Cheap Chinese Oscilloscope Preamp Tear Down/Review – Is It Any Good?

[gby]

I have had a need for scoping low level signals and been looking for a solution.  I have gone over the following EEVBlog threads:

   https://www.eevblog.com/forum/testgear/preamplifier-for-noise-measurements/
   https://www.eevblog.com/forum/testgear/quick-review-and-teardown-of-alphalab-lna10-low-noise-amplifier/

Unfortunately these choices are either fairly expensive, even used Ebay, or seemed to no longer be available.

I recently ran across an inexpensive Chinese unbranded “oscilloscope preamp” that I thought I would give a try.  Seems to be widely listed on Aliexpress, Ebay, etc for $60 on up deliverd.  In all cases the listing title was:

    10M Bandwidth Oscilloscope Differential Probe Signal Amplifier

I initially ordered from two different Aliexpress dealers and in both cases after listing the item as shipped a while later they asked me to cancel the order.  So, I am guessing that cheap Aliexpress shipping has trouble with the built in Lithium battery.  I finally got it from Banggood at:

   https://www.banggood.com/10M-Bandwidth-Oscilloscope-Differential-Probe-Signal-Amplifier-for-Weak-Electrical-Signal-Measurement-with-Metal-Shell-p-1959444.html?cur_warehouse=CN&rmmds=search

This post is a combined
  -   Tear down
  -   Review
  -   Request for help to finish reverse engineering and fixing

The published specifications of this T100 preamplifier are:

   Single-ended input voltage: -10V to +10V
   Differential Input Voltage: -10V to +10V
   Single-ended input impedance: 2M
   Differential input impedance: 4M
   Output Voltage: -10V to +10V
   Magnification: 1x/10x/100x
   Bandwidth:
      G=1 -- 10M
      G=10 -- 10M
      G=100 -- 2M
   Common mode rejection ratio: >80DB
   Noise : 5nV/√ @1KHz
   Power supply: built-in rechargeable lithium battery
   Charging interface: USB TYPEC
   Battery life: about 5 hours
   Working temperature: 0°C-40°C
   Working humidity: ≤90%RH
   Power consumption: ≤0.2W

TEAR DOWN:

The attached picture shows the item along with tear down front and back of the single internal PC board.
Assembled
Top side
Bottom side
Schematic
In reverse engineering it I have most of the circuitry traced out.  There are some mysteries/guesses because some the IC’s have the markings sanded off.  The high points are:
1.   Analog section is a single instrumentation amplifier with input direct from the input BNC’s and output direct to output BNC.
2.   Power supply is lithium battery to
     a.  Dc-dc boost to make +14.6Vdc
     b.  Dc-dc inverter to make -14.6Vdc
3.   Linear regulators to make +/-12Vdc
Lastly, see the attached partially finished schematic for this device.


REVIEW:

The Good:
  -   Compact
  -   Runs from internal rechargeable battery so easier to avoid noise from ground loops
  -   10 MHz BW for 1x and 10x with 2 MHz for 100x
  -   Reasonably low noise within caveat listed below

The Bad: (missing functionality)
  -   No ac coupling option
  -   No adjustable dc offset
  -   No adjustable low pass filter
  -   No adjustable high pass filter

The Ugly:
  -   -14.6 Vdc switching supply couples 70 mVp-p noise on the output which is way to big.  It must be fixed.
  -   Input pins direct to input BNC with no limiting impedance so more than 12V and device dies.
  -   Input impedance is set by 2 Meg resistor….would be better at 1 Meg so could use standard 10x scope probe.  Can use 10X probe and get known 5.5x atten at dc but standard probes don’t have enough ac compensation range for this pre-amp’s input capacitance.  Changing to 1 Meg will fix that.
  -   1x and 10x step response have HUGE overshoot.  Probably from stray capacitance on the gain adjust pins.
      o   1x has Po = 1.26 for 0.5 Vpp step or 252%!
      o   10x has Po = 0.70 for 0.5 Vpp output step or 140%!
      o   100x has no overshoot
  -   Output directly drives the output BNC and with a long cable the step response overshoot gets worse.  Included 0.5m cable adds 10% more overshoot.


NEXT STEPS:

1)   It would be great if anyone recognizes the instrumentation amplifier so that we could look at the data sheet.  It is:
  -   8 pin industry standard pinout instrumentation amplifier
  -   In SOIC package
  -   Gain equation is G = (1 + Rg/9.9k)
  -   Has 10 MHz BW at 10x
I can find many potential matches with the gain equation fixed/internal gain resistor at 24.7k or 3.0k.  But I have not been able to find one with 9.9k.

2)  The -Vdc switcher noise needs to be cleared from coupling to the output.  It does not get to the linear supply voltages so somehow this signal is direct coupling input the analog signals.  I highly suspect it is layout coupling.  Also, these noise spikes don’t change amplitude with changes in the gain so my suspicion is the VREF pin on the instrumentation amplifier that is tied to ground plane but near the -14.6V dc-dc supply output.

3)  Change the 2Meg input load R’s to 1 Meg so that 10x probes can work.  Probably just solder another 1 Meg R right on the bottom side at the input BNC’s.

4)  Is there a fix for the huge step response overshoot on 1x, 10x?  Probably have to wait until we find the data sheet for the instrumentation amplifier.

5)  Add a switch to get input ac coupling or maybe three-way switch and have DC, 1 Hz, 500 Hz with the last position used when there ac line noise you need to eliminate.  Any suggestions for a tiny three-way two pole switch that might work?

6)  Adjustable BW limit.  At least lower to 1 MHz and 100 kHz.  Easiest way to do this is on the output but it really wants to be done on the input which is hard to do and maintain high CMRR without insanely matched components and placement/layout.

7)  Any other thoughts from the EEVBlog community?

[thm_w]

Good info.
There is a BG review:

Very basic device, not a real lab instrument. Actually non-inverting input is the inverting one. 🤭 Metal shielding case is not grounded at all. On 100x range is a hughe phase shift and not compensated 50mV DC output offset. No frequency compesation at all, just basic opamp. +- 14.5V switching supply rails for opamp generates a lot of noise, no linear suppl. High figures of output noise in usable range, which is already very narrow. No real manual or measurement graphs available. Disapointed.

Here an alix link: https://www.aliexpress.com/item/3256804188345764.html
For the BG price, I think you could build a better device. Of course its going to take a lot of effort to design.

Why do you feel low and high pass are needed in hardware?

[Someone]

7) Any other thoughts from the EEVBlog community?

Don't bother wasting time on it, way too many problems to reverse engineer debug and then fix.

[gby]

Low pass and high pass in HW is needed if interfering signals are large enough to cause clipping.  Once you clip you can't get the signal back no matter what post processing you do in the digital scope.  When using 1x probe/connection it would not take much ac line frequency leakage into the signal to cause saturation if the scope is set below 10 mV/div and you want to see below 100 uv/div.

[thm_w]

Low pass and high pass in HW is needed if interfering signals are large enough to cause clipping.  Once you clip you can't get the signal back no matter what post processing you do in the digital scope.  When using 1x probe/connection it would not take much ac line frequency leakage into the signal to cause saturation if the scope is set below 10 mV/div and you want to see below 100 uv/div.

How does Alphalab get away with 1000x gain when there is a low pass filter only?
I didn't see a 60Hz notch in there.

[gby]

How does Alphalab get away with 1000x gain when there is a low pass filter only?
I didn't see a 60Hz notch in there.

To my knowledge no general purpose oscilloscope preamp incorporates a line frequency notch filter.  Setting the high pass to 500 Hz will drop 60 Hz by almost a factor of 10.  So, if your signal of interest is in the 5 kHz and above range, then 500 Hz high pass can be useful.

To me the only really fatal flaws in this T100 preamp are no ac coupling (so can't look at ripple on a dc supply) and the noise coupling from the minus supply dc-dc inverter to the output.

Someone, as for "...wasting time on it" I would say that it is a lot less effort to work on fixing this rather than starting from scratch with a clean sheet design.  Also, it is a hobby after all and sometimes just wanting to solve a problem is enough .

[raquenaengineering]

Hey,
I just got the same device you're showing on there.
Seen from another perspective (possible improvements)
User experience improvements:
- Not clear what the LED is for.
- No "charged" LED.
- No "power on" LED.
That made me believe I got a broken device to start with.

Questions:
- Would a 4 layer PCB with power and ground planes at 4 help for the noise?

[TopQuark]

I had this device, I don't recommend it.

Basically, the SMPS with inadequate output filtering swamps the noise performance of this thing, and as a result you can't resolve low level signal any better than a scope with a half decent front end, which makes the whole thing worthless imo.

The layout and circuit design of this thing is either half-arsed or designed by someone without adequate expertise. The supposed low noise performance of this thing is not due to clever design of some sort of discrete amplifier, but rather from an off the shelf amplifier chip. So the supposed value of this thing really is just the power supply and battery management, which they did a terrible job designing, so I don't see the point of this thing at all.

If you want to resolve low level signals that you can't see with your scope's noise floor and don't need BW pass  a few MHz, I suggest you just stick a decent in-amp along with the front end you need (filtering, ac coupling etc.) in a shielded box and hook two 9V batteries to it. It will work a lot better than this thing. AD8429 if you need selectable gain, AD8428 if you are ok with 2000x fixed gain and don't need overall BW over 3.5MHz.

[David Hess]

The shown transient response is horrible.

[schmitt trigger]

Someone, as for "...wasting time on it" I would say that it is a lot less effort to work on fixing this rather than starting from scratch with a clean sheet design.  Also, it is a hobby after all and sometimes just wanting to solve a problem is enough .

I agree with you on this one. It is fun to improve a shitty product and make it to stink a little less.

I would suggest to you something: To completely eliminate the power supply and replace it with a pair of A23 batteries. Those are 12 volts, thus your dinamyc range would decrease accordingly And they are also very low capacity. But at least this simple hack would allow you to better see what the actual S/N for the amp itself.

Depending on your findings, you may decide, or not, to add a connector to an external PSU.

[switchabl]

There are not actually that many fast inamps on the market. The AD8421 would seem to fit the bill (https://www.analog.com/media/en/technical-documentation/data-sheets/ad8421.pdf). If you take a look at fig. 52-56 in the datasheet, you see that it basically works as expected. It isn't really compensated for G=1 (or G=10).

The unterminated BNC cable essentially looks like a 100pF/m capacitive load at low frequencies, so the phase margin is reduced further. A 50 Ohm series resistor at the output would likely help with that. It doesn't solve the core problem though.

But to be honest: if you enjoy tinkering with this thing that is hard to argue with. If the goal is to have a working preamp, I seriously believe that it would be easier to start from scratch.

[rvalente]

After a few more months of use, have you experienced any issues?

[gby]

I would say the only thing to add to the original review is that the device self discharges the battery when just sitting there and off after a couple of weeks.  So, basically any time you want to use it you have to charge it up for an hour or two first.

I haven't used it much due to the fatal flaw of the switching supply noise coupling to the output.

[BennoG]

for those who want to reverse engineer here are some x-ray images of the board.
they are where 16bit tiff files converted to 32bit png files.
I am also drawing it in kicad but have little time for it.

Benno

[BennoG]

As promised the kicad files including pcb layout.

ToDo:  values of resistors and capacitors.
ToDo: multiplier resistors for the opamp.

The main problem is the output of the amplifier runs under 1 of the coils of the switching power supply.
I have cut the traces and short them to the ground plane. And use a thin coaxial cable from the output of the amplifier under the accu and then to the output.
This decreases the switching noise a lot.

Next step is to decouple the power supply from the LDO's in some way.

Benno

[5U4GB]

for those who want to reverse engineer here are some x-ray images of the board.

Just out of curiosity, that's not something that most people have sitting in their home lab, how are you getting these?

[BennoG]

Well I develop vision and industrial automation software as a small company in the netherlands.
And by coincidence because i find it a interesting, the speciality is becoming more and more to x-ray (over 20 years by now).
I have a special room here in the house with 6mm steel wall and on top of that 3 mm lead and behind that concrete.
I have a permit for x-ray development over here to 160kV and 30mA.
For the pcb I used a special microfocus x-ray tube at 50kV and 0.3mA and a 0.4 mm camera.
I we have a project that you see inside steel then you need more power like 100 kV or even 160 kv and more current for example 10mA.

P.S. if you buy a bunch of tulips the chance is about 90% that software I wrote looked at the length of the tulip with x-ray an bunched them at the same length with a machine that I wrote the software for.