Opa Alice build noise floor

[AUDIO FREQ]

Here is mine. Hissy hiss hiss

https://cdn.imagearchive.com/groupdiy/data/audio/90/90272-9499738494cba48efaa86320e94a92c4.wav

 

[oscarhuang]

I removed his 12v Zener diode. It can be said to be very quiet now.

 

[jp8]

I removed his 12v Zener diode. It can be said to be very quiet now.

Was it noisy before removal of the zener?

I know zeners are noisy, but that doesn't mean that noise will enter the audio path thanks to the excellent PSRR of the OPA. So I fail to see a relationship between removing the zener and low(ering) noise, if that was the message you wanted to convey.

I would not remove the zener. It keeps the ground return current DC, preventing potential Common Mode signal distortions.

Jan

 

[oscarhuang]

The hizz noise can be clearly heard before removing the Zener diode. After removing the Zener diode, the noise floor is reduced a lot. Maybe increasing the filter capacitor can achieve the same job.

 

[jp8]

OK, then I'd assume there is no, or insufficient decoupling on the voltage divider, which provides the 1/2Vdd reference voltage ("virtual ground").

Jan

 

[Khron]

Maybe increasing the filter capacitor can achieve the same job.

Not if it's directly in parallel with the zener...

 

[jp8]

@oscarhuang : can you share the exact schematic of your circuit? In the original OPA Alice circuit, there are two elcaps that would effectively eliminate zener noise. The ones encircled in the picture.

Jan

 

[rogs]

.... I would not remove the zener. It keeps the ground return current DC, preventing potential Common Mode signal distortions.

I've not heard this explanation before ....
Whilst a zener is useful in the regulation of the capsule polarisation voltage in this circuit, I cannot see why it is used to regulate the DC supply to the Op-amp(s)?
The 'half rail' supply should help maintain optimum op- amp bias, and the current drawn by the 2 amplifiers (c.2mA each) will restrict the maximum supply voltage to the amplifiers to c.30v , which is well below the maximum limits.

As far as I can see, all the zener does here is draw extra current, restrict headroom and - as is being discussed here - possibly introduce extra noise?

I've not come across this 'ground return current' problem you're describing before?

 

[Cqwet Dbdfte]

Using a Zener in Low Noise circuit is not a good idea.
There are plenty of linear series regulators to choose from, or even an LM431 or LT1431.
https://www.eevblog.com/forum/projects/tl431-common-base-amplifier-stability/msg599580/#msg599580

Even the old LM317HV may be useful.

 

[rogs]

Using a Zener in Low Noise circuit is not a good idea.
There are plenty of linear series regulators to choose from, or even an LM431 or LT1431.
https://www.eevblog.com/forum/projects/tl431-common-base-amplifier-stability/msg599580/#msg599580

Even the old LM317HV may be useful.

My point is, why regulate the op-amp DC supply at all? ... OK, it does need to be effectively de-coupled of course, but it doesn't need to be regulated.

The passive - separately decoupled - 'half rail' voltage divider will maintain the optimum bias point.

All a zener or regulator will do is restrict headroom, and draw extra current from the phantom power supply....

 

[Cqwet Dbdfte]

The OP amp has 36V limit, the Phantom is 48V.
How much output swing is needed?
Schematic does not seem to center the supply...

 

[Khron]

the Phantom is 48V.

... fed through a pair of 6.8k resistors, which will drop SOME voltage, at any non-zero current.

Schematic does not seem to center the supply...

"Center"?

 

[Cqwet Dbdfte]

... fed through a pair of 6.8k resistors, which will drop SOME voltage, at any non-zero current.



"Center"?

The DC operating point for the OPAs...
Should be at midpoint of the single ended supply.
Nevermind, I just noticed the green reference symbols.

 

[rogs]

The OP amp has 36V limit, the Phantom is 48V.
How much output swing is needed?
Schematic does not seem to center the supply...

As Khron says, the 6k8 resistor in series with each leg of the phantom power supply - plus the 2k2 resistor in series with the feed to the op-amp supply - will ensure that the supply can't exceed the maximum DC for this type of op-amp.
With the 2 amps shown in OPA Alice circuit, each leg of the phantom power supply will need to supply c.2mA as a mimimum. (Probably at bit more, if a voltage multiplier is being supplied as well?).
So the voltage drop fom 48 v with 6k8 and 2k2 in series with each leg will be I X R .... 0.002A x 9K = 18v.
Thus the maximum DC supplied to the op-amps will be 30v..... Probably less in practice.

The half rail - 'center' - supply is derived from the equal value resistors - sadly partly covered by the red line in the schematic above.
The attached OPIC schematic - which doesn't use a zener - shows R6 and R7 - together with decoupling caps C4 and C6 performing that task.
Seems to work pretty well.....

 

[Cqwet Dbdfte]

Once the OP amp starts drawing current its supply will have a voltage drop. Before this happens the 48V is hitting the OP amp.
If the decoupling cap is big enough the charge up time will reduce the startup voltage, keeping it below 36V.
Likely the part can take more, 36V may be a power limited number, not secondary breakdown, wich would have a few more volts before failure, maybe 45V is OK.

 

[rogs]

Once the OP amp starts drawing current its supply will have a voltage drop. Before this happens the 48V is hitting the OP amp.
If the decoupling cap is big enough the charge up time will reduce the startup voltage, keeping it below 36V.
Likely the part can take more, 36V may be a power limited number, not secondary breakdown, wich would have a few more volts before failure, maybe 45V is OK.

The absolute max limit on the spec sheet for opa164* series is listed as 40V. I would think the main 47uF decoupling cap being charged through the feed resistors should ensure that voltage is never likely to appear at the op-amp?
In the OPIC schematic the feed resistors are 10k, so it's even less likely.

The extra headroom that is provided by not using a zener can be quite useful.... @kingkorg made some interesting observations on that in this thread: https://groupdiy.com/threads/ldc-k67-capsule-output.79855

 

[Cqwet Dbdfte]

To be sure a TVS , or Zener diode could be used to clamp the voltage to a safe level, above the "working" level, as to not interfere with the OP amp.

 

[Matt Syson]

You could fit a 30 Volt zener as 'security and it would probably never see actual 'use'. the resistors are 47??? They could be a lot larger as the input CURRENT into the fet inputs is approaching infinite, this would mean the cap on the new 'centre' could be reduced in value.
There is no obvious protection against input (output) lines getting accidentally shorted while in use such as if it is fed via a patchfield or dodgy mic cables where the voltage stored on the output caps gets stuffed into the Op amp outputs. Then what abpout incoming RF protection (small chokes) and the little capacitors shown.
Having the zener conducting (regulating) does provide a constant reference against variations in 'gain' with changing polarisation voltage. Although 6K8 is a 'standard' value for phantom power resistors some Neve desks use 15K and possibly other variations and of course the standard Voltage tolerance is plus and minus 4 Volts.

 

[Cqwet Dbdfte]

If the stored energy in those caps is an issue maybe some megohm bleeders to ground?
A Zener clamp could be 36V.
Inductors in the output, very close to output pins, with those decoupling nF caps right after, or a Pi filter to keep RF levels low.

 

[ccaudle]

If the stored energy in those caps is an issue

I believe that would likely refer to fault conditions, e.g. shorted cable which will cause a high-voltage pulse to propagate across the capacitor.
See for reference:
https://thatcorp.com/the-48-volt-phantom-menace-returns/