt you will ever blow out even an 0402 on start up. But do
you have any trouble finding ESR data for your ceramic caps? I would
probably go with a 1210 size rated for a bit extra voltage (say, 10 or
16 volts, rather than a 6.3 volt unit) to make it more likely that the
ESR was low. And no Y5V or Z5U. X7R or X5R only.
 
T

Terry Given

Jan 1, 1970
0
John said:
Terry Given wrote:




The current lasts a very short time and is linited by the regulator,
so I doubt you will ever blow out even an 0402 on start up. But do
you have any trouble finding ESR data for your ceramic caps? I would
probably go with a 1210 size rated for a bit extra voltage (say, 10 or
16 volts, rather than a 6.3 volt unit) to make it more likely that the
ESR was low. And no Y5V or Z5U. X7R or X5R only.

Hi John,

Very true wrt regulator. OTOH a load short will dump the cap into the
series R.....hows that for paranoid :)

I always measure the ESR. And some manufacturers provide good data (eg
TDK, AVX). Good call wrt voltage rating, and especially the dielectric.
Although by now there should be no excuse for SED patrons to use the
evil Z5U/Y5V dielectrics (except as transducers :)

Cheers
Terry
 
G

Greg Neff

Jan 1, 1970
0
Hi John,

Very true wrt regulator. OTOH a load short will dump the cap into the
series R.....hows that for paranoid :)

I always measure the ESR. And some manufacturers provide good data (eg
TDK, AVX). Good call wrt voltage rating, and especially the dielectric.
Although by now there should be no excuse for SED patrons to use the
evil Z5U/Y5V dielectrics (except as transducers :)

Cheers
Terry

One reason why Z5U/Y5V can be a problem is temperature range and
tempco. In my (industrial) world I have to use X5R or better.

In my experience using small resistors in series with ceramics works
well. We are prohibited from using chip tantalums, and the big
hermetic tantalums tend to have a fairly high ESR.

================================

Greg Neff
VP Engineering
*Microsym* Computers Inc.
[email protected]
 
R

Rich Grise

Jan 1, 1970
0
Sorry, I should have specified through-hole. 0.1" radial lead spacing.

Aside from the TL75L05 being quite a bit cheaper, I need something in a
TO-92 package - or at least, something that'll fit in the same footprint.

According to
http://www.revistapcecia.com.br/downloads/datasheets/LM1117.pdf
page 20, the TO-263 package has little leads on nominally .1 centers -
problem is, they're ground, output, input, which is about as
ass-inside-out from a 78L05 as you can get.

Good Luck!
Rich
 
J

John Woodgate

Jan 1, 1970
0
(in said:
Yep - and that s*cks big time. Seems I _need_ to put a big nasty
electrolytic cap on my LM1117 in stead of the nice 22uF ceramics I
thought I could use :-(

Why not put a very low value resistor in series with the ceramic?
 
T

Tilmann Reh

Jan 1, 1970
0
Scott said:
Sorry, I should have specified through-hole. 0.1" radial lead spacing.

Aside from the TL75L05 being quite a bit cheaper, I need something in a
TO-92 package - or at least, something that'll fit in the same footprint.

--> LP 2950
 
L

legg

Jan 1, 1970
0
Hi John,

Very true wrt regulator. OTOH a load short will dump the cap into the
series R.....hows that for paranoid :)

Not so paranoid. Supply rail mechanical-intermittent short is the kind
of fault that should cause no damage to a power supply. No small smd
film resistors are really appropriate in series with larger caps - but
below 220uF and at low voltage, is less than 100mJ per pop.

RL
 
A

Anders F

Jan 1, 1970
0
Greg Neff said:
Terry Given wrote:
[...] I normally use X7R caps & series R's, also ensuring the peak pulse
power rating of the resistor can hack it (IOW not an 0402 :).
In my experience using small resistors in series with ceramics works
well.

Whart size are you using for what cap sizes? Any tests done?
I actually also considered using a 0805 ferrite. But I'm not quite sure it
will better the EMC on the LDO input anyway...

/Anders
 
G

Greg Neff

Jan 1, 1970
0
Greg Neff said:
Terry Given wrote:
[...] I normally use X7R caps & series R's, also ensuring the peak pulse
power rating of the resistor can hack it (IOW not an 0402 :).
In my experience using small resistors in series with ceramics works
well.

Whart size are you using for what cap sizes? Any tests done?
I actually also considered using a 0805 ferrite. But I'm not quite sure it
will better the EMC on the LDO input anyway...

/Anders

We have thousands of boards in the field using this approach. Our
boards are put through extensive testing because they are used in rail
vehicle applications. What we usually do is use 1 ohm 1206 resistors
in series with 2.2 to 4.7 uF ceramics (X5R or better). We use
multiples of these RC networks to get the desired capacitance and ESR.

================================

Greg Neff
VP Engineering
*Microsym* Computers Inc.
[email protected]
 
A

Andy

Jan 1, 1970
0
Scott said:
Sorry, I should have specified through-hole. 0.1" radial lead spacing.

Aside from the TL75L05 being quite a bit cheaper, I need something in a
TO-92 package

What about L4931CZ50?
 
A

Andy

Jan 1, 1970
0
Anders said:
Because the LM1117 dislikes ceramic caps? ;-)

/Anders

The datasheet for REG1117 specifies only the upper boundary on the ESR
and the lower boundary on the capacity. Does it mean that one can add
the ceramic caps on the output rail without a stability problem, once
the base 10uF ESR<0.5 Ohm tantalum capacitor is there?

-- Andy
 
Top