48V microphone phantom voltage problem

IC desoldered. On further checking there is no 10K , cap straight to opamp input. But on that line , some distance from the IC ,as space constraints , a diode topcode A7W. Probably a dual diode BAV99, checks out that way with DVM-D check. The common is connected to the input+cap and the diodes connected reverse-wise to + and - rails which are +5 and

-7V, which I suppose clamps to less than the +/-10V of the specs, but fast enough?

I just noticed in the 2122 datasheet "When the supply is less than +/-10V, the maximum input is equal to the supply " not supply voltage and 0.6V let alone any next to instantaneous overshoot voltage beyond that. Looks like another design flaw with this

2012 model

N_Cook-

My thought is that the 33uF capacitor is leaking. The diode to the negative supply breaks down at less than 12 Volts (or less than 48 Volts if the other diode is open), acting as a noise generator. See if removal of the capacitor makes the noise go away.

Fred

Isn't it obvious the caps are the source? Leaky as a sieve;clamps are useless.. If you insist on 48V phantom and electrolytics, then do a LOT of homework..testing all makers of various voltage ratings and capacitance values. Quite a matrix. Oh, to help reduce the problem, use 100K for one tenth the capacitance and substitute ceramic caps.

Yes I had , as routinely the problem, already replaced the caps. Replaced with a new opamp and noise generator has gone. I had mis-traced the path to the IC and the caps are directly connected to an input. I cheated and removed the 48V option, internally, as never used anyway.

Just put a pair or zeners or bi-directional TVS diode on the front end?

Jamie

Even without the 10k you at first suspected isn't the 48V Phantom supplied via two 6.8k resistors - should limit current to ca 7mA?

piglet

I guess you do not know how to read and reason. Diodes CANNOT fix noise.

They *can* prevent the front ends from getting blown into next week.

yes indeed, it was looking for 2x 6.8K that led me to the relevant opamp in the first place. Does the 7mA limitation overide the 0.6V overvoltaging of an input? As the designer was very restricted in space (the "protect" diodes were some distance away) perhaps mutually reversed zeners in one SOT23 don't exist and he chose 2 diodes tied to the supply rails instead

The 6k8 would limit the cap charging current to 7mA so that is unlikely to have caused damage but someone may have later connected a low-Z between input pin and ground causing a very high cap discharge current. But that op-amp could have got damaged when inv and non-inv inputs got pulled too far apart, datasheet spec differential input voltage -/+0.5V but unclear if there are back to back anti-parallel protection diodes on-chip. Maybe some of the diodes you see on the board are to clamp the diff input voltage?

piglet

**At first switch on, after that the 33uF electros become fully charged and dangerous.

If an input pin suddenly shorted to ground during plugging up etc, negative

48 volts appears at an op-amp input with the capacity to deliver many amps of peak current. Instant death for the BJTs involved.

The simplest fix is adding diodes direct from each op-amp input to both DC rails plus series resistors ( say 22ohms ) between the electros and the diodes. The DC rails need to have large value electros ( and or zeners) across them to absorb the 2A spike current.

BTW: zener input clamping only works if the DC supply voltage is greater than the zener voltage - which is fine if you never switch the power off and let the DC rails fall to zero.

Anecdote:

Diodes direct from input pins to DC rails are not always a good idea. Years ago, I saw a Roland 31 band graphic equaliser ( used in live music situation) where most of the dual op-amps inside were destroyed - plus one shorted 4002 diode at the input. Urgency and poor light had made a speaker lead look just like signal lead.

Plus the lack of a 1 kohm series resistor.

.... Phil

** Bingo.

** That *dual* op-amp is a odd ball special:

Odds on, the pre-amp circuit uses both halves in an instrumentation amp configuration.

.... Phil

Not simple white noise, more like a wind noise where the wind was gusty, ie not a constant level. Problem gone now, opamp changed and 48V source disabled internally

Wasn't worried about noise, the subject of over driving the input and causing damage was more to order, where did you get the idea of removing noise?

Jamie