Thanks Rick.
What do you actually use to stabilise the CFB Opamp or do you not have any capacitance in your feedback loop?
I just use the I/V circuit that Walt Jung had in his article in The Audio Amateur issue 2 of 1992. It should be in his web site. There is a .002uf across the 1.82K feedback resistor.
Its been working great for 22yrs. Check out Walt Jung's web site. Walt told me they don't like a C to ground on the input, but no warning about the feedback C.
The closed-loop stability of CFA opamps often seems to benefit from some R in series with any feedback C. I presume, simply to limit feedback at high frequencies.
Hi Rick,
The AD811 is a great opamp! It was designed after the AD9610 that was one of my favorite opamps (but much to expensive). In fact I still have over 600 unused AD811's in parts drawers.
I just wanted to point out that we used the AD811 specs for a baseline to make the LME49713 a better CFB opamp for Audio, not video like the AD811 was designed for. Have you tried the LME49713 especially in the HA metal can? Do you need any AD811's?
Best Ex-National Audio Group (now gone) Regards,
audioman54/mark
The AD811 is a great opamp! It was designed after the AD9610 that was one of my favorite opamps (but much to expensive). In fact I still have over 600 unused AD811's in parts drawers.
I just wanted to point out that we used the AD811 specs for a baseline to make the LME49713 a better CFB opamp for Audio, not video like the AD811 was designed for. Have you tried the LME49713 especially in the HA metal can? Do you need any AD811's?
Best Ex-National Audio Group (now gone) Regards,
audioman54/mark
Hi Mark, yes I really like the 811 too, in fact I use one as the line stage of my preamp with a JFET follower feeding it, and it sound great. Thanks for the info on the 49713 I did not know it had roots from the 811. Thanks also for the offer on the 811's, if I need some I know who to call, you have a great stash there. I run my 811's at + & - 12v with a clip on heat sink and they have been very reliable. I have not tried the LME49713 at all, but I bet it would sound good!
I also use 811 (after dual differential 1704's) followed by differential to single-end LM6172 stage.
The capacitor around 811 (to ground, or in a feedback loop) may sound ok and it may even work okay, but the only way to make sure the stage does not oscillate is to check for oscillations with a CRO.
Nick
The capacitor around 811 (to ground, or in a feedback loop) may sound ok and it may even work okay, but the only way to make sure the stage does not oscillate is to check for oscillations with a CRO.
Nick
I agree with this sentiment. I have tried both the AD811 and the LME49713 in IV duties as an integrator and the better noise/distortion spec of the LME49713 is clearly audible.
The only issue is that the LME49713 is a lot more difficult to stabilise in this application with capacitance in the feedback path. I use Low Q, high Z ferrite beads in series with Pin 2 as a workaround instead of R and I still only achieve marginal stability with the LME49713 whereas the AD811 is completely stable.
I have not tried the metal can version. Why would the metal can perform any different to the normal SOIC package? I take it that the listed specification is exactly the same?
I suspect the fact that the metal can is in effect a Faraday cage with in effect feedthrough capacitors for the leads. Gives better RF immunity. But there are other theories too.
mark:
is the output stage bias current of the LME49713 comparable to that of the AD811?
many believe that is one of the things that sets the AD811 apart from other parts.
mlloyd1
is the output stage bias current of the LME49713 comparable to that of the AD811?
many believe that is one of the things that sets the AD811 apart from other parts.
mlloyd1
I believe the 811 idles at 15ma, that is why I run them at +-12v for less heat in the package. This was recommended to me by Walt Jung.
Another good thing about the 811 is that it has an open loop bandwidth of 21Khz. The entire audio range.
Another good thing about the 811 is that it has an open loop bandwidth of 21Khz. The entire audio range.
We did a lot of measurements and could find no difference between the cans and the dips on our AP2 Cascade 192's. However in double blind listening tests the cans were chosen to be clearly better by an impressive percentage of the participants. In fact even Bob Pease picked the cans over the dips...IN THE ANALOG POWER SUPPLY REGULATORS I DESIGNED. The signal path for those tests were already all Metal Can (HA) devices.
I do miss Bob! He felt the difference between the cans and the plastic parts (dip and SO) had to do with A) bigger bond wires inside the can B) Better thermals due to the mounting of the die to a larger thermal mass inside the can or C) Better thermals because the die could "breathe" as Bob put it. He said that was because the bare die inside the can was not constrained for thermal transients by the rigid plastic housing for the SO and dip parts. It is possible all three of these items contributed to the sonic improvement. He also felt that the minor shielding effect of the can (which is not a Faraday shield) was not a contributing factor to what we heard in our tests. Having done extensive testing in the past with various types of shields, including actual Faraday shields which include both electro-static and electro-magnetic shielding, I agreed with him.
Best out of work in Silicon Valley for well over a year now (thank you TI for killing the National Audio Group) Regards,
audioman54 / Mark
I do miss Bob! He felt the difference between the cans and the plastic parts (dip and SO) had to do with A) bigger bond wires inside the can B) Better thermals due to the mounting of the die to a larger thermal mass inside the can or C) Better thermals because the die could "breathe" as Bob put it. He said that was because the bare die inside the can was not constrained for thermal transients by the rigid plastic housing for the SO and dip parts. It is possible all three of these items contributed to the sonic improvement. He also felt that the minor shielding effect of the can (which is not a Faraday shield) was not a contributing factor to what we heard in our tests. Having done extensive testing in the past with various types of shields, including actual Faraday shields which include both electro-static and electro-magnetic shielding, I agreed with him.
Best out of work in Silicon Valley for well over a year now (thank you TI for killing the National Audio Group) Regards,
audioman54 / Mark
Oh I forgot to mention those tests were with both the LME49710's and LME49713's in the signal path and the power supplies. We were not able to get the LME49990's into the cans before we were all let go.
I believe the metal can IC's have much better acoustic resonance isolation, compared to plastic or even ceramic case. The plastic and ceramic IC's are firmly soldered to the PCB and hence -> are part of the PCB plane that resonates a lot. All mechanical vibrations are fully transferred to the plastic/ceramic DIL (or similar) cases, where the metal can is more resilient to these type of resonances.
Even the pins are better mechanically isolated in a metal can IC's (surrounded by rubber/plastic o-ring), compared to plastic / ceramic IC pins that can freely resonate with the PCB.
I also believe and am sure I'll get corrected if I'm wrong, that metal can IC's die is suspended in a jelly-like high viscosity liquid, inside the metal can, which is absolutely perfect way to isolate a die with a gain, i.e. the IC from wide-spread mechanical vibrations / oscillations.
Nick
Even the pins are better mechanically isolated in a metal can IC's (surrounded by rubber/plastic o-ring), compared to plastic / ceramic IC pins that can freely resonate with the PCB.
I also believe and am sure I'll get corrected if I'm wrong, that metal can IC's die is suspended in a jelly-like high viscosity liquid, inside the metal can, which is absolutely perfect way to isolate a die with a gain, i.e. the IC from wide-spread mechanical vibrations / oscillations.
Nick
Sounds like metal cans are the way to go, if the part you want to use is offered in metal. This makes me glad I saved some metal can 723 regulators, (off topic I know).
I am tempted to replace the 4 X 811 with LME49713's to give it a try, but apart from $15 a pop (!!) (metal can version) + shipment to Australia totaling $100 in cost, what you wrote re issues when trying to stabilise LME's is probably an additional reason to re-think this sudden urge.
Re beads, I don't use them, but can suggest pure silver wire instead of anything else when using the beads. I think Rick mentioned in one of his previous posts that he used steel wire with beads, but a copper or even better - silver, will give much better results. It is not important that the wire has magnetic properties.
Nick
I have never used steel wire, just copper. I use just 3 or 4 loops, if you use too many you get capacitive coupling between the windings and it will affect the loss at RF in a negative way. Any L before the 811 helps lower the RF going into it from the 1541's fast output waveforms.