On the issue of sources, interference and import substitution

Documentation is like sex: if it's good, then everything is great, and if it's not very ... it's still better than nothing.

Recently, MyYoungCollega chose a source for the implementation of a network adapter, and in the process of this work, interesting and useful (although not always happy) facts were discovered that I would like to share with readers.



First, let's decide what parameters the manufacturer of the power supply must provide us with in the documentation (and ensure their implementation).

Pnp: in this case, the country of origin of the source and its manufacturer are completely unimportant.



Primary static parameters, without which the use of the source in development does not make sense at all:

1. The operating range of the input voltage (minimum / maximum operating voltage), in the case of a variable input voltage - also the frequency (and its boundaries).
2. Output voltage and guaranteed current.
   
   More static parameters, which should also not be forgotten:
3. Input voltage limits are the maximum allowable voltage (when the source is off, which does not damage it) and the shutdown voltage (if any). It is also necessary to consider compliance with the power supply standard in terms of permissible short-term surges and dips, as well as the presence of protection against polarity reversal (of course, except for the variable input).
4. Output load capacity including temperature derating, minimum load requirements.

Dynamic parameters:

1. (-), , , / .
2. , , .
   
   
   
   : , .

Let us consider the situation with interference in more detail - modern, high-efficiency (and therefore pulsed) sources contain in the spectrum of produced interference a fundamental component that coincides with the conversion frequency and its odd harmonics, which quite themselves get into the radio range, starting from 1 MHz. Fortunately, this noise is differential in nature and is well suppressed by standard filters, despite their not too high frequency.



Unfortunately, in our time, additional components have been added. First of all,

this is switching noise, an inevitable consequence of the high efficiency of modern power supplies, from which follows the need for low switching losses, which leads to very sharp edges on the keys.



In itself, this is not a disaster, but from the desire to reduce the size grows another cause of troubles in terms of interference and it is called printed transformers. First, compared to wound transformers, they have a significantly higher inter-winding capacitance, resulting in much more unpleasant common mode noise. Secondly, the design of such a transformer excludes the possibility of using a grounded shield, which leads to a significant increase in interference propagating over the air. Summing up, it can be argued that the problem of interference for modern small-sized highly efficient power supplies is, as they say, in full growth.



3. Additional functionality: protection against overvoltage of the output current and its nature, the presence of protection against over / undervoltage at the input, protection against overvoltage at the output, thermal protection, additional functions (external feedback, remote control, status indication, the presence of a management and control interface, etc.) are rarely decisive when choosing a specific source, but will often be a pleasant bonus, confirming the correctness of your choice.



My regular readers are probably already worried - several paragraphs have already been read, but "Yaroslavna's cry" has not begun, has it happened to the author (after all, the coronavirus is rampant). I want to reassure you, it is about to start, bear with me a little, I have not changed my (perhaps not the best) habit of writing about what I did not like, I start to scold the documentation of specific manufacturers.



Let us now consider some specific manufacturers and see how the documentation provided by them for (in general, not bad) IP meets the high requirements of the author. We will look for a source for an input alternating voltage of 220V 50Hz (the author knows about the new standard of mains power, but allows himself to give the well-established name) after the rectifier (ignoring the requirement of European authorities) for an output power of 200 W, an output voltage of 24V. As objects of comparison, we select a domestic manufacturer and a world-renowned company.



To begin with, consider the products of the Murata company (I can already imagine how the workers of this company shuddered after reading the beginning of the paragraph and knowing my ruthless nature). Well, completely in vain, I'm going to build my story in a direction from good to terrible (which prompted me to post) and finish it wonderful. So the source of production of this company (I will not indicate the specific type, it does not matter) is chosen as an example of good development, the source parameters are quite acceptable, the documentation is decent, all the necessary parameters are indicated explicitly, including data on the guaranteed (in the sense of not exceeding ) noise level as a short line relative to standard compliance. The only drawback of this product (except for the input voltage, which is not in the range) is its manufacturer,which makes the chances of finding a product on the MOS list very elusive.



Pnp: many readers understood in the interests of which department the enterprise where the author of the post works produces products, all those who did not guess can only envy.



Nearby there is a source from the equally well-known company TRACO and almost everything can be said about him as mentioned above. I took them just to show that decent documentation exists in principle, because it will begin now ...



Now consider the documentation for the source of the Vicor DCM series.



The products of this company have always stood somewhat apart due to the fact that the operating frequencies of the converters exceeded (sometimes multiple) that of other manufacturers. This circumstance led to a significant increase in the specific capacity of the IP firm and caused a certain exclusivity of the products.



PNP: for example, the source in question has a specific power per unit volume almost 3 times higher than the previously considered analogue, see the table at the end of the post



If you really need an ultra-compact power source and do not mind having to pay 2 times more for it, then Your choice is certain, if not, you have options.



The downside to compactness is the increased heat generation per unit surface area suitable for heat removal. To neutralize the influence of the latter factor, the sources of the series under consideration provide for the removal of heat both from the upper part of the source body (through the radiator) and from the lower part (through the board), for which special thermo-pads are provided on the lower part of the case, which cannot but please the designer ...



Pnp: It should be noted that in the latest developments of the company, the conversion frequency is not only not increasing, but even decreasing. Personally, I believe that this was due to the use of silicon carbide devices, but I may be wrong, however, the fact is.



But back to our ramsLet's finally look at the documentation. We are looking for the indicator of interest to us - the level of interference, it is likely that it will be significant. And we see that a reputable company simply does not provide ANY data about this parameter of the released source. It does not provide absolutely anything, from a word at all, not even that of pictures of the spectral density of interference, but there is not a single phrase about compliance with any applicable standards.



PNP: when I talk about a drawing, I mean exactly a drawing, because when the measurement conditions are not indicated on the art object placed in the documentation and (most importantly) there are no lines that are guaranteed to limit the measured parameters from above, but instead there is the phrase "Typical values", then this is a picture, or a drawing, or an illustration, but in no way a part of the technical documentation for a product, which determines its guaranteed parameters.



Instead of this data, the documentation (in the application manual, not for a specific device, but for a series) has a whole section called "Common-Mode Noise Filter Design Steps". In it, we are asked to perform the following steps:

1. «Determine the AC ripple magnitude over the frequency range of interest using the EMI spectrum analyzer and compare the measured AC ripple magnitude with the EMI standard limit line», , « »,
2. ( ) , ,
3. .

Guys, are you serious (and I didn't know it was possible)?



Pnp: from this moment, and until a specially designated moment, the term "you" I will apply not to my readers, but to the employee of the company, although it is possible (and unlikely) that these two categories may overlap. Well, at the same time to those readers who will express themselves in the comments in the style of "Why is this, everyone does that ...".



Well, firstly, the very idea of ​​measuring interference on a specific IP instance (and making some far-reaching conclusions based on these measurements) seems to me as creative as it contradicts the principles of the classical engineering approach. What should I do if a slightly different (higher) interference rate is observed on another IP instance? Can I project the data obtained under specific conditions for development with a different nature of the load? You can think of many more questions of a similar kind that no one will have an answer to (I definitely do not have them, but I probably do not know everything).



And so, by the way, the equipment and the necessary instruments for carrying out such measurements (if we are talking about a measuring device, and not about an indicator), does not cost at all at every workplace and not even in every laboratory, since it does not cost a little money ...



Secondly, your calculation method (implemented in the program) strongly resembles the calculation of a spherical horse in vacuum, since it does not take into account the presence of an input capacitance at the MT, as such, and implies (for calculating the filter Q-factor) zero losses both in the MT and in filter elements, which is somewhat contrary to my practical experience.



Thirdly, why are you so sure that the noise in your source is of an exclusively differential nature, since the proposed filter circuit is a little less than completely useless in the fight against common mode noise.



Necessary note: I have not yet managed to publish this post (usually they lie in my unfinished folder for a long time), and the company has already taken measures - in the new version of the specified document on page 37 in the same section in paragraph 1, we are now asked to independently determine the in-phase and the differential component of the interference (again with the help of direct measurements of a specific instance), then - as before (only the link to the program was removed, well, thank God).



PNP: A note in the light of the above, is it not time to separately indicate both components of the interference in the MT documentation (at least for a spherical vacuum, since one of them depends strongly on a specific device) in order to more efficiently design a filter if necessary.



A logical question arises - how could such a documentation be released by a company that has been on the market for several years. The only answer that comes to my mind is that in the products under consideration, in terms of interference, everything is so bad that marketers advised not to place data in the documentation, but to use a completely "ingenious" move with shifting the work of the development department onto the shoulders of the IP consumer.



My colleague expressed the opposite opinion - in this series of sources everything is so good in terms of EMC that the employees of the company considered it completely redundant to include the relevant data in the documentation. The point of view is not uninteresting, since the level of interference in a finished product can (and will) change depending on the design features of the latter and, if the source's own interference is really small, then it is difficult to give any guarantees without taking into account the specifics of use. Since I do not have a specific instance of the source, it is not possible to confirm one of the hypotheses with the measurement results. However, the experience I already have with the sources of this company does not at all testify in favor of this hypothesis.



Well, the cherry on the cake (there must be a cherry on the cake, whoa) for some reason, the link to the instructions for use is not indicated in the reference data for the device, and we must guess about its importance. I must declare that the document itself is made quite soundly, there are recommended switching circuits and they have elements for reducing the level of interference, and of both types. A lot of useful information is also given, such as: a device state diagram (which is rare), time diagrams of transient processes and many other interesting and practically valuable recommendations, so if the need to familiarize yourself with this document were directly and unambiguously spelled out in the main documentation, then I would only be glad. Moreover, the indication should not be in the last paragraph (a rare bird will fly to the middle of the Dnieperreads the date to the end) and not in a note in small print (an even rarer bird reads notes), but in the very first lines of the specification and in bold print (although there are birds that will not be helped by this, but medicine is powerless here).



P np: for some reason, in the documentation for an older series of sources of the same company (Mini), this was exactly what was done and the link to the application manual was on the first page of the documentation, which means that the company was able to do this before. Even now, the manual is not hidden from us, a reference to it is given on the page of the IP series, but this is still worse than in the documentation.



To summarize: in the documentation for this particular series of power supplies, there is no data on interference (both conducted and radiated), and the only recommendation that I can give to developers is do not use these sources at all, since you cannot guarantee anything in terms of EMC if such the requirements are laid down in your TK.

Pnp: if they are not set, then do not take up such a development, since they will still pop up at the very last moment in the most unpleasant (hard) form and it will be the developers who are to blame.



Although I must admit that the primary parameters of the source look very attractive.



Well, or the second option - use the classic noise suppression schemes (they are given right in the application manual), counting on the most unpleasant case (since you (here I am again addressing the readers) do not know anything about the source parameters, then I even I don’t know what would be the worst case) put a multi-link filter, well then ask yourself - I don’t need such a small-sized source, if I still have to add filters to twice its volume, wouldn’t it be better to take a source from another company, which in its documentation gives quite specific guarantees.



And now, in accordance with the previously outlined plan, I present to you an ideal (in terms of documentation - I will write about this clause later) series of MDM-P power supplies from AE-DON. So that I would not be suspected of bias, see my old posts, where I used to scold this company a little. Why slightly - because I understand very well in what conditions they develop and produce their products, but in this case there is no reason for condescension and discontent.



The point is, first of all, that on the website of this company you can find (apart from leaflets and tables) technical specifications (TU) for this series of sources. Moreover, it is not an extract from the TU posted, as it has become fashionable recently (and as it was before with this company) and you do not need to issue a request for access to materials, and you do not need to buy the TU (in principle, money is ridiculous, but how long you have to wait until it gets to you), but you just go to the site and download, just some kind of fairy tale, it turns out that it was possible.



Why this approach is important and valuable - again, we are talking about interference. First of all, the TU clearly indicates the compliance of this parameter with the requirements of GOST B 25803 in various inclusion methods. Further, since the TU is a direct document, the corresponding sections contain the measurement methodology and connection diagrams, which allows you (if necessary) to reproduce them and make sure that a specific copy of the source corresponds to the declared parameters and that your problems with interference during product testing are not caused source, as such (stand-alone).



One can only support with both hands this practice of providing documentation and wish it the widest possible distribution to other products of domestic manufacturers.



Unfortunately, the (perfectly defined) source parameters themselves are not quite ideal. In particular, the power density is 4 times inferior to the first and second of the considered sources, and the second and generally 11 times. However, if it is true what the developers of the company write and they really implemented the latest series on domestic components, then this is a significant success for the company. Yes, to get a device on domestic (no fools) components that is only 3 times inferior (in the sense of more) in terms of volume to an imported analog is an achievement, such is the harsh reality in which we live.



Well, as for the topic of the post, it's also not easy. The noise level of sources with a power of up to 160 W without a filter is indicated (the phrase is somewhat sly if you look at the measurement scheme) - in accordance with curve 2 of GOST. Correspondence to curve 1 is indicated in the case of using a source together with a filter manufactured by the same company at a power of up to 240 W. The correspondence to curve 3 is indicated for sources with a power of more than 240 W without a filter (again, we look at the measurement scheme and make sure that the filter is still there, it is simply recruited from loose material). This is where the information ends, I admit that the company does not produce filters for high throughput power, but it would be possible to bring the filter circuit necessary to reduce the interference of high power sources to the level of curve 2 and even to 1.



Yes, of course, such data are not the subject of technical specifications in its classical sense, but they could be placed in the reference information section. But this is my evaluative opinion, expressed in private, in any case I express my gratitude to the developers from AE-Don for what has already been done and you can go to the application of this series with your eyes open, perfectly understanding what exactly awaits you. By the way, this company also has the MDM-R series, which is almost not inferior in power density to eminent rivals (well, of course, except for VICOR, no one competes with them in this parameter), it is a pity that the input does not suit our task.



Parameter comparison table