In which case you have no basis for any claim of similarity based on those graphs.
Of course I have no 'proof' as science isn't concerned with proof, that's the domain of mathematics. But seeing as those graphs ostensibly show the LT1083's load transient response is of the order of 20dB better than LT317 I'd say @Markw4 has some objective basis for his claim it sounds better in that particular application.
Also your claim of "order of 20dB better" has no basis as the graphs are not comparable due to mismatching capacitors. Why do you think there would be 20dB difference with similar load regulation spec?
Edit: besides Markw4's claim included modern LDOs many of which have far better load transient responses than LT1083.
Edit: besides Markw4's claim included modern LDOs many of which have far better load transient responses than LT1083.
The output cap is noted to improve transient response, 10uf for LT1083 vs 0 or 1uF on LT317 graphs, you don't think that might have something to do with it?
It seems reasonable to me that LT would wish to paint their parts in the best possible light in their DSs. So to me its a fair comparison to contrast the dotted lines in both instances.
Because that's what the graphs are showing. The load regulation spec omits details (what's the frequency?) so I'm going in favour of the graph with the more relevant details.
Another relevant observation here is the load regulation spec is at constant die temperature. But that's not a real-world situation - the LT1083 has better thermal conductivity die to case (which is shown in the DS).
Go ahead and show those in comparable contexts then.
@lasercut - that might indeed have something to do with it. So for LT317 to be similar to LT1083 in transient behaviour we'd need to see how it does with 10uF output cap - does anyone have that data?
Last edited:
What context might that be? Markw4's comment did not specify any.
Edit: here is one modern LDO.
Attachments
Clearly worse than LT1083 - 22mV deviation on the negative going transient vs less than 100mV for LT1083 with 15X the current step size. To be fair that's not quite apples to apples as the initial current is 100mA for LT1083 as against 10mA here. Output caps are the same.
As this discussion has been about DACs maybe you could estimate how real-world your statement was.
Same output cap though probably lower ESR as its ceramic. Current step size 200mA and 15-20mV deviation. Clearly poorer than LT1083 but pre-load current 50mA vs 100mA.
Got any examples that beat LT1083?
You still have not shown that LT1083 even beats LM317 in similar circumstances. DACs rarely need much current so your arguments are moot. But you are free to try LT1083 with ES9038Q2M or any other DAC.
There's a good reason for that, I'm not here to show that. I'm here to explore why you spout such obvious falsehoods.
So any evidence to support your more recent claim that recent LDOs have 'far better' transient regulation than LT1083? Should I hold my breath?
As I said earlier LT1083 (and LT317) have limited bandwidth. This discussion is about DACs. How well do you think LT1083 (or LT317) will cope with load transients or noise in DAC chips? The examples I showed (with maybe exception to the third) have wider bandwidth so I trust that their load transient responses are better with DACs.
I accept that my point concerning the thermal resistance of a regulator isn't particularly relevant in a DAC application. But here you're saying arguments, plural. So which other point(s) of mine do you consider moot?
Do you consider a reg's transient performance irrelevant for the simple fact that 'DACs rarely need much current' ? If so I'd be interested in your reasoning.
Its clear that the cap on the reg output is handling the high-speed stuff. Load transients would be characterized by current slew rates would they not? The LT1083 is characterized with a 7.5A swing in under 5uS. That corresponds to 7.5mA in under 5nS - I'd hazard that's in the same ballpark as a demanding DAC chip. What current slew rates have you found on your favoured modern LDOs?
You're free to trust them as you wish. But your arguments seem to me to be very hand-wavy rather than rooted in the physics. With 500mS/div on the first one I can't see how that's particularly 'wide bandwidth'.
Like I said you are free to use LT1083 with cap on reg output with ES9038Q2M or any other DAC chip.
I have used modern LDOs very successfully with ES9038 and AK449x chips so my trust is based on real-world experience.
What was your criterion for 'success' ? Did you (for example) make any measurements of noise at the LDO output to see if you were getting the expected DS performance?