EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: Trurl on June 14, 2024, 12:11:07 am
-
Is there a safe and simple method to convert a floating "wall wart" to be "grounded"?
Grounded "brick"supplies are much larger than the typical phone/USB charger "wall wart", so I'd like to convert a "wall wart" to be grounded if possible.
[ Reason For Conversion Need ]
I just realized that the "wild" fluctuation in my ICL7107 based "ammeter" caused by some power sources are all "floating"(2-prongers). Grounded supplies don't cause these fluctuations in the ammeter.
I noticed a difference and fluctuating ground potential difference between the ammeter monitored device(a server PSU) and the floating supplies that power the ammeter, in the range of about 60~130mV(higher in the "floating" supplies). This ground potential difference and fluctuation between the supplies made the ammeter's readings fluctuate as well.
For more details refer to the thread below:
https://www.eevblog.com/forum/beginners/current-sense-output-disrupting-ammeter-powered-with-most-smps-but-not-all()/msg5539861/#msg5539861 (https://www.eevblog.com/forum/beginners/current-sense-output-disrupting-ammeter-powered-with-most-smps-but-not-all()/msg5539861/#msg5539861)
Cheers~
-
No. Most floating wallwarts either have no ground pin or a dummy plastic one simply to open the protective shutters in some countries mains sockets, so there is no good way of making a ground connection to the socket the wallwart is in. The plastic housing is also commonly ultrasonic welded, so cant be opened to make circuit changes without compromising its integrity, and if you decide to remove the PCB to remount in a different enclosure more suitable for modification you will find they nearly all have no mounting holes and little physical clearance between the edges and adjacent components so cant simply be slid into PCB guides.
You have an X-Y problem. The maximum permitted leakage current from a Class II wall wart (0.75mA) should *NOT* be causing your digital ammeter module to misbehave, but due to your modifications its flowing through some part of the ICL7107 input network, thus producing an AC voltage superimposed on the DC voltage you are trying to display. This may be overloading the ICL7107 inputs, resulting in partial rectification and your observed negative readings and also possible latchup and even permanent damage . Grounding out the leakage current at the auxiliary power supply is not a reliable solution. If any other part of the circuit gets grounded you will have a ground loop inducing noise (hum), and it will also remove the isolation needed for the meter supply. Its like putting a band-aid over a deep infected wound! A more appropriate solution would be to use an isolated DC-DC converter module to power the meter from the server PSU while still keeping it floating.
-
Usually the secondary side of "wallwarts" are floating, and you can connect either of the outputs to earth to create either a "+5V" or "15V" power supply.
But it won't work with your ICL7106 / ICL7107. Those old IC's can not measure a ground referenced signal. The inputs have to be at least 1V above the "gnd" level of the IC, and it also needs 500mV headroom on the power supply rails. It is very common for these meters to be powered from separate floating power supplies.
I assume you need to add some kind of filtering or extra decoupling for your meters to work properly. It's also possible the design around your meter is flawed, or maybe even you have some fake IC's that don't meet their specification at all. If you have an oscilloscope, you can start by checking whether there is excessive noise or oscillations in your circuit.
-
I do not recommend it.
-
Thank you all, for your considerations...
As grounded laptop/monitor "brick" SMPS supplies work with the meter, I searched online for "laptop power supply schematic" seeking to find out how the these units are grounded internally.
I didn't find many leads, but I did find a YouTube video teardown of a grounded LiteOn SMPS. At about 06:24 into the video a hand drawn schematic is presented, in which the internal grounding is shown along the bottom of the schematic(screen capture attached below video, so you can skip the video, although it is amusing with the presenter's voice(e.g. around 5:14): "BLOODY HE~LL~! What DISASTER happened HEEE~RRRE~":-DD).
https://www.youtube.com/watch?v=-vBlqSMwVh0&ab_channel=DiodeGoneWild (https://www.youtube.com/watch?v=-vBlqSMwVh0&ab_channel=DiodeGoneWild)
Schematic
[attach=1]
I'm wondering if a similar grounding(with decoupling capacitors and resistors) is done internally to a wall wart, it could be converted to "grounded" like the laptop/monitor grounded supplies. I checked the resistance of my "good" 12V grounded supply between the output's negative and the input ground and it was about 1.2K BTW(the schematic above has a 1.2k towards the AC input + a 1M towards the DC output negative).
Note: I've tried connecting a Murata 5V-5V DC-DC converter to the 12V server PSU via a 5V regulator in between, but again there is a ground offset which results in a higher reading at the meter(without the fluctuations seen with non-grounded supplies). The meter is not damaged, as it works fine with grounded supplies.
The fact that a 2-prong wall wart does not have a third physical conductor at the AC input is not really a big issue. I'm more interested in whether the grounding concept like that shown above in the LiteOn unit would work on a wall wort, and what considerations need to be taken for the decoupling capacitors(type, values etc.)
Cheers~
-
Removing the Y capacitor that provides the HF return path between the secondary and primary sides of the wallwart SMPS, (and is responsible for nearly all the leakage current), is not simple, and may be unsafe due to housing damage while accessing it. You cant convert to the three pin mains input laptop/monitor PSU setup without doing that. There may not even be room to safely retrofit the grounded PSU EMI suppression network. Directly grounding the two pin wallwart output is undesirable as discussed above.
Look for an old wallwart with a line frequency transformer in it rather than a SMPSU as their leakage current is much, much lower. They are easily identifiable as compared to modern SMPSU wallwarts, they are very heavy for their size. Odds are, one will work OK without grounding its output.
I'd first revist the DC-DC converter and check carefully that neither the supply current to it nor the output current from it was passing through the meter shunt as that could account for the observed offset.
-
How do you know it's causing the problem?
For test purposes, try connecting the negative side of your power supply to a main earth wire: does it solve the problem?
I suspect this is an EMC issue.
-
How do you know it's causing the problem?
For test purposes, try connecting the negative side of your power supply to a main earth wire: does it solve the problem?
I suspect this is an EMC issue.
Holy schmidt! I just checked with a 5.2V-ish 2-prong wall wart(with a ferrite core) connected directly to the meter, and with the wall wart's negative connected to one of the power strip's ground conductors, the meter's fluctuation absolutely vanished! The reading is steady as a rock! I checked the ground potentials and in this case there is about a 8~9mV "steady" difference(in this case the wall wart's ground is lower than the server PSU being monitored by the meter). Without the "grounding" there is like up to 130mV fluctuating difference. The slight offset results in the meter's reading being slightly higher than it should be, but it does confirm that the "wild fluctuations" in the meter are definitely related to grounding.
Can you elaborate further if pertinent in this case how an EMC issue would be related, and what a solution might be in that case? Thanks!
-
Removing the Y capacitor that provides the HF return path between the secondary and primary sides of the wallwart SMPS, (and is responsible for nearly all the leakage current), is not simple, and may be unsafe due to housing damage while accessing it. You cant convert to the three pin mains input laptop/monitor PSU setup without doing that. There may not even be room to safely retrofit the grounded PSU EMI suppression network. Directly grounding the two pin wallwart output is undesirable as discussed above.
Look for an old wallwart with a line frequency transformer in it rather than a SMPSU as their leakage current is much, much lower. They are easily identifiable as compared to modern SMPSU wallwarts, they are very heavy for their size. Odds are, one will work OK without grounding its output.
I'd first revist the DC-DC converter and check carefully that neither the supply current to it nor the output current from it was passing through the meter shunt as that could account for the observed offset.
I don't intend to alter any existing Y capacitors.
I just tested with 3 different transformer wall worts and they essentially produced the same results as with the 5.2V SMPS wall wort I mentioned in the previous post (the ground offset and its fluctuations are directly reflected in the meter's readings, and when the negative is grounded to the power strip's ground conductor, the ground offset is zeroed or near it, the fluctuations in the readings stop, and the readings are slightly higher than they should be due to the slightly lower ground potential, but pretty close.)
There is no shunt in the circuit(not that I've placed anyway), because the "ammeter" is really just a voltmeter with a reference set to 2.5V(taken directly from a TL431 in to the ICL7107's REF HI, instead of the usual 1V via the onboard voltage divider) so that the server PSU's 25mV/A CS(Current Sense) output signal can be directly input into the ICL7107's IN HI without a voltage divider at the input.
I still truly appreciate your comments...
-
...
I'd first revis(i)t the DC-DC converter ...
I did just that today and instead of trying to "ground" the 5V regulator's negative, I tried "grounding" the DC-DC converter's output side negative to the server PSU's ground tab(not the power strip's "earth" conductor) and the meter came back to life!(Just like how "grounding" a secondary supply's negative to the power strip's "earth" conductor eliminated the meter's fluctuations).
The DC-DC converter is tiny, about the size of a blue colored 5k potentiometer, so I think I'll go with the DC-DC converter which is what I'd been hoping to use all along.
I'm so glad I can thank you for success! :-+
In the photos below, on the breadboard's center there is a LM7805 5V regulator and in front of it is the tiny DC-DC converter(a Murata CRE1S0505SC). The DC-DC converter's OUTPUT negative is linked back to the PSU's negative. The multimeter displays the CS(Current Sense) output voltage at no load(quiescent voltage 31mV) and at about 2A(31mV + 50mV). The meter's IN LO offset of 31mV negates this amount at the IN HI and so 50mV divided by 25mV/A =~ 2A(1.9 shown but both clamp meter and ICL7107 meter were slowly alternating between 2.0~1.9).
[attach=1] [attach=2]
Cheers!
-
I realized that despite the size advantage of a tiny DC-DC converter, the particular one I have has more disadvantages:
1. It is NOT regulated.(!)
2. It requires a minimum load of 10% of max load for the output voltage to be within specs. Otherwise, the output voltage hovers around 7V! (This is too high for the ICL7107)
It's possible to setup a minimum load with a couple of 470 Ohm resistors in parallel, but there was still ripple in the output even with a small cap. As any fluctuation in the input supply voltage directly impacts the reference voltage and general stability of the meter, this is the deciding factor.
I'd be better off in this case to just connect a switching 5V wall wart directly to the meter, but of course I'm gonna opt to use one of my boosted chargers(7.3~7.5V) + an LM7805 regulator as the supply for the meter as this keeps the input voltage for the meter tight and reduces the wasted power and heat from the regulator.*
* I noticed that going from a 12V supply to the LM7805 causes much heat dissipation(the attached heatsink got toasty hot), and the required minimum load at the DC-DC converter end doesn't help. The issue would be even worse if the output of the server PSU is boosted to 14V which is intended later.
Cheers~
-
Add a 5V low drop-out regulator such as the output of the DC-DC converter.
https://diotec.com/request/datasheet/di78l00zab.pdf
-
Add a 5V low drop-out regulator such as the output of the DC-DC converter.
https://diotec.com/request/datasheet/di78l00zab.pdf
I do have a similar part - a 2950A CZ-5.0 (5V regulator, brand?). According to ONSemi's datasheet, it requires only a 1uF cap(which I have on hand) on its output for stability, and its "typical" current limit is 220mA(300mA max) which is higher than the 200mA max for an ICL7107 based meter, so I think I'll give it a try and report back.
Thanks again for your support!
Cheers~!
-
[attachimg=1]
[attach=2]
The two 2950A CZ-5.0 regulators I have are providing about 4.989V/5.00xV without the meter connected, and this voltage drops to about 4.7xxV/4.8xxV with the meter connected, but this drop doesn't seem to affect the meter's performance.
I'll have to fiddle with this some more to make sure this method is solid and the meter doesn't fidget even if strange stuff is thrown at it, but thanks for the pointer!
Cheers!
P.S.
So far, I've just briefly tapped the yellow load-side negative wire, to different twist tied points on the load resistors for roughly loading of 2A/3A/6A, and the clamp meter's readings are matched(to the lowest digit) immediately by the ICL7107 meter(very responsive and accurate too). (!)
If there are some other methods of testing this setup for weaknesses, I'm all ears!
-
4.7V is probably within the power supply voltage tolerence of he meter.
How much current does the ICL7107 based meter use?
-
With the meter's power input wires in series with the multimeter, apparently only like 35mA!(?)* A lot less than I'd expected, as the ICL7107 datasheet mentions max load of around 200mA.
Update: After considering all possible combinations of working options, I'm leaning towards powering the "ammeter" with a "boosted" wall wart charger(boosted from 5V to 7.3V) connected to the 2950A CZ-5.0 as the "5V regulator" (instead of an LM7805), as this method provides enough voltage and current for the meter without the efficiency/heat dissipation concerns.