Linear's LT3080 Regulator

[jackinnj]

I've had these around since they were introduced, finally came into a situation where it would have been good to employ -- but.....

Here's Z for 10V / 100mA a.c. current is 50mA


Here's Zout when the "Lower Set Resistor" option is chosen -- 499R/49.9K/10K per the figure in the datasheet:


Bypassing the with 47nF/100R at the 10K junction:


Bypassing with 470nF:


Bypassing with 4.7uF:

[Conrad Hoffman]

That's the reason for not leaving resistance wire lying around the bench. It's too easy to pick up a piece and use it for hook-up wire.

Seriously, that does seem like sorta poor performance. Does the data sheet suggest the parts are out of spec, or is that what to expect?

[jackinnj]

Conrad, I "standardized" the Z-measurement system with a 0.100R Panasonic so I'm pretty sure about the result. The data sheet is silent with respect to output impedance (as are most.)

Will check for PSRR and noise next.

[Gopher]

That's really interesting jack. There are clearly a lot of parameters that can affect the output impedance and noise of an IC regulator, not least of which is output current draw which you haven't investigated yet. TNT Audio did a few atricles on a similar topic.

Have you compared this one to the old 317 workhorse at the same Iout?

It's clear that just following the data sheet may not bring the best out of a regulator. Could be the start of a major thread here to optimise regulators.

[jackinnj]

There are a lot of charts of PSRR, Noise and ZOut on Walt Jung's site from his (and Jan's) articles "Regulators or High Performance Audio". Here's some data on the LM317 -- the chart is pretty close to those WJ produced:

[Conrad Hoffman]

I just noticed that it's an LDO reg. They generally don't have as good specs for fairly obvious reasons, so unless I need an LDO, I don't use one.

[jackinnj]

Lower gain of the PNP transistor I guess, and the need for compensation -- this being said I've found some LDO's which sound really good despite measurements which are only so-so.

[qusp]

I disagree, the LT1963A, LT1764A, LT3032 and LT1763 all perform quite well and the LT1083 1084 1085/1033 LDO 3 pins are drop in lm317/337 with somewhat better performance as well as lower dropout

you just have to remember that many of them function very much as part of a system and their dynamic performance is a function of the output capacitance

[jackinnj]

Add the LT3015 to that list -- it's the negative rail complement to the LT1963A -- and one of the great tweaks you can do with these is to remote sense the load.

The LT1963A is limited to max 20V input voltage, so some circuit modification is necessary.

[tomchr]

Originally Posted by Conrad Hoffman

That's the reason for not leaving resistance wire lying around the bench. It's too easy to pick up a piece and use it for hook-up wire.

With a Kelvin connection (as I'm sure jackinnj is using) the resistance of the wire shouldn't matter.

Originally Posted by Conrad Hoffman

Seriously, that does seem like sorta poor performance. Does the data sheet suggest the parts are out of spec, or is that what to expect?

I am currently using the LT3080 in a high voltage regulator application and have performed extensive simulation on it. The sims correlate well with lab measurements as can be seen in This Thread. I'm seeing the same 60-ish mOhm output impedance across the audio band.

It seems the regulator was optimized for operating with just Rset to ground.

That said, I have to take issue with 60 mOhm being "crappy performance", though. By the time the regulator is connected to the load, the PCB traces and/or wiring will add up to several times that. This is why remote sensing was introduced. Unfortunately, I don't think it's possible to do remote sensing with a floating regulator.

If you don't mind switching regulators, I suggest looking at the LM2267x-series from National Semiconductor (now TI). I use one in my Universal Filament Regulator. Its performance is rock solid with about 1.5 mOhm output impedance. You can take the feedback voltage at the load (remote sensing) for minimum output impedance. That's a topic for a different thread, though.

But all in all, I find the LT3080 to be a quite rocking regulator. I challenge anyone to beat its ripple rejection! In my high-voltage setup, I ran it with 50 Vpp (16 V RMS) ripple on the input, and measured 20 uV RMS AC on the output.
It also seems to jive quite well with the simulation results obtained in LTspice.

~Tom