How to improve the output noise of a laboratory supply using LTspice

hello and welcome back now in this video I want to look at the problem that I've had for quite some time and you might also have had this problem or still have it it is related to laboratory power supplies and the ripple they generate now the supply works quite well I mean you don't notice any problem by using it on its own but if I connect it to an austere scope that's something interesting happens so this is the noise that my otoscope generates on its own so right now my probe is connected to the ground lead and I'm reading about 8 millivolts of ripple so this is quite a key for this type of G POSCO now watch what happens when I connect this thing my power supply so I'm only connecting this to the ground input so not the signal input and I'm getting more than 100 volts of ripple which is quite frankly unacceptable especially if you're trying to make low noise measurements I mean you can't even analyze the power supply or anything one that you're working on because of holder noise that the inertial power supply is generating and the odd thing about this noise is that it's not necessarily the output noise of the laboratory supply because to measure that you would need to be probing both of its inputs it's noise coming through the ground system somehow and I sometimes get this problem sometimes I don't now in the past I've used this massive filter to filter out the noise but in the meantime a box fell over it and it kind of broke into pieces so I'm not going to be using this thing anymore and this is not the right way to fix the problem the sad thing is you only need one component with a single component you can get rid of most of this ripple you just need to know where to put it so if you're curious where this trip is coming from how you can simulate it using your tea spice and of course how it you fix it that keep watching so let's start things off by talking about the power grid and how you get electricity from the grid into your device basically the typical electrical connector has free wires the life the neutral and the protected Earth which can also be found under different names like earth ground or something else now when you're getting energy from your grid into your unit you're using either all three of these wires or just do so what I got here is an example connector from my country in which I only have two wires the live and in neutral and then we have this sort of connector in which you have the live the neutral and also this protected ground so the free wires go into your unit and the actual power comes through the two wires or live in the neutral then inside you probably have one isolation transformer and from this you're powering your device you rectify it and so on and this is where you get your local ground inside your unit now the protected Earth is there mostly as a safety measure so this might be connected to the housing if the housing is made from metal like is the case in washing machines or the power supply let's just check this with the multimeter set to deep continuity function so we take the plug that the power supply is connected to and if we check D protected Earth it's not connected to the ground so the output ground of the supply but it is connected to the casing so through this screw we have a direct connection to the protected Earth but usually it's not connected to the local circuit ground but there are some exceptions from this rule of course now if you have two units powered from the same grid and these two units interact like let's say your computer is sending audio information to your amplifier then usually the ground is interconnected between these two and that's not a problem since this local ground is completely isolated from the electrical grid now as I said in certain cases this protected earth is not really completely isolated one device that is built in this manner is the Oslo scope so normally every Oslo scope will have its local signal ground reference to the earth ground so if we check the continuity with the protected area on the Oslo scope while it's not connected to the casing since that's plastic but it is connected to the ground of the probes so all of the outputs of the Oslo scope have the ground directly connected to D protected Earth in the power plug since all the measurements you want to make should have both very stable reference which is the earth ground now when you connect your also scope to any other circuit like the power supply you're connecting this completely floating round to the earth ground now this normally should not be a bad thing I mean this protected earth is there to protect you so it's supposed to do good things what is it now to get a better understanding of the requirements and the specifics of designing circuitry intended for being supplied from the electrical grid I highly recommend this document from our integrations basically they go over some power supply the sunny techniques so this document is specifically written about power supplies and how you design them for electrical noise emissions and safety so it covers the various emission standards that such devices needs to comply with how you're measuring these what are the various safety principles and requirements behind it this type of equipment the various ways in which you can connect the AC mains supply so two wires or three wires and then how you're going to build your EMI filter and then they go over the various tip my sources in flyback supplies so both differential noise and common mode noise now in our case we don't really want to design an input filter or anything we just wish to first of all understand how such a supply is built now this document is great not just because it covers switching supplies but because it covers flyback type of switching supplies the exact type of supply that my laboratory power supply is so here they are giving an example in which two wires are used to supply the circuitry but we're more interested in something like this so this is not the supply that I have but it's very similar so basically we're supplying the units through the live and the neutral wires we've got our input filter rectifying bridge the transformer and then on the secondary side a rectification diode and then an output capacitor and then the circuit can supply something else now I previously mentioned common mode noise and differential noise now in short differential noise usually goes between the two input or the two output wires so for example we have differential noise between the live and neutral wires so it's no it's that goes through the live wire and then back through the neutral or if we look at the output of the supply it goes out the plus wire and comes back in Fruity ground wire and common mode noise is something a bit more difficult to explain basically it's a kind of noise that comes in through both supply wires and goes out through both supply wires and then close it around in a more complicated manner now inside the flyback supply we have an interesting source of common mode noise because of the connection between the primary and the secondary so the two sides are galvanic Li isolated there's no direct connection between them and this is done for safety reasons but the noise in the primary side couples to the secondary through stray capacitance inside the transformer and then this noise is supposed to close back so the no it should be kept inside the power supply shouldn't go outside through capacitors connected from the ground back into the ground of the primary side so in this case c8 s c7 now in between these two capacitors is also the protected earth and we'll get back to this so before starting to simulate the circuit let's do one more thing and that is try to measure the noise that our supply is generating so we can start by looking at the common mode noise so common mode noise can be defined as the noise that comes out of the supply and closes back through the earth ground so we can measure the noise and we know that the otoscope is connected to the earth ground directly so I can simply connect to the output of the supply and we can see that there's quite a bit of noise here so first of all we can see this 50 Hertz oscillation so basically this is the mains voltage but we always see a 5 volt peak-to-peak remaining from it and if we zoom into this thing we can see something that's quite characteristic to fly back supplies so this thing is quite high voltage peak to peak we can see the switching and then thus oscillation going on so this will be a thing that will be interesting to try to simulate now another thing that's interesting to notice is that we're getting the exact same thing if we measure the plus output so this common mode noise is coming out of both of the outputs of the supply and we're getting the exact same thing from either now just like before if we connect the ground of the probe to the ground of the supply and we short-circuit the probe we're gonna get this thing so this is an oscillation going at around 400 kilohertz and it's modulated at around 70 kilohertz so most probably 70 kilo hertz is the switching frequency of the power supply and then this 400 kilohertz is some sort of ringing or oscillation going on through the system now an interesting thing to notice about this so right now we're getting around 90 100 millivolts of peak-to-peak noise now let me show you something basically I've unplugged the supply and if I take a bit of electrical tape and isolate the earth connection so don't try this at home this is just for experimentation purposes so this is highly not recommended so right now I a stated B earth connections but the supply it can still be supplied and now if I plug it back in so the supply starts working again but the noise has disappeared so the connection is still there what this is telling us is that the noise the 400 kilohertz noise has something to do with the earth ground because by removing the earth ground we're left with only around 15 16 millivolts of noise so quite decent and if I try to measure now the differential noise so the noise coming between the outputs of the supply we get the same 26 something millivolts so the supply is not that bad I mean it's differential noises quite small it's just that the common mode noise is large and now we can start simulating using ltspice and try to find a way to reduce the noise that the supply is generating without isolating the earth ground since this is not a solution let's see what we can do about this now I won't be similar in the entire circuit because it will take 4 simulate and because I don't have the actual static in the supply I mean I know it's supply but I don't perhaps the exact schematic so to begin with what I got here is a simulation involving the electrical grid so I got my power supply then the reporters of time itself is represented by having the input filter some rectification diodes capacitor and a load and then the secondary side again has a voltage source output capacitor and then the capacitors linking both of the primary and the secondary side to the protected earth on the otoscope side I have my protected earth connection and then the basic representation of the resistance is found in G Oster scope so my input divided by ten resistor network and then one ohm resistor to represent the wiring going through the ground system so let's start off by simulating connecting D also scope to D negative output from the supply and if we do this we get nano volts of voltage so we can't really read anything it's nowhere near the extra measurement that we've done so something must be missing and the missing element is the stray capacitance inside of the transformer so because of the way the transformer is being built there is a bit of capacitance between the primary side and the secondary side now if we include this so basically putting in a hundred Pico farad capacitor the actual value might be higher or lower depending on how the transformer is built and now it will assimilate this we see something completely different this time we have the exact same shape that we have in real life so right now the simulation looks exactly like the measurement basically this is the 50 Hertz motif from the main supply pass through the diode Network now the voltage is slightly smaller than the one in real life but this has to do with the capacitors and components that I didn't really place inside the supply but the shape is correct now the next thing we can look at is what happens when we disconnect deprotected er so we've seen that this helps but it's not something recommended and if we simulate we can see one of the reasons why using the same setup I'm no longer getting the one fault or two volts or 5 volts of ripple I'm getting up 300 volts of ripple basically the mains power supply is directly coupled through this three capacitance into the Oslo scope and in the simulation I'm getting roughly three hundred ten volts and in reality it's around 360 volts so my mains voltage might be a bit larger than this 310 you can never trust the electrical company so now moving on reconnecting things let's see what happens if we actually add the ground connection of the Oslo scope if we look at the output I mean we're seeing a bit of ripple but it's extremely small so it's in the millivolt range it has nothing to do with the actual measurement so the thing that we're missing at the moment is the actual fly off supply so until now I just simulated the power supply as the load on the input side and as a generator on the output side I didn't really simulate the fly back supply itself and for that I will make a completely different simulation because of the way it slows down the simulation if I simulate everything starting from the electrical grid so the important thing that we learn from this part is that we have this circuit Basten's and this is something that we need to include in our next simulation so what I've got here is basically the same thing only this time I'm supplying directly from a 300 volt DC supply I got a generic flyback circuit with stray capacitance at it between the primary and secondary side I also have the capacitors going from the supply lines both on the input and on the output to the detector and this time if we simulate and look at what we're getting on the output we're getting a waveform that is very similar to the one that we actually measured when we zoomed in to our first measurements so basically this is what I measured on a connected only the tip of the ausco scope without the ground lis now to make the two a bit more similar let me just invert this and now the two waveforms are almost identical so if we compare this to when the switch is switching we can see that during the moment the switch is turned on we have the spike then the coil discharges and then it oscillates until the next switching cycle so exactly what we're seeing in real life now with this circuit if I now connect the ground lead of the otoscope to the output and look at the output voltage I am starting to get a bit of noise much higher ripple but it still doesn't look like the actual measurement now there's one thing missing here and that is that in the simulation the detector is connected to the oscilloscope through an ideal piece of wire now in reality it's not an ideal wire it's about 2 or 3 meters of electrical cable that has some resistance which let's say we can neglect but it does have some inductance which we cannot neglect so if I simply add let's say about 5 or 6 microhenry of inductance the waveform changes completely and this time the waveform does look like the measured thing so basically what we're seeing here is energy coming from the switching supply couple 3d transformer for the stray capacitance and then causing a ringing through the particular supply network because of its inductance so the voltages are not exactly the same but the phenomenon is being reproduced correctly now another interesting thing you might noticed right now is that I'm measuring 200 something millivolts and this voltage is coming across this internal one on resistance of the hasta scope and that means that there's a hundred something milliamps of current going through there so what we're seeing in the hasta scope when we connect the ground is that current is coming from the supply discharging into the protected Earth and this is not something that should happen so this noise this energy is coming from the switching node so from the switching action of the supply and this is coupling through the stray capacitance and normally what would happen is that this energy would close back the capacitors inside the supply so he has quite high spikes of current going through this output capacitor but not all of the current goes through there some of it is coming through the protected earth and causing us problems so this is basically giving us an indication of what's going on and how we can fix it so our problem though root of the problem is that that supply is noisy and that the initial capacitors which were placed in there to keep the noise inside the supply are not doing a very good job basically when I connect my ausco scope through the protected earth the impedance of the connection has a closed value to the impedance of the output capacitor so the capacitor from the output to the protected Earth inside the supply now there's two ways to keep the noise inside that supply regardless either increase the impedance of this external connection this is where disconnecting the protected of helps since this way the noise can no longer be coupled through the external circuit but we're not allowed to disconnect the protected earth since that's not safe we can add a very big inductance on the line so let's say if I add 10 milli Henry then the noise coupling fluid will be neglected also my crams but again this is not something that you can really do especially since the supply is a 5 amp supply so you need to make a 10 milli Henry inductor that can handle 5 amps now of course 10 milli Henry is not really necessary but you would need a high inductance to actually filter this out but a better thing that can be done without actually overcomplicating the supply it's work on this output capacitor so the capacitor that's connecting the output to the protected Earth now at the moment it's at around 20 nanofarads but let's say we increase this to 400 nano farad's by doing this we didn't get rid of the noise but we did reduce it substantially so we went from hundreds of millions to tens of millions so at least a 10 fold reduction in noise can obtained just by increasing this capacitor the only thing to look out for when doing this is that this capacitor needs to be rated for high voltages so at least 400 volts that's why you can't really put any sort of value here so if you would put the micro farad it would be better but 10 microfarads at 400 volts is a huge capacitor now there's one thing to mention here so in case you're planning to modifier supply to add this thing the manufacturer of the supply whoever it is it's not really responsible if you break your supply so please proceed with caution any sort of modification to the supply could end up destroying it so modifying any sort of equipment you need to at your own risk now let's have a look at what's inside this thing so basically this is the circuit it's not very complicated so on the left side you've got your input mains power circuitry input filter input capacitors rectifying bridge and so on so don't mess with this area since this works with high voltages and of course only open the device if it's unplugged so be safe about things then you got your transformer and then your secondary side so here's the rectifying diode for the secondary first filtering capacitor a permanent load an inductor filter another electrolytic capacitor and here on top I'm not sure how well it's visible we've got our first capacitor which is connecting the ground to the earth so this screw is connected to the casing which is connected to the protected Earth and then we got a bit of wiring going to the output so here are the two connector plugs that go outside and here again we have some capacitors which link the two wires to the protected Earth so now the question comes with my massive capacitor where do I put it do I put it right here on the board do I put it down here at the connector plugs is there a difference between the two positions so I think we need to stimulate things in a bit more detail to find out where exactly to put this thing and we can simulate this by splitting up our initial 20 nano farad capacitor into two thousand of our capacitors and adding let's say hundred nano Henry of inductance for the wiring in between so if we add it at the beginning we have around fifty millions of big current and a bit of noise if we add it at the end we get slightly more current so what we're seeing from this is that the closer dub short-circuiting capacitor is to the noise source the better it will act so one way of doing this is either closest to the noise source or speaking of the capacitor into two separate capacitors which have the value so I added the capacitor from the ground to D protected Earth so this is the 470 nano farad capacitor but I also added an extra resistor so this is a 4 mega ohm resistor and the whole point is that you don't want this capacitor to charge up with certain voltage that could discharge and create some damage so it's common practice to also add such a resistor to unload the capacitor but a high value resistors are not hundred ohms or something now of course this capacitor is not sitting very well here so some sort of proper fixation will be need but for now let's just see if this actually helps if the theory and simulation actually looks like what we have have in real life so I'll just have to put things back together now so now let's see if that modification worked I remove the tape from the earth connection so the power supply is properly earth this time now if I take my otoscope probe connected directly to the ground and run the noise almost completely disappeared so we're still getting some common mode noise but this time it's much more smaller than before so we went from deep nearly a hundred millivolts down to twenty thirty something so now the supply is form or usable now if I disconnect the otoscope so just to see where my noise floor is we can still see that connecting to the power supply still adds a bit of problem so the solution isn't perfect but we're getting there I mean it's much better than it used to be before now another thing we can try this point is work on this inductance so if we would have a larger inductance and we would receive late this we would notice that current is going down so we're building a better and better filter the hardest inductances and at the moment this inductor is only formed from the wiring existing between the power supply board and the output connector now you could have inductors in here but they would need to be rated at 5 amps at least and that's not a very cheap inductor the second thing you can try is that the sort of thing so this is a ferrite clamp this can be found in all sorts of equipment simply clamped over supply wires and the way the set it's by increasing the inductance of the wire by adding this ferrite over it now if you don't have this sort of thing you can simply use a ferrite ring so this again can be found in all sorts of circuits usually there's a bit of wire around it but you can simply take the wiring from your supply and just thread it through this thing the more times you thread it the more inductance you're getting and this way you can increase the inductance of the wires and create the better filter so that's the theory let's see how this thing actually works in real life so what I did was extend the supply lines so from the board to the connectors and add this ferrite ring so basically I want the supply lines around is ferrite being a few times to create a basic common mode choke so I added inductance on both wires so both the plus and the minus and by putting them on the same piece of ferrite created the common mode choke so now let's see if this actually helps so now with the final inductor added let's just see the noise that the supply generates so we're still getting a bit of noise but it's a bit lower so we went from around 30 millivolts down to around 20 millivolts so this common mode choke did actually help and we're getting the same thing on both outputs so in the end I managed to reduce the noise of the power supply quite substantially just by having some very simple measures and this is all thanks to properly modeling and understanding where the noise is coming from so just recap we went from this sort of noise so this is on a 20 million volt per division scale binding the capacitor we went down substantially so most of this noise simply disappeared just by the addition of the extra capacitor and then when we added also the common mode choke we went down even more so you might not see a very big difference between quit or without the common mode choke but you can see that a lot of these little spikes so most of the high frequency noise has been reduced so all in all if you have a noisy supply you can try out these measures but of course be careful with what you're doing so don't get electrocuted since the power supply usually works at high voltages and try not to break the supply because any sort of modifications you will do inside of it is at your own risk with that being said hope you got some useful information after this leave your thoughts in the comments thank you for watching make sure to subscribe to be up-to-date with all my latest videos and see you next time bye bye [Music]

Channel: FesZ Electronics

Views: 12,402

Keywords: fesz, electronics, diy, hobby, electronics tutorial, tutorial, smps, switch mode power supply, power supply, laboratory supply, ltspice, simulation, noise, ripple, common mode, diferential mode, ax-3005ds, 3005ds, electromagnetic compatibility

Id: VkdtESI6C74

Length: 30min 39sec (1839 seconds)

Published: Wed Dec 11 2019