When something switches quickly, it generates a broadband RF burst (Fourier theory tells us that something narrow in the time domain is wide in the frequency domain).
The RF will generally be both radiated into the air and conducted in the AC wiring.
Unless your system is close to the switching unit, then probably you are mostly hearing the results of RF that is conducted through the AC wiring. But maybe it's possible that the wiring also radiates some of the RF, at some distance from the actual switch.
Anyway, the first thing I would try is low-pass filtering. There are two main locations at which to try filtering, initially.
One location for an RF filter is across the AC Mains, before your power transformer, i.e. across the primary winding's connections of the transformer. There are special capacitors that are safe for use across the AC Mains. Soldering one across the transformer's primary connections might be enough. There are also more-complicated filter topologies, and some ready-made power-line filters, that you could try. But I am not enough of an expert to try to tell you what to use, there.
The second location for an RF filter is the signal input to the preamp. You can easily install a low-pass RC passive RF filter, there. It's just a resistance in series with the signal input and a capacitor to ground, downstream from the resistor. (But if you are using the inverting input, then you shouldn't connect a cap to ground directly from the inverting input. So you use two resistors, and connect the cap to ground from between the two resistors.)
You want to put the -3dB "cutoff" frequency of the filter at about ten times the highest frequency that you don't want to affect. So 220 kHz or more should be OK.
For an RC lowpass filter, the cutoff frequency is
f = 1 / (2 π R C) and
C = 1 / (2 π R f)
So you can just choose an R value and calculate the C that is needed.
Note that C is in Farads and f is in Hz. (So multiply the resulting C value by 1000000 to convert it to uF.)
Example, with 1000 Ohms and 220000 Hz:
C = 1 / ( 2(3.14)(1000)(220000))
C = 723 pF (0.000723 uF)
You could use a standard value like 680 pF.
Or, pick a C value and calculate the needed R value:
R = 1 / (2 π f C)
Example for C = 0.01 uF and 220 kHz:
R = 1 / (2(3.14)(220000)(0.00000001)
R = 72.3 Ohms
Using a slightly-different value for R will not matter very much.
----
You should have an RF filter on your signal input, anyway. So I would install that, first, and leave it there, whether or not it solves the popping problem you have right now.
(Another location for RF filtering is at the DC power inputs for the opamp. You should already have sufficient decoupling capacitors from the power pins to the load ground. Installing a small resistance (33 Ohms?) in series with each DC rail, just before the first decoupling capacitor, might help.)
If RADIATED (i.e. over the air) RF is causing the problem, the first things to check are the wires for the input signal and ground. They must never be apart or separated from each other. Otherwise you have made an antenna. They should be tightly twisted together ALL the way from end to end. Shielded twisted pair cable would be even better, with the shield connected to the chassis at one end, and to nothing at the other end.
Also, the wire pairs to and from the transformer should be tightly twisted, everywhere.
RF can also get in through the output (speaker cables) and through everything else.
There are techniques for both filtering and shielding that are discussed in Chapter 7 of ADI - Analog Dialogue | Op Amp Applications Handbook .
The RF will generally be both radiated into the air and conducted in the AC wiring.
Unless your system is close to the switching unit, then probably you are mostly hearing the results of RF that is conducted through the AC wiring. But maybe it's possible that the wiring also radiates some of the RF, at some distance from the actual switch.
Anyway, the first thing I would try is low-pass filtering. There are two main locations at which to try filtering, initially.
One location for an RF filter is across the AC Mains, before your power transformer, i.e. across the primary winding's connections of the transformer. There are special capacitors that are safe for use across the AC Mains. Soldering one across the transformer's primary connections might be enough. There are also more-complicated filter topologies, and some ready-made power-line filters, that you could try. But I am not enough of an expert to try to tell you what to use, there.
The second location for an RF filter is the signal input to the preamp. You can easily install a low-pass RC passive RF filter, there. It's just a resistance in series with the signal input and a capacitor to ground, downstream from the resistor. (But if you are using the inverting input, then you shouldn't connect a cap to ground directly from the inverting input. So you use two resistors, and connect the cap to ground from between the two resistors.)
You want to put the -3dB "cutoff" frequency of the filter at about ten times the highest frequency that you don't want to affect. So 220 kHz or more should be OK.
For an RC lowpass filter, the cutoff frequency is
f = 1 / (2 π R C) and
C = 1 / (2 π R f)
So you can just choose an R value and calculate the C that is needed.
Note that C is in Farads and f is in Hz. (So multiply the resulting C value by 1000000 to convert it to uF.)
Example, with 1000 Ohms and 220000 Hz:
C = 1 / ( 2(3.14)(1000)(220000))
C = 723 pF (0.000723 uF)
You could use a standard value like 680 pF.
Or, pick a C value and calculate the needed R value:
R = 1 / (2 π f C)
Example for C = 0.01 uF and 220 kHz:
R = 1 / (2(3.14)(220000)(0.00000001)
R = 72.3 Ohms
Using a slightly-different value for R will not matter very much.
----
You should have an RF filter on your signal input, anyway. So I would install that, first, and leave it there, whether or not it solves the popping problem you have right now.
(Another location for RF filtering is at the DC power inputs for the opamp. You should already have sufficient decoupling capacitors from the power pins to the load ground. Installing a small resistance (33 Ohms?) in series with each DC rail, just before the first decoupling capacitor, might help.)
If RADIATED (i.e. over the air) RF is causing the problem, the first things to check are the wires for the input signal and ground. They must never be apart or separated from each other. Otherwise you have made an antenna. They should be tightly twisted together ALL the way from end to end. Shielded twisted pair cable would be even better, with the shield connected to the chassis at one end, and to nothing at the other end.
Also, the wire pairs to and from the transformer should be tightly twisted, everywhere.
RF can also get in through the output (speaker cables) and through everything else.
There are techniques for both filtering and shielding that are discussed in Chapter 7 of ADI - Analog Dialogue | Op Amp Applications Handbook .
What is an AD815 preamp? Is that based on the Analog Devices AD815 amplifier? What is the input impedance? I'm guessing high. Is the amp and power supply regulators properly bypassed? Do you have a scope to see what is misbehaving?
G²
Yes. I was going to ask about the bypassing and decoupling caps but forgot.
The power pins of the op amp must have a small cap to ground, which must each be connected extremely close to the pin; more than 2 mm is probably too far away. The other end of each cap should go to power ground, as close as possible. These caps help to prevent high frequency instability, which would possibly occur because of the hidden positive feedback loop(s) through the power rails that most transistor amplifiers have.
You also must have some larger capacitors from each power pin to the load ground, for decoupling. They supply the fast transient currents and prevent those from inducing voltages across the inductances of the main power and ground rails. The decoupling caps should also be connected as close as possible to the power pins. If multiple cap sizes are used, the smallest ones should be closest to the pins.
The power pins of the op amp must have a small cap to ground, which must each be connected extremely close to the pin; more than 2 mm is probably too far away. The other end of each cap should go to power ground, as close as possible. These caps help to prevent high frequency instability, which would possibly occur because of the hidden positive feedback loop(s) through the power rails that most transistor amplifiers have.
You also must have some larger capacitors from each power pin to the load ground, for decoupling. They supply the fast transient currents and prevent those from inducing voltages across the inductances of the main power and ground rails. The decoupling caps should also be connected as close as possible to the power pins. If multiple cap sizes are used, the smallest ones should be closest to the pins.
Last edited:
Another possibility:
If your AC Line voltage dips lower, momentarily, when a large load switches on, maybe you are hearing the sound caused by your regulator(s) going out of regulation.
Where is the schematic?
What is the rated transformer output voltage, what kind of rectification is used, what is the total reservoir capacitance per rail, what are the nominal rail voltages after the regulators, what is the nominal load impedance, what capacitance values are installed from the AD815 power pins to ground, and what is the rated maximum load current per rail? You would also need to know the clipping voltages of the AD815, i.e. between each power pin and the output pin, under your operating conditions.
To diagnose whether or not the headroom for the regulators is the problem, it might be simpler to just temporarily try a much larger reservoir capacitance per rail, after the rectifiers, unless you are already cutting it too close and have very small ripple amplitude. Or you could instead try a transformer with secondary voltages that are a few volts higher.
Do you have an oscilloscope, or a "True RMS" meter?
Last edited:
Why an AD815? The suggested uses
APPLICATIONS
ADSL, HDSL, and VDSL Line Interface Driver
Coil or Transformer Driver
CRT Convergence and Astigmatism Adjustment
Video Distribution Amp
Twisted Pair Cable Driver
The input bias currents and offset Voltages are a bit high for low noise preamps. 400 mA currents and 100 MHz seems a little excessive for a preamp.
Amplifiers at this speed usually do better on 4 layer boards with a ground plane to keep it under control. I'd be leary of a 2 layer board and can almost guarantee you bad results on a single layer.
There are many chips better suited as a 'preamp'. For driving audio balanced line the SSM2142 or DRV134 are better suited.
G²
Thanks all... I think it would be the problem from the AC power line.
The bypassing and decoupling caps done but far more 2mm.... I think the length from the pins would be around 7mm something.
The 10ohm wire-round resistance add between like:
DC in-> caps -> 10ohms -> caps -> LM series reg chip -> caps with bypass -> opamp
I will try to using the isolated transformer to see any improvement... Thanks guy!
The bypassing and decoupling caps done but far more 2mm.... I think the length from the pins would be around 7mm something.
The 10ohm wire-round resistance add between like:
DC in-> caps -> 10ohms -> caps -> LM series reg chip -> caps with bypass -> opamp
I will try to using the isolated transformer to see any improvement... Thanks guy!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.