mrkva

EIN measurement
« on: April 29, 2018, 08:13:59 AM »
Dear GroupDIY,

having hard time to find a easy to follow tutorial on measuring EIN of a
recorder.
So far I gathered this:

– measuring 150 ohm resistor (between 2 and 3 pin on XLR) seems to be a standard
– measuring should be done at highest possible preamp gain (seems like
recorders offer best results at that point)
– results should be constrained to a specific bandwidth of interest
(e.g. 20 Hz – 20kHz)

My question is – how to get from a bandwidth limited recording of a
150 ohm to the final EIN level?
Here is a list of “home-measured” recorders that inspired me:
http://www.avisoft.com/recordertests.htm

Thank you very much,
Jonas


squarewave

Re: EIN measurement
« Reply #1 on: April 29, 2018, 09:33:15 AM »
It could be difficult to measure the EIN in a way that could be compared to that table. There are so many things that could throw off the measurement by a few dB (or by a lot more). For one thing, how do you verify that the gain is really what the device claims on it's little LCD display? Normally you would measure it by feed in a very small signal (with a very carefully crafted shielded attenuator) and then simply compare that to the level coming out. But being a recorder you don't have access to the output. You only have what was recorded into a WAV file. So unless you go through some clever gain measurement procedure, you have to just assume that whatever it says the max gain is is accurate and just set to max. So let's say that is +60 dB. Then determining EIN is simple: you would just terminate the input with 150 ohms (but you would probably have to put a 150 resistor into the shell of an XLR housing because any exposed wires could pickup noise) and then record that (silence), import that into some software that will tell you the noise level and finally just subtract the gain from that. So let's say the software says your "silense" noise level is -66.5 dB which after a gain of 60 dB means -66.5 - 60 = -126.5 dB.

mrkva

Re: EIN measurement
« Reply #2 on: April 29, 2018, 11:05:03 AM »
Thanks, great point! I guess I will have to assume the manufacturers are not too far off wit the max gain setting then.

I have one more question though:
I grabbed a random 0,5 second silence in a recording as a test, ran it through the "amplitude statistics" in Adobe Audtion and received several different measurements. Which one is the one I would use in EIN measurement?

Thanks again!


squarewave

Re: EIN measurement
« Reply #3 on: April 29, 2018, 11:41:52 AM »
It's hard to say but Total RMS Amplitude is vaguely what you want maybe. So add (err subtract actually) the gain to that.

But the software is another source of failure because a sound card interface would have to be calibrated or the numbers are basically meaningless. It is very possible that different software programs even with the same interface could give you different numbers and a deviation of a few dB makes an EIN measurement basically meaningless. Unfortunately unless you have a fancy audio analyzer and not just something that is basically a glorified sound card, you are limited to comparative measurements. I use a QA400 audio analyzer and despite the fact that the software is designed specifically for it, I don't really trust the numbers from it. It is known to be a little off. It really is just a glorified sound card so I only really use it for comparative measurements. Meaning I collect data, make a change to the circuit and then test again and compare. Anything more than that an you need an Audio Precision analyzer or similar which costs more than a black market kidney.

mrkva

Re: EIN measurement
« Reply #4 on: April 29, 2018, 12:14:02 PM »
Yeah, just realised that those are dBFS values and I need a verified relationship between dBFS and dBu (or dBV) to obtain the exact value.
For example I've asked about that at Sound Devices and was told that with their products (at 0 gain) it is 0dBu = -20dBFS. That would mean that their specified EIN for MixPre-6 recorder of -130dBV (-128dBu) should be measured at -148dBFS...? I am ignoring the fact that it is A weighted value for now, not sure how to compensate for that.

Thanks a lot for the tips and the patience :)

Re: EIN measurement
« Reply #5 on: April 29, 2018, 01:25:34 PM »
Measuring EIN at maximum gain will give you a useful number, but many high gain amplifiers have multiple gain stages, so the EIN will vary with gain. When you measure with high gain, the EIN mostly depends on the first stage's performance, and the second or subsequent stages noise contribution, which may not vary with gain, will be diluted by the first stage noise.

So, it may be useful to also measure noise at lower total gain, since the often fixed output stage amplifier noises will not be diluted by the high first stage noise at high gain. If you find that you use a specific gain like 30dB or whatever for your work, that may be a more useful overall number than at some huge gain like 65dB or more that you probably won't use often.

And, measuring output noise directly, and not converting it to an EIN might also be more practical, since you can then convert that to dBFS easier.


 

Related Topics

  Subject / Started by Replies Last post
8 Replies
3197 Views
Last post October 08, 2004, 04:24:07 PM
by
3 Replies
1978 Views
Last post April 02, 2007, 12:52:22 PM
by Arrigotti
0 Replies
1173 Views
Last post April 16, 2007, 02:34:46 PM
by bcarso
0 Replies
923 Views
Last post August 03, 2009, 08:29:17 AM
by ENS Audio