It's telling you that various data sheet sites also have various data sheets. They are NOT all the same.
So, if you're really interested in some device, you'd do well to visit severasl data sheet sites, hoping for the best.
Sorry, but that's the way it is. Also try ALLDATASHEET.COM - Datasheet search site, Datasheet search site for Electronic Components and Semiconductors and other semiconductors., if you haven't already.
Hi,
Sure. But also look at the circuit. Your Hafler Amp is SO MUCH clamped all over the place, only the IPS and VAS see any serious voltgaes, no matter what you do the amp. The Ip's cannot see more than +V + 1V if the transistors are operating. Some reserve does little harm, but no need to get paranoid.
Fault conditions are another story. Something I learned the hard way.
Nowadays when I encounter a slid state amp with a blown output stage I religiously change ALL (I mean ALL) semiconductors in the channel, even (especially) those that test fine. I learned that every time I do not do that, I get to see the POS Amplifier again, which I prefer not to...
Which are most relevant depends.
However what few datasheets show are how the parasitics change with changing operating conditions (read changing signal). Modern transistor models for BJT's tend to be Gummel-Poon model based and use H-Parameters, though they still miss significant effects, sadly most Spice models and datasheets have a tendency to be very primitive in what they reveal and hence are next to useless...
The solution is to fully qualify parts for all parameters and to do it for a representative sample... Much work.
I said the japanese have BETTER PARTS FOR A SPECIFIC PURPOSE, compared to the US/Eu generica, which seems all one can get these days (the really horrorshow stuff went out of production ages ago). They also tend to cost more.
Beta in general
Beta linearity around the operating point
Early effect (high early voltage = good)
Fbeta for current driven transistors and Ft and Cbo for voltage driven ones
My first question would be "what is your source impedance".
If the answer is either "variable" or "high" or you would expect more than few dozen mV of RF on the base/gate the answer would be "use a Fet".
A Fet may be gainfully used in front of a BJT stage as follower with it's tail returned to the BJT's common, the combo of FET follower and bipolar voltage amplification can offer better linearity and source impedance independence than either one alone.
A Fet follower into the emitter of a BJT (falsely called "Rush Cascode" - it ain't no cascode) is also quite interesting.
Input stage tends to run at low voltage change with high collector voltage, so early effect issues as such are low, HOWEVER signal levels are low too, and common mode levels are high, so we may still have an Early problem.
So you want a BJT with very high Beta, very high collector voltage (it causes high early voltage) and the two are mutually exclusive but have workarounds. Or you could use a tube...
Basically you are asking complex questions and they never have simple answers.
Ciao T
Sure. But also look at the circuit. Your Hafler Amp is SO MUCH clamped all over the place, only the IPS and VAS see any serious voltgaes, no matter what you do the amp. The Ip's cannot see more than +V + 1V if the transistors are operating. Some reserve does little harm, but no need to get paranoid.
Fault conditions are another story. Something I learned the hard way.
Nowadays when I encounter a slid state amp with a blown output stage I religiously change ALL (I mean ALL) semiconductors in the channel, even (especially) those that test fine. I learned that every time I do not do that, I get to see the POS Amplifier again, which I prefer not to...
Which are most relevant depends.
However what few datasheets show are how the parasitics change with changing operating conditions (read changing signal). Modern transistor models for BJT's tend to be Gummel-Poon model based and use H-Parameters, though they still miss significant effects, sadly most Spice models and datasheets have a tendency to be very primitive in what they reveal and hence are next to useless...
The solution is to fully qualify parts for all parameters and to do it for a representative sample... Much work.
I said the japanese have BETTER PARTS FOR A SPECIFIC PURPOSE, compared to the US/Eu generica, which seems all one can get these days (the really horrorshow stuff went out of production ages ago). They also tend to cost more.
Beta in general
Beta linearity around the operating point
Early effect (high early voltage = good)
Fbeta for current driven transistors and Ft and Cbo for voltage driven ones
My first question would be "what is your source impedance".
If the answer is either "variable" or "high" or you would expect more than few dozen mV of RF on the base/gate the answer would be "use a Fet".
A Fet may be gainfully used in front of a BJT stage as follower with it's tail returned to the BJT's common, the combo of FET follower and bipolar voltage amplification can offer better linearity and source impedance independence than either one alone.
A Fet follower into the emitter of a BJT (falsely called "Rush Cascode" - it ain't no cascode) is also quite interesting.
Input stage tends to run at low voltage change with high collector voltage, so early effect issues as such are low, HOWEVER signal levels are low too, and common mode levels are high, so we may still have an Early problem.
So you want a BJT with very high Beta, very high collector voltage (it causes high early voltage) and the two are mutually exclusive but have workarounds. Or you could use a tube...
Basically you are asking complex questions and they never have simple answers.
Ciao T
By the way, did anybody try modern Chinese copies, how close they are to originals?
300PCS 2SC1815 C1815 GR Transistor TO92 NPN | eBay
300PCS 2SC1815 C1815 GR Transistor TO92 NPN | eBay
And what about such parts, how close they are to originals?
2SA1943 2SC5200 15A 300V Hi-Fi Amp Power Transistor NEW | eBay
2SA1943 2SC5200 15A 300V Hi-Fi Amp Power Transistor NEW | eBay
Hi,
At AMR we test "2nd source" parts fully to destruction for several samples against data obtained from identical tests on original parts.
For example, we use a known thermal impedance heatsink on power transistors,calculate what the DUT should handle safely given tolerances and datasheets and apply that.
So far AMR has declined to give business to Chinese foundries making 2nd source parts as a result, we use SGS, NXP and Fairchild Semi's, essentially exclusively, with the odd Toshiba and Renesas parts taking over.
Given the continuing trends in semiconductors (good parts are becoming progressively unavailable) we are changing design criteria''s to allow the safe use of Chinese cr@p despite their failure to adhere to their published data-sheets, which involves going from 70...100% safety margins with original parts to 300%+ for the Chinese fakes combined with a lot of praying to Buddah.
Ciao T
At AMR we test "2nd source" parts fully to destruction for several samples against data obtained from identical tests on original parts.
For example, we use a known thermal impedance heatsink on power transistors,calculate what the DUT should handle safely given tolerances and datasheets and apply that.
So far AMR has declined to give business to Chinese foundries making 2nd source parts as a result, we use SGS, NXP and Fairchild Semi's, essentially exclusively, with the odd Toshiba and Renesas parts taking over.
Given the continuing trends in semiconductors (good parts are becoming progressively unavailable) we are changing design criteria''s to allow the safe use of Chinese cr@p despite their failure to adhere to their published data-sheets, which involves going from 70...100% safety margins with original parts to 300%+ for the Chinese fakes combined with a lot of praying to Buddah.
Ciao T
Hi,
I'd suggest 300% derating. But buy a few pairs and test them to destruction, then you will know.
Ciao T
I'd suggest 300% derating. But buy a few pairs and test them to destruction, then you will know.
Ciao T
The good sign is, they don't hide behind fake "Toshiba" stamp. I actually intend to use them on 0.5A idle current (1A peak), and what I most care about, is dependence of beta on current.
Hi,
We do not bother with the riff raff who actually make fakes. These you better derate 1000%....
Buy a few pairs and test them fully. You will know.
So far I have found the alternatives that are not pure ripoff severely wanting (lets not talk about those that are). Maybe you'll have better luck.
As an exmple, I tested a nominally taiwan TIPXX at rated power, the SGS part failed at around 150%, the 2nd source one at 75% of datasheet values... The 2nd source part had much worse beta linearity too.
As said, buy a handfull, test them.
If you do place a volume order use escrow with "no quibble return clause". Then spot-test what you get.
Ciao T
We do not bother with the riff raff who actually make fakes. These you better derate 1000%....
Buy a few pairs and test them fully. You will know.
So far I have found the alternatives that are not pure ripoff severely wanting (lets not talk about those that are). Maybe you'll have better luck.
As an exmple, I tested a nominally taiwan TIPXX at rated power, the SGS part failed at around 150%, the 2nd source one at 75% of datasheet values... The 2nd source part had much worse beta linearity too.
As said, buy a handfull, test them.
If you do place a volume order use escrow with "no quibble return clause". Then spot-test what you get.
Ciao T
I had enough trouble persuading a certain author that either the Aldridge or the Boxall circuits even worked, because the drafting was in the "old school" tradition of showing the signal path unadorned with bias circuitry. Finally, once obtained (thanks to Samuel Groner), Boxall's article showed precisely the Larson-Baxandall stage, in one figure.
I don't think I'll attempt to argue now that tubes already did everything
Of course at least with Boxall/Larson et al., the lack of a complementary polarity and low operating voltages precludes at least that specific circuit.Brad
Tvr, who said that you should 'google' the data sheets? I SAID that you should find a SEMICONDUCTOR HANDBOOK (used) through the internet, if necessary, so that you can have the full range of data on many parts you are interested in, like the 2N5551, or the 2N5401.
Now please understand that while I have almost all the data sheets, and almost 50 years of design experience with the devices, I am only trying to help, not 'criticize' you unduly.
I have been 'criticized' here for my lack of complete understanding of quantum mechanics, etc.. This is so, because when I graduated with a BA in physics, I had not seriously studied nuclear physics, just Newtonian physics slightly updated.
I, like you are doing with amplifier design, about 15 years ago, really tried to learn nuclear physics on my own. I bought a lot of books from many of the greatest scientists over the last 150 years, concentrating mostly, in this time period, to 'bootstrap' myself up on the quantum mechanical explanations of how things really work. Of course, I 'improved' a little bit, but unfortunately nobody, including some of my critics here, (you know, the PhD physicists) attempted to steer me in the right direction. They would say things like: 'You are clueless' or some such, without helping me find the path. I finally sort of gave up trying.
Keep on trying, tvr, but please try to understand and accept what T, BC and I are trying to help you with. At least, we are trying to help.
Now please understand that while I have almost all the data sheets, and almost 50 years of design experience with the devices, I am only trying to help, not 'criticize' you unduly.
I have been 'criticized' here for my lack of complete understanding of quantum mechanics, etc.. This is so, because when I graduated with a BA in physics, I had not seriously studied nuclear physics, just Newtonian physics slightly updated.
I, like you are doing with amplifier design, about 15 years ago, really tried to learn nuclear physics on my own. I bought a lot of books from many of the greatest scientists over the last 150 years, concentrating mostly, in this time period, to 'bootstrap' myself up on the quantum mechanical explanations of how things really work. Of course, I 'improved' a little bit, but unfortunately nobody, including some of my critics here, (you know, the PhD physicists) attempted to steer me in the right direction. They would say things like: 'You are clueless' or some such, without helping me find the path. I finally sort of gave up trying.
Keep on trying, tvr, but please try to understand and accept what T, BC and I are trying to help you with. At least, we are trying to help.
Hi,
Well, being someone from behind the Iron Curtain I am not familiar with those gentlemen and I do not intend to retake all of this simply for historical accuracy. Sorry.
We are talking fundamental principles here, not devices.
Then again, the US/WT-PTO seems to have taken complete leave of any sense anyway so you can patent circuitry that was public domain 50 years ago...
Ciao T
Well, being someone from behind the Iron Curtain I am not familiar with those gentlemen and I do not intend to retake all of this simply for historical accuracy. Sorry.
I don't think I'll attempt to argue now that tubes already did everything
We are talking fundamental principles here, not devices.
Then again, the US/WT-PTO seems to have taken complete leave of any sense anyway so you can patent circuitry that was public domain 50 years ago...
Ciao T
No need to apologize. I will continue to strive for historical accuracy, duly mindful that once a name gets associated with a circuit, or when a phrase is imprecise but understood by most anyway (like "linear beta" used when "constant beta" is meant), change is nearly impossible.
And I argued along fundamental principles to no avail with the particular individual.
And yes, the patent system is absurdly dysfunctional and has been for quite a while. And patent protection is useless to the small inventor anyway. I may have mentioned a recent one, being highly touted as a breakthrough by its co-inventor, a power amplifier supposedly that we've all been impatiently awaiting. The only problem is it doesn't work very well, and it's prior art dating back at least to 1991. Another amplifier designer who's been mentioned in here said "the primary claims are suitable as a wrapper for dead fish".
Brad
John,
John, with the greatest respect, I think you do need to drink either better vintage wine or drop the occasional button of peyote.
For me personally a mixture of Lysergic acid diethylamide with a smidgen of benzoylmethylecgonine taken with halve a bottle of Möet et Chandon works quite well for mind expansion.
But what heck, I'm generation Tekkno and used to drop Acid at the Tresor in Berlin in the early 90's, when I did not happen to "drop records" or "dropping rhymes" (Plattenleger/Moderator/Rapper)...
Current physics try to bridge the gap between Newtons billiard balls and Heisenberg's clouds of probability, while mostly missing the main point.
Nothing like theory before practice...
Ciao T
John, with the greatest respect, I think you do need to drink either better vintage wine or drop the occasional button of peyote.
For me personally a mixture of Lysergic acid diethylamide with a smidgen of benzoylmethylecgonine taken with halve a bottle of Möet et Chandon works quite well for mind expansion.
But what heck, I'm generation Tekkno and used to drop Acid at the Tresor in Berlin in the early 90's, when I did not happen to "drop records" or "dropping rhymes" (Plattenleger/Moderator/Rapper)...
Current physics try to bridge the gap between Newtons billiard balls and Heisenberg's clouds of probability, while mostly missing the main point.
Nothing like theory before practice...
Ciao T
- Status
- Not open for further replies.