ok.. its sure im certainly in for some flaming with this.. but i cant do elsewise.. i have built an opamp preamplifier (that is, with ic's...) here it is: what can i do to get it better? (only gainstage shown.. supply ordinary LM stuff) resistors cheap metal films hand matched to 0.1% (holco's on the way), decoupling through Elna RJH. pot alps 27mm.. so?
Attachments
Hi,
I could think of a few things....
Howzdat for freaky?
Basic design is an amalgam of stuff I played around with before but with a battery supply, PSU where also tried at some point in time, in a Phonostage. Circuit drawn by Carlos M.
Recommend placement of the feedback resistors (all Op-Amp's) as SMD Resistors soldered directly at the pins. Powersupply is per supplied chip, wiring between regulator out and supplied chip must be minimal and no supply decoupling should be applied. If critical apply a 100pF SMD COG Cap between the positive and negative supply of the amplifier chip.
Pot should be ideally 1K log which is difficult at such low values. Alternatively a 220 Ohm Linear should work okay with a 470 Ohm "build out" resistor, please use cermet track pots if you can, if not carbon. Also, then you need to compensate the gain loss. Maximum output Z with 470R & 220R Pot is around 200R, with 300R & 1K Log it is 230R, both at maximum volume.
Sadly, transformer based attenuators as passive linestage are notably better still....
Sayonara
I could think of a few things....
Howzdat for freaky?
An externally hosted image should be here but it was not working when we last tested it.
Basic design is an amalgam of stuff I played around with before but with a battery supply, PSU where also tried at some point in time, in a Phonostage. Circuit drawn by Carlos M.
Recommend placement of the feedback resistors (all Op-Amp's) as SMD Resistors soldered directly at the pins. Powersupply is per supplied chip, wiring between regulator out and supplied chip must be minimal and no supply decoupling should be applied. If critical apply a 100pF SMD COG Cap between the positive and negative supply of the amplifier chip.
Pot should be ideally 1K log which is difficult at such low values. Alternatively a 220 Ohm Linear should work okay with a 470 Ohm "build out" resistor, please use cermet track pots if you can, if not carbon. Also, then you need to compensate the gain loss. Maximum output Z with 470R & 220R Pot is around 200R, with 300R & 1K Log it is 230R, both at maximum volume.
Sadly, transformer based attenuators as passive linestage are notably better still....
Sayonara
You are doing fine, costiss!costiss said:
Very good OPamp OPA627 (peranders has just bought some)
Metalfilm matched resistors
ELNA caps
ALPS potentiometer
And I am sure you will get a good Power Supply.
You already got the best boys as assistance!
Congratulations!
/halo - wish he had some Good Components at home
- maybe time to place an order????
Kuei,
Thanks for an interesting design. I have two questions about
the regulator part.
1) It seems reasonable that all the regulators share the
circuitry connected to the positive inputs (ie. the reference
voltage circuitry). Do you agree on this?
2) Would you consider it sufficient that they also share the
capacitors on the preregulated rails, or should these be
per regulator and as close as possible to them?
Just like a previous poster I am also a little bit curious about
the JFET buffer. I assume the purpose is to make the input
impedance invariant of the attenautor. However, I see that
they use a fixed Vgs of 0V and have no feedback. I assume
this particular topology is intended to be very linear, but I
do not quite see why or how (although it is a bit late and I
am tired, I admit). I also guess they have to be well matched.
Thanks for an interesting design. I have two questions about
the regulator part.
1) It seems reasonable that all the regulators share the
circuitry connected to the positive inputs (ie. the reference
voltage circuitry). Do you agree on this?
2) Would you consider it sufficient that they also share the
capacitors on the preregulated rails, or should these be
per regulator and as close as possible to them?
Just like a previous poster I am also a little bit curious about
the JFET buffer. I assume the purpose is to make the input
impedance invariant of the attenautor. However, I see that
they use a fixed Vgs of 0V and have no feedback. I assume
this particular topology is intended to be very linear, but I
do not quite see why or how (although it is a bit late and I
am tired, I admit). I also guess they have to be well matched.
Hi,
Yes. If you need for example for a Phonostage two gainstages per channel you can use dual Op-Amp's, assuming the layout remains tight.
Doing this on PCB's almost literally means using two PCB's arranged with the two soldersides touching, locating each regulator chip output pin directly next to the supply pin driven by it. In short, think 3D not 2D.
The latter. The smalles (physical & value) capacitor closest to the chip then the larger ones. The PSU Reg Chips do need to be stabilised against oscillation by having their supply pins well decoupled.
Well, the input impedance IS invariant of the Attenuator, but as the circuit is following the SPEEDO Axiom and the INVERTING Axiom the input impedance is rather low. The two Idss @ 15V matched (and thermally connected) J-Fets simply buffer the low input impedance of the Op-Amp and provide a high impedance input, in fact pretty much as high as you like (1M no sweat)
Actually, the topology is not intended to be very linear, it actually is. Driving 3V RMS into 1k will give around 0.008% THD, or better noted -82db THD, and this THD is rather low order and thus of low subjective sonic impact.
Yes, one could have used an OPA627 connected as follower, but somehow this simple J-Fet buffer struck me as more "Zen", just as the J-Fet MC stage on the Phonostage that this linestage was matched with (namely the "analogue addicts phono") seemd a better idea than a gain of 20 AD797. I guess it's a question of "designers fingerprints".
Ciao T
Yes. If you need for example for a Phonostage two gainstages per channel you can use dual Op-Amp's, assuming the layout remains tight.
Doing this on PCB's almost literally means using two PCB's arranged with the two soldersides touching, locating each regulator chip output pin directly next to the supply pin driven by it. In short, think 3D not 2D.
The latter. The smalles (physical & value) capacitor closest to the chip then the larger ones. The PSU Reg Chips do need to be stabilised against oscillation by having their supply pins well decoupled.
Well, the input impedance IS invariant of the Attenuator, but as the circuit is following the SPEEDO Axiom and the INVERTING Axiom the input impedance is rather low. The two Idss @ 15V matched (and thermally connected) J-Fets simply buffer the low input impedance of the Op-Amp and provide a high impedance input, in fact pretty much as high as you like (1M no sweat)
Actually, the topology is not intended to be very linear, it actually is. Driving 3V RMS into 1k will give around 0.008% THD, or better noted -82db THD, and this THD is rather low order and thus of low subjective sonic impact.
Yes, one could have used an OPA627 connected as follower, but somehow this simple J-Fet buffer struck me as more "Zen", just as the J-Fet MC stage on the Phonostage that this linestage was matched with (namely the "analogue addicts phono") seemd a better idea than a gain of 20 AD797. I guess it's a question of "designers fingerprints".
Ciao T
Thanks for the answers Kuei.
However, on question number 1 I am still not sure if it is me
not understanding the answer, or you misunderstanding my
question? I repeat both my question and your answer here
to avoid potential further confusion.
After thinking for a while I came to the conclusion that you
probably assumed separate preregulated supplies for the
channels, in which case your first paragraph makes sense.
I am not sure how the second paragraph relates to this,
however.
However, on question number 1 I am still not sure if it is me
not understanding the answer, or you misunderstanding my
question? I repeat both my question and your answer here
to avoid potential further confusion.
Christer said:
Kuei Yang Wang said:
After thinking for a while I came to the conclusion that you
probably assumed separate preregulated supplies for the
channels, in which case your first paragraph makes sense.
I am not sure how the second paragraph relates to this,
however.
Hi,
No. The LM6181 used to "regulate" the PSU for the actual Amplifier (also an LM6181 of course) is a high speed operational Amplifier. It requires it's supply lines to well decoupled locally, hence all the capacitors on the supply line are per chip.
The pre-regulated supplies can be shared of course, as can be the references, though this makes layout akward.
Sayonara
No. The LM6181 used to "regulate" the PSU for the actual Amplifier (also an LM6181 of course) is a high speed operational Amplifier. It requires it's supply lines to well decoupled locally, hence all the capacitors on the supply line are per chip.
The pre-regulated supplies can be shared of course, as can be the references, though this makes layout akward.
Sayonara
costiss said:
Perhaps balance the input impedances seen by the opamp´s both inputs.For more info look here:
Article by Walt Jung
Another article
One more
And the last
Nicke
This would been have true 20-30 years ago. DIN-norm?frez said:
With todays sources, read CD-standard,
I would use 22kohm or even lower!
-----------------------------------------------
NE5534 which is a rather old one OP, can drive 10Vrms/600ohm max.
2Vrms into 1kohm would not be much.
Not even two loads 1kohm parallell= 500ohm
would upset NE5534 very much.
The standard, soon to come will be <= 10 kohm.
That is halo's prediction.
47 kohm is long time ago outdated, in practice.
/halo - low impedance rails - more current - less disturbance
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.