Significantly reduce oscilloscope noise with a low-cost low-noise differential probe.
Vložit
- čas přidán 7. 02. 2023
- At low frequencies, such as audio signals, oscilloscope noise is usually 100x or more higher than what is possible with a low-noise differential amplified probe. These low-noise probes are usually very expensive (not to be confused with high-voltage differential probes, which are inexpensive but quite noisy). Now AlphaLab Inc. manufactures a very low noise differential amplifier/probe that is much less expensive-- the LNA 10. It's used for signals 1 MHz and below. Gains are 10/100/1000x, with a tunable low-pass filter 1Hz-1MHz.
www.alphalabinc.com/product/l... - Věda a technologie
Wow, that's an awesome tool at this price. It's on my short list!
Nice setup, good demonstration. Thanks
Thank you!
Awesome, congrats.
impressive
That is the perfect little 'calibrated' differential front end. My only question, is the distortion at 10x comparable to LM4562? I didn't see any specs on the website. I was going to build one myself - but for the price - I'd be crazy to match it.
edit: I have a GW Instek MDO2404eg scope. ~10mv p-p full scale12bit ADC, 50uV/1MHz dc coupled, 40uV ac coupled/20kHz digital filter. It has dual 14bit DDS that can be set as tracking notch around DUT for distortion measurement. Built in spectrum analyzer mode - what scope are you using in your demo?
The scope in the video is one of those el-cheapo Siglent ones. (I always thought that was a bad name--- too much like "Soylent". I wonder if it's made with real people.) The + and - input initial 10x op amps are opa4209. They are fixed at 10x gain, then are subtracted from each other, and that single-ended signal is gained by 1x, 10x, or 100x.
@@PattyDung That's a better choice for 'scope probe' 1M input. SOLD Thanks Duck Man, you should see my 'herd' 😎
Will this help me see the actual noise in a circuit, that is reducing the probe noise?
I am trying to see the actual noise in my circuit rather then using any kind of filtering to clean it up visually.
Yes, it's the real signal. For example, you can capture a single trace and analyze how much noise is present (and if a signal is present, how much signal is there also). For typical audio frequency work, you can resolve down to a microvolt, even with a single trace. (no signal averaging needed).
@@alphalabinc ok, so just to be sure, when i am now trying to measure noise in my circuit the probes are likely inducing noise and masking the actual noise of the circuit and this would help with the probe noise and let me see the real noise in my circuit?
I need to measure my voltage with an adc and really need to see/know the actual noise.
Just want to be sure before i buy, not an expert by any means :)
@@HerraHazar Please describe your setup (what you are measuring) in more detail. You can use the email in the ABOUT section of this channel. Not sure if an email address can be listed here in the comments, but it's mail (AT) alphalabinc (D0T) com
@@HerraHazar I'm guessing that you're doing audio frequency work. Then yes-- typical (non-differential) probes will introduce far more noise than the true noise in the circuit that's being tested. A low-noise differential probe, like the LNA 10, will eliminate that extra noise.
@@alphalabinc I am reading a dc voltage, moving between 170mV to about 3V in about 60-80mv increments, so about 37 increments into a 12 bit adc.
The increments are not 100% linear as I am implementing this into 1970's technology (and psu) so to recognize/read each increment I need very low noise in my design.
The adc reading is fluctuating quite a bit so i have some noise but with a regular scope probe it seems impossible to know exactly the noise on my signal.
I guess i could program the mcu to show me the noise in mV but knowing the real world noise pre adc would be immensely helpful.
Or just use two probes and math function to sum ch1 and inverse of ch2 for the poor man. But I'll admit that is one good unit to have around
It doesn't work with digital scopes. They digitize each of ch 1 & ch 2 first (most scopes are 8 bits, which is 256 levels). Then they're subtracted digitally. When trying this technique, I get even worse performance than a 256:1 CMRR that I expect, possibly because they sample the two channels at slightly different times or the lowpass filter is only applied after subtraction.
Very interesting. Can you try with a 4 ch scope with Ch1&Ch3? Mine has only 2 ch. Also, I never had such low noise demands but CM affects everyone...
I suspect many scopes use the same digitizing module for pairs of channels@@alphalabinc
@@OctavMandru Actually the best demo of the problem is to try it yourself by connecting both channels together to a 5V 1 KHz sine wave and doing the scope math subtraction. (Of course the gain must be set so both channels are on screen; otherwise they will clip). In theory you should see only a small math signal, but in fact the math signal typically jumps up and down randomly at high frequency (regardless of the filter setting) by typically 12 mV. A 12-bit scope that applies the filter before subtracting should work better, but I don't think a scope like that would have a reasonable cost, if available at all. 4 channels won't help if the scope is 8-bit. I've tried doing the math subtraction with several of the digital scopes while I was doing general audio circuit troubleshooting, and they were all terrible, which is one of the reasons I decided to produce this differential amp.
@@alphalabinc thanks for the reply. Will actually try that when I have the time but I have no doubt you are correct, especially with a square wave (although squares are not really audio domain given all the high f content).
Regards from Romania!
The cable you use for the scope is lousy. Use a better quality cable and the noise will be as weak as your wonderbox.
I do like the name Wonderbox. Anyway, you can't see microvolts out of these cheap digital scopes without some kind of amp, and common mode noise is also often a significant problem that is not fixed just by using top-quality cables. Good cables can shield pickup inside the cable from time-varying electric and magnetic fields and can avoid microphonic effects from flexing the cable, but other than having the best possible ground connection, they can't solve common mode noise. This noise often comes from magnetic pickup from the loops outside the cables, and even if the cable grounds were superconducting, it would still be present. Another interesting thing to note, if you are using one of these cheap scopes, is put a shunt on both ch 1 & 2 . Then set both at 1 V/div and use the math function to subtract: ch 1 minus ch 2. The result should have been a nice quiet straight line, but in fact these kind of scopes have a lot of noise in this configuration. A low noise amp with high CMRR can greatly improve the differential performance.
Ok, You can do as you see fit, I just don't have your problems. :-)