A novice scope user - measuring noise?

[phrarod]

I'm working on a pre-amp project. I have the audible noise down to pretty much zero. If you stick your ear into the speaker driver you hear a slight tube rush.

Now I want to get that "jet black background" that I've heard on some of the best pre-amps. When I use a scope what is the best way to measure noise? Is it just a spikey wave form?

I have a 1:1 and a 30:1 probe on a HP 1740A I'm borrowing.

[Enzo]

Turn the noise up and look at it. That is what it looks like. At lower levels, the noise will be smaller on the screen, so increase the vertical sensitivity.

Be warned that in x1 probe mode and turning the vertical sensitivity way up, you will also pick up local AM radio stations, and other noise in the environment. You need to learn what is out there so you don;t assume it is coming from your amp when it arrives on your screen.

[SY]

Fine noise measurements like that aren't really the bailiwick of scopes. You want a spectrum analyzer (like a 24 bit sound card and appropriate software) or a broadband AC voltmeter.

[phrarod]

When you say turn the noise up - its static noise. I do notice I get more 'noise' with the 1:1 and it probably is picking up something aside what's in the pre-amp.

SY. That's what I worried about. Out of curiosity you mean I can put a sound card into my computer to measure this? Do you have suggestions? I'm Mac based.

[multisync]

This is obvious, but when you are switching your scope probe from 1:1 to 10:1 are you also multiplying/dividing by 10 the display you see on the screen or is the scope doing it for you. If your scope is set at 5mv/ division and your probe is 10:1 your display is 50mv per division. Be careful. I can tell my scope what the divider is 10:1 or 1:1 and the scope display is updated accordingly, but if switch back and forth from 1:1 to 10:1 the scope never knows the difference and the signal display will be larger. Also remember you will see pp display not rms or A or C weighted readings. It will look a lot worse than if you use a limited bandwidth low Z AC volt meter. I was fighting what looked high frequency oscilations for a few days that seemed to come and go randomly when I realized the oscilations only showed up when a certain CFL was on. A few hits to the head with a hammer later made me see the light.

[Conrad Hoffman]

Meaningful noise measurements are a lot harder than most people realize. First, you need to define a bandwidth, say 20-50,000 Hz. Next, you have to specify the type of measurement, usually RMS. To say anything about audibility you need to know the spectral distribution of the noise. Right off it should be obvious that a plain scope isn't really set up to do any of this, and for any good preamp a typical scope won't be sensitive enough. Usually one starts with a very quiet amplifier to get the noise up to a decent level, followed by a filter network to limit the measurement to the necessary bandwidth and finally, a wide band RMS meter with sufficient crest factor to accurately measure the waveform. There is a rough technique for getting a close-enough measurement with a scope by fiddling with the intensity and looking for a specific area of the noise band. Can't remember exactly, but it was published quite a few years ago probably in EDN and is still around on the 'net.

[oldmanStrat]

and my 2 cents...

my little tek scope has FFT built into it - so I would think you could get a lot of useful information measuring the outputs.

but - whenever I'm looking for noise - I always take a baseline measurement to rule out picking up ambient noise. 60hz is everywhere and without really good probes and measuring differentially it's hard to tell if what you are seeing is real or not...

[Poindexter]

Whut Stuart sez.  The problem with scopes is that they are just not sensitive enough to test really quiet gear.  Go out on the 'bay and get an AC millivoltmeter.  I got a Millivac MV812A that's a lab-grade piece of gear for less than it cost to ship, and its most sensitive setting is 1mV full scale. I can measure reasonably accurately down to a couple tens of microvolts if I unplug my cordless phone.

He's also right about ambient roobish in the general environment.  When I'm measuring down there, I can affect the reading by moving my hands around the DUT, where my body is, stuff like that.

I have actually found the MVM a more generally useful tool than the scope.

Aloha,

Poinz
AudioTropic

[phrarod]

Thanks! I'll go on the bay and look for a Millivac MV812A. Are there other AC millivoltmeter models that will do?

[ChrisA]

Originally posted by phrarod
Thanks! I'll go on the bay and look for a Millivac MV812A. Are there other AC millivoltmeter models that will do?

You can build one. The basic idea is you rectify the signal and use that to charge a cap then measure the voltage on the cap. But it's all in the detals.

I built one of these. It really can allow a simple meter to measure down into the environmental noise. The metal can and home made probe provide good sheilding too. for almost zero cost. It says "RF probe" but it works below RF too if the coupling cap is reasonable
http://www.kk5na.com/kk5na_files/AccupManual.htm

Using two diodes like this is a smart, self correcting trick.

I built mine using a standard BNC connector in place of the home made probe tip. This way I can connect a coax cable to the device under test. and stand back. Those of use using tubes don't to reach inside by hand.

But you may be dealing with thermal noise. Tubes run hot. No way around that. I remember working on a scientific camera system where to reduce the noise we chilled the entire device to cryogenic temp. Find a way to run your tubes at 50 or 100 below zero and they will quiet down nicely.

OK seriously now. If you heat a resistor it will make noise. You can't expect your amp to beat the heated resistor. So measure a resistor and use that as a baseline. You do not need absolute measurements, relative is good enough. So you can get by with home made uncalibratd test equipment. Lots of good high temp resistors around for free too. light bulbs solder irons and what not.