Richard-
I remembered why you should not defeat the switch. If you plug in a transistor with the power up there is a good chance it will be destroyed since the base/gate drive will be all the way up.
In the different collector currents try them with the base resistor set to 10 K Ohms. With bipolars the noise current can be significant and goes up with drain current.
I remembered why you should not defeat the switch. If you plug in a transistor with the power up there is a good chance it will be destroyed since the base/gate drive will be all the way up.
In the different collector currents try them with the base resistor set to 10 K Ohms. With bipolars the noise current can be significant and goes up with drain current.
I just checked on new ones -- it is still 50v for the -A version.
I knew they were very linear from curve tracer tests back in they day at LLNL....but am surprised they are so quiet also.
-RM
Couldn't find the mod'ed circuit, probably didn't feel it was worthg saving - this was the Amplifier TPC, dated December 28, 2010. Noted on the original schematic is THD20 = 0.00176, which I confirmed.got a reference, article, posted measurements?
Originally Posted by fas42
An often referred to amplifier circuit put up on the forum by Bob Cordell, with excellent theoretical performance, loses orders of magnitude accuracy when attached to something like real world parts. Plus, when the circuit is pushed a bit in terms of current delivery, slewing, etc, it again is easy to identify areas in the circuit where misbehaviour starts to rear its ugly head - and this is with relatively 'perfect' components ...
First thing I did was to drop the load to 2R, still with perfect supplies; introduced major instability, THD well over 1% - signaling marginal behaviour.
Back to 4R load, and introducing a parasitic resistance of 0.1R in the supplies; this simple, realistic value yielded a THD20 of about 0.05%.
Then add some stray inductance in the rail lines - start with 100nH, not much impact, raised to 200, still OK, but at 250nH some more ugliness appeared: HF oscillations on the voltage rail, and THD20 was now about 0.07%. If the wiring from the supplies to the circuit was not very tightly controlled it would be easy to get that 250nH ...
This hadn't even considered the transformer, rectifier side of the equation ... I didn't feel there was any point in pursuing it further.
low Z dynamic
I check their website (Bella and LJ), I can not find them. I don't know about Emerad.
You mean low? Due to mobility issues PNP's squeeze out a lower rbb in almost any process for equal area.
2n4401/4403 have low rbb, but also fairly low Vceo (40V) and Va, as well as low beta (double digits) - so they're somewhat limited in scope for LTP and VAS duty. BC807/817-40 have fairly low rbb, but also high beta.
The best-sounding LTP BJTs are still the oldie-but-goldie Japanese pairs of 2sc1845/2sa992 and 2sc2240/2sa970 (and their SMD derivatives), which have a combination of moderately low rbb, moderately high Va and high beta.
The best-sounding LTP BJTs are still the oldie-but-goldie Japanese pairs of 2sc1845/2sa992 and 2sc2240/2sa970 (and their SMD derivatives), which have a combination of moderately low rbb, moderately high Va and high beta.
I hate to say it, but you guys are really novices with measuring bipolar transistors for noise, and even linearity.
The 2N2219 devices have been in use since the mid '60's. However, they were bested by the 2N4401/3 combination in 1968. Perhaps, some confusion today comes from manufacturers now sharing a common substrate for a number of 'vintage' parts.
The Rbb' is the key for low voltage noise.
It can be shown that ideally, a bipolar transistor has a basic noise equivalent to 13 ohms when operated at 1ma. This rises to 130 ohms at 100ua, and so forth. This cannot be easily changed as it is related to the intrinsic transconductance of the device. It is not the inverse of the transconductance, but one half that value, so at 1ma, the inverse gm might show 26 ohms, but the noise will be 13 ohms as it is derived from another equation.
Then you have to add the Rbb'. For about 10 years, the 2n4403 with an Rbb' of perhaps 30 ohms, was the device to use for moving coil cartridges, but then in 1978, the Japanese made a number of superior devices that had low Rbb' as low as 2 ohms. This completely changed the noise picture, and obsoletes the discussion here, as what is being discussed at the moment is an almost 50 year old observation, and rather simplistic as well.
Of course, on the other side of the problem is BETA, and how it holds up, especially under low current conditions. This is where the 2N2219 fails, as well has having non optimum Rbb' for low noise input. This is why many parts, often compromised for Rbb', actually work better for higher impedance sourced inputs, where operation at 100uA or less is mandatory. You have to keep the beta up when running low currents. The 2N2219 is not specified to do that very well.
The 2N2219 devices have been in use since the mid '60's. However, they were bested by the 2N4401/3 combination in 1968. Perhaps, some confusion today comes from manufacturers now sharing a common substrate for a number of 'vintage' parts.
The Rbb' is the key for low voltage noise.
It can be shown that ideally, a bipolar transistor has a basic noise equivalent to 13 ohms when operated at 1ma. This rises to 130 ohms at 100ua, and so forth. This cannot be easily changed as it is related to the intrinsic transconductance of the device. It is not the inverse of the transconductance, but one half that value, so at 1ma, the inverse gm might show 26 ohms, but the noise will be 13 ohms as it is derived from another equation.
Then you have to add the Rbb'. For about 10 years, the 2n4403 with an Rbb' of perhaps 30 ohms, was the device to use for moving coil cartridges, but then in 1978, the Japanese made a number of superior devices that had low Rbb' as low as 2 ohms. This completely changed the noise picture, and obsoletes the discussion here, as what is being discussed at the moment is an almost 50 year old observation, and rather simplistic as well.
Of course, on the other side of the problem is BETA, and how it holds up, especially under low current conditions. This is where the 2N2219 fails, as well has having non optimum Rbb' for low noise input. This is why many parts, often compromised for Rbb', actually work better for higher impedance sourced inputs, where operation at 100uA or less is mandatory. You have to keep the beta up when running low currents. The 2N2219 is not specified to do that very well.