Hi,
I never captured a certain flag in the endless struggle: discrete solid state, opamp, tubes... when it comes to phono amps.
While I like the cleanliness and dry but powerful bass response of opamps circuits, I also want colourful mids and 'right' trebles I often get with discrete circuits.
So I thought of an hybrid. It turned out to work as intended: kickin' bass and swingy cymbals.
I wasn't able to hear problems due to the single gain stage topology, noise is very low as well.
I like the split riaa technically (some fine adjustments still on the todo-list), and maybe it plays its part in the received sonics.
The range of usable opamps is limited due to the huge gain it has to deliver. So far, I have only tried OP27, OPA637 and AD797, the latter being the taste winner for me, OPA637 is nice as well, but softer and less pronounced in the bass, OP27 tends to be squeeky.
So far, I don't use coupling caps or servos, I made R6 partly adjustable, and the remaining DC (a dozen of mV) is snipped of by the poweramp's input cap.
It lacks the negative aspects some people feel opamp-circuits have sonically. I like to hear your comments on possible design flaws it certainly has, but with a smile, due to the fantastic sound 😉
Have fun,
Rüdiger
I never captured a certain flag in the endless struggle: discrete solid state, opamp, tubes... when it comes to phono amps.
While I like the cleanliness and dry but powerful bass response of opamps circuits, I also want colourful mids and 'right' trebles I often get with discrete circuits.
So I thought of an hybrid. It turned out to work as intended: kickin' bass and swingy cymbals.
I wasn't able to hear problems due to the single gain stage topology, noise is very low as well.
I like the split riaa technically (some fine adjustments still on the todo-list), and maybe it plays its part in the received sonics.
The range of usable opamps is limited due to the huge gain it has to deliver. So far, I have only tried OP27, OPA637 and AD797, the latter being the taste winner for me, OPA637 is nice as well, but softer and less pronounced in the bass, OP27 tends to be squeeky.
So far, I don't use coupling caps or servos, I made R6 partly adjustable, and the remaining DC (a dozen of mV) is snipped of by the poweramp's input cap.
It lacks the negative aspects some people feel opamp-circuits have sonically. I like to hear your comments on possible design flaws it certainly has, but with a smile, due to the fantastic sound 😉
Have fun,
Rüdiger
Attachments
Hi,
what about moving R13 & C2 to after the FET follower? Would it perform as well?
maybe change to 160r and 470nF.
How well matched to the two quad FETs need to be?
what about moving R13 & C2 to after the FET follower? Would it perform as well?
maybe change to 160r and 470nF.
How well matched to the two quad FETs need to be?
Hi Andrew,
one thing to try would be to add another follower in front of R13/C2. What would you think is the advantage of 160/470n for R13/C2?
For J1-J4 I use 2SK389/2SK109. Even with these, the offset has to be trimmed out. It's one of the penalties with this insane amount of gain.
They are biased at 6.9mA.
(the rest of the setup is AT33PTG cart, SAEC 407/23 tonearm and Sony TTS-4000 player)
Rüdiger
one thing to try would be to add another follower in front of R13/C2. What would you think is the advantage of 160/470n for R13/C2?
For J1-J4 I use 2SK389/2SK109. Even with these, the offset has to be trimmed out. It's one of the penalties with this insane amount of gain.
They are biased at 6.9mA.
(the rest of the setup is AT33PTG cart, SAEC 407/23 tonearm and Sony TTS-4000 player)
Rüdiger
lower output impedance to better drive the cables and next stage. But, will it sound as good?Onvinyl said:
What if the follower/buffer is simply moved inside the feedback loop?
Do you need to implement W.Jung's composite (opamp) feedback to get this to work?
Have you got an OPA228 to try? Some like them and others don't.
OK, I see. No, I don't I want such an high output impedance.
I have no OPA228. I will try a LT1028 I have in my parts box.
It might indeed be interesting place an additional follower before R13/C2 and include it in the loop. On the other hand, it needed some trying to place the caps of C1 in a way the opamp could still control the feedback loop well, so might introduce trouble. A discrete diamond buffer would be nice as well, but is even more complex.
Rüdiger
I have no OPA228. I will try a LT1028 I have in my parts box.
It might indeed be interesting place an additional follower before R13/C2 and include it in the loop. On the other hand, it needed some trying to place the caps of C1 in a way the opamp could still control the feedback loop well, so might introduce trouble. A discrete diamond buffer would be nice as well, but is even more complex.
Rüdiger
Nice amplifier, Rüdiger
I wouldnt change much.
I have used similar design for x100 gain microphone amplifier (9 VDC battery supply).
I would maybe only replace those two 2SJ74 in input stage,
with two (2) 2SK170 used as jfet-CCS constant current source
for the LTP input pair.
Great circuit.
And good FET performance/ FET sound, I imagine 🙂
I wouldnt change much.
I have used similar design for x100 gain microphone amplifier (9 VDC battery supply).
I would maybe only replace those two 2SJ74 in input stage,
with two (2) 2SK170 used as jfet-CCS constant current source
for the LTP input pair.
Great circuit.
And good FET performance/ FET sound, I imagine 🙂
Hi,
I added a servo. Works very good. I have to figure out the maximum working value for R16, since it's crucial for proper performance (too high = DC might not be nulled, too small=noise and THD would rise)
An issue is the fact, that I wasn’t able to implement the distortion cancellation caps according to the AD797’s datasheet without getting huge overshoot in the output waveforms. Any clues what might cause this?
Lineup, what would you think is the benefit of replacing the 2SJ74 (2SJ109 in reality)?
Rüdiger
I added a servo. Works very good. I have to figure out the maximum working value for R16, since it's crucial for proper performance (too high = DC might not be nulled, too small=noise and THD would rise)
An issue is the fact, that I wasn’t able to implement the distortion cancellation caps according to the AD797’s datasheet without getting huge overshoot in the output waveforms. Any clues what might cause this?
Lineup, what would you think is the benefit of replacing the 2SJ74 (2SJ109 in reality)?
Rüdiger
Attachments
Really is not sure what benefit.
But my intuition tells me to do this.
One thing is of course that 2SJ109 is NOT one P-Channel version of 2SK389.
If anyone ever did imagine this ....
Is also in my KISS philosophy. (Keep It Simple)
Why add 2 active transistors, when I can make it work as well without them.
Okay, they do some job.
But one commonly used current source would be my choice,
and was my choice for my microphone amplifier,
... if not using some Pure Resistor Biasing!!
Regards
Lineup
But my intuition tells me to do this.
One thing is of course that 2SJ109 is NOT one P-Channel version of 2SK389.
If anyone ever did imagine this ....
Is also in my KISS philosophy. (Keep It Simple)
Why add 2 active transistors, when I can make it work as well without them.
Okay, they do some job.
But one commonly used current source would be my choice,
and was my choice for my microphone amplifier,
... if not using some Pure Resistor Biasing!!
Regards
Lineup
onvinyl:
that's pretty clever. where have i seen that input stage before?
😉
i can't help feeling like something is being wasted by not somehow using signals out of the drains of the 2sj74, but i don't know how to address that feeling ...
have you compared different input stage topologies? for example, using a conventional, resistor biased differential stage of single polarity njfets (either 2sk369 or the precious 2sk146 comes to mind)? this was done a lot in the 80's to varying degrees of success, but there were no op amps around like the AD797 or OPA637 to follow it up.
mlloyd1
that's pretty clever. where have i seen that input stage before?
😉
i can't help feeling like something is being wasted by not somehow using signals out of the drains of the 2sj74, but i don't know how to address that feeling ...
have you compared different input stage topologies? for example, using a conventional, resistor biased differential stage of single polarity njfets (either 2sk369 or the precious 2sk146 comes to mind)? this was done a lot in the 80's to varying degrees of success, but there were no op amps around like the AD797 or OPA637 to follow it up.
mlloyd1
Hi mlloyd,
similar topo's are sometimes used for mic amps. Bernhard Vogel discussed something like that (but with bjt's) in his 'noise-bible'
There would be a whole world of different input stages possible, but this one did sim extremly well.
I have a bunch of relativly beefy 2SK369's I might try someday.
lineup, 2 pfet's are not more than 2 nfet's for a current source...
Rüdiger
similar topo's are sometimes used for mic amps. Bernhard Vogel discussed something like that (but with bjt's) in his 'noise-bible'
There would be a whole world of different input stages possible, but this one did sim extremly well.
I have a bunch of relativly beefy 2SK369's I might try someday.
lineup, 2 pfet's are not more than 2 nfet's for a current source...
Rüdiger
Rüdiger,
I would get rid of that push-pull symmetry.
😀
I would get rid of that push-pull symmetry.
Maybe it`s time to find out and take a position.
Why would it, just for your sake?
Compared to what?
😀
What does the J2 and J4 FET pair do?
I've seen quite a few circuits that use a FFET pair in front of an OP-amp (makes sense, especially for MM), but you have used a balanced complemetary front end . . . . but not used the bottom half. Have you considered replacing J2/J4 with a current source? you could use a JFET for this as well. Linear Systesm make a very nice low nolise D JFET (LSK389) which would be great for this job.
Anyway, good luck with your project.
I've seen quite a few circuits that use a FFET pair in front of an OP-amp (makes sense, especially for MM), but you have used a balanced complemetary front end . . . . but not used the bottom half. Have you considered replacing J2/J4 with a current source? you could use a JFET for this as well. Linear Systesm make a very nice low nolise D JFET (LSK389) which would be great for this job.
Anyway, good luck with your project.
Onvinyl said:
It is true that using a pair of JFET for CCS, would not reduce the number of Semiconductors.
But there is a bit different with Active Transistors.
Than used only passively, as in a CCS.
I see CCS as being supporting component, as are resistors.
In that sense that such will inflict signal in microscopic way.
My original statement was
I have not done FFT harmonics analysis of your Push-Pull input
compared to my Single input differentail N-JFET pair, biased by CCS.
Maybe there is something nice from your setup?
I also said, that P-Channel will have other parameters, than N-Channel.
I am quite sure, when N+P work together, there will be some harmonics created,
that is not for one well matched N-pair, like 2SK389.
I have done some investigations, when using 2SK170 2SJ74 as a 'complementary' input.
The result I have also presented in this Amplifier Topic
Complementary Class A Preamp using BC368-369
See this post ..
http://www.diyaudio.com/forums/showthread.php?postid=1572579#post1572579
Schematic:
http://www.diyaudio.com/forums/attachment.php?s=&postid=1560865&stamp=1215949714
Blue = 2SK170BL
Red = 2SJ74 BL
As you see, image shows the AC Voltage/Phase angle
across those two 390 Ohm Drain resistors.
The result I have also presented in this Amplifier Topic
Complementary Class A Preamp using BC368-369
See this post ..
http://www.diyaudio.com/forums/showthread.php?postid=1572579#post1572579
Schematic:
http://www.diyaudio.com/forums/attachment.php?s=&postid=1560865&stamp=1215949714
This graph is very interesting.
Blue = 2SK170BL
Red = 2SJ74 BL
As you see, image shows the AC Voltage/Phase angle
across those two 390 Ohm Drain resistors.
Attachments
Hi Lineup,
these are *sim results*, so probably indicative, but not more:
Spectrum for 20Hz, input 0.5mV (*without* output buffer):
Number [Hz] Component Component
1 2.000e+01 3.817e+00 1.000e+00
2 4.000e+01 3.061e-07 8.019e-08
3 6.000e+01 2.072e-07 5.427e-08
4 8.000e+01 6.600e-08 1.729e-08
5 1.000e+02 3.716e-08 9.735e-09
Spectrum as above but with 3x2sk369 and a resistor current source in the front end:
1 2.000e+01 3.812e+00 1.000e+00
2 4.000e+01 1.225e-06 3.215e-07
3 6.000e+01 8.367e-07 2.195e-07
4 8.000e+01 7.221e-07 1.894e-07
5 1.000e+02 5.645e-07 1.481e-07
EDIT: not different with ideal current source loading
Conclusion: Front end's file closed, as it concers me.
Rüdiger
these are *sim results*, so probably indicative, but not more:
Spectrum for 20Hz, input 0.5mV (*without* output buffer):
Number [Hz] Component Component
1 2.000e+01 3.817e+00 1.000e+00
2 4.000e+01 3.061e-07 8.019e-08
3 6.000e+01 2.072e-07 5.427e-08
4 8.000e+01 6.600e-08 1.729e-08
5 1.000e+02 3.716e-08 9.735e-09
Spectrum as above but with 3x2sk369 and a resistor current source in the front end:
1 2.000e+01 3.812e+00 1.000e+00
2 4.000e+01 1.225e-06 3.215e-07
3 6.000e+01 8.367e-07 2.195e-07
4 8.000e+01 7.221e-07 1.894e-07
5 1.000e+02 5.645e-07 1.481e-07
EDIT: not different with ideal current source loading
Conclusion: Front end's file closed, as it concers me.
Rüdiger
Hi,
I'm still happy with this design. My understanding of the input stage (which might be weak, anyway) is, that small irregularities get canceled out by the push-pull arrangement (and sim backs this up as well) while others seem to think that it introduces such irregularities.
Rüdiger
I'm still happy with this design. My understanding of the input stage (which might be weak, anyway) is, that small irregularities get canceled out by the push-pull arrangement (and sim backs this up as well) while others seem to think that it introduces such irregularities.
Rüdiger
mlloyd1 said:
Yes, though it *might* be, that the drain loads are just 500R, and the C3 (the servo C) might have just 2.2µ. (too lazy to open the box at the moment) The drain res's of the 2sk389 (2sk170 in schemativ of #7) need to be trimmed to get the offset low enough.
What I will try out sometime is to buffer the AD797's output and/or replacing it with some of the new opamps from national. The LT1028 gave me oscillations, the OPA637's work well but did not sound as good.
Rüdiger
- Status
- Not open for further replies.