Built a Differential O-Scope Amp

[blasphemy000]

I was in need of a differential scope probe for use in building my SMPSs but at the present time I was unable to afford even the cheapest of differential probes as the cheapest new one I found was $360 dollars and the cheapest used one I found was $200. Now I know that this differential amp I built is not up to par with a professionally manufactured one as far as bandwidth and CMRR goes, but it does function well for what I'm doing as it at least allows me to view the waveform between two non-ground points.

I built this probe using a TL074 quad op-amp which has a slew-rate of 13V/uS. Input attenuation to the amplifier box is X100 so that with a 600V input to a probe gives only 6V to the op-amp's input. I'm achieving this attenuation for each probe by using X10 probes(9M Impedance) coupled with a 100K/560pF input impedance of the amplifier box. I know this isn't ideal, its not completely accurate, and it only gives a total of 9.1MOhm input impedance. I do have 2 X100 probes on order, they should be here soon, and once they get here I will be changing the input stage of my differential amplifier to match the input stage of my scope giving a proper X100 attenuation.

The amplifier box has connections for 2 scope probes. I left the GND of these probe connectors unhooked that way if I accidentally forget and hook up the ground clip of a probe it won't short out and explode. The amp box does have a central earth-ground post located below/between the two probe connectors so that if I am working on a completely floating circuit, ie: battery powered or a completely isolated power supply such as the 16V supply I posted in the DIYSMPS section, I can attach the earth-ground lead to the ground of the floating circuit and make the circuit earth-referenced. The amp will work without this lead attached to a floating circuit, but connecting it does reduce the noise in the circuit. If the circuit I am working on is already earth-referenced, ie: the mains side of a SMPS, then this GND/Earth connection on the differential amp is left unconnected and both of the probes may be connected to points away from GND/Earth.

I've posted some pictures, the schematic, and circuit board layout of the current(first) version of this amplifier. In the picture of the scope trace, I have the GND/Earth lead connected to the GND of my 16V floating supply to give it an earth reference, then the each of the two probes are connected to the two outputs of a TL494 running at 50KHz. Now that I am able to measure the difference between both outputs I can easily see the amount of dead-time that is between the outputs, where as before I was only able to measure a single output and was unable to see the deadtime. I have tested this amplifier with the probes connected directly to the mains voltage and I've captured the waveform of a D-S across a MOSFET in an ATX PSU.

The amplifier runs from +9V/-9V power that is supplied by two 9V batteries located inside the box.


View attachment Differential Board.pdfView attachment Differential Sch.pdf

 

[dinithm]

great work.
I'll try this soon ,thanks

 

[blasphemy000]

Like I said, it is _FAR_ from 100% accurate and the bandwidth leaves a lot to be desired, but so far it has proved to be a helpful device.

I ordered some samples from TI.com of their THS3201 op-amps. They have a slew-rate of 6700V/uS so once those arrive I'm going to redesign the circuit using them and see if I can't improve the bandwidth. I'm sure they will be overkill as my scope only goes to 5MHz anyways.

Also, if anybody does build this, it isn't shown on the schematic, but the inputs are each connected to Tektronix X10 scope probes. These probes have an internal 9MOhm resistor in parallel with a 3.5-7.5pF trimmer capacitor and these two items absolutely must be there. Without them you will blow this circuit to pieces and most likely damage your scope if the inputs of the TL074 short to the outputs. In fact I think I'm going to modify the schematic to include these items just to be sure everyone understands properly.

 

[blasphemy000]

Here is the updated schematic to include the RC combination that is inside of the two probes I am using with this amplifier. I cannot stress enough how important the RC combination inside of the probe handle is. This amplifier will self-destruct if you attach full mains voltage to the input without the RC combo inside the probes.

Also, the 560pF caps that are on the input of the circuit board are rated for 1KV at minimum.

View attachment Differential Sch 2.pdf

 

[MicrosiM]

very nice work.

Thank you for sharing :UP:

 

[blasphemy000]

You're very welcome. I've learned quite a lot from this site, and when I start building the massive welding SMPS I'm going to be undertaking soon, I'm sure I'm going to need some help fine-tuning it, so as soon as I get all the parts collected and the prototype boards made(going to have them professionally done) I'll probably be asking lots of questions.

 

[blasphemy000]

Just made some slight modifications to this device. I've removed R3 & R4 because they serve no purpose since the input stage op-amps are configured as voltage followers. And I've replaced the TL074 with a LT1365CN. This vastly improved the bandwidth of the device.

 

[phaedrus]

Just to be assured :
Output+/Output- will go to single channel of the oscilloscope ?Could you verify with your x100 scope probe.
Thanks.

 

[blasphemy000]

You need a piece of coax cable with a BNC connector on one end, take the unterminated end and strip back the insulation. The outer "shield/ground" of the coax goes to OUT- and the center conductor of the coax goes to OUT+, then the BNC connector plugs directly into your scope. Please make sure you completely understand the updated schematic from POST #4 and not the one from the first post. Make sure you are using probes with the same internal components as is in the schematic. If the internal components of your probes are different values it will throw off the voltage divider and you could damage your scope with too much input voltage. If you are unsure if you've constructed this circuit properly then instead of hooking the output connection to your scope you can "test" your build of this circuit by connecting the output to a volt-meter but instead of leaving your volt meter floating, connect the negative side of your volt meter to an earth connection, this will simulate the type of connection that is found in your scope. Then you can apply different input voltages to the 2 input probes and read the output on your volt-meter to make sure the 100X division is functioning properly. IE: 10V across the input probes would yield 100mV reading on your volt meter. Also be sure to try AC inputs as well as DC inputs to make sure that both the resistor and capacitor voltage dividers are both working properly.

Also, I'm only using 10X probes with my amplifier but I'm getting 100X attenuation because I'm not using a standard termination on the input of the amplifier which is reflected in the schematic. My 10X probes were designed for a 1M-Ohm termination but my amp terminates them into 100K which changes the DC attenuation from 10X to 100X, the 560pF termination cap is also non-standard and yields 100X AC attenuation.

One more thing to note is that if you look at the picture of my completed amp box, there is a grounding post between the Probe connectors(my scope doesn't use BNC connectors, its a really old scope). This ground post is connected to the OUT- pin as well and when hooked to the scope it's also earth-grounded. This is used when a circuit is completely floating(ie: powered from batteries, or a fully isolated power supply with NO earth connection) so that you can earth-reference the circuit you're testing. This reduces noise across the two input probes and also assures that the voltage dividers can properly sink the current of a fully floating circuit. If the circuit you are testing is already earth referenced then you don't need to worry about the grounding post connection and either or both of the two probes may be attached at any point away from ground.

I'm sorry if my explanation of these things isn't very clear but this is the best way I knew how to describe the workings of this circuit. If you are in anyway unsure of how this circuit works or you are unable to understand my testing procedures, and you connect it to your scope and measure a large voltage and your scope explodes, I cannot be held responsible. I've tested this circuit on my own scope and I've measured very low voltages(1-20V) and I've also measured voltages directly from the mains coming out of the wall(120VAC) and I've also measured voltages of switching stages connected to rectified mains voltage and a voltage doubler(340VDC) without any issues so I know this works the way I have it configured. Thank you.

 

[KX36]

Thanks for sharing!

The 2x 9V batteries is a good way of powering this for low noise and good isolation, but a low battery indicator might be a good idea. One small modification I'd suggest is to tie off the unused opamp by making it a voltage follower with the +IN tied to a mid bias point (ground since yours is split supply). That's a standard thing on opamps as the high gain and open loop can cause the output to oscillate and be a noise source. If you make your input a 900k/100k divider you should have the right 1Meg input for your 10x probe and get 100x out of it for the inamp.

There are of course instrumentation amplifiers you can buy off the shelf instead of making your own from general purpose opamps. There is also a 2 opamp inamp topology as well as the 3 opamp one you've used, but it has a min gain of 2. The performance of your circuit depends on the matching of resistors etc, and a laser-trimmed inamp would perform better because of this.

I might have a go at making a wider bandwidth differential probe adapter with an AD8130. Not an inamp but a relatively cheap high speed differential to single ended converter. At first glance it looks suitable. With +/-12V supplies, gain is flat to 110MHz, common mode range is +/-10.5V and differential mode range is +/-2.5V which as you know would be multiplied by the input attenuator. It should also work on +/-9V with the same diff mode range but lower bandwidth and common mode range.

Regards
Matt

 

[malch]

KX36 any chance you could make a schematic of what you posted, especially about the unused opamp.? This mod would be with R3 and R4 missing?