LM1085: low noise vs stability?

[patwen]

For general digital use, the LT1085 and LM1085 cannot be beaten. However, in order to get the best performance out of them, what kind of output capacitor is recommended?

I am thinking of OSCON together with some SMD ceramics, but I am afraid that the ESR is too low and will make the LM1085 oscillate (common problem for LDO device regulators?).

What is you opinion/experience on this issue? What type of capacitor? At what value?

Thanks

Pat

[Fred Dieckmann]

Has not been a problem for me. Go for it. Three teminal regulators are very stable with low ESR caps. Go read the LT1085 data sheet for somemore background on this.

Fred

[hifiZen]

I don't think you'll have a problem with the OSCONs, but depending on the regulator used, you'll need to double check if you need diodes around the reg to prevent damage when the power supply is turned off, and the large value caps discharge (could take a path back through the reg...).

As far as decoupling caps go, I think you definitely want the SMD ceramics there, as close as possible to the IC (directly beneath is best) to provide a low-impedance path and small loop size for the supply noise currents. However, there are various views on the use of low-ESR electrolytics. Some feel that using low ESR types privides best stabilization of the rail voltage. Others feel that using low-ESR 'lytics adds secondary low-impedance paths to ground, which can interact and cause oscillations and high(er)-Q resonances with the series trace inductances and the other low-ESR paths through ceramic caps and other 'lytics. The proposal is to use standard lytics for this purpose, since their higher ESR will tend to damp and absorb energy, while still providing the low frequency stabilization of the supply voltage. HF currents are then more restricted to the short, low impedance paths through the ceramic chip caps.

I am somewhat undecided on the issue, as I have not done any trials to compare the two approaches, but I'm instinctively leaning toward the use of standard grade lytics with digital circuitry, and low-ESR types for analogue circuits. If for nothing other than the cost savings, I'll stick with standard 'lytics for digital circuitry. I also like to use lots of ferrite beads to block HF currents from straying beyond a small subsection of a circuit. My general goal is to keep the ground plane as free as possible from HF return currents, where avoidable, and to prevent return currents from flowing where sensetive clock or data traces run, or at the very least, I try to get perpendicular return current paths to minimize the interactions.

[Peter Daniel]

Check this thread: The mods never end, adding Goudreau's triplets
especially the link on Goudreau's triplets.

[jwb]

ESR in electrolytics capacitors is inversely proportional to the size of the case. For example, the Panasonic FC series 6.3V caps start out around 2Ω in the smallest 4x7mm package, and end up around 0.014Ω in the largest 18x40mm package. ESR is directly proportional to rated voltage.

With the Linear LT1962 regulators I have been using recently, the ESR can be anywhere from 0 to 3Ω. There is no minimum ESR once the capacitance gets over 10µF. I typically will put 100µF at the regulator output, so there is no problem.

These same regulators need no protection diodes. Modern technology really is wonderful.

[Fred Dieckmann]

"but I'm instinctively leaning toward the use of standard grade lytics with digital circuitry, and low-ESR types for analogue circuits"

Many of the low ESR caps are designed for switching power supplies and digital circuits and are optimized for low parasitic inductance as well as low series resistance. I think you may find that low ESR high frequency caps such as the Panasonic FCs, Black Gates, Oscons, Poscaps, and other Organic semiconductor caps yield equally large sonic improvments for digital and analog circiuts Values of 100uF and smaller can often be used on digital circuits without the addition of ceramic caps in parallel. The non OS DAC am using has no ceramic or film caps on the supplies and sounds great.

"My general goal is to keep the ground plane as free as possible from HF return currents, where avoidable,"

I thought whole point of a ground plane was for the return of HF currents while minimizing ground bounce and other voltages series with the ground references for the digital ICs. That's what Howard told us anyway. I will not argue that high frequency grounding is as much art as science though. And there are many case be case approaches depending on the design goals, cost, simplicity, signal integrety, and EMI concerns. Mixing digital, analog and transient grounds on telecom boards will give you whole new level of respect for the art of grounding. That is if you don't lose your sanity in the process.

http://www.sigcon.com


Fred

[mirlo]

You might consider the "specialty polymer" SMD electrolytic caps from Panasonic. I suspect they have similar to better ESR and lower inductance (due to smaller size) than OS-CONs.

http://www.electromarkinc.com/pdfs/pana061401.pdf

http://info.digikey.com/T023/V5/0603.pdf

Not cheap though.

[Fred Dieckmann]

Try the Sanyo Poscaps The Pansonics are way expensive and do not have better specs as I remember.

http://www.sanyo.com/industrial/comp...scap_home.html

Fred

[mirlo]

Good suggestion. Where do you get them?