I have a pair of LCD-2's I have been playing with. They are very good. However they need power and a very low source impedance. I'm not sure why they are so sensitive to source Z but even a few Ohms will seriously degrade the bass. They also need a lot of drive. I built a test amp using the LTC1010 that can drive them to 6V RMS which seems to be enough. 2V RMS is not enough. iPod- forget it.
I will look at the peak current playing loud to see just how much they seem to demand.
I will look at the peak current playing loud to see just how much they seem to demand.
Demian, JC, jcx, et al, is there a simple yet effective way to raise a typical 2V input to 6V without resorting to an op-amp (assuning a high load impedance and looking for relatively low distortion using a simple-ish stage with only local / degenerative feedback)? Just thinking about voltage gain here, not current gain. I know I could probably do it with a tube, but would like to know a good solid state way.
Hummm, just re-read my question and it's not clear. Raising a 2V signal by x3 (x4, etc.) without high order distortion seems fundamental to many parts of audio... preamplifiers, headphone amplifiers, etc. . I understand about single transistor common emitter / common source circuits but they seem high-ish distortion, so wondered what solutions might be available without the circuit getting too complex?
The least complex and highest performance would be an NE5534, but you don't want an opamp.
Take a look at the Liniac:
http://www.linearaudio.nl/Miscellaneous/JLH_LINIAC.pdf
jd
Hi John,
After reading this I noticed from looking at the BF862 data sheet that capacitance (Cis) is much, much more linear at very small bias currents. If you're paralleling lots of JFETs like you meantioned in your phono preamp, I'm guessing, you bias the JFETs in a range where the capacitance is more linear?
At least for an input stage that doesn't require high Id, could work well, unless the device is not very linear at low bias currents.
Ever onward, hopefully. Well, now that we know that complementary differential jfets are very useful, how can we make them even better, and reduce some of the inherent problems with nonlinear capacitance? Well, as many already know, we can CASCODE.
There are two different approaches to cascode that can be used. The first is the series cascode that just sets another device, fet, tube, or bipolar transistor over each input device that isolates the input device from generating Miller multiplication of the input capacitance, and isolates the input device from any load aberrations, such as power supply noise, etc. There are also two kinds of biasing for the series cascode. You can fix bias or you can 'self bias' when you use a jfet or a tube, as the cascode part. One important rule with 'self bias' is to use a high Vp part or high Idss-low Gm fet. They are plentiful and cheap.
The other, now popular approach, is the folded or complementary cascode. This is where you use a complementary active device to make the cascode. Unfortunately, this takes a third 'part' which can be as simple as a load resistor that provides current for both the input and the complementary part, or it could be a 'current source' of some type. This sort of design has both good and bad points. First, it often limits the voltage drive, due to the fact that the source resistor has to be a reasonably high value. However, it has the advantage of being a single gain system, capable of having O volts in and 0 volts out. Look, Mom, no caps!
Many commercial designs use a current source, instead of a resistor as the current provider for the stage. However, one must be VERY CAREFUL with current sources as loads, as they amplify their own self noise, sometimes overwhelming the input stage noise, as they sum their noise contributions together. This is tricky, and I could trip up on a specific circuit model, but this is often a problem with using current sources as loads. Enough for now.
There are two different approaches to cascode that can be used. The first is the series cascode that just sets another device, fet, tube, or bipolar transistor over each input device that isolates the input device from generating Miller multiplication of the input capacitance, and isolates the input device from any load aberrations, such as power supply noise, etc. There are also two kinds of biasing for the series cascode. You can fix bias or you can 'self bias' when you use a jfet or a tube, as the cascode part. One important rule with 'self bias' is to use a high Vp part or high Idss-low Gm fet. They are plentiful and cheap.
The other, now popular approach, is the folded or complementary cascode. This is where you use a complementary active device to make the cascode. Unfortunately, this takes a third 'part' which can be as simple as a load resistor that provides current for both the input and the complementary part, or it could be a 'current source' of some type. This sort of design has both good and bad points. First, it often limits the voltage drive, due to the fact that the source resistor has to be a reasonably high value. However, it has the advantage of being a single gain system, capable of having O volts in and 0 volts out. Look, Mom, no caps!
Many commercial designs use a current source, instead of a resistor as the current provider for the stage. However, one must be VERY CAREFUL with current sources as loads, as they amplify their own self noise, sometimes overwhelming the input stage noise, as they sum their noise contributions together. This is tricky, and I could trip up on a specific circuit model, but this is often a problem with using current sources as loads. Enough for now.
Last edited:
Then it will amuse you to see my "Beast With a Thousand
Jfets" which is an output stage with 588 pairs of 2SJ74 and
2SK170 as complementary followers.
I'll try to bring it to BAF for you to see.
Originally Posted by Nelson Pass
Then it will amuse you to see my "Beast With a Thousand
Jfets" which is an output stage with 588 pairs of 2SJ74 and
2SK170 as complementary followers.
I'll try to bring it to BAF for you to see.
So, this will idle with a combined IDSS of something like 3 to 6 amps. : )
Maybe more of a Beauty with a Thousand JFETs. Glad someone is doing this.
-
(Does Nelson usually post his schematics?)
Last edited:
Their specification says 50 Ohms, and 91 dB at 1 mW. So 6V RMS will be about 720 mW and 170 peak mA. That will be about 119 dB. Should blow your head off !!!
Maybe their published specifications are just wrong.
I've seen measurements for three pairs of Audez’e headphones indicating that 90 decibels requires 0.111, 0.134, and 0.119 Vrms. Which is like 124 dB at 6 Vrms.
Still shy of loudnesses achieved at concerts of Manowar, My Bloody Valentine, and Gallows (> 130 dB). Not sure, but crescendos of large orchestras will probably approach 124 dB.
-
Maximum output: 133dB, 15W
Audez'e LCD-2 --> Wikiphonia
Last edited:
- Status
- Not open for further replies.