hello. i build composite headphone amp. ad797 inner loop. but it oscillate. maybe slew rate too much? how do i slow down slew rate? this inverting input- so cap from output to inverting input? 100pf for start? thank you.
i p2p this. works with 811. but i socket them both so i can change different opamps. i want slow down 797 so that it work with other opamp.
First I'd try removing the 22pF cap, and instead adding a series RC network directly across the 797 input terminals.
Start with a 100 Ohm resistor in series with 2.2nF. If this helps reduce the oscillation, we can further adjust the values.
BTW, make sure that the input is always terminated, either with the signal generator, or a short to ground.
Change its place - put it between pin 6 and pin2 of AD797. See what you get and adjust its value if needed.
across +/-? wow, value very high. i will try, thank you 🙂
i will try 22pf in someplace else, too. 🙂
Try without the 22pF first. The RC network adjusts the HF noise gain of the amplifier.
Also - sockets are a big problem with opamps like these!
Those decoupling caps need the shortest, most-direct return to power 0v possible; your 0.1uF caps need to have zero lead length. Sockets add inductance, and ruin things; without really 'tight', well-considered layout I am not surprised you have oscillation problems.
My suggestion - take the AD797 out and try something with considerably less GBW product: NE5534, OPA604 or similar - and get the circuit working as nice as you like. Then develop it for a faster opamp, and refine the physical layout using the lessons learned.
Those decoupling caps need the shortest, most-direct return to power 0v possible; your 0.1uF caps need to have zero lead length. Sockets add inductance, and ruin things; without really 'tight', well-considered layout I am not surprised you have oscillation problems.
My suggestion - take the AD797 out and try something with considerably less GBW product: NE5534, OPA604 or similar - and get the circuit working as nice as you like. Then develop it for a faster opamp, and refine the physical layout using the lessons learned.
Yes, ideally an all SMT layout with a solid ground plane would be best in this case.
definitely lose the lead cap 22 pF in the outer feedback
even without the lead C the net loop gain seen by the 797 is effectively unity due to the AD811 local feedback giving it Av +2 - but with the AD811 added phase shift
which is a problem since close reading of the datasheet reveals that the 797 isn't exactly unity gain stable even by itself
if its OK for your headphones I would increase the outer loop gain to at least 4 to get back some margin - keeping the 811 local gain at +2
noise gain compensation could make the existing circuit work but I like to see it with defined AC gnd - so I'd put the series RC from the -in to gnd
and it helps to look at the 797 datasheet gain/phase curve and actually design the noise gain and frequency breakpoint
the noise gain R should be selected with the outer loop feedback R, and a eye on the gain/phase curves - noise gain of ~10 is conservative but about what you get with the existing circuit values and 100 Ohms when the 811 local gain of 2 is included
whatever the R the RC constant should almost certainly be bigger - you want the noise gain corner frequency to be at least an order of magnitude below the loop gain intercept - and it looks to me like you should be aiming for ~1-2 MHz so the RC should be closer to 1 usecond than rayma's value
of course the +in of the 797 should be well damped by 1-2 MHz too - so a 2nd series RC should be added from +in to gnd, of ~100 Ohms but with a smaller C to not roll off audio so much
I'd use OPA826 or OPA1641 today for the input op amp - low noise for fet input op amps, much more convincingly unity gain stable, enough slower than the AD811 to mostly ignore the added phase shift
and you're wasting some of the AD797 noise potential anyway with kOhm feedback and source R
even without the lead C the net loop gain seen by the 797 is effectively unity due to the AD811 local feedback giving it Av +2 - but with the AD811 added phase shift
which is a problem since close reading of the datasheet reveals that the 797 isn't exactly unity gain stable even by itself
if its OK for your headphones I would increase the outer loop gain to at least 4 to get back some margin - keeping the 811 local gain at +2
noise gain compensation could make the existing circuit work but I like to see it with defined AC gnd - so I'd put the series RC from the -in to gnd
and it helps to look at the 797 datasheet gain/phase curve and actually design the noise gain and frequency breakpoint
the noise gain R should be selected with the outer loop feedback R, and a eye on the gain/phase curves - noise gain of ~10 is conservative but about what you get with the existing circuit values and 100 Ohms when the 811 local gain of 2 is included
whatever the R the RC constant should almost certainly be bigger - you want the noise gain corner frequency to be at least an order of magnitude below the loop gain intercept - and it looks to me like you should be aiming for ~1-2 MHz so the RC should be closer to 1 usecond than rayma's value
of course the +in of the 797 should be well damped by 1-2 MHz too - so a 2nd series RC should be added from +in to gnd, of ~100 Ohms but with a smaller C to not roll off audio so much
I'd use OPA826 or OPA1641 today for the input op amp - low noise for fet input op amps, much more convincingly unity gain stable, enough slower than the AD811 to mostly ignore the added phase shift
and you're wasting some of the AD797 noise potential anyway with kOhm feedback and source R
Last edited:
jcx, your post is spot on but maybe not what the OP needs 😎
Why not placing a resistor of say 20-50k from pin 6 to pin 2 of the '797. Less influence of the '811 phase shift.
Jan
Why not placing a resistor of say 20-50k from pin 6 to pin 2 of the '797. Less influence of the '811 phase shift.
Jan
it was popular to speculate about Walt Jung's extra feedback R around the input op amp aiding stability - but it turns out that it really doesn't until you've reduce loop gain so much that you might as well have not bothered
it seems like a bad idea on all counts - doesn't improve anything, does limit the gain available to correct the output
the AD797 is a specialist op amp - can't be used safely without paying attention to its peculiarities
I did play with a transistor level sim a while back - it does show some of the datasheet oddities:
http://www.diyaudio.com/forums/soli...-blowtorch-preamplifier-1770.html#post1809194
it seems like a bad idea on all counts - doesn't improve anything, does limit the gain available to correct the output
the AD797 is a specialist op amp - can't be used safely without paying attention to its peculiarities
I did play with a transistor level sim a while back - it does show some of the datasheet oddities:
http://www.diyaudio.com/forums/soli...-blowtorch-preamplifier-1770.html#post1809194
hello thank you much for your feedback.
the trick i need for ad797 slow down because i also want take that opamp to make something like this amp- miniref. http://www.diyaudio.com/forums/chip-amps/184165-miniref-schematic-pcb-layout.html
also no want smt pcb because i have no skill, already make p2p breadboard, and want to see how ad797 sound compare against other opamp.
i know slower slew rate, different sound ad797, but distortion profile same, and distortion profile sometimes make or sometimes break sound. 😀
the trick i need for ad797 slow down because i also want take that opamp to make something like this amp- miniref. http://www.diyaudio.com/forums/chip-amps/184165-miniref-schematic-pcb-layout.html
also no want smt pcb because i have no skill, already make p2p breadboard, and want to see how ad797 sound compare against other opamp.
i know slower slew rate, different sound ad797, but distortion profile same, and distortion profile sometimes make or sometimes break sound. 😀
Using the AD797 point to point isn't a good idea. At least use perf board with a ground plane.
- Status
- Not open for further replies.