Low-distortion Audio-range Oscillator

[davada]

Originally Posted by 1audio

I would make a small coil of maybe 10-20 turns on a 1/2" to 1" former as a pickup. Couple it directly into the 50 Ohm cable (the mismatch won't matter). If there is something in distress it will radiate and the coil will pick it up. You could sniff all around the box to find the unhappy parts and no electrical connection to make a mess of things. Also good for getting depressed about power supply radiation and how its nearly impossible to eliminate.

You can buy these coils but they are way expensive.

Great idea Demian. It just gave me another. Remember those balun transformers the ones used for 75 ohm cable to FM receivers and very old TV's. 75 ohm to 300 ohm. We still have a bunch of those at work. DC won't pass through them and it's a lot closer to 600 ohms than 75. It also offer some input protection to the analyzer. We have inline 75 ohm attenuators at work as well. I can slip these in on the 75 ohm side. The input sensitivity is 0dBV - 1mV into 75 ohms for cable systems. 60 dB of padding is not a problem. This will give me 9KHz to 1 GHz.


Thanks,

[Samuel Groner]

Not every interference is of electromagnetic nature. With what I'm fighting in my lab in the 100 kHz to 1 MHz appears to be mostly an electrostatic field.

Samuel

[RNMarsh]

I must go into the test equipment rental business --
I have here three analyzers which can be used (covering 20Hz to 6.6GHz); I put one on the 339A which covers 150Kz to 1Ghz and just connected the output (339A output 'off') directly into the spectrum analyzer input. Looked up and down.. out to 100MHz and stopped... didnt see anything but random noise. Did same next with an HP3400 and just measured the rms noise level.... which is fairly high by conventional terms. However, low enough to be useful for FFT to below the THd of the oscillator. BUT... I did notice that the input is super sensitive to picking up just a hand wave near the input port. This can indicate a high input Z. The schematic doesnt indicate such. However, the input does have an inductor in series right at the input terminal. That could do it. Going to investigate this more.... input Z/sensitivity. Thx-RNMarsh

[davada]

Originally Posted by RNMarsh

I must go into the test equipment rental business --
I have here three analyzers which can be used (covering 20Hz to 6.6GHz); I put one on the 339A which covers 150Kz to 1Ghz and just connected the output (339A output 'off') directly into the spectrum analyzer input. Looked up and down.. out to 100MHz and stopped... didnt see anything but random noise. Did same next with an HP3400 and just measured the rms noise level.... which is fairly high by conventional terms. However, low enough to be useful for FFT to below the THd of the oscillator. BUT... I did notice that the input is super sensitive to picking up just a hand wave near the input port. This can indicate a high input Z. The schematic doesnt indicate such. However, the input does have an inductor in series right at the input terminal. That could do it. Going to investigate this more.... input Z/sensitivity. Thx-RNMarsh

The input Z is supposed to be 100K.

There is something wrong with my 339a. This is a repair.
I can't continue with mods with a problem like this going on. Anything I do maybe meaningless.

It could be something simple like a poor ground somewhere. It's very likely a ground trace burned open at or near the input or monitor output. It's used gear after all and I have no idea of what it's been through.

Fortunately, I did repair work for many years. I just don't like having to repair my own stuff.

[Samuel Groner]

The HP 3400 is surely a high-Z input.

Samuel

[RNMarsh]

I placed a 10K across the input binding posts and it still acts like a high Z (?) Weird. -RNM

[RNMarsh]

did you notice the oscillator common is not the same common as the analyzer? Grnd 1 and grnd 2. Where did you ground the coax from the osc to the front panel bnc?

[davada]

Originally Posted by RNMarsh

did you notice the oscillator common is not the same common as the analyzer? Grnd 1 and grnd 2. Where did you ground the coax from the osc to the front panel bnc?

I haven't connected the BNC yet. It's just an intention.

Yes I know they are different grounds. The BNC should be connected to analyzer gnd I think.

We have enough interconnections it easy to isolate each section.
With the notch filter disconnected the meter goes to zero and is not effected by the monitor ground being connected to an external ground. The problem seems to be at or before the notch filter. I'm betting on input damage.

[davada]

This is interesting.

With the notch filter disconnected the meter, in distortion mode, rests at 0.0016% with the sensitivity set at -80dB FS and all filter in. There is not distortion signal at this point. Therefore this is all noise from the distortion amplifier, the auto set level, the filters and meter amplifier. The noise floor measures about -90dBV on the QA400. Or is it noise?

The oscillator is still running at -10dBV. When the oscillator range switch is switched to the off position the meter drops to a zero reading. So how is the oscillator signal getting to the remaining circuits? the path is disconnected. If there is a another path this explains why the meter never really drops to the distortion and noise levels we measure by other means.

Cheers,

David.

[davada]

Here is the auto st level amplifier with the input cable disconnected. Clearly we have some oscillation seen at the 10KHz to 20 KHz range. The noise is about +10dB above the floor of the QA400.