Okay, many of you have probably already seen this at one time or another, but here are the recommendations from the LT1085 spec sheet.
"For all operating conditions, the addition of 150uF aluminium electrolytic or a 22uF solid tantalum on the output will ensure stability."
There are then some words about being able to get away with smaller values if an adjustment pin cap is not used, but that the (above) values will cover the use of an adjustment cap. It then goes on to say:
"Output capacitance can be increased without limit and larger values of output capacitor futher improve the stability and transient response of the LT1083 (/5) regulators."
So the implication is that the output cap does not have to be tantalum, and that's the only one where they call out a specific type. The adjustment pin cap and the input cap are just shown to be polarized (is this really necessary?), with the former needing to have an impedance less than the value of R1.
This tells me that the adjustment pin cap just needs to be a high quality low ESR type like an Os-Con or a BG NX Hi-Q, and both the input and output caps could be electrolytics with maybe just a styrene bypass or an RC bypass. Thoughts?
Interestingly, the spec sheet then goes on to show a number of different designs, some with low value (10uF) input caps, but a couple with huge (50,000uF, 10,000uF) input caps without any smaller bypass values.
"For all operating conditions, the addition of 150uF aluminium electrolytic or a 22uF solid tantalum on the output will ensure stability."
There are then some words about being able to get away with smaller values if an adjustment pin cap is not used, but that the (above) values will cover the use of an adjustment cap. It then goes on to say:
"Output capacitance can be increased without limit and larger values of output capacitor futher improve the stability and transient response of the LT1083 (/5) regulators."
So the implication is that the output cap does not have to be tantalum, and that's the only one where they call out a specific type. The adjustment pin cap and the input cap are just shown to be polarized (is this really necessary?), with the former needing to have an impedance less than the value of R1.
This tells me that the adjustment pin cap just needs to be a high quality low ESR type like an Os-Con or a BG NX Hi-Q, and both the input and output caps could be electrolytics with maybe just a styrene bypass or an RC bypass. Thoughts?
Interestingly, the spec sheet then goes on to show a number of different designs, some with low value (10uF) input caps, but a couple with huge (50,000uF, 10,000uF) input caps without any smaller bypass values.
AndrewT said:
I suspect by the need for higher density (smaller size for the same capacitance), is all. The ESR of a tantalum is lower than a comparable electrolytic, but perhaps mostly due to its internal construction.
I've seen tantalums fail as brand-new units on a motherboard, short-circuit. I doubt if the manufacturer of the mboards didn't test these before they left the factory, so it appears that the tantalums sometimes "just fail". Perhaps not having a bias voltage on them is bad?
Cheers
This is a report about tantalums I have found to illustrate the issue
http://members.aol.com/sbench102/caps1.html
and the main cap comparison
http://members.aol.com/sbench102/caps.html
http://members.aol.com/sbench102/caps1.html
and the main cap comparison
http://members.aol.com/sbench102/caps.html
lndm, I've seen this before. While I have no doubt about the measurements, per se, I am unsure about what appears to be the anecdotal association between how these caps measure and how they sound. There is, for example, in diyhifi a blistering thread about the use of teflon caps in digital circuits. One guy who put them in made it very clear how much better it made his system sound. But everyone else was throwing rocks, saying that in this case a mica cap is way better because of much lower THD. In nether case did anyone really try to associate either the sound or the measurements with the supporting measurement or sound (if that makes sense). This hysteresis page actually bothers me more, because it seems to be saying that high hysteresis caps sound a certain way becuase of that, with no real supporting evidence on the listening side.
Ok tonyptony, thats fair, only hope I get it right.
Umm, for example when you watch the behaviour of a device as it is actively engaging a signal, it is best to see it behave with a linear characteristic. When the resultant is curved (like a banana) it shows the component under those circumstances will tend to produce assymetrical distortion (even order).
The thing that concerns me about the tantalum test is the sharp corners indicating some sort of rapid internal changes in characteristic. On the first photo, these corners are around X=0 to -1 , Y=0 to 1 on the graticule.
I don't claim to know precisely what this will sound like but I have been trained to avoid this unless linearity is not an issue. I expect the spectrum of distortion would be somewhat ugly.
Umm, for example when you watch the behaviour of a device as it is actively engaging a signal, it is best to see it behave with a linear characteristic. When the resultant is curved (like a banana) it shows the component under those circumstances will tend to produce assymetrical distortion (even order).
The thing that concerns me about the tantalum test is the sharp corners indicating some sort of rapid internal changes in characteristic. On the first photo, these corners are around X=0 to -1 , Y=0 to 1 on the graticule.
I don't claim to know precisely what this will sound like but I have been trained to avoid this unless linearity is not an issue. I expect the spectrum of distortion would be somewhat ugly.
Bob Pease discusses hysteresis and "soakage":
http://www.national.com/rap/Application/0,1570,28,00.html
http://www.national.com/rap/Application/0,1570,28,00.html
An externally hosted image should be here but it was not working when we last tested it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.