Differential preamplifier for oscilloscope

[rfguy2020]

Hello,

I would like to measure weak differential signal with oscilloscope. I have Rigol DS1054 and I was able to get some signal out by subtracting two channels, but as my signal is in level of few mV, it is not very good solution.

My signal is around 10kHz - 600kHz. I have fairly significant common mode noise around 4-10MHz, which I need to filter out. Amplitude on my signal is around 1mVpp.

I have been looking for some low cost preamplifiers like Alphalab LNA10 and Chinese T100. Those are a bit limited and I have been thinking to make my own preamplifier as it looks fairly simple. Has anyone done this project before or is there some ready made schematics of such board? I would need at least differential input, adjustable gain and low pass filter.

[ch_scr]

In theory, all you'll need is a differential amplifier. In practice, look at the CMRR of Opamps in the 4-10MHz range. That's before the additional perfomance hit from impedance/phase mismatch in the two input paths... So it's (quite) a project, not insurmounteable, but probably you'll learn something. Do you need any gain? How big is the voltage swing on those signals? Maybe putting a (faster) fully differential amplifier at the input and a (slower) single ended one behind it, that's able to drive 50R with the required amplitude, you might get some benefit in "stacking" the CMRR's like that?

[rfguy2020]

In theory, all you'll need is a differential amplifier. In practice, look at the CMRR of Opamps in the 4-10MHz range. That's before the additional perfomance hit from impedance/phase mismatch in the two input paths... So it's (quite) a project, not insurmounteable, but probably you'll learn something. Do you need any gain? How big is the voltage swing on those signals? Maybe putting a (faster) fully differential amplifier at the input and a (slower) single ended one behind it, that's able to drive 50R with the required amplitude, you might get some benefit in "stacking" the CMRR's like that?

My differential signal is around 1mVpp so I need probably 1000x gain. On the other hand, my common mode noise is level of 10mVpp.

[tszaboo]

It's not that difficult to build an amplifier like that. You start with an Instrumentation amplifier, like the classic INA128, select a gain for it that will amplify the signal enough, and give it a dual supply. The REF pin should be either ground, or driven by a low impedance source. The gain is selected by one resistor, you can place a dip switch to select the multiple ones, maybe a pot. The power supply can be two 9V batteries. Look at the gain/bandwidth graph, and make sure you don't use too much gain for your INA.

[David Hess]

I could not find the Chinese T100, but what did you find limiting about the Alphalab LNA10?

The old Tektronix AM502 will do what you want, and they usually work despite their age.

In theory, all you'll need is a differential amplifier.

Oh, I like that idea, and it is simpler than a complete instrumentation or difference amplifier.  A two operational amplifier differential amplifier has a common mode gain of 1, and adjustable differential gain, and only requires three resistors.  Pick some operational amplifiers with reasonable common mode rejection at 10 MHz and you are all set.  Let the oscilloscope do the subtraction.

[tszaboo]

I could not find the Chinese T100, but what did you find limiting about the Alphalab LNA10?

The old Tektronix AM502 will do what you want, and they usually work despite their age.

In theory, all you'll need is a differential amplifier.

Oh, I like that idea, and it is simpler than a complete instrumentation or difference amplifier.  A two operational amplifier differential amplifier has a common mode gain of 1, and adjustable differential gain, and only requires three resistors.  Pick some operational amplifiers with reasonable common mode rejection at 10 MHz and you are all set.  Let the oscilloscope do the subtraction.

Sure, keep in mind, with discrete components, your CMR is always limited by resistor accuracy. For 1% resistors it's like -40dB, while a proper integrated INA will do -140dB depending on your gain. So either hand pick components, or use 0.01% parts. Or use an INA, which isn't more difficult TBH.

[rfguy2020]

I could not find the Chinese T100, but what did you find limiting about the Alphalab LNA10?

T100:
https://www.alibaba.com/product-detail/T100-Differential-Probe-1X-10X-100X_60816853012.html

LNA10 is probably ok for my purposes, but it would be nice to have some option for higher gain and filter adjustments.

[David Hess]

Sure, keep in mind, with discrete components, your CMR is always limited by resistor accuracy. For 1% resistors it's like -40dB, while a proper integrated INA will do -140dB depending on your gain. So either hand pick components, or use 0.01% parts. Or use an INA, which isn't more difficult TBH.

But here the limit will be at higher frequencies where even if external resistive dividers are compensated, the common mode rejection of the amplifier falls.  The oscilloscope itself has limited common mode rejection at higher frequencies when used in add and invert mode.

This is why I like the idea of using only a two operational amplifier differential amplifier.  It boosts the common mode rejection of the following stages while being insensitive to common mode signals itself, although the operational amplifiers selected still need good high frequency common mode rejection performance.  At least unity gain stability may be relaxed.

LNA10 is probably ok for my purposes, but it would be nice to have some option for higher gain and filter adjustments.

The Tektronix AM502 has all of that, but is old and would require a power supply mainframe.

If I did not already have an AM502, I would build something from scratch as described earlier with a pair of operational amplifiers.

[sonpul]

I have been making and using an amplifier on SSM2019. The gain is 40dB. I use a multiplier of 0.01x

[TimFox]

The simple four-resistor-plus-op-amp differential amplifier does have a CMRR limited by the resistor matching.
To overcome this, since only ratio match is required, you can replace the fixed resistor to circuit ground with a resistor and trimpot in series:  for DC CMRR, adjust for negligible output change when applying a battery to both inputs simultaneously;  for AC CMRR, use a fixed capacitor across the feedback resistor and a trimmer capacitor in parallel with the resistor to ground.
Stanford Research Systems makes some very nice differential preamplifiers (a bit pricey), aimed at users of lock-in amplifiers.
Before that, Princeton Applied Research made similar units, some of which are available at reasonable price on eBay.
The SRS and PAR units normally have switchable hp and lp filters.
A true differential amplifier is almost always a better choice than computing the difference between two input channels.

[jonpaul]

See Tektronix diff amp model AM502 , manual at w140, Bama archive.

https://w140.com/tekwiki/wiki/AM502

Also Oscilloscope Vertical Amplifiers  (Circuit Concepts Series) Tektronix: Bob Orwiler

Jon

[sonpul]

I’ll add a few screenshots of SSM2019 working together with Siglent SDS1202X-E. 100kHz For the correct RMS value, see Stdev.

100kHz 1x probe



100kHz 1x probe Eres



100kHz SSM2019

[David Hess]

The simple four-resistor-plus-op-amp differential amplifier does have a CMRR limited by the resistor matching.

There is going to be confusion if confusing terms are used.  We should at least agree on definitions.

Differential amplifiers have differential inputs and outputs.  Difference and instrumentation amplifiers have differential inputs and single ended outputs.  Instrumentation amplifiers have high impedance inputs which minimize the effects of source impedance.

Texas Instruments among others likes to confuse the issue, but resist it.

[TimFox]

At least the first eight definitions of "differential amplifier" that come up on Google state "A differential amplifier is a type of electronic amplifier that amplifies the difference between two input voltages but suppresses any voltage common to the two inputs" (Wikipedia), and some others use the same definition for a "difference amplifier".
A push-pull, balanced, or "differential" output is not required, but may be used.

[rfguy2020]

I have been considering to use AD8129 with their evaluation board:

https://www.analog.com/media/en/technical-documentation/user-guides/UG-133.pdf

AD8129 is not probably lowest noise, but it might do the job?

[tszaboo]

The simple four-resistor-plus-op-amp differential amplifier does have a CMRR limited by the resistor matching.

There is going to be confusion if confusing terms are used.  We should at least agree on definitions.

Differential amplifiers have differential inputs and outputs.  Difference and instrumentation amplifiers have differential inputs and single ended outputs.  Instrumentation amplifiers have high impedance inputs which minimize the effects of source impedance.

Texas Instruments among others likes to confuse the issue, but resist it.

I don't find it confusing, though I usually use the term 'fully differential amplifier' for the amplifier that has differential input and output (even though usually those are high speed).
So why is CMR an issue? Let's say you put together an amplifier out of 1% resistors, and it has a CMR of -40dB. You want to use this amplifier to amplify 1mV signal by 1000. If your input signals are 0V and 1mV, your output is 1V, everything nice. Let's say your input is 1V and 1.001V. Your expected output is 1V, but because of resistor mismatch, you get 11V (-40dB is 1/100). And this might happen at DC, or it picks up 50Hz noise, or your signal has other sources of common mode noise. There are ways to improve this, for example ferrite cores and placing your wires through it, but ultimately, it's the best to use very much matched resistors.
And you can either select these by hand, get expensive 0.001% resistors, or get one of these nice ICs, where these are internally matched, often times laser or electronically trimmed. But ultimately it's characterized, which is important. You don't get surprised, when the output is 11V, because the IC has -100dB CMR.
And you might be familiar with this, and maybe there is no new information for you in my post.

I have been considering to use AD8129 with their evaluation board:

https://www.analog.com/media/en/technical-documentation/user-guides/UG-133.pdf

AD8129 is not probably lowest noise, but it might do the job?

Looks like a very useful board for this purpose.

[ch_scr]

(...)
I don't find it confusing, though I usually use the term 'fully differential amplifier' for the amplifier that has differential input and output (even though usually those are high speed).
(...)

An example for a low speed, high voltage, cheap-ish (and precision to boot!) fully differential amplifier is TI's THP210

(...)
but ultimately, it's the best to use very much matched resistors.
 (...)

To automate such selection tasks in general, I've made up a python script.For an arbitrary example of "tightest 8 values from the list of measured examples", it looks like this:

Code: [Select]

grouping_optimizer.py 8 6.932 6.939 6.9155 6.8915 6.905 6.909 6.936 6.943 6.9355 6.9355 6.933 6.943
Average value 6.93712
Error sum 3928.14
Average error 491.02 ppm
With these elements and corresponding errors:
(6.932, 6.939, 6.936, 6.943, 6.9355, 6.9355, 6.933, 6.943)
['-738.78', '270.28', '-162.17', '846.89', '-234.25', '-234.25', '-594.63', '846.89']It can also be asked for a specific target value as well:

Code: [Select]

grouping_optimizer.py 8 6.932 6.939 6.9155 6.8915 6.905 6.909 6.936 6.943 6.9355 6.9355 6.933 6.943 --target 6.918
Average value 6.92519
Error sum 15394.62
Average error 1924.33 ppm
With these elements and corresponding errors:
(6.932, 6.9155, 6.905, 6.909, 6.936, 6.9355, 6.9355, 6.933)
['2023.71', '-361.38', '-1879.16', '-1300.95', '2601.91', '2529.63', '2529.63', '2168.26']It works happily with multiple elements of the same value, can also show the iterations and handle priority elements much like the resistor divider ratio optimizer.
With this and an grid of measured examples on a piece of paper, quite a tight match is easily achived, for small batches. The "divider optimizer" can be used in the same vain to automatically get multiple "equal ratio dividers" from "excessive examples of random measured resistors". If they are measured at the same time with a warmed up instrument under stable conditions, the absoulte errors (like calibration) cancel out. As a bonus the "arbitrary ratio" of the divider comes free, within the limits of gaussian distribution* of the resistor examples  (*not guaranteed to be gaussian)

[hpw]

I would like to measure weak differential signal with oscilloscope. I have Rigol DS1054 and I was able to get some signal out by subtracting two channels, but as my signal is in level of few mV, it is not very good solution.

My signal is around 10kHz - 600kHz. I have fairly significant common mode noise around 4-10MHz, which I need to filter out. Amplitude on my signal is around 1mVpp.

I have been looking for some low cost preamplifiers like Alphalab LNA10 and Chinese T100. Those are a bit limited and I have been thinking to make my own preamplifier as it looks fairly simple. Has anyone done this project before or is there some ready made schematics of such board? I would need at least differential input, adjustable gain and low pass filter.

I am looking for such differential gear as 0.01Hz to xx-Mhz is my target as to measure ADC/DAC power noise/figures as most ADC/DAC are internally cap. switcher so the today's LDO regulators may at higher freq. on the limits.
Even shunt regulators have some limits. While simulation tell not the full trough.

Or may the new 12 bit DSO's using averaging, may help in this requirement! But still a differential probe is a must have.

Hp

[leowood]

This is the T100 freq response looks like, from 1M to 10M (they rated the useful BW up to 10M when gain setting is X1，)

So there will be a serious design flaw, the 5M gain will be 3x high than 1Mhz, terrible

[hpw]

I have been considering to use AD8129 with their evaluation board:

https://www.analog.com/media/en/technical-documentation/user-guides/UG-133.pdf

AD8129 is not probably lowest noise, but it might do the job?

Well, it looks for me, that we / I need a LNA one for 1/f noise and one for RF noise 

Currently on DIYAUDIO https://www.diyaudio.com/community/threads/ad8428-low-noise-preamplifier.407270/ as for the 1/f project with 2000 gain chip as AD8428.

As for the RF LNA & AD8129 testing: BW RF range ends soon about 20..200Mhz and CMR gets down to 40..50dB. Some other project as seen on TI forum.

Important will be to protect the opamp input as using ADC & Capacitors as connecting various power sources as capacitor to charge and discharge.
In addition the connecting cable should not pickup any RF  so shielding is must apply.

Hp