EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: rx8pilot on May 02, 2015, 11:49:03 pm
-
Hi all, A few months back I got a Tektronix ADA400A which is a differential amplifier for my scope. One of the goals was to be able to measure across sense resistors to capture a current trace directly from a sense resistor.
In my case, I am in the neighborhood of 50mV or less. I have tested the unit on various known points and get the expected result - cross checking with a DMM. The amplifier agrees very closely with the DMM numbers on known DC. When I try to measure across a .005 Ohm sense resistor, it shows nothing. The DMM gives me the expected values as I change the current but the ADA does not.
Any ideas?
-
How is everything earthed? What the common mode voltage relative to earth on the resistor?
-
The scope is earth grounded but the DUT is floating. Strange thing is that I can accurately measure a number of other spots on the same DUT/setup successfully. From 8mV to 5V, that all agree.
-
You need to earth the DUT and make sure the common mode voltage for your differential amplifier is not exceeded out of principle ... this is not an isolation amplifier.
-
I am still trying to understand this problem. Keep in mind, I am not a degreed' engineer so there may be a concept that I am missing.
I deliberately powered the DUT from a isolated supply to reduce the chance of an 'oops' moment while probing. The differential amplifier only amplifies the difference between the two probe tips so no ground is ever presented to the DUT. The amplifier has an adjustable offset as well as a limit indicator if it is over voltage. It would seem that if some common mode voltage was present, it would show up on the output to the scope, right? What I am seeing, is a ZERO output where my DMM indicate 8.5mV - which is the value I calculated to be correct.
The question is how would grounding the DUT make a difference?
-
Like he said, common mode range. The amplifier doesn't amplify if the absolute (not relative / differential) voltage is all crazy. By explicitly floating the equipment, you're practically guaranteeing that!
Don't worry about degree; tons of degreed engineers remain ignorant of things like common mode noise, electromagnetism, shielding, etc.
Tim
-
Common mode voltage continues to baffle me - has not yet clicked.
I will ground the DUT and take a look followed by spending some time understanding why it works. Thank you for the tips!
-
The differential amplifier only amplifies the difference between the two probe tips so no ground is ever presented to the DUT.
No it does not, that's what an isolation amplifier does ... the differential amplifier gives you the difference between the oscilloscope ground referenced voltage at one probe and the oscilloscope ground referenced voltage at the other probe, both those voltages have to be less than the maximum common mode voltage of the differential amplifier. If the oscilloscope is earth grounded and the voltages being probed relative to earth are unknown it's unknown whether the differential amplifier is within it's maximum common mode voltage. It might work, it might not (as you noticed).
You probably don't need the differential amplifier at all ... you could probably just float the DUT and simply use the oscilloscope ground clip and probe to take your "differential" measurement. The reason people don't use this method all the time when it's possible to float the DUT is that it is not truly differential, the impedance added into the circuit by the ground clip and the probe tip are different. Majorly the power rails of the circuit are capacitively coupled to the mains through the transformer, whatever point you clip the ground clip into now has to drive that capacitance because the rails are being forced to move relative to earth (galvanic isolation does not help here, batteries do though, but even then the ground plane has parasitic capacitance to earth).
The differential amplifier you are using is for sensitive measurements where such capacitive loading inside the circuit is unacceptable and meant to be used with both the DUT and the scope earth grounded ...
PS. if you want to take a differential measurement in a mains voltage circuit you should use high voltage differential probes ... and still earth both the DUT and the scope.
-
Ok, that all makes sense. Thank you for the detail.
I setup a quick test with the DUT grounded and the result is the same. Measured the Earth to DUT ground at 7v, then after tying the earth ground to the DUT of course it was zero. The result of measuring across the sense resistor was the same - ~900uV DC offset regardless of the the current flowing through the resistor. I re-checked some other voltages to compare with my DMM and they match up - just not the sense resistor. The voltage at the resistor is 14V or so relative to the return/gnd.
Maybe I should create a block diagram of what I have setup to better communicate.
-
Updates with success......
As I continued to look at this test setup, I realized that the electronic load is grounded. When I previously thought the DUT was floating it was actually grounded through the electronic load. I decided to float the DUT and use my passive load box - resistors and capacitors not connected to anythying. With everything isolated from ground except the scope I get the results I was expecting. It is very clean and matches what I calculated to the results to be which is encouraging.
While I am super happy I can now see transient currents in my circuits - I am still confused about the previous suggestions. Should the ground reference method have worked?
-
Voltage is always a difference.
A differential amplifier might have an output which is a function of the input difference, given that other constraints are satisfied.
When those constraints are not satisfied, it ceases to amplify, or be differential, and things aren't right.
If you look at the standard "long tailed pair" circuit, you'll notice it won't work if the inputs are too low (NPNs won't be forward biased = no bias current flows = output difference goes to zero), or too high (one or both saturates C-E or C-B, and output difference again goes to zero). Moreover, input bias current varies when these limits occur (goes to zero when turned off, or acts like a diode to +V when saturating "overhead").
Differential signals are frequently touted as some magical cure-all for signaling purposes, but this is rarely the case, and willfully ignores the necessity of common mode (or simply normal mode -- what each signal vs. ground is doing, in and of itself). In the case of amplifiers, normal mode limitations apply. In the case of signals (including RS-422, USB, Ethernet and many more), common mode impedance typically needs to be well defined, otherwise unintended signals bounce around and exceed the common mode range. USB takes care of this with complete shielding (and any breaks in the shielding will cause big problems), Ethernet by using isolation transformers.
TLDR: if you ever felt weird that you're putting voltage on a wire, even though you've supposedly been told that it's the difference (i.e., between two wires, not to ground) that matters... your instinct is right and everyone who says otherwise is wrong.
Tim
-
No it does not, that's what an isolation amplifier does ... the differential amplifier gives you the difference between the oscilloscope ground referenced voltage at one probe and the oscilloscope ground referenced voltage at the other probe, both those voltages have to be less than the maximum common mode voltage of the differential amplifier.
This nugget was key to figuring this one out. The setup I have is to measure the voltage drop of a current sense resistor. So, both probes are probing the high-side of the circuit which for the test was about 14v. With no ground reference at all, I could see the correct result. When the DUT was grounded the output was a constant 900uV or so.
After thinking through it and referring to the specs of the amplifier I finally put together what was happening. Grounding the DUT is certainly the right way to set this up. The problem is from the amplifier being overloaded on each side of the differential amplifier (as pointed out by Marco). The 900uV that I would see, was the difference in the clipping points on the amplifiers. I did not understand what the implication of the 'common mode range' was and that created the confusion. The max common mode range of the amplifier is 10v and I was presenting it with 14v. Great lesson embedded in this exercise.
Now I can float the DUT and measure the current, but I need to be more clever to be able to use the remaining 3 channels of the scope along with the differential amplifier. I was thinking of using a precision resistor voltage divider to get the current sense voltage inside the common mode range of the amplifier. That way I should be able to ground reference the DUT to see all the other signals at the same time.
Thank you Marco and Tim for the information. It is a key concept that was holding me back.
I roughed in a circuit that should get the differential amplifier in range while being referenced to ground. A 50% divider was chosen to make it easy to correlate in the end - just double the result. 1k was chosen to be a 'stiff' enough divider to not be influenced by the measurement and have no impact on the in/out. Should be good in the very low ranges <few mV where noise overwhelms the measurements.
Do you think this is a good solution?
(https://solidcamera.box.com/shared/static/qxlvdm41z6x75wcnwd9efit4rzm30hai.png)
-
Note that the two resistor pairs need to be ratio-matched very precisely -- as precise as your measurement and CMRR (common mode rejection ratio) dictate!
For the 1/2 ratio shown, this shouldn't be a big deal; 1k, 1% or better resistors aren't a pain.
It's very tempting to make wide range diff amps this way, using large ratios to measure high voltages, but the precision (especially for high value resistors) is often uneconomical, and a better method is chosen (such as a proper isolation amp, or for AC, using transformers instead).
In the case of a current shunt, you can also get modest precision (to DC) using a Hall effect sensor.
Tim
-
I think most high voltage differential ("isolated" - note how some manufacturers never say that) are actually high input impedance very high common mode range diff amps and not real isolation amps (which are even more difficult to design and manufacture than a high input impedance very high common mode range diff amp).
Although it might be an option for a DIY probe, something like a buffer amp, transformer for HF and ADC+digital coupler+DAC for LF. But that of course requires a high side power source, which can be tricky.
-
The precision resistors at .01% are cheap enough for a 1-off fixture so that should be ok. Since the voltage will never be over 20v, no need for any high ratios. I am probably going to build a little interface that will house this little divider.
The sense resistor in the schematic is actually part of the DUT. The DUT is a power management and monitoring circuit so there are 7 sense resistors on the PCB. The goal is to be able to see the transient currents at any point. Current probes would be too disruptive to the circuit and too difficult to use on a production PCB since a wire would need to be soldered in/out for a test.
Inrush current is one of the top concerns that I am trying to see. The DUT will drive loads that are totally out of my control. I will be acquiring samples of the worst case scenario loads to characterize and design my circuit to handle them.