Active Differential Signal (eg Audio) Limiter with Switched Attenuator

[marks2c]

Protection for audio analyser balanced input: ie 100V-Line speakers, valve HT etc to 400V (ref XLR pin 1).
Uses bi-directional depletion-mode MOSFETs to safely limit input current to below 3mA.
Uses bi-directional shunt clipper to limit analyser input voltage to +/-20V (+/-10V optionally) wrt to XLR pin 1.
CAD files are attached to the PDF (download via the attachments panel).
Attenuator SMath calculation sheet is in there too.
SMath and KiCad are free to use and awesome.

[dmendesf]

Very nice. Do you have any ideia about bandwidth? Thinking about other uses for the circuit...

[marks2c]

Very nice. Do you have any ideia about bandwidth? Thinking about other uses for the circuit...

Small-Signal Bandwidth
GHz BW should be possible because under normal small-signal conditions the depletion-mode MOSFETs are not 'active' (ie they are just a low-Ohm resistor, and are not switching).

As such, small-signal bandwidth is only limited by the physics of the PCB layout/substrate etc (ie should be good beyond several GHz as you can buy RF transistors good for 4GHz in the same SOT223 package).

Protective Function
Here is detail FYI (voltage is red, current is blue with current limiting of 1.5 mA):

BW=0.35/rise time. Given 2.5ns that would be about 140MHz (with XLR connectors etc...). With less stray capacitance much faster protection should be possible.

The ringing was down to non-ideal power supply leads/shunt/XLRs etc.

Other Performance
With high-impedance loads this limiter/clipper circuit is essentially transparent both distortion-wise and noise-wise.

[dmendesf]

Thanks for the info. Will assembly one for testing purposes.

[marks2c]

How did it go?
Did theory and practice align?

[dmendesf]

Haven't bought the transistors yet, but it's on my list...

[marks2c]

With a sharp enough spike,  the gate capacitance causes gate over-voltage (failure).

The fix is easy, and retains the excellent 'no-difference' audio performance.

[Calvin]

Hi,

I think your circuit might run into gnd-related issues i.e. hum loops.
After the AES recommendations for balanced signal transmission Pin1 should only be connected as a screen to PE.
It should not have a connection to signal-gnd, apart from a single point ... typically a lownoise powersupply point.
Balanced studio gear regularly feature a gnd-Lift switch (if not audio-trannies), to break hum loops between several devices.
If You need to have a dedicated signal-gnd between Pin2 and Pin3 You might consider using 4-pin XLRs.

regards
Calvin

[marks2c]

I think your circuit might run into gnd-related issues i.e. hum loops.
After the AES recommendations for balanced signal transmission Pin1 should only be connected as a screen to PE.
It should not have a connection to signal-gnd, apart from a single point ... typically a lownoise powersupply point.
Balanced studio gear regularly feature a gnd-Lift switch (if not audio-trannies), to break hum loops between several devices.
If You need to have a dedicated signal-gnd between Pin2 and Pin3 You might consider using 4-pin XLRs.

Thanks for the input Calvin. Yup, would be better to not have the centre of the attenuators' shunt resistors connected to 0V (a job for the next version). This is easily fixed by removing the shunt resistors and replacing them with no mid-point connection to 0V.

Otherwise, nope. Humm is about either:
1) a lack or balance and/or poor CMRR in the presence of EMI cutting the conductors of the balanced signal, or
2) humm current flowing in ground in a ground-reference signal path.

Neither are a problem here (balance is maintained, clippers are massively high impedance (ie zero current) unless >9V (or 18V) is present (ie beyond small signal conditions).

Remember that:
A) the limiter is intended to be mid-cable. Good practice is for connecting the screen at only one end or the other: no power-related currents should be flowing, and any capacitive signal currents would cause close to zero effect (small current, low impedanec, high-CMRR) etc.
B) the screen (pin 1) is present only to provide electrostatic protection (ie capacitive): it is transparent to magnetic fields, which rely on balance and CMRR to prevent humm.
Thank you for contributing.