That's crazy to overvoltage any amp component. This is why we DIY ...
no profit incentive to skimp on de-rating.
60-70V rails - all 100V decoupling
12V for all servo's /buffers (the zeners).
I've noticed that even shorting out .... or slipped probes , reverse polarity
errors - all the screw-ups that could possibly be made .....
I've maybe got a bit scared (when it did not work) - almost NEVER a puff of
"magic smoke".
Spilled liquid - nothing , slipped probe .... one puff of smoke.
A close direct lightning strike did take out some BC550's once , but
the amp survived.
Very easy for us hobbyist's to make amps that will survive till
the electrolytic's fail in >20 years.
PS- what I give the forum .... I usually will burn out on purpose to
make sure no dangerous , catastrophic "trend" is present.
ANY design that runs components close to thermal or electrical limits
for some imagined sound quality advantage is hogwash.
Or , a bad design that runs hot to minimize thermal drift.
Some designs will even be manipulated to compensate for cheap
thermal solutions , manufacturers recommend "breaking in/warming up"
I can't accept that , should be ready in seconds 1001 , 1002 , 100ma Re -
period !!
OS
OS
no profit incentive to skimp on de-rating.
60-70V rails - all 100V decoupling
12V for all servo's /buffers (the zeners).
I've noticed that even shorting out .... or slipped probes , reverse polarity
errors - all the screw-ups that could possibly be made .....
I've maybe got a bit scared (when it did not work) - almost NEVER a puff of
"magic smoke".
Spilled liquid - nothing , slipped probe .... one puff of smoke.
A close direct lightning strike did take out some BC550's once , but
the amp survived.
Very easy for us hobbyist's to make amps that will survive till
the electrolytic's fail in >20 years.
PS- what I give the forum .... I usually will burn out on purpose to
make sure no dangerous , catastrophic "trend" is present.
ANY design that runs components close to thermal or electrical limits
for some imagined sound quality advantage is hogwash.
Or , a bad design that runs hot to minimize thermal drift.
Some designs will even be manipulated to compensate for cheap
thermal solutions , manufacturers recommend "breaking in/warming up"
I can't accept that , should be ready in seconds 1001 , 1002 , 100ma Re -
period !!
OS
OS
Could not agree more. If I want something to work at 20V reliably, I wouldn't use an 18V part- or, for that matter, a 20V part.Well dont everyone get all too excited. I know well reliability issues (CE-UL-repair warentees... been there and done that). I sure am curious though that they dont go up in smoke at 19v and beyond. If no one wants to find out if they are reliable at higher voltages... no problem. I'll do it and run it for >1 month that way.... just to know.... and take a look at them. Then leave it that way for a year or until I get tired of it.
Sure, I would be scared to run product that way too if I didnt try it that way first. Now cap voltage rating is another matter..... many budget electro caps are about to smoke at the rating.... others at +50% rating. Films -same same. However, mil spec is very under rated. REL-CAP for example high-pots each and every cap to 2X and some films to 3X their rated max. (one reason they are larger than others for same voltage). They can easily be used at rated voltage forever at home room temps.
It is just amazing to me that so many transistors and IC can actually still work 'normally' at much higher PS voltages than Absolute Maximum Voltage rated number. Curious.
THx-RNMarsh
Sure, I would be scared to run product that way too if I didnt try it that way first. Now cap voltage rating is another matter..... many budget electro caps are about to smoke at the rating.... others at +50% rating. Films -same same. However, mil spec is very under rated. REL-CAP for example high-pots each and every cap to 2X and some films to 3X their rated max. (one reason they are larger than others for same voltage). They can easily be used at rated voltage forever at home room temps.
It is just amazing to me that so many transistors and IC can actually still work 'normally' at much higher PS voltages than Absolute Maximum Voltage rated number. Curious.
THx-RNMarsh
Last edited:
If you have the facilities to do that then it would make an interesting conversation piece.
We have to remember that the spec takes account of not just voltage but also allowable temperatures and power dissipation at a given output current delivery. While it may not fail for months (or years) at any given over-voltage situation, that may well alter if its also subjected to its max operating temperature and current delivery.
any idea where that info is published for the opamps we might use? Especially breakdown voltage?
I did find a comment by WJ in his book that Process related breakdown voltage for opamps is >50v but is very opamp specific. So maybe +/- 24vdc would work for many opamps but factoring in worst case deviations et al.... +/- 18vdc makes sense as a catch-all safe number.
We can find out which opamps are >50v breakdown alluded to by trail and error -- or The Great Smoke Test to weed-out the marginal/weak ones at the deviation edge..
THx-RNMarsh
Last edited:
I would like to relate an incident that was told to me by a former PA mixing engineer for the Grateful Dead. I spent late 1972 and all of 1973 designing electronics for the 'Wall of Sound' that the GD toured with for a few years.
He told me that he was doing a concert, and everything at first was going great. Then all of a sudden part of the system broke down. Apparently, one of the Sallen-Key 3 pole filters that I designed with a pair of complementary jfets, broke unexpectedly. Why? Well I was operating +/- 24V power supplies with 40
V jfets. He lamented that it would have been better to have slightly more distortion, and have avoided this breakdown. This happened 40 years ago or so, but it was a good lesson for me. Many of us read a simplified data sheet on discrete parts and perhaps IC's, and only see a few key numbers, but there are hidden leakages, etc that can affect the overall reliablity, even though it might work for months or years at the limits of its rating.
He told me that he was doing a concert, and everything at first was going great. Then all of a sudden part of the system broke down. Apparently, one of the Sallen-Key 3 pole filters that I designed with a pair of complementary jfets, broke unexpectedly. Why? Well I was operating +/- 24V power supplies with 40
V jfets. He lamented that it would have been better to have slightly more distortion, and have avoided this breakdown. This happened 40 years ago or so, but it was a good lesson for me. Many of us read a simplified data sheet on discrete parts and perhaps IC's, and only see a few key numbers, but there are hidden leakages, etc that can affect the overall reliablity, even though it might work for months or years at the limits of its rating.
likely the part was a diffusion doped fet, due to the process variation typically the same mask/family parts would be tested/selected/binned by the factory for V rating
Gordon Audio MicPre people claimed to use factory screened discrete jfet at 2x the nominal part V rating
as process controls improve I expect the distribution narrows, parts are rated much closer to their limits today
I know by direct experience some recent decade +/-12 V op amps died when plugged into +/-15
Gordon Audio MicPre people claimed to use factory screened discrete jfet at 2x the nominal part V rating
as process controls improve I expect the distribution narrows, parts are rated much closer to their limits today
I know by direct experience some recent decade +/-12 V op amps died when plugged into +/-15
What sorts of applications are people worrying about needing greater-than-specified rails? A perusal of Samuel Groner's opamp testing (plus most any datasheet) doesn't show material benefits in term of performance (and sometimes degraded performance) at max recommended voltage. Rarely do we need to swing rail-to-rail with opamps, save a phono input on overload.
Saving on a regulator?
Saving on a regulator?
Gotcha. I thought there was something else at play here. Apologies for misunderstanding the direction of the question.
25 milliohm equivalent. That's one hell of a transformer, Weiss domains are likely bribed to stop screeching "Barkhausen" at room temperature.
To name a single-source part: how many AD797 units do you need to measure, to arrive at a reasonable and defensible answer to the question: Is the distribution of breakdown voltages of AD797s Gaussian, yes or no?
Of course it would be more difficult with multi-fab and or multi-company devices like OP27.
You'd want to know the mean and the standard deviation, before blindly running unscreened parts at higher-than-datasheet voltages. If you screened them, or if you pay "We will take the blame Laboratories" to screen them, that's a different kettle of fish. You want nondestructive screening and, if possible, nonharmful screening.
Of course it would be more difficult with multi-fab and or multi-company devices like OP27.
You'd want to know the mean and the standard deviation, before blindly running unscreened parts at higher-than-datasheet voltages. If you screened them, or if you pay "We will take the blame Laboratories" to screen them, that's a different kettle of fish. You want nondestructive screening and, if possible, nonharmful screening.
Last edited:
composite/multiloop op amp circuits let you pick and choose for great input characteristics, and good output properties separately
ADA4870 for instance as a output works from 40 V total supply, swings 1 A out
LME49860 is 44 Vsupply rated
with a little attention to compensation, composite multiloops can realize Vgain in the output op amp local feedback so the input op amp could be operated from lower Vsupply, possibly subregualted and still drive the output op amp to its higher V rails
compensation is easier with much faster output op amps
but if you really need Vswing, the 140 Vsupply LTC6090 ~ 1 MHz unity gain frequency could still be useful at audio
ADA4870 for instance as a output works from 40 V total supply, swings 1 A out
LME49860 is 44 Vsupply rated
with a little attention to compensation, composite multiloops can realize Vgain in the output op amp local feedback so the input op amp could be operated from lower Vsupply, possibly subregualted and still drive the output op amp to its higher V rails
compensation is easier with much faster output op amps
but if you really need Vswing, the 140 Vsupply LTC6090 ~ 1 MHz unity gain frequency could still be useful at audio
- Status
- Not open for further replies.