Tubes vary quite a bit when we look at their noise floor. Buyer beware!!
We recently got in a batch 6072 tubes and I didn't pay extra for them to be noise tested. I am having regrets about that decision now. As I was assembling some mics for customers I was noticing some noise, way down there, in my headphones, but it was there nevertheless.
After I do all the electrical and acoustics tests on a customer's mic, I always end with a headphone listening test. You know the kind. Come on, admit it. Everyone does it. The ones where you say some of your favourite mic testing words into the mic and listen to yourself. We all do it. And you know what? It's a valid test to find gremlins in a mic. Microphonic parts show up when you tap the mic; you might hear buzz or hum that machine based testing does not pick up.
Waiter, there's a hiss in my soup
So I hear noise... broadband noise, as in like white noise. It's way down there, but I hear it. A very quiet Hissssssssss...
Now I always do a simple noise check when I test the electrical characteristics of the preamp, but as it turns out that test is not sensitive enough. (We will be changing that test in our mic manuals from now on)
So I decided to spend some time finding a way to compare one tube noise to another.
In the old days I would just connect a Vaccum Tube Volt Meter (VTVM) to the output of a device and reduce in the input attenuator on the meter until I saw noise dancing on the meter needle. That gave you a quick and dirty way to know if the device was behaving normally.
But for this test I want to get an idea of the spectrum of the noise. Mid-range and high frequency noise will cut into human hearing far quicker than lower frequencies. So using the same test jig that we use to get a frequency response, we cut the input signal to the mic preamp, and turned the input gain to maximum. That's right 60 dB of gain. And guess what, we saw the noise. As clear as the reverb on a Perry Como record. There was no escaping now!
Test the test method
So the next thing we did was to see what an open input looked like compared to having a tube connected. I needed to be sure that our digitizer input was indeed quieter than a tube microphone. Whew! It was quieter. Not by very much, but enough that we could trust our measurments.
Now remember that the gain on that mic input is turn up to maximum. That means that with 60dB of gain the noise is still 75 dB below "0" on this scale, at 1Khz. Pretty impressive.
With some trepidation I pulled out one of my treasured GE, double mica, black plate NOS 6072 tubes (hand made by Edwin Armstrong I'm sure) and inserted it into a FAR 251 style prototype mic that uses a Jensen output transformer. (more on that in another article ...)
I waited one minute for the tube to warm up and ran a sweep.
Thank goodness it was pretty quiet. That one cost me real money!
Now I should tell you that I had some other NOS tubes that I put into this test setup and let's just say they won't be going into any mics that I use. Sorry to be the one to say it but not all NOS tubes are good.
So then I spent the long and boring afternoon, plugging tubes in the mic, connecting the preamp to the test jig, waiting for them to warm up and running the test. I had purchased 20 tubes and I had many others in the 6072 drawer, so it took some time...
... hours go by...
I will cut to the chase" as they say. We found that the variation in noise floor on the tubes was quite wide. And if we use our GE NOS 6072 as the reference and let's say we decide no tube should be more than 4 dB noiser than that reference tube, we have a situation where fully 50% of the tubes we bought cannot be used in our microphones.
Now your requirements may vary. And of course if you use your mics for loud sources you can probably tolerate up to 6dB more noise that that Golden reference tube. The lesson here is simply that all tubes are not created equal.
"Caveat Emptor" (Buyer beware)
So here is a graph of the best ElectroHarmonix 6072A tube and the worst. Just for the record, that is a 16dB difference between 1500 Hz and 2Khz!
You can compare these two plots to the GE tube above. Some of the EH tubes compare very favourably.
Chinese tubes are good?
I had to show you this one. Here is the result from a Chinese **12AT7 tube pulled out of an old Nady TCM-1050. Not too shabby eh? The out of band noise (higher than 20Khz) noise is pretty high but in the mid range and low end it is excellent. I'm just saying... you can't tell a tube by it's cover.
**The new 1050's I see use a 12AX7. Nowhere near as good as this one
And one more thing
I should warn you that when you turn up the gain this high you can find other things too. I discovered that my test setup was being affected by the power transformers in my powered loudspeakers!
That's right, the magnetic field from the power supply in the speakers is creating some of that *180 Hz bump you see in the graphs. I am working on fixing that as well.
One step forward and half a step back... but we are making progress.
*3rd harmonic of North America's 60Hz power system
If you have any comments or questions on this article send us a note at: