How to eliminate ground loop noise from USB for microphone circuit

[Fluffyboii]

Hi,
I have this problem for a long time now. Whenever I power something from USB that has analog circuitry inside, high pitched annoying random noise invades the circuit. I was able to eliminate some of it by using a BJT capacitance multiplier but some high pitched random stuff still appears on the output. I learned that this is most likely from ground loop issue and there are some devices that can be bought that are especially build to eliminate this issue. Only problem is that these devices are usually expensive and hard to find.
It is also said that this issue can be eliminated by having the circuit galvanic isolated.
For example DiyPerks electret mic circuit runs from USB power. But his diy microphone has no annoying noises at the end. Only logical reason can be that the NMA0515 isolated boost converter is eliminating it. Finding this device locally is a nightmare though. I can only buy it from china and unknown sources. He also has 4400uF capacitance connected directly to USB power. Wouldn't this cause sparks at the USB port when you plug it in and ruin it overtime.

Regular non isolated boost converters are cheap but they don't fix the issue. For example this is the one I got which directly converts 5V USB up to 26V. I soldered two pole low pass capacitance multiplier with 2000uF and 2.6K resistors (3.3K parallel with 12K because 12K was too much and caps were taking too long to charge) which doesn't really help. Seems like generally adding low pass filters to the circuit that will be powered doesn't really work to fight against this noise.

I have multiple transformer cores so I can wind a transformer for making a galvanic isolated boost converter. Any simple topologies for making a boost converter that will eliminate USB noise? Most of the galvanic isolated DC-DC boost circuits are too complex for my liking. If it can go up to 48V it would be nice to make a USB powered Phantom Power supply. But a lower voltage like 25V would work as well. My objective is making a low noise USB power supply that I can use for my future USB powered projects.

For lowest noise possible I found out that using a simple transformer to lower mains voltage, full bridge rectify it and use something like 7809 regulator with lots of filtering works really well. But I don't really want to make it work from mains and I need higher voltage.

I also found out that when using 2 pin electret capsules with build in JFETs and powering the preamplifier circuit from batteries and shielding the circuit with metal case some noise still appears when gain is increased. This noise doesn't change with changing the op amp from generic TL081 to something else. Which only leaves the JFET itself as the only possible noise source. The capsule I tested was this head part from some random microphone I found for cheap:

audiogrus pinned electret preamp circuit. All metal resistors, additional two pole input low pass with 10ohm resistor and 100uF capacitors, decoupling caps + one directly at the op amp inputs. As I said I put it inside a metal contained and grounded both the electret capsule shield and the container itself. Still did not get low noise signal out of it.

rjenkinsgb made this simple amplifier that can run off the mic line voltage and it performs much better with a cheap two pin small capsule. It basically has no noise compared with the one on top at similar output levels. I am not sure how can it works so well with just one active element.

I also just purchased a 34mm bare electret capsule accidentally thinking it was a true condenser one (they look very similar). Default configuration on sale page is this:

Can I configure it as a 3 terminal electret instead (is this technically 3 terminal). From my understanding and testing in real life when I have some negative feedback at the internal JFET (degeneration resistor) the gain of the capsule decreases but SNR ratio increases. Then I can amplify it with a regular Op Amp and have much better Gain/Noise ratio since Op Amp performs better than a single JFET amplifying most of the signal. Looks like the picture at the bottom is how people usually wire bare electret capsules. JFET directly connected to the capsule. He then connects it to the circuit at the first image having resistors both at the drain and source.

I also want to add that I wanted to use this capsule with rjenkinsgb's microphone upgrade circuit. He took the trouble of sending one to me and I really appreciate it. I am sure it works really well but the fact that I accidentally bought an electret capsule and me having no XLR USB interface with Phantom supply makes it complicated.
Can I leave the red circled area unoccupied on the PCB and directly connect the middle positive plate of the bare electret to Gate of Q4 JFET and ground its negative plate to circuit's ground.

Sorry for posting so many different thing in one topic. I just can't get this out of my head.

 

[Fluffyboii]

Microphones must be connected with shielded audio cable that is not an antenna wire picking up interference.
Google finds no information about a 596S Jfet.
Here is how the Jfet inside the electret mic should be biased:

2sk596s N channel Jfet commonly found inside electret mic casules. Adding giga ohm range bias resistor doesn't change anything. I believe capsule itself or jfet has some leakage current that acts as the bias resistor. What happens if we add a source resistor to jfet while still biasing the gate to ground. Does adding feedback and losing some gain help linearity that way or is it invalid for electrets. I will put the whole thing in a metal mic case so it will be shielded. Testing it shows that it works as it is but I can not say anything on how optimal it is.

 

[audioguru]

If the Jfet has some leakage current to bias the gate at 0V then what prevents drain to gate leakage current from saturating the Jfet which prevents it from working?

Measure the voltage at the drain. It should be near half the filtered supply voltage feeding the drain resistor.

 

[Fluffyboii]

If the Jfet has some leakage current to bias the gate at 0V then what prevents drain to gate leakage current from saturating the Jfet which prevents it from working?

Measure the voltage at the drain. It should be near half the filtered supply voltage feeding the drain resistor.

Vdrain was about 5.7V and when additional Giga ohm resistor is connected it would rises to 6V. By making the Drain resistor 20K instead of 10K I used with 7.8V Vdd (9.1V zener diode supply with filtering causes it to drop about 7.9-7.8) I was able to set Vdrain about 3.9V which is near middle. I put 1K trimpot as source resistor and set the drain voltage to about 3.9V when capsule ground was removed so that it doesn't fix the drain voltage.
Source resistor doesn't effect the voltage at the drain when capsule ground is connected. If I remove source resistor all together gain increases a lot but I rather have the gain low and have better characteristics, less distortion etc.

 

[audioguru]

A Jfet is "depletion type" which conducts when it has zero volts bias.

Lookup "Linkwitz Electret Mic Mod" which allows an eletret mic to operate with low distortion in very loud places like inside a drum or in a guitar. It converts the Jfet from the original common source with distortion when loud, to common drain (a follower).

 

[audioguru]

Deleted a double post.

 

[Fluffyboii]

A Jfet is "depletion type" which conducts when it has zero volts bias.

Lookup "Linkwitz Electret Mic Mod" which allows an eletret mic to operate with low distortion in very loud places like inside a drum or in a guitar. It converts the Jfet from the original common source with distortion when loud, to common drain (a follower).

I see, Jfet is like a depletion mosfet. Which makes sense considering I always thought mosfets infinite input resistance was useless in some cases because some kind of resistor network was always needed to bias it so making a true op amp with infinite input impedance would be impossible with regular mosfets.

I can wire the capsule like in the Linkwitz mod for sure by shorting the drain resistor and keeping the source resistor about 10K and wiring the PCB to get the signal from source instead of drain. Although for the use case of using this as a vocal microphone with not too much varying input signal level, makes me think losing that much gain isn't a good tradeoff.

For receiving the signal what should I do. I have a ICM7660 IC I finally bought. I can convert 5V USB power to -5V and use it with an op amp like TL072 with it. The USB noise problem still remains though since I don't have galvanic isolation. It only needs around 2-10 times gain as it is, because output level is already quite high out of the mic PCB. If I didn't needed single ended input for my computer then I could just wire the XLR to a 3.5mm jack and use it as it is without headaches. Since I don't have a microphone interface I need an instrumentation amplifier or a simple differential amplifier. I see that people use "THAT" brand audio op amps but those doesn't exist here.

For obtaining galvanic isolation to eliminate ground loop issue can I use a 555 timer with inverted output to get very short pulses to feed into a isolation transformer like I tried with the flyback driver. Issue is that circuit didn't worked anything other than a flyback transformer. I have yet to find a reliable way to get noiseless power from USB. At lest using a phone power brick instead of computer USB port gives me usable results. Getting power from the Phantom supply is not possible unless I modify it since its output drops a lot already by powering the mic (44V to 23V) due to 6.8K resistors.