I haven't researched the 'surge voltage' rating, so can't answer.
From decades of servicing some observations.
Higher environmental temp will shorten life of all electros....the typical 1000/2000 hr rating is for max quoted operating temp, ie 85C or 105C and lifetime dramatically increases with lowering of operating temp.
Larger diameter cap means better sealing of the rubber bung and longer service life...I have rebuilt a rack wall of 25 years continuous operation amplifiers where all caps 10mm and above diameter measured ok, but all 8mm and 5mm diameter caps were bone dry.
Larger physical size/diameter for a given capacitance also usually equates to lowered internal ESR which doubly helps lifetime due to lowered internal heating and better dissipation due to increased surface area.
So, for maximal lifetime keep temps low, use the largest diameter/highest voltage caps that will physically fit and/or are within BOM budget.
Dan.
From decades of servicing some observations.
Higher environmental temp will shorten life of all electros....the typical 1000/2000 hr rating is for max quoted operating temp, ie 85C or 105C and lifetime dramatically increases with lowering of operating temp.
Larger diameter cap means better sealing of the rubber bung and longer service life...I have rebuilt a rack wall of 25 years continuous operation amplifiers where all caps 10mm and above diameter measured ok, but all 8mm and 5mm diameter caps were bone dry.
Larger physical size/diameter for a given capacitance also usually equates to lowered internal ESR which doubly helps lifetime due to lowered internal heating and better dissipation due to increased surface area.
So, for maximal lifetime keep temps low, use the largest diameter/highest voltage caps that will physically fit and/or are within BOM budget.
Dan.
Last edited:
What kind of amplitudes did you find at <<1 hz repetition rates wrt say 10 Hz ?.
Dan.
1/f behavior held up, so the standard mathematical relationships would be applicable.
Lytic lifetime typically doubles for every 10C decrease from the rated temperature. So a 2000 hour 85C cap at 45C will last 32000 hours, or 3.7 years. At a more casual usage of 8 hours a day (albeit full power) it will be 11 years. A 105C cap here will last 44 years. Assuming my calculations are correct.
An ATX power supply I just opened up uses 200V caps. That's slightly under 20% mains tolerance. Based on some searches, puncture voltage for lytics is typically 15% higher than the rated voltage, unless you're using very high voltage caps.
An ATX power supply I just opened up uses 200V caps. That's slightly under 20% mains tolerance. Based on some searches, puncture voltage for lytics is typically 15% higher than the rated voltage, unless you're using very high voltage caps.
See http://www.cde.com/resources/catalogs/AEappGUIDE.pdf
esp page 9.
The over voltage rating is there to help you with various ways the applied voltage may exceed, temporarily, the rated voltage on the cap. Note also there can be ripple voltage on the Dc which needs to be considered in using a cap near its rated dc voltage.
A respectable company such as CDE will adhere to standards as much as possible or tell you exactly how they do their tests. When you buy a cap from some countries/mfr, they do not spec the standard and may rate a part any way they think will help it sell based on price usually. In such cases, leaving a margin of safety is prudent for reliability. Similarly, their life rating (hours at temp) should be viewed with suspicion.
-RNM
Last edited:
Accuphase's latest EQ amp C-37 uses 2SC6000 on IPS of its MC headamp.It is not claimed low noise transistor, but may have low rbb'.
Unfortunately, there is no complementary PNP so it is not symmetric circuit but normal LTP which is not lowest noise design.In addition generally NPN have higher noise than PNP ,I heard.
I notice that a marketed "ultra low noise" Jfet , like the 2sk2394 - has a 1db
noise figure.
Searching all through this forum for "low noise" brought me to the
"vinyl" (phono preamps). Some one mentioned 2sa1015/sc1815.
I have 100's of those.
IT has a .5db NF rating. Better than some fet's.
The issue is Ic , BJT's get noisy when hot - the NF rating was taken with
<500nA Ic. Not really a problem , most of my circuits favor @1ma -
I like cool !!
A tip for low noise BJT's .... a low Z feedback network does wonders.
OS
noise figure.
Searching all through this forum for "low noise" brought me to the
"vinyl" (phono preamps). Some one mentioned 2sa1015/sc1815.
I have 100's of those.
IT has a .5db NF rating. Better than some fet's.
The issue is Ic , BJT's get noisy when hot - the NF rating was taken with
<500nA Ic. Not really a problem , most of my circuits favor @1ma -
I like cool !!
A tip for low noise BJT's .... a low Z feedback network does wonders.
OS
Surge ratings were common on old higher voltage electros. I don't know the history behind it although I suspect it could be from when valve stages were common and resistive HT feeds could rise way above normal operating conditions until the valves were up to temperature and drawing current. Typically you might see a 375 volt part with a 450v surge rating... as big a difference as that.
No modern part should be subject to voltages above its marked value. Its an absolute upper ceiling that must not be exceeded. That includes peak ripple voltages. Leakage currents increase with temperature and applied voltage... the two together can shorten cap life considerably.
All things being about equal otherwise, the base region in a PNP will have higher mobility majority carriers (electrons) and lower resistance than an NPN's base region.
For two general purpose bipolars mentioned nearby, the 2SA1015 has about a 30 ohm rbb', and the 2SC1815 a rather-higher 50 ohms. When you run the devices at a current high enough to make the emitter resistance small in comparison to rbb' (re shows ~half-thermal noise, so at 1mA the noise resistance is about 13 ohms due to it) then you will be limited by these rbb' noise sources. More current may make things slightly better for voltage noise, but at the expense of higher base current and its associated shot and excess noise, which pushes impedances down if the current noise is not to start spoiling things.
The question here is what to use when you are in the "borderline" range. Assuming no greater variations in the grid power supply, I plan on +/- 56V PSU lines for the current stages (40V on secondaries). Toroidal transformer regulation is just 3% difference from Load off to full Load on (as per the declaration).
I had a case like that once before, on an amp project I was not very happy with, so it's questionable how much i can trust my findings. I used 80V caps furst (70V caps are rare in Europe, just 63V which are reasonably priced, next come 80V which are more expensive, and 100V which would have been an overkill). Anyway, using 80V caps the sound I got was a little "loose", not defined well enough. Repeating the exercise with 63V caps, the sound actually improved a little bit. In both cases, same type of electrolytic, same value, same source, just different voltage.
My local grid voltage is nominally 220 VAC, but 80% of the time it measures at 224 VAC. I have never measured more, but sometimes I measure less.
So, what to use? 63V which should be fine, or 80V which is playing it super safe?
I had a case like that once before, on an amp project I was not very happy with, so it's questionable how much i can trust my findings. I used 80V caps furst (70V caps are rare in Europe, just 63V which are reasonably priced, next come 80V which are more expensive, and 100V which would have been an overkill). Anyway, using 80V caps the sound I got was a little "loose", not defined well enough. Repeating the exercise with 63V caps, the sound actually improved a little bit. In both cases, same type of electrolytic, same value, same source, just different voltage.
My local grid voltage is nominally 220 VAC, but 80% of the time it measures at 224 VAC. I have never measured more, but sometimes I measure less.
So, what to use? 63V which should be fine, or 80V which is playing it super safe?
If you want to dink to that, you may have had one too many already
But seriously, where would such an effect hide in measurements? The thought is appealing, but as always in such circumstances, you have to be extra careful in proving/disproving the notion to prevent confirmation bias.
I think I read that even the earth rotation / length of day has 1/f.
The ever rising noise density is no problem because of the ever falling
bandwidth over which it applies.
Because the noise comes from the BJT base current. But then if you're using a pot at the input, at half volume you may still have plenty of noise from the input side base current.
Because we don't measure transients in dynamic (music) mode.
I you agree that different devices, with similar measurements can 'sound' different, and even devices with distortion numbers so little than they are miles under our threshold number, you will leave the camp of the pure objectivists.
If you look at this thread: sound-quality-vs-measurements
you will see that they are just talking about cooking and kitchen equipment.
Because we don't know how to measure the music and our pleasure, and not even what exactly means our traditional measurements VS what we can hear or not (ears + brain), the only serious way to use our measurement instruments is to use them to help-us, while we design, to go in the right direction, and some numbers as minimal threshold.
Because this hi-fi game is a make believe one, you have to rely on your own ears to figure out how good this illusion works for you.
It is like this. You have to be a scientist while you design, and an artist while you judge the quality of the devices you will use or produce.
Any position witch is not balanced between the two is, omho, just stupid.
Sure, but do the lifetime specifications by reputable manufacturers hold up here? I think we can all agree that derating is an objective practice, and that it can be calculated. So, if I've calculated the cap target lifetime based on working temperature and ripple current, and ensure that the rated voltage is not exceeded except within the surge tolerance allowed by the manufacturer, then what else is left to derate?
As Mooly remarked, this is a remnant of the tube era, when voltages could go all over the place during start-up.
For a modern use, simply ignore the surge rating.
As long as the voltage remains within safe limits, applied voltage has a tiny effect on life, contrary to temperature for example.
Some manufacturers have published derating factors of voltage combined to temperature, but they are a minority, and the effect is mostly anecdotal compared to other factors
In general, the forming voltage of (modern) E-caps is somewhere between Vr+15% and Vr+50%.
It can be evaluated by applying a low, constant current and monitoring the terminal voltage: at first, the voltage will steadily increase above the rated voltage, then things will slow down, stabilize or even revert: the voltage you read at that time is more or less the one that was used to manufacture the aluminum foil.
This test is safe enough, provided you stay with low currents (100's of µA, depending on the cap size obviously) and stop as soon as a steady state or reversion is noticed
- Status
- Not open for further replies.