Do you think, that these extremely high 200dB gain can be realized in reality? When feedback loop would be closed, composite system instability may needs to reduce this gain, or to make some phase/gain correction tricks...
I have the fully discrete opamp working prototype, which has the open loop gain about 125..130dB @1kHz. It was not so easy to get it stable.
Victor, honestly, I don't know, and I imagine keeping both fast opamps stable (even if individually they're well-behaved) might require inordinate effort. Especially considering the discrete opamp you've already built. I was more suggesting it if you were still in the brainstorming-mode.
Obviously, the more forward gain you have, the easier it will be to stabilize, but a "slower" front-end chip would make things much easier, as the amount of phase contribution by the DSL line driver would be minimal.
Obviously, the more forward gain you have, the easier it will be to stabilize, but a "slower" front-end chip would make things much easier, as the amount of phase contribution by the DSL line driver would be minimal.
The current feedback amps do have a really wide bandwidth and often have a lot more drive available. However stacking high gain on high gain, especially with current feedback in the chain is really dicey. I would use the CF amp as the output with its gain set to 20 dB? It should be really low distortion and have a very low output impedance as well as essentially flat gain and phase across the high gain opamps range. The use it as a buffer at the output of the main opamp. This works pretty well. Try to keep both opamps having some gain so you don't crash into the phase margins. I did similar with LM3886's and got very low distortion, plus isolation of the opamp from the thermal stuff going on inside the LM3886.
For those of us who might not have the knowledge or background as y'all,
can you explain what the extra bandwidth get's us?
From the AD797 with 8 MHz
vice the ADA4896 with 65 MHz?
Does it mean an increase in stability when we used it in the audio band?
Or
does it mean an opportunity for more noise and RF to enter the circuit
creating havoc?
can you explain what the extra bandwidth get's us?
From the AD797 with 8 MHz
vice the ADA4896 with 65 MHz?
Does it mean an increase in stability when we used it in the audio band?
Or
does it mean an opportunity for more noise and RF to enter the circuit
creating havoc?
I must be confused. On Victors plot with the polystyrene capacitors no distortion was shown. That would indicate to me that the passive components were the distortion limiting factors.
The noise level would be from the opamp. So I don't think increased gain will provide improvement.
To reduce the noise would require perhaps either a lower noise front end with gain or paralleling amplifiers.
Using four parallel amplifiers should give 6dB lower noise. That would I think then allow the passive components to be the limit again.
The noise level would be from the opamp. So I don't think increased gain will provide improvement.
To reduce the noise would require perhaps either a lower noise front end with gain or paralleling amplifiers.
Using four parallel amplifiers should give 6dB lower noise. That would I think then allow the passive components to be the limit again.
I think a detailed noise analysis is needed before determining that the opamps are the source. I have been burned there. Its inverting so the series resistor is a limit as well as is the agc and the reactive leg etc. This is where the spice noise analysis helps.
The AD797 is significantly quieter than the lme49720 except where it isnt.
Extra midband gain usually comes with a larger gain bandwidth product. In an oscillator that out of band noise is usually rejected quite well.
Sent from my LG-H811 using Tapatalk
If one had to choose to replace a capacitor in the signal path (vs power supply) of an amp and the choice was between film or multilayer ceramic, which would be recommended. And the obvious question is why.
I have a vintage amp from the 70s and it is full of ceramic disc caps and one of them cracked. Trying to replace with modern day quality part(s).
All the capacitor questions recently made me think of posting this.
I have a vintage amp from the 70s and it is full of ceramic disc caps and one of them cracked. Trying to replace with modern day quality part(s).
All the capacitor questions recently made me think of posting this.
It depends on the value and physically if you can fit a better cap. The best ceramic (COG/NP0) is better than many film caps, but they are much larger than lower grade ceramic caps and come in a limited range of values. If not COG, you are better off with any kind of film if you can fit it.
I only just read this and am very sorry, always enjoyed his writing and web pages. May he RIP. Enjoyed reading his obituary and learning a bit more about his life.
These values would be perfect in SMD COG/NPO size 1206. Better than film IMHO. What voltage rating do you need?
Jan
Open loop gain mainly drops down in every opamp depending of the rising frequency. If we compare opamps with the similar open loop DC gain, but different bandwidth, then the opamp with the wider bandwidth has more resources (higher open loop gain on HF) for to reduce distortions on high frequencies, when the feedback loop is closed.
It sure would be nice to break off threads that discuss specific equipment mods. I had to work backwards for about 15 minutes just to figure out what was being discussed here.
According to some old measurements, third harmonic produced by AGC FET regulator is on approximately -110dB level across this FET , when signal across this transistor is 0.5% from the maximum output signal level. The Wien bridge stage reduce this harmonic for some dB. This probably means, that this contribution maybe around -160dB in the output signal.
Harmonics contribution from the opamps maybe on -165...-170dB level, without a hard load.
At this time I cannot see these levels in the noise. The software runs with the maximum averaging. Needs to reduce the noise, needs to use other regulator for the AGC (probably the optocoupler NSL-32). New amps instead of LME49720 may reduce noise and also distortions. Maybe I can make new amps test at the end of next week. Now the printed boards are in manufacturing. Will be the BF862 at the input.
Do you have a possibility to increase the FFT resolution? This can also be used to see a small signal in the noise. What FFT size are you using?
The LME49720 may be difficult to beat for this. The challenge is to find an op-amp with low voltage noise and low current noise as well as a very low distortion. In terms of the noise the OPA209 (or OPA2209 for a dual) might be interesting. But I don't know how it compares in terms of distortion. The bandwidth is lower, which is of course not ideal.