Brian Roth
Well-known member
Just imagine a vinyl cutting system designed with response from "DC to Daylight" and the RIAA curve that keeps boosting at 6 dB/octave until our sun explodes <g>.
Bri
Summing amp?
- Thread starter living sounds
- Start date
Help Support GroupDIY Audio Forum:
Bri
freshtapescent
Well-known member
- Joined
- Jan 25, 2014
- Messages
- 51
Well at least until the fuses pop.
A not immediately obvious benefit of operating virtual earth sum amps at high noise gain is the increased stability from attenuating the negative feedback.
Stability is not a typical problem while there is a temptation to use under-compensated amplifiers in the sum amp socket for a little extra open loop gain. They can get dicey if say less than three inputs are assigned to a 5534 sum amp with no compensation cap (there are ways to deal with that too but this veer is enough for now).
JR
Or the cutter head.
living sounds
Well-known member
What's the consequence of overcompensating an NE5534? 22pf between pins 5 and 8 is what's needed for unity gain, however, in my console every NE5534 is compensated with thrice that, 68pf, instead (exception: mix bus amp). With this much compensation, is there even an advantage over the NE5532?
Overcompensation reduces Aol (open loop gain) and slew rate. With a good test bench you might measure higher distortion and phase shift in high noise gain sockets.
Speculating, they may have increased the compensation to prevent instability.
JR
Speculating, they may have increased the compensation to prevent instability.
JR
living sounds
Well-known member
Somehow there is instability even with this much compensation, but absent local decoupling. Are the non-inverting fader buffers with caps added for low-pass filtering (post #34) prone to instability?
Frequency response. I always check that the frequency response is flat to 20K Hz. I've been bitten by that before. Not specifically with a 5532 but certainly other op amps.
living sounds
Well-known member
Sure, but my specific question is how the compensation cap between pins 5 and 8 of the NE5534 affects slew rate, unity gain bandwidth etc.
did you check the specification data sheets? They may have Aol plots including open loop phase response.
JR
Isn't unity gain bandwidth another word for frequency response? That's directly related to slew rate, no?
no... unity gain bandwidth is the frequency where the open loop gain falls to 0dB or unity. 5534 is typically 10MHz unity gain bandwidth.
To insure stability with negative feedback circuits the gain must fall below unity gain "before" the forward delay/phase shift equals 180' making the negative feedback, positive feedback that can release the magic smoke from oscillation.
not strictly. Dominant pole compensation creates an open loop transfer function that looks like a one pole integrator. That compensation scheme slows slew rate with a larger integration capacitor. Not all op amps use dominant pole compensation. If you look under the hood of 5534 the compensation cap is across an internal differential pair. Slew rate with 0pF compensation cap is 13V/uSec. Slew rate with 22 pF compensation cap is 6V/uSec.
JR
Are you saying you don't have local decoupling ? If so then put some in. 100n or anything near you have to hand tbh. It works honest
Do you have the ability to measure noise on the power supplies? The power supply rejection ratio decreases with frequency, so if current to the load causes the power supply to droop quickly enough, that can feed through to the output. The input stage will try to correct, but at high enough frequencies there is too much delay so the power supply ripples feed through to the load. That is a little bit different case than the traditional oscillation caused by phase shift through the feedback connection.
tl;dr almost all the discussion you see about stability start with the assumption you have clean power supplies, which is going to require good high frequency decoupling.
living sounds
Well-known member
Thanks. That kind of information is not in the current TI NE5534 datasheet.
So with 22pf you get less slew rate than with the (equally unity gain compensated) NE5532... interesting. I guess with an 68pf compensation cap performance is closer to an RC4558?
living sounds
Well-known member
The console doesn't have much with regards to local decoupling. I've only had one instance of excessive distortion with a certain (mid 90's Phillips) 5534 in one spot that was cured by adding local decoupling. I'm still trying to find the best decoupling scheme for the console.Are you saying you don't have local decoupling ? If so then put some in. 100n or anything near you have to hand tbh. It works honest
living sounds
Well-known member
Sure.
The main linear PSU is very powerful, stable and has very low noise. Some voltage gets lost in the cable on the way to the console, but audio power then is regulated again on every module, with a sufficient drop margin. Each module has a thick iron ground bar, but not every op amp has decoupling caps very close to it.
I had to look it up, the 5532 is specified for 9 V/uSec, rc4558 is spec'd 1.7 V/uSec
JR
living sounds
Well-known member
Yes. A diagram plotting compensation capacitance vs. slew rate/unity gain bandwidth for the NE5534 would be nice though.
