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

Thanks, will do!

Thank you for the kind words! Will do!

Thanks, will do!

I'm happy you enjoyed it! I tried to give a bit of background information to explain why the "fix" actually helps.

Yep. I hired an actor. Not a very good one...

I just checked and in my country its 230V according to en.wikipedia.org/wiki/Mains_electricity_by_country I guess 220 is an older value (at least for my country)

Hello Piet! You are right, for measuring low noise, you can use a spring or something, and do away with the long ground lead. I usually make these from pen springs, since that way they are elastic and hold better on the probe end. Measuring low noise is quite an interesting and difficult topic.

@@FesZElectronics I usually just wrap a thin wire around it. Sometimes even solder it. And yes it a very interesting and difficult topic. I once even noticed a pretty significant change by just connecting to a different ground point on the PCB. Direct soldering it to the board can also make a huge difference. Crazy

Hello Piet! Well I guess if the ground on the PCB is not very uniform, then you can get more or less noisy spots. And of course of you are working with high currents such effects will be even more visible. This will be especially noticeable on 1 or 2 layer boards. I'm not sure how safe soldering the wires to the oscilloscope probe is; I mean it might damage the probe. On the other hand, if you have a probe with detachable tip or some sort of proper adapter, and solder to that, then it should be no issue. Are you using a special kind of probe for such measurements?

I'm happy you liked it!

Thank you for the kind words! There are a lot of books out there, what exactly would you be interested in? Electronics is a very broad subject, is there any particular part that you would like to find out more about?

@@FesZElectronics you re right ! I would like to know what are your recommandations about power electronics and about magnetic design :) thank you :)

Well I guess as a general recommendation I suggest things written by the people that make the IC's and materials - especially application notes made by the major manufacturers (for power supplies Texas Instruments, On Semiconductor, Maxim etc.) (for magnetic materials TDK, Epcos, Kemet, Vishay etc.). Nobody knows more about the object than the person (company) that manufactures it. Its in the best interest of the manufacturer to inform you how to use their product so you buy it and use it properly; even if the product is good but its miss-used, the buyer might think its the manufacturers fault. If you are interested in books then again the best material is the one made by the people who work on this topic - the engineers from the major companies (not random internet people or university teachers) - even though they all know what they are saying, the engineer will tell you the practical side and practical problems while the university will tell you the theoretical side and the ideal conditions. As authors, I have 2 that spring to mind - Cristophe Basso (from OnSemi) and Ray Ridley (from Ridley Engineering), but there are of course may others out there.

Ok :) thank you very much ! :) thank you for this detailed reply ! :)

I thought putting a common mode choke in the psu output was in fact the last thing he did.

@@Graham_Wideman Yes, he put a common mode choke in the PSU output which is great, but there will be pickup in the large inductive loop formed between the power supply and the scope based on the fact that they are both tied to the ground wire in the house wiring. This length of this loop could be as short as the combined length of the two power cords or it could include a lot of additional wiring in the walls. Of course it's the loop area, not the length which is important, but it's easy to keep the length short and run the wires together to minimize this area...

@@ats89117 Sure, the large inductive loop is an issue. But your original comment "probably use of a large CMC on the output of the supply would help" seemed to imply that he had not put in a CMC on the PSU output, which he had. Or were you saying the one he used isn't large enough?

@@Graham_Wideman Looks like I was confused, but the combination of the grounded power supply and the grounded scope is likely to be causing a lot of the problem. It's similar to the issues you frequently have with audio distribution and the solution is probably the same, i.e. transformer isolate the scope input when measuring the power supply noise. The CMC will be effective at higher frequencies, but probably not so much in the < 20 MHz band he was probably looking at.

Normally, that should not happen. The noise measures that I added work only at very high frequency, much higher than the supply feedback loop should be able to handle. Anyway, it is a good practice though to verify the power supply at the end, since any modifications added after the original design was validated can have one effect or another.

Hello! Yes, that would also work; I considered adding an inductance also on the PE pin of the power supply (that would keep the noise inside the supply better). I saw this approach in some schematics, but its quite rare. I did not go on this route because it might interfere with the way the supply is filtered; I mean the conducted EMI is supposed to be certified when releasing a mains supplied product. By working on the supply output I did not affect the mains connection on either of my devices.

Great video, thanks! The point is that the external ground shouldn't affect the supply noise, but in a low cost supply it does. Most quality lab supplies provide a separate Chassis Ground terminal. You can add a cap or a short wire external to the supply to reduce Common mode noise. The best lab supplies also use shielded transformers with one or more copper shields between the primary and secondary. This works for both linear and switching supplies. Unless you are expert at transformer design and fabrication, this is one thing that separates DIY and low-cost lab supplies from the good stuff.

It would be interesting to see the effect. Not sure how big the effect might be since the current shunted to ground is high frequency.

Hello SaihoS1! Thank you for these observations! Lets take things one by one; The first observation is completely true - I forgot to add the K statement, the 2 inductors L3/4 are not coupled but should be. This was a mistake on my side. I did not really have this in focus, since this is not the main topic addressed. Regarding the second topic, I will try to take the various points one by one. The 470nF is between output GND and PE line. I did not touch the high voltage primary part - the original X, Y caps and CMC are not altered. Between the output GND and high voltage region you still have the transformer isolation. In the absence of a PE connection in the socket, my output is connected to the live wires, but trough my 470nF cap in series with the built in 2.2nF Y caps; so you should only get the unpleasant tingling (not what would happen trough 470nF only). I just took the supply apart to re-check - the original capacitor between output GND and PE are 2x10nF and from output VCC to PE another 10nF all 1KV rated, but not X or Y (no special marking present on them other than the value and voltage rating). The proper Y caps other than being blue have a set of certification markings. The X2 cap is 2500V peak rated and 275AC ( ~380V DC) continuous rated. Not as much as the original 1KV of the ceramics, but worst case, the capacitor gets shorted and the output GND is directly connected to PE line - this should not be a safety problem. In case the circuit connected to the output is at a high voltage AC in reference to PE and the PE is not grounded because of faulty wiring the impedance was increased only ~17 times (470+30/30 nf). Not a nice thing of course, but its less bad as connecting the oscilloscope GND to the output circuit (the osc gnd is directly connected to PE). I'm not sure if you agree to this way of thinking, I may still be wrong of course. Let me know what you think.

​@@FesZElectronics So, in addition to K, the upper terminal of the capacitor C2Y should be connected to another terminal L4. And now let's look at the main issue. You have provided a link to AN-15. See following IEC950 Definitions (page 5, CAPS LOCK provided by me): (1.2.8.2): Secondary circuit: A circuit which has no direct connection to primary power (except through PROPERLY SELECTED Y-capacitors) and derives its power from a transformer. From this it can be seen that only Y-type capacitor is needed between the primary and secondary part of the circuit. Is PE the primary part of the circuit? I have always considered what it is. Firstly, PE and N are in electrical contact, and N undoubtedly refers to the primary circuit. "The neutral wire will be treated as an ungrounded AC mains or separately phased line conductor requiring the same safety considerations as any AC mains line conductor." (See page 7) Secondary, note the definition of SELV: (1.2.8.5): Safety extra-low voltage (SELV) circuit: A secondary circuit which is so designed and protected that UNDER NORMAL AND SINGLE FAULT CONDITIONS, the voltage between any two accessible parts, or between one accessible part and the equipment protective earthing terminal for class I equipment, does not exceed a safe value. It assumes that the SINGLE FAULT CONDITION will not create a hazard. However, I have already drawn your attention to the case when the voltage on the PE can become dangerous. This is the case when the PE wire loses contact with the real ground (SINGLE FAULT CONDITION - breaking PE wire). In this case, two input Y-type capacitors form a capacitive voltage divider and the voltage across PE becomes equal to half the voltage between L and N. It would seem that there is nothing to worry about since these capacitors have a small capacity, but consider the case when the PE wire is broken not in your device, but for example at the entrance to a large building. And this is a very real situation when the conductor simply rotted in the ground. In such a house, many devices are connected and each has a pair of Y-type capacitors at the input. It turns out that they are all connected in parallel and make up a large capacity. If in your power supply the capacitor between PE and SELV was 1-2.2 nF, then this would not be a problem. But if there is 470 nF, then it can pass a significant current. I do not presume to give a simpler example than a lightning strike, when the 'Peak Impulse Voltage applied before endurance' 5kV for Y2 or 8kV for Y1 can be exceeded. However, such requirements are laid down in the standard. I believe that your power supply did not comply with EU safety standards even before the alteration, since instead of a Y1-type capacitor, a conventional 10nF 1kV capacitor was installed. The Chinese simply saved money. However, their capacitor looks very similar to the correct Y capacitor. Do you think this is a coincidence? Of course, I am not an expert on electrical safety certification either. This is my personal understanding. And this is a subject for discussion.

Hello SaihoS1! These are very good points you are making, but rather than discuss the exact legal definitions, I would like to show you some examples: Here is a schematic I found - this is the service manual to an Agilent supply from April 2000 ( sites.fas.harvard.edu/~phys191r/Bench_Notes/A1/agilent_e3610a.pdf ) Page 16 has the schematic, which is quite nice, and page 15 has some of the component values - I would like to point your attention to C4 and C5 connected between output and the PE connection. From a value point of view 100nF the total capacity from output to PE is 200nF so its close to my implementation. From a rating point of view, the manual does only specify "ceramic 500V". These capacitors may be Y class rated though - no proper order code is given. But the point is that with a fairly recent design (from a respected manufacturer), you still have large capacitors between output and PE. I guess there are other schematics with even bigger values, but I just could not find them. Also in the AN-15 document, in figure 40 and 44 you have a capacitor between PE and the secondary (C8) which regardless of value (100pF in Fig 40 or 100nF in fig 44) is not marked X or Y, but all other safety relevant components in the schematic are. Regarding both the lightning strike and building PE line not connected, the worst case is that the extra 470nF is shorted, and the GND is connected to PE or you have leakage current coming from PE. Now for any electronic equipment the has a metal casing, the casing has to be connected to PE (like PC, washing machine, guess even the fridge, and so is the case of the supply - metal outer shell directly connected to PE) - I don't know the exact norm, but you can easily check with a multimeter. So if the PE becomes dangerous, you have some really large surface areas that cause real problems when you grab it - that is far more dangerous than a thin wire coming from a circuit.

I guess the biggest safety issue regarding large capacitors between GND and PE is that if there is a voltage difference, the capacitor will charge up, and then its possible to discharge it trough the human body which would cause an unpleasant experience. For that I did place a large resistor (~4Meg) over the capacitor, just so it does not charge easily, and it discharges once the device gets disconnected.

@@FesZElectronics I am not lost. There was just a lot of work. Please wait until the weekend.

You are right, I didn't clearly show that measurement - I took it off camera. But the setup was like at ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-VkdtESI6C74.html ; I isolated the EG but unlike in the video, I only connected the probe tip, not the probe ground.

You are right - but this depends on the ferrite material. You can chose a material which behaves resistively at the frequency of interest - a lossy inductor which has imaginary inductance; but even this material, at low frequency will have real inductance. "Low" and "high" frquency are of course material dependent.

@@FesZElectronics Indeed. A combination.

Hello! I usually provide simulation or other related video files as patron only content.

oh and that's a film cap right? any particular voltage rating or cap type you'd reccomend?

Well the capacitor (in parallel with an extra 1Meg resistor) is between the ground of the supply and the PE (that is directly connected to the heat sink and that screw). The capacitor is a high voltage 275VAC film type. I guess its an old X capacitor that I recycled.

I would ask if all the supplies where the same type? Usually linear converters are far less noisy than switching converters - for example in my supply that I was testing, most of the noise was occurring because it was a switching converter.. Anyway congratulations on making the least noisy supply! I guess that benchtop supplies are some of the few things that can actually be made better than the commercially available ones.

@@FesZElectronics Yes the same type - mine and others just normal linear power supplies (1 of 3 was defectiv with faulty or damaged potentiometer and probably capasitor but another one was ok). Thank you :) I wonder also visibly my power supply looks better on the oscilloscope but what I have noticed that my PSU has Vpp like 600 mV but I don't see it (I mean I see almost straight line with my PSU even when changing time division etc. Anyway it looks and for sure is more stable than commercial ones (maybe because in my psu I don't care about final prise so I put everything I need to make it work..)

Indeed, in commercial products, every corner that can be cut, will be... so if you spend just a little bit more on better quality components you will be surprised with the results you can obtain.

I think that will interfere with the measurement since the proprieties of the cable will change. Also, the problem is not on the oscilloscope side, so there is no point in trying to filter the oscilloscope, its better to filter the actual noisy circuit.

Hello! Well the closest mains earth is in the power supply already, so that could have been used. The reason why I didn't do that is as you said ground loop trough the oscilloscope. Also the capacitor works at higher frequencies only; my problem is the switching frequency at ~70KHz and harmonics even higher; with the capacitor I am only acting on this part of the spectrum. There main difference that occurs by using the capacitor though is that the power supply circuit is still left floating so if you connect the power supply + pin to the oscilloscope ground you do not get a short circuit. This way also if you have 2 power supplies that you want to connect to get differential voltage (connect the + of one supply to the GND of the other) you will not have problems trough the mains earth.

FesZ Electronics Thank you for clarification. Can you make a video just focusing on how to measure CM or DM noise in practice and how to test and verify whether the noise we see is CM or DM? Because if the receiver is diff ended the noise is due to imbalance but in scope or single ended receiver case the noise is due to CM currents passing through DC ground. This is what I believe. It would be great if you can clarify this confusing topic in your way. Just a request thanks for the uploads which both show the theory and practice together.

Hello! I will try to make a video about this at some point, I wrote down the topic. The main difference between CM and DM is the way the current closes, and its all a matter of perspective. For a mains powered supply the DM is between the 2 outputs (so something that goes out of +terminal and into -terminal is DM) and CM is something that goes trough both or one of the outputs (+ and - terminal) and returns trough the mains supply either one or all of those pins (N, L and PE). The main use of knowing if its CM or DM is to know what to do about it. Adding a capacitor between terminal + and - can only help with DM, it has no effect on CM since it does not affect the current path. The same way the capacitor I added in the video has no effect on DM. If you have a 2 channel oscilloscope connected to 2 terminal that is capable of doing a Ch1-Ch2 math function the math output will be the DM noise and the individual Ch1 and Ch2 will be the CM on each of the outputs.

you should obtain the same result - it doesn't matter where you brake the PE loop.

If you need low noise, and especially if the supply is "switch mode" type - then most often you will need filtration. But of course "low noise" is a relative term, there is no universal solution.

@@FesZElectronics excellent point that noise is relative and that switching supplies will always have a level of noise. Thank you.

Well the scope already has an EG connection, I pointed that out during the video. Changing to a linear bench supply would work in this particular case, but the root problem is common to any mains supplied SMPS, or in other words any modern supply. The main point of the exercise was to understand the noise generated by this type of supply and what see what are the ways of simulating and fixing it.

I guess its a combination of 2 factors - every extra component costs - especially this power supply is a low cost one, so price is critical; second is compliance - if the design is compliant with legislation needs, why bother doing something better. If you want to make a very low noise, high performance powers supply, it will not be cheap :D