High Fidelity Performance Op-Amp.
No need to make too complicated
when such good results with fairly simple 3 stage circuit
Set up this amplifier design model
in your simulator
and you will see!
And why not layout your own Hifi Op-Amp PCB
and make a few small killer OPamp devices of your own.
Version 1 of mine used 17mA current total
at +/-15 Volt DC
and tested for practically 0% THD distortion, in various loads and output voltages.
Fourier Analyse shows very nice even spectrum of harmonics.
Regards
Lineup Audio Lab Productions
No need to make too complicated
when such good results with fairly simple 3 stage circuit
Set up this amplifier design model
in your simulator
and you will see!
And why not layout your own Hifi Op-Amp PCBand make a few small killer OPamp devices of your own.
Version 1 of mine used 17mA current total
at +/-15 Volt DC
and tested for practically 0% THD distortion, in various loads and output voltages.
Fourier Analyse shows very nice even spectrum of harmonics.
Regards
Lineup Audio Lab Productions
Attachments
mod_evil said:
I tested simulation only with using 2SK170GR input JFET:s
For the Current sources, J3 / J4, I used BF245B.
Bipolar transistors, Q1-Q5, are my good spice models of BC560C / BC550C.
The output are some good TO39 transistors I have models for.
These 2 Output stage TO39 NPN, I have here at home
because they came from some old junk circuits PCB
that I have found at our electronics recycling station.
I got them for a few cents.
I say, what transistors you use, is not so critical.
As long as they are normal good audio small JFET / Bipolar.
This is why I did not specify any special transistors in my circuit.
But I recommend you try to use like those currents I have in circuit.
Because these currents will work!
I did add one little compensation cap. Somewhere ...
I use Electronics WorkBench, EWB, MultiSim9
with good spice models for transistors I downloaded from internet
and also found here in forum.
http://www.electronicsworkbench.com/
Regards
lineup
I will as usual be back with a Version 2.
After further testings in extreme loads.
But as the Version 1 shows so very good behavior
I can not imagine there is very much final trimming needed.
Input JFETs, 2SK170GR should be as good MATCHED as possbile.
Best solution for this is of course use a DUAL JFET.
With 2 matched jfet in same case.
Like 2SK389 and several other.
There is another dual matched pair, with very low noise and very good matched devices.
But I do not remember the name. (maybe J404 or something like that???)
There are some extremely fine such N-JFET Pairsexcellent for Audio Amp input.
JFET also have almost no input bias current which can cause offset
when you want to use different resistors in input/feedback
------------------------------
For fine tuning of open loop output DC offset
I used R5, which can be replaced with 500 Ohm trimpot.
Use for example my recommended power supply +/-15 Volt.
1. You attach both IN+ and IN- to GND.
2. You use 100/1.000 Ohm for load. From OUT to GND.
3. Use no feedback from output = Open Loop setup.
4. You use this trimpot (R5) until output is close to 0.00 Volt
5. Measure this resistance of trimpot.
6. Replace the trimpot with a resistor with same value = R5
Until I come back with eventual updates
Regards
lineup
Seems like a workable idea. I would degenerate the current mirror for less noise. With 2SK170's in there it's not so much of a problem but with lower gm JFETs you will get a lot of noise gain from the mirror to the input. The ouput stage has some limitations too. I wonder how this would work with an auto biasing complimentary N/P FET follower?
You also might want to cascode J1 and J2, 2SK170's have lots of excess gate current at 15V
BTW I tested some BF862 Philips' JFETs and was pleasantly surprised, <1nV/rt-Hz @ 1 mA and a corner well below 100Hz. The data sheet almost reads right on the 2SK170. These cost me 15 cents each. SMT only though and same excess gate current (impact ionization from the short channel).
You also might want to cascode J1 and J2, 2SK170's have lots of excess gate current at 15V
BTW I tested some BF862 Philips' JFETs and was pleasantly surprised, <1nV/rt-Hz @ 1 mA and a corner well below 100Hz. The data sheet almost reads right on the 2SK170. These cost me 15 cents each. SMT only though and same excess gate current (impact ionization from the short channel).
Hi Lineup
I like the common base VAS. I like it better (IMHO) than a common emitter VAS. Also (IMHO) I like differential VAS as well. Refreshing it is too see some new ideas. I have one too.
Since I don't have a really good simulator, I usually just build a test circuit...it's more fun anyway
Haven't made this yet, what do you think? Bias of first stage is dependent on 2nd stage CCS. Just another idea to try....
I like the common base VAS. I like it better (IMHO) than a common emitter VAS. Also (IMHO) I like differential VAS as well. Refreshing it is too see some new ideas. I have one too.
Since I don't have a really good simulator, I usually just build a test circuit...it's more fun anyway Haven't made this yet, what do you think? Bias of first stage is dependent on 2nd stage CCS. Just another idea to try.... Attachments
Justcallmedad said:
You wouldn't have a measurable problem unless you did high impedance sources. The BF862 is a lone anomoly as far as I can tell, Philips needed one for some car radio application so they made their own. They must make them in mass quantities since a tube of 50 cost me $6.95.
I got my boards back just today from PCboardexpress for my own 'nice little preamp' based on a couple of JFETS and a dual opamp in one of the early SSM configurations. I fit it all on a tiny board to slip into a microphone body. I'm using huge resistors on the input so the coupling can be just some silver micas, hence my sensitivity to the excess gate current.
Oh yes, it’s a typical example where Igsx could be harmful, is it for a condenser microphone?
Why silver mica caps, they aren’t so well regarded today in the “audio world” often styroflex caps (polyester) are preferred…
Sorry for this short “aside” lineup, your design is quite interesting, current mirrors could be 2SA1349/2SC3381 and 100-200R as degeneration resistors, some more current in the first stage would be better 2/3mA per leg (with SK170/SK389 parts).
Why silver mica caps, they aren’t so well regarded today in the “audio world” often styroflex caps (polyester) are preferred…
Sorry for this short “aside” lineup, your design is quite interesting, current mirrors could be 2SA1349/2SC3381 and 100-200R as degeneration resistors, some more current in the first stage would be better 2/3mA per leg (with SK170/SK389 parts).
can't get it to work in LTSpice...... has a negative 14.5v offset, 10mv of signal (with 100mv in), and according to the behavior, the + and - inputs are reversed (it oscillates with a gain of 10 with the left fet used as noninverting input).... i used the 2n5484 model for the fets, and 2n3904/06 for the bjt's. instead of led's i used 3 1n914's in place of each led. i get the proper current in the fet C/S legs, but the output section is only running 900pA, and the VAS is running 10mA through R4, Q5 collector current is 8mA, R5 is 8mA. the voltage at the base of Q6 is a dc average of -14.05v. R3 and R6 are 330 ohms to set the fets at 2mA, r5 and r4 are at 1k, r7 is 1k. as i said the inverting and noninverting inputs are reversed, if i apply feedback to j2, it oscillates, but it is stable when feedback is applied to j1.
does anybody have an extended standard.bjt and standard.mos for LTSpice? i find my parts libraries somewhat limited with the stock LTC version. for an amp design i'm working on i had to "fake" a complement mosfet pair. i've figured out how to modify the model libraries, but not all the necessary info is always available on the device data sheets.
Testing session #2
I haven't changed anything from Circuit in my first post.
But today I did some more testings.
I wanted to see if I could press this 2K170GR JFET Op-Amp
to show any distortion, that is worth the name.
This was not easy.
For a pre amplifier circuit, line level,
I think you would agree that the following test is pretty Heavy Duty.
Supply is still +/-15 Volt.
I use 1 kohm for load.
Gain is unity, 1 = buffer.
Input = Output: +/-10 V peak.
Test frequency is standard 1 kHz sinus.
+/- 10mA peak current, to within 5 Volt of supply voltage
would be a difficult task for most op-amps.
If even they could take that level at input stage.
But no problem for this one!!!
Distortion is actually slightly below -100dB now = Visible.
But it is not much. THD is showing 0.001%
Fourier Analysis is like what can make anybody happy!
At least I am very much pleased with a spectra like this one.
See Attachment
Regards, 30th January 2007
lineup
I haven't changed anything from Circuit in my first post.
But today I did some more testings.
I wanted to see if I could press this 2K170GR JFET Op-Amp
to show any distortion, that is worth the name.
This was not easy.
For a pre amplifier circuit, line level,
I think you would agree that the following test is pretty Heavy Duty.
Supply is still +/-15 Volt.
I use 1 kohm for load.
Gain is unity, 1 = buffer.
Input = Output: +/-10 V peak.
Test frequency is standard 1 kHz sinus.
+/- 10mA peak current, to within 5 Volt of supply voltage
would be a difficult task for most op-amps.
If even they could take that level at input stage.
But no problem for this one!!!Distortion is actually slightly below -100dB now = Visible.
But it is not much. THD is showing 0.001%
Fourier Analysis is like what can make anybody happy!
At least I am very much pleased with a spectra like this one.
See Attachment
Regards, 30th January 2007
lineup
Attachments
