Class 2 SMPS - Removing the Y capacitor - Good design solution wanted !

[pedrocelli]

I'm new in this blog and doing DIY projects related to vintage computers for many years. Often, I've stumbled over the the electric shock effect, when working with class 2 SMPS together with other earth grounded instruments or computers. It's about the Y capacitator in almost every SMPS connecting secundary and primary side. Especially in class 2 (not earthed) PSU, I can't understand, why the existing design tradeoff resulting in a typical "touch current" and high "touch voltage" is so commonly accepted. When connecting these PSU or its connected devives to other circuits, it may cause problems or even damages. How can it be, that there is no good solution for class 2 PSU ?

I understand from the authors at Electrical Engineering Stack Exchange that it's purpose is to suppress the high frequency common mode signal comming from the flyback transformer. It is necessary to put capacitors between the input and output side of the power supply with a capacitance substantially higher than the capacitance in the flyback transformer. This effectively shorts out the high frequency and prevents it escaping from the device.
(see https://electronics.stackexchange.com/questions/216959/what-does-the-y-capacitor-in-a-smps-do )

Dr. Jochen Jirmann is decribing the high frequency part in very detail in his interesting article I've translated into english.(attached PDF file)
However why do we have to live with this tradeoff ? I would rather like to remove the Y capacitator and add some components to the secundary side for high frequency supression and having a totally isolated secondary output line, no humm no electric shock.
Probaly some PSU designers had some good ideas in the past, I would rellay like to know. Or any good designs found by reverse engineering of well performing class 2 PSU ?

[T3sl4co1l]

The only alternative is a well shielded transformer.  That would take heroic effort to solve, so, you're better off just clipping a ground lead on there and using a three prong cord like everything else.

Why do they put up with 'touch current'?  Because it meets the safe limit, end of story.  Is it annoying?  Is it still potentially unsafe?  Sure.  There's arbitrariness, compromise, in most standards.

If you can find a medical grade power supply to replace it, those use shielding and minimal size Y caps to meet more stringent standards.

Tim

[Ian.M]

The problem is that HF coupling from the chopper transistor end of the primary to the secondary.  Without the Y capacitor between the 0V side of the primary DC bus and output 0V, and without an inter-winding screen, you'd get far more leakage current at the chopper frequency than is permitted at the line frequency as it is typically at least three orders of magnitude greater than the line frequency.

If the PSU has a three wire input including Ground, the output 0V can either be directly grounded or the HF current returned to ground via a capacitor then the Y capacitor can be fitted between the 0V side of the primary DC bus and Ground to complete the return path for the HF inter-winding leakage current.

The only other options are a foil inter-winding screen returned either to Ground (with a Y cap to the 0V side of the primary DC bus) or to the 0V side of the primary DC bus, or a split bobbin design with much greater primary-secondary spacing, but both push up the transformer cost so are typically only seen in medical devices and other applications where a low capacitive leakage floating supply is required.

[pedrocelli]

Thank you for the suggestions and explanations. What I had in mind was a kind of "add-on" to a Class 2 PSU on the secondary side when the Y-cap was removed. Just enough to block RF transmission to the output lines.
Connecting the Y-cap to an incoming ground line ... does this really eliminate the secondary RF component or will the ground line be just a good antenna?

[T3sl4co1l]

Well, that's the point of the Y cap, it is already "enough" to deal with EMI.  Hopefully.

Connecting it to ground instead of secondary, guarantees full EMI on the secondary. That's a non-starter.

Tim

[PChi]

I think that the only solutions to look beyond the common low cost switch mode power supply.

I was under the impression that supplies rated for medical use had higher insulation standards and lower leakage currents and (unfortunately) a higher price so might be better.

[Ian.M]

If the Y capacitor from the 0V side of the primary DC bus goes to ground, then there must also be a capacitor from secondary 0V to ground to provide the return path for the switching frequency leakage current through the transformer's inter-winding capacitance.    However splitting the Y capacitor that way so its a series pair with the mid-point grounded eliminates nearly all the line frequency leakage current by grounding it out.   A RF choke between the capacitor center point and the ground lead may be required to further reduce EMI.

[NiHaoMike]

Change the plug to polarized, then put the Y cap from output ground to neutral.

[Ian.M]

Unfortunately, in many countries, there is either *NO* consistent polarisation for wall sockets, or the plugs with ground are reversible.

[Johnny B Good]

 Hi, I'm a refugee from the " FeelTech FY6600 60MHz 2-Ch VCO Function Arbitrary Waveform Signal Generator" thread where, amongst many other topics about the various shortcomings of this otherwise excellent "cheap 'n' cheerful" gem of an AWG, this business of half live mains voltage 'touch current leakage' problem has been a major concern harking way back to before even this thread was started.

 I was late to that party too (joined it only a couple of months back after it had been active for over a year!). I've pretty well dealt with most of the cost effective improvements that took a 75 quid 60MHz (sine limit) dual channel arbitrary wave generator to a levels of performance more in keeping with a 700 quid Siglent AWG (it now represents a total investment of 103 quid). However, after upgrading the crap dual CFB opamp with a pair of THS 3491s, fitting a much needed cooling fan and replacing the crappy commodity 100ppm 50MHz XO chip (obviously chosen by Feeltech on cost alone) with a 0.1ppm TCXO, I've finally turned my attention back to the commodity 10W 5v/+&- 12v smpsu board (which I've already 'improved') with regard to this issue of the Y1 cap induced half mains live leakage.

 This leakage issue was one of the first 'problems' noted by the early reviewers who tackled it by either replacing the psu completely with home brewed analogue concoctions of their own (there was ample space to accommodate the bulk (only just, in some cases!) of a 50/60Hz 10 or 15 VA mains transformer and a PCB packed with bridge rectifiers, huge smoothing caps with 78xx and 79xx regulator chips and their heatsinks shoehorned onto the PCB(s). Other, less ambitious solutions were to replace the figure of 8 mains socket with a 3 pin connector (clover leaf or IEC13/12) and wire an extra 10nF cap in series with the existing Y1 cap, taking the junction to the mains socket earthing pin (with or without an "Earthed/Floating option switch").

 None of those "Solutions" appealed very much to me for several reasons. Firstly, since the whole signal generator only weighs a mere 681g (with additional fan and TCXO board fitted - it was even lighter to start with!), 'upgrading' from a 2 wire figure of 8 plug ended mains lead to a three core plug ended one, would make it a shining example of the phrase, "Allowing the tail to wag the dog." writ large in the physical reality of such a setup. :-( Secondly, such protection against frying Ds UT from high voltage transients relies strongly on the existence of a sound protective earth which can go open circuit through any variety of Acts of Sod (including forgetfulness if one has foolishly included an "Earthed/Floating option switch" without the inclusion of a 10K resistor across said switch <which resistor is yet another plaything for Sod's amusement>).

 With all of that in mind, I've been rather reluctant to charge straight in (like a bull in a china shop, so to speak), "upgrading" to a 3 pin mains socket to add a protective earth so have spent fruitless hours scouring Ebay and then every other resource searching for an SMPSU that uses the recommended, "costs an extra penny to implement", screened transformer (electrostatic shield between the high voltage primary and the SELV secondary windings to solve this issue at a stroke).

 It seems the much tighter earth leakage requirements in various of the United States which are only now just beginning to be incorporated into the regulations concerning domestic electrical goods hasn't yet started to take enough effect on the manufacturers of such small (5 to 75 watt) SMPSU powered kit and wallwarts to persuade them to invest the extra penny right now on the cost of the transformers they're buying in. Even if some SMPSU manufacturers are being sensible about the whole half mains voltage leakage issue and gearing up so as not to be caught 'napping', it'll probably take a few years for them to appear on Ebay at commodity prices so I've given up hope of buying a 'ready made' solution and taken to considering alternative ways to effectively null out the unwanted leakage voltage.

 Initially, I thought of adding a small 1VA 240 to 120 volt isolating transformer where I could create a half mains live in anti-phase with respect to the "neutral" line (an arbitrary concept with class II PSUs) which I could link to the SELV common rail via a matching class Y capacitor (typically 1 or 2.2nF) to null out the 50/60Hz fundamental half live voltage.

 However, I had what I thought was a stroke of genius, namely adding a separate smpsu to serve the 5v requirement, leaving the original smpsu board to handle the +/-12v rails so that by connecting the separate 5v smpsu to the mains supply in anti-parallel, the commoning of the two SELV grounds would result in a neutralised earth leakage. Sadly, a quick test revealed the flaw in my thinking by demonstrating that the direct bridge rectification always results in the same phase of this half live voltage wrt the neutral regardless of which way round you connect it to the mains supply so, no neat solution there and thus it was back to the original transformer idea which I've yet to actually test but which does stand a good chance of providing an effective enough solution.

 I mention all this to save others the bother of testing "The two smpsus connected to mains in anti-phase earth leakage solution" and having their hopes dashed. The 1VA anti-phase half live transformer is practicable but the problem lies in obtaining such a specialised 1VA sized PCB mount transformer at an economic price.

 The closest to 'cheap' is to buy a pair of more usually specced step down transformers and link them 'back to back' via their secondary windings. This is the way I'm going to proceed, using 6 and 2.2 VA 240 to 22v transformers I just happened to have handy for initial experimentation. Ideally, I need half mains live voltage in antiphase but full mains voltage in antiphase will do for now - I can try using half the capacitance value of the original 1nF to compensate for the excess of voltage (a pair of 1nF caps in series should do nicely).

 What concerns me the most is the distortion of this half live leakage voltage which will prevent a perfect nulling out. Hopefully, the residual harmonic voltages will be insignificant enough to be discounted as a hazard to the Ds UT (they'll certainly not be a "Touch Voltage" issue). Since I'm initially doubling up on 1VA PCB mount transformers, it has occurred to me that I could add a suitable opamp to drive a 6.3 volt secondary on a 120v transformer to generate an inverted replica of this distorted half live voltage waveform to completely null out this troublesome mains leakage voltage. Although this will add the complication of an amplifier,  it'll save using a pair of cheap and cheerful 1VA PCB mount transformers or getting hold of a more obscure 240 to 120v 1VA PCB mount unit at a price premium and, possibly more importantly, allow me to tune out any inconvenient fractional phase shifts using an RC phase shifting network to get a perfect anti-phase bucking waveform.

 The reason why I don't already have results to report is that I'm in the middle of researching the problem right now as I type and haven't yet got to the point of lashing up the pair of transformers into the required configuration. I only discovered this thread as a result of my searching for a similar solution to my own. Having found this thread, I thought it best to impart my findings so far and report what I'm planning on doing before risking electrocution in the name of solving the class Y capacitor earth leakage issue.

 This might be my last testament on the subject if it all goes fatally wrong. If you hear no more from me and are intrigued enough to follow through on my transformer derived anti-phase solution, take note and be extremely careful to avoid *exactly* replicating my experimental procedure. :-)

 Hopefully, I'll be able to follow up with a proven solution in a day or two, give or take the usual IRL distractions. Meanwhile, anybody who feels this idea is worth following up on, is, naturally, more than welcome to try it out for themselves. :-)

 Regards, Johnny B Good

[Zero999]

Johnny B Good,
I don't see how your anti-phase earth leakage idea can work. Please post a schematic.

Wouldn't it be easier and safer to simply power the SMPS off a small isolation transformer? The type used in a shaver wall socket will probably do the job.

[Johnny B Good]

Johnny B Good,
I don't see how your anti-phase earth leakage idea can work. Please post a schematic.

Wouldn't it be easier and safer to simply power the SMPS off a small isolation transformer? The type used in a shaver wall socket will probably do the job.

 Hi Zero999,

 That's an excellent question. Rather sadly for me, this idea of using a tiny transformer to provide a Hi Z anti-phase voltage to buck the nasty half live leakage current, also falls foul of the direct fullwave rectification that dooms my (not so) clever idea of combining the the mains supply connections of a second smpsu in "anti-phase" so that the leakage currents null out at the commoned ground connections on the LV side. I now have some inkling of what it's like to be Baldrick[1]. 

 Your reply, btw, materialised in this thread during the 30 second window between refreshing the page to check for any replies and my actually starting to post a follow up. I got a warning message in the send box, advising me that I might want to review my posts. Unfortunately, there seems to be no cancellation/abort mechanism so I clicked the back button and refreshed the page. For all I know, there's a post reply window now sat in Limbo on the server. Anyhow, I'll forget that imponderable and cut back to the chase...

 I should have realised from my test with a second smpsu that the buck transformer idea would also fail for exactly the same reason. I've mulled the results of my transformer testing over and it's finally dawned on me just what the real problem is. In essence, if you want to reduce this leakage voltage to a safe level (circa 5v rms) without the benefit of a safety earth connection, you need some means of identifying which way round the mains live and neutral connections are so that you can link the 0v rail to the actual neutral via a 100K resistor (1nF Y cap case, a 2n2F cap would need a 47K resistor).

 It's possible to detect which is the live (hot) and which is the neutral in the absence of an actual safety earth connection by using a very high input impedance detector referenced to a small probe antenna (a few square inches of foil placed clear of any high voltage influence within the case of the double insulated smpsu powered gadget - in my case, a small AWG) to sniff the ambient electric field in lieu of an earth connection. This can then switch the 100K safety resistor to the identified neutral connection to load down the 80 volt half live (as measured with a 10Mohm DMM), typical of such kit plugged into a UK 240v mains outlet, reducing it to around the 5vrms mark for the sake of any "At risk of esd" device under test. The virtue of using such a high value resistor being that should the automatic routing of the resistor become confused to the point of connecting to the live instead of the neutral, the touch current is limited to a mere 2.4mA, eliminating the risk of electrocution. The user might get an unpleasant shock but they'll at least survive the experience to investigate the failure.

 However, I've now reconsidered the idea of using a polarized three core mains lead with safety earth connection after checking out a couple of my spare ex- laptop charging brick trefoil plug ended mains leads. Whilst not all such leads use thinner and more flexible three core flexes than those typically found sporting the bulkier C12/13 connectors used by desktop PCs and the like (including my Siglent SDS1202X-E DSO), I do seem to have at least two such leads flexible enough to not 'wag the dog' quite so much as I'd feared.

 Whilst it's possible to solve the 'Touch Voltage" issue without adding a safety ground wire by using some clever electronic add-on, the upgrade to an earthed solution now has a lot more going for it than I'd originally thought (largely on account of its pragmatic nature versus all the other possible and much more difficult to implement solutions I've given serious consideration to).

 So, all I have to do now is check out my junk pile collection of salvaged 'useful bits', recovered from scrapped ATX PSUs and laptop charging bricks, to see if I happen to possess any already blessed with a plug in flylead to match the existing plug in headers on the PSU board. If I'm lucky, I might be able to save a bit of time on the soldering work. I'm planning on using a 10 or 15 K resistor to link the 0v rail to the safety earth. This will reduce the leakage voltage to less than a volt which is more than sufficiently low enough to eliminate the ESD risk and neatly avoid potentially troublesome 'earth loop issue'.

 At the end of the day, there is no easy way to eliminate the undesired effects of the class Y EMI suppression capacitor in class II smpsus without the benefit of a transformer which includes the electrostatic shield layer between the HV and LV windings to allow it to truly float free of an earthing connection[2]. In my case (Feeltech FY6600-60M AWG), I have the extra complication of having to mount a supporting bracket to strengthen the rather flimsy rear panel against the much higher insertion/extraction forces mandated by the trefoil plug and socket upgrade. It's all extra effort I was hoping to avoid by a "More Brain, less Brawn" solution I'd imagined I'd be able to come up with, hence my delay in dealing with the problem.

 The answer to the question posed in the thread title is this: You can't really eliminate the Y capacitor unless you're prepared to accept the resulting increase in RFI (and even then, there'll still be some residual 'touch voltage' via the 50 to 100pF leakage capacitance between the primary and secondary windings).

 Short of finding a screened version of the transformer to replace the unscreened original or else the capability to take the transformer apart in order to rewind it so as to add a screening foil yourself[3], the pragmatic solution is to modify it to use a 3 pin plug in mains lead so as to ground out the touch voltage on the low voltage output terminals (with or without a 10 to 15 k resistor to eliminate concerns over adding an earth loop risk). It turns out that the pragmatic approach is the best solution after all in this case. Who knew? Well, I'm a little older and wiser after all that faffing around with back to back transformers so I certainly know it now! 

][1] A  "Black Adder" reference.

[2] Seemingly the preserve of the much more costly "Medical Grade" wallwarts.

[3] Yes Sir! I reckon I must have considered every conceivable option there was to consider, including that one but, after stripping apart a few spare 12 and 15 VA smpsus for spare bits and some inspiration, it became obvious all too soon that this was a step too far, even for me.

 I'd had visions of unwinding a dozen or so turns of secondary windings (doable without splitting the core), laying a screening foil down and rewinding the secondary windings. Unfortunately, it seems the common winding practice is to lay down the LV secondaries first and then wind the HT primary windings on top. It's one thing to rewind a dozen or so turns of relatively robust enamelled wire but dealing with a hundred or so turns of very fine enamelled wire was another world of pain I felt best not to visit (I know my limitations).

 Harking back to your comment regarding using a shaver wall socket isolating transformer, I'm not entirely sure such a commodity item would necessarily provide enough isolation at the microamp leakage level required to sufficiently reduce the ESD risk. Even if you use a high grade 10VA or so transformer (and this is something else I gave plenty of thought to), there'll still be some leakage to deal with unless you sacrifice some efficiency to improve the isolation performance using split bobbin construction to separate the primary and secondary windings more effectively (71% efficiency was a figure I saw for one of these transformers), so a significant heat load if it's installed into something like that FY6600 AWG, especially if still reliant solely upon the original, ill thought out, passive ventilation (the fan upgrade was not so much a 'luxury' as it was an 'essential' item that really should have been fitted as standard to be begin with).

Regards, Johnny B Good

[Johnny B Good]

 I'm following up my own post simply to report that I finished the class II to class I earthed psu modification on my FY6600 late Saturday evening. I could only find the one C6 socket, an ex-laptop charging brick type that slides into location rather than the easier to deal with panel mount screw on styled type. This just meant I spent a lot more time manufacturing the reinforcing/mounting bracket from a piece of 1mm thick scrap of aluminium sheet.

 I'm quite pleased with the result (non of that flexing of the rear panel when inserting and removing the original 2 pin C8 mains lead connector). Also, the 90v touch voltage has been loaded right down to just 500mV by the 11K earth drain resistor (3x33k in parallel for a bit of redundency) which all but eliminates the risk of ESD damage to sensitive devices under test.

 A useful but predictable side benefit was elimination of the loss of output from the generator from ESD events created whenever connecting any earthed equipment to the generator at just the wrong part of the mains cycle. It may also have cured another, slightly less annoying issue with the left and right arrow digit select buttons but I won't know without further testing or extended use. Suffice to say that this pragmatic cure for the earth leakage touch voltage seems to have been the best way forward despite my initial misgivings about this type of modification.

 Even if you can source or create a mains leakage free class II PSU, there'll still likely be a 50 to 100 Megohm half live leakage path simply to prevent destructive static charge build up on the ELV side so don't get your hopes up of getting a perfectly isolated class II psu, regardless of type, any time soon.

 Regards, Johnny B Good

[Zero999]

I thought shaver transformers are wound on double bobbins for safety reasons? In any case how lower leakage current are you after?

How about using a capacitor, rather than resistors to earth? That would remove any DC path to earth, yet shunt the AC leakage current and has the advantage of lower impedance, at higher frequencies.

[Johnny B Good]

I thought shaver transformers are wound on double bobbins for safety reasons? In any case how lower leakage current are you after?

How about using a capacitor, rather than resistors to earth? That would remove any DC path to earth, yet shunt the AC leakage current and has the advantage of lower impedance, at higher frequencies.

 A dedicated 110v shaver socket transformer probably will  use a split bobbin construction for safety reasons in the damp/wet environment of a bathroom. The extra losses, maybe 3 or 4 watts, aren't any problem in that situation and actually a blessing in the circumstances to help keep damp at bay.

 Unless I'm very much mistaken, the 110v output winding will be centre tapped to connect to the safety earth so as to create a bi-phase 55v supply with no neutral return (both wires in the shaver plug will be 55v 'lives' to reduce the risk of electrocution through body contact with earthed objects to meet the safety standards required in a wet room environment. If you're going to use such a transformer to power a smpsu designed to work over the mains voltage range from 90 to 265vac here in the UK, the reduced 110v will reduce the half live mains leakage voltage by a factor of two. Whether the balanced nature of the supply will provide another 50% reduction should you choose to connect the centre tap earthing point to an actual protective earth or not is not something I've given any consideration to. It might or might not help but I Have a feeling, in the light of recent experience, it won't help. Obviously, this is something you can test for yourself.

 You'll probably get the most benefit in this case by allowing the secondary to float free of any earth connection, assuming you're just looking to avoid the need for a three core cable connection to the PSU and all but eliminate the troublesome half mains voltage leakage without introducing troublesome hum loops or unwanted conduction of the switching harmonics into the mains wiring, perhaps to radiate and interfere with portable LW/MW/SW AM radios you (or your immediate neighbours) might have in regular use around your(their) home(s).

 As for my choice of an 11K earthing resistor, rather than, say a 100nF capacitor as others have used, let's just say I know my own mind on this. If the smpsu board had used a shielded transformer with a 2n2F Y cap connected to the inter-winding shield instead of to the common end of the secondary windings (effectively the zero volt common ground return for the + 5 and 12 and -12 volt rails), the worryingly high touch voltage would have been reduced to a vanishingly small value with less destructive capacitive charge (circa 50pF with perhaps 10v peak versus 1nF and 170v peak) potentially able to damage the more fragile parts of any kit through ESD transients when having test signals injected into it, then there wouldn't have been any perceived need for such mitigation measures.

 As it happens, Feeltech elected to go with a commodity three rail smpsu board rather than order one with a screened transformer costing just a penny or two more to manufacture, so we have a potentially serious problem in the event that anyone using such a signal generator needs to inject test signals into random locations inside a possibly very expensive piece of kit sat on the repair bench. The absence of a ground loop is a handy feature but not when it comes with a hefty half live mains voltage riding on the BNC ground return.

 My aim with upgrading from a 2 pole C8 connector to a 3 pole C6 connector was simply to provide access to an earth point by which to link the BNC grounds via a carefully chosen 11K resistor so as to knock the half live leakage back from 120vrms right down to just half a volt rms without creating a low impedance ground loop circuit to complicate the business of injecting test signals into various electronic devices, including audio gear.

 In this regard, the 11K resistor is virtually an open circuit compared to the one ohm or so resistance of the signal generator test lead ground connection to the ground or common return point of the device under test. A capacitive link to the safety earth connection would be of no benefit whatsoever in this usage case - it would actually be a hindrance and very probably inject unwanted switching harmonics into the mains wiring as I've already pointed out. I'd still like to test the odd bit of radio kit from time to time, so needlessly polluting the local environment with RFI is an activity best avoided in this circumstance. In short, I chose to use a resistor instead of 100nF capacitor because the capacitor, to my mind, is not the best solution in this case.

Regards, Johnny B Good

[Zero999]

I can assure you shaver sockets are completely floating. The one in my house has both a 240V and a 120V output. The point is that touching one side of the output can't give you a shock. I think you're confused with building site transformers, which do have a centre tapped to earth output.

Regarding the resistor vs capacitor debate, they each have their up/down sides. I've seen both. Bench top power supplies and lab equipment tend to use capacitors rated to a few hundred volts, so the power supply can be floated with respect to earth at reasonable voltages. That 11k resistor would get very hot if if had  a couple of hundred volts across it. Laptop power supplies and IT equipment often have a resistor connected to earth. If it were me, I'd do both: a capacitor with a 1M or so discharge resistor across it

[Johnny B Good]

I can assure you shaver sockets are completely floating. The one in my house has both a 240V and a 120V output. The point is that touching one side of the output can't give you a shock. I think you're confused with building site transformers, which do have a centre tapped to earth output.

Regarding the resistor vs capacitor debate, they each have their up/down sides. I've seen both. Bench top power supplies and lab equipment tend to use capacitors rated to a few hundred volts, so the power supply can be floated with respect to earth at reasonable voltages. That 11k resistor would get very hot if if had  a couple of hundred volts across it. Laptop power supplies and IT equipment often have a resistor connected to earth. If it were me, I'd do both: a capacitor with a 1M or so discharge resistor across it

 Thanks for the info regarding shaver socket outlets being completely floating. My only knowledge of such transformer safety measures comes from what I've read about those building site transformers. Considering the very light electrical load and the use of potted split bobbin isolating transformer construction, I guess the risk of one side of the secondary winding 'silently' developing an earth contact fault must be considered astronomically low enough to make it a safer option than the 55-0-55 bi-phase system used with building site transformers to power 110v power tools.

 I can see why you were giving this option serious consideration.  BTW, do you have any idea what sort of VA rating these shaver socket transformers have? In the case of my modified FY6600-60M signal generator, this can demand almost a full 10W from its little smpsu board (the rating label states "10W max") so I'd be looking at an isolating transformer with at least a 10VA rating, preferably 20VA  if you wish to preserve the full 90 to 265 volt universal voltage rating without adding a 120/240 volt selector switch (you'd use it as a 1:1 isolating transformer using the 240v windings to simply pass on whatever the mains voltage happens to be, hence the need, in this case for ideally, a 24VA rated transformer to cater for the higher I squared losses in the windings when operating at the low end of the voltage range in this mode).

 Obviously, if you're prepared to fit a voltage selector switch, you can get away with a 12VA transformer and use the 110v secondary to feed the smpsu with a floating mains supply voltage - just how often would anyone ever need to adjust between mains voltage standards to the point where this becomes an onerous task? 

 Regarding bench supplies, they're normally class I earthed with the outputs floating wrt to the case metalwork so they can simply tie the 'zero volt' rail and all output terminals to the earthed case via 100nF caps with a static bleed resistor to eliminate potentially damaging static build up under conditions of extremely low humidity. My cheap AWG, otoh, was never blessed with an earthed case to begin with, relying on the class II rating of its integrated smpsu to make the complete unit a class II double insulated device.

 Unfortunately, the half mains live 'touch voltage' that results is of some concern in such T&M equipment where a moment of forgetfulness over linking the BNC shield connection to the DUT via a bnc cable or a separate ground linking cable could result in damage to the item being tested. The concern may be more imagined than real but, nevertheless, the fact that such a scenario could play out given half a chance for Sod's Law to be invoked by a moment of carelessness, does legitimise the concern. Since I don't plan on stacking a bunch of these up in synch to generate 80v p-p signals, the issue of burning out the 11K earthing resistor won't normally arise, Since I don't wish to create unwanted earth looping paths for HF harmonics, the resistor is all I require to effect a cure for the half live 'touch voltage' issue. Obviously, that won't suit everyone's needs but it suits mine just fine. 

 If the original question is in regard of wallwart smpsus, the Y capacitor can only be relocated to connect to a screening foil between primary and secondary windings in the transformer instead to the 0v point of the output rails. That being the case, the practical solution is limited to spending extra cash on a "Medical Grade" replacement wallwart.

 Regards, Johnny B Good.

[Zero999]

The issue with the resistor overheating is due to the device under test being at a high voltage, relative to earth. The main advantage of the capacitor is the device being tested can float at any voltage, up to the capacitor's maximum rating.

Shaver transformers are normally rated at about 20VA, although they won't be rated to that continuously. Dual voltage operation is standard. The one in my house has separate sockets and a mechanism which ensures only one can be used at a time. I haven't taken it apart, but I'd expect it has two 120V secondary windings and a switch to select between series and parallel configuration, depending on whether the 240V or 120V socket is used.

I had a quick search for 25VA isolation transformers.
Triad VPS230-110
Available from Mouser and Farnell. Lookst good, double bobbin and low capacitance.
https://uk.farnell.com/triad-magnetics/vps230-110/transformer-2-x-115v-25va/dp/1610413
https://www.mouser.co.uk/ProductDetail/Triad-Magnetics/VPS230-110?qs=%252bfhPKvh1fqM%2fIMKduQ8CRw==

Triad VPT230-110
Toroidal. Probably higher capacitance, but will be more efficient. US stock so available from Farnell at high cost or Mouser, with a long lead time.
https://uk.farnell.com/triad-magnetics/vpt230-110/transformer-toroidal-2-x-115v/dp/1785731?gclid=EAIaIQobChMIlZq7ypjy3wIVqrftCh2WqQLMEAAYAiAAEgL2sPD_BwE&mckv=s4rYYLzOH_dc|pcrid|99299485688|kword|vpt230-110|match|p|plid||slid||product||pgrid|5285736368|ptaid|kwd-24784252128|&CMP=KNC-GUK-GEN-SKU-MDC
https://www.mouser.co.uk/ProductDetail/Triad-Magnetics/VPT230-110?qs=wkKrz7WmEgNNNgyT8w4YqA%3D%3D&gclid=EAIaIQobChMIyMfwz5Hy3wIVo7XtCh3yIQbZEAAYASAAEgJLFPD_BwE

Bel Signal A41-25-230
Similar to the first one. Available from Digikey and Mouser.
https://www.mouser.co.uk/ProductDetail/Bel-Signal-Transformer/A41-25-230?qs=sGAEpiMZZMvwUzoUXIIvyZ8riR3Pta28%252b8Y6KbSbVRk%252bKxx4UgnHTw%3d%3d
https://www.digikey.co.uk/product-detail/en/signal-transformer/A41-25-230/595-1303-ND/953173

Another possibility is to buy a cheap shaver socket or salvage one from a skip and take the transformer out of it. Some are cheaper than buying an isolation transformer. The only issue is whether it will be able to power the SMPS continuously, without overheating.

https://www.tlc-direct.co.uk/Products/PU8900.html?source=adwords&ad_position=&ad_id=315107931576&placement=&kw=&network=u&matchtype=&ad_type=&product_id=PU8900&product_partition_id=297072075353&version=finalurl_v3&gclid=EAIaIQobChMI7dGoqZry3wIVDJztCh1ErAEjEAQYBCABEgKNE_D_BwE
https://www.directtradesupplies.co.uk/product.php/193005/ansell-dual-voltage-shaver-socket--white-?gclid=EAIaIQobChMIwImru5vy3wIVRofVCh0nzwF8EAkYASABEgJrgfD_BwE
https://cpc.farnell.com/unbranded/701875/shaver-socket-dual-voltage-white/dp/PL15991?mckv=sKmngMpNY_dc|pcrid|224681503328|kword||match||plid||slid||product|PL15991|pgrid|47203068935|ptaid|pla-559841591210|&CMP=KNC-GUK-CPC-SHOPPING&gclid=EAIaIQobChMIiejRrZvy3wIVSrftCh18vghaEAkYDiABEgJ95fD_BwE

Regarding dual voltage operation: I don't see any issue with having a jumper or switch. How often are you planning to take it abroad? A 1:1 230V transformer will work, but bear in might the current draw on 120V will be greater, so a 230V transformer running on 240V will only have half the VA rating. The power factor of the SMPS might also be poor, so even if it only uses 10W, it still might need a 20VA transformer.

[Johnny B Good]

 I'll address the several points of concern that you've raised, one by one.

 Firstly, the resistor overheating risk in this case doesn't (or damn well shouldn't!) apply since only an idiot would elevate the signal generator's ground rail above the 50v dc level typical of telecoms kit. In general connecting a signal generator's ground via the BNC connector shield to a device is always only done on a device which has its common rail floating or at zero dc/ac potential.

 Accessing test points within the DUT (old valved gear for instance) that have elevated potentials (the more likely scenario) are normally dealt with by using either capacitive coupling or else an isolation transformer. I certainly won't be taking such risks when using test gear on equipment containing destructively high voltage potentials which may also pose an electrocution hazard just as any EE  worth his salt would not. I would have much greater concerns than that of overheating or burning out a half mains live suppression grounding resistor in such cases.

 If you need to power a 10W rated class II smpsu via a 20VA peak rated shaver isolation transformer, it makes a lot of sense to use a voltage selector switch to prevent the smpsu compensating for the halving of the voltage by a doubling of the current draw. You can either wire the secondary windings in series to supply 240v or else in parallel to supply 120v to the smpsu and add a voltage selector to switch the split primary into either a 240v series  or a 120v parallel configuration to match the local supply voltage. As we've both observed, "How often would anyone need to switch between mains voltage standards as to make this an onerous task?"

 As for the examples you linked to, the only viable contenders were the split bobbin types. Their voltage regulation (an indicator of their efficiency) might be poor compared to the toroidal type but the leakage capacitance stands a better chance of being 50pF or less whereas the toroidal type may well have 500pF or worse leakage capacitance making them no solution to the problem caused by the accursed 1000pF Y cap in the typical 10W rated class II smpsu.

 I happen to have a 30 year old example of the 24VA split bobbin breed which is still in manufacture today in precisely the same PCB mount form right down the the plastic shell having been produced by the same injection moulding machinery. The difference being that it has a pair of 15.4 volt 350mA and a a pair of 10.3 volt 700mA secondary windings which would nicely supply, after rectification and smoothing, a set of switching regulator modules to provide the +/-15v and the +5v rails, obviating the horrendous losses of the analogue regulators used by others who'd replaced the original smpsu board inside their FY6600 AWGs with analogue PSUs of their own design.

 However, having tracked down the current manufacturer of that transformer, I was able to download their data sheet on this family of mains voltage isolating transformers and spotted the 71% efficiency reference which rather put me off any further consideration of its use to eliminate the class Y cap leakage issue. I'd originally guestimated a 12VA figure some 30 years ago when I first acquired it but, having had access to the data sheet, I've been able to identify it as a 24VA unit and could estimate the current ratings of the secondary windings with some confidence.

 It may yet see use in a future psu project. :-) I might even wire it up to a mains connector to test its leakage capacitance if curiosity ever gets the better of me when I've nothing better to do but, for now, there's no burning need for that information now that I've found a simpler solution to the mains leakage issue.

 Regarding your concern over the poor power factor of such class of smpsus that aren't required to meet the PFC requirements that apply to smpsus with 75W or higher wattage ratings, I wouldn't worry. The Power Supply Utility (and the smpsu for that matter) will thank you for thoughtfully adding a measure of PFC in the form of a high leakage inductance transformer with its circa 100 ohm per 120v winding to further take the sting out of the smoothing capacitor induced narrow angle rectifier conduction current spikes. In view of the obvious use with such rectifier/smoothing circuits on the 15 and 10v secondary windings version in my possession, I should think the same conditions apply wrt such a rectification and smoothing technique even at mains voltage levels. The losses in the transformer will be slightly increased but not enough to raise any concerns imo.

 Regards, Johnny B Good.

[Briain]

Hi Zero999

A while back, I bought myself a Feeltech FY-3224s signal generator and it had the Y cap issue (from memory, unloaded it measured 85 Volts on the BNC screen, WTR mains earth) so I bought a Triad Magnetics N-68X (50 VA isolation transformer) and with me being in the UK, I used the primaries in series to give me a 230 V to 110 V step down isolation transformer (I also fitted a switch whereby I could leave the outputs floating, or mains earth reference them). With the switch open the BNC screen now shows as having about 5 Volts present, WTR mains earth (though before building it, I'd noted that it was significantly more than 5V when reversing the 110V connections from the transformer to the SMPS). This was pretty 'cool' in that with it weighting 0.77 Kg, it gave the tester some weight (so it didn't slide back when pressing a button, though more weight would be even better); attached photo shows a picture of the tester, with a piece of hardboard added below the transformer to spread the load, such that the plastic doesn't so easily shatter if it's dropped a small distance).

So, it all appeared to be working very well (I bought the transformer back in 2017, but I only got around to fitting it a couple of days ago) but after it had been on for a few hours, I noted that the top of the plastic case was quite warm, so I opened it up and discovered that steel band surrounding the transformer core was sitting at well over 50 deg C (and the Feeltech only draws about 2.5 Watts) so that was not quite so cool (sorry about the pun). I thought about adding a vent gauze to the case, but with the case lid completely removed, it still slowly rises to quite a high temperature (I gave up the test when it crept beyond 40 deg C; the probe was not well coupled to the metalwork so I suspect both temperatures were probably even higher than was indicated) so I cannot say that I'm too impressed with that particular Triad transformer. Indicative measurements - just using one of these cheap and cheerful domestic power meters, so I am not sure how accurate the readings are - showed that when sitting off load, the 50 VA Triad N-68X transformer consumed 4.9 W, whereas my big, yellow 750 VA 'site transformer' consumes just 5 W when it is sitting off-load.

During my Internet dredge for a suitable replacement (there are surprisingly few available and most are all too tall to fit into my case) I stumbled across this thread and the 25 VA Bel (A41-25-230 ) that you identified in your post looks to be well suited to my own requirements (so thank you for that). I have just ordered one and I will post back in a few days to let everybody know whether it runs at a more sensible temperature than that toasty Triad.

Briain
(GM8PKL)

NB The new transformer will only weigh 0.57 Kg, so I'll maybe add about 400g of lead to kelp keep the tester stationary when pressing the buttons on the front panel (there is another thread about these units - and the Y capacitor issue - and someone else has also added lead). :-)

[shakalnokturn]

Should it actually be called a "Y capacitor" in this case (primary to secondary on a class 2 SMPS)?

[Briain]

Hi

I've always called it Y when it sits that position (as it would have to be one rated as being a Y capacitor), but it is entirely possible (possibly even probable) that I've always been entirely wrong. 

Anyhow, I fitted my new transformer, though as it was physically smaller (and thus lighter), I replaced the hardboard load spreading sheet (shown below the transformer in the photograph attached to my previous post) with two sheets of lead flashing material (and the added mass is very effective).

As to the performance, whereas the Triad N-68x ran at 40 deg C off-load (with the plastic case wide open) and >50 deg C with it on light load (i.e. the FY-3224s running, but not with terminations on the outputs) and with the case closed, the Bel transformer runs at just 31 deg C with the case open (and off-load) and just 38 deg C with the case closed (same load as described for the Triad experiment) so I am delighted with that result. In both cases, the temperatures were taken after > 1hour running time and ambient room temperature was 20 deg C.

With a high impedance meter (a Fluke DMM) measuring between the BNC and mains ground, the indicated Voltage was 3.2 V (actually, it only dropped by 0.2 V with the mains switched off; I added a switch -the white one in the back panel - to the primary side). As I had an ESP strap sitting right next to me (which contains a total of 4 M Ohms resistors) as an experiment I used that to link the BNC outer to mains ground and the Voltage dropped to 1.2 V, so I'd call that a great result. Pondering it further, I think I'll actually fit a 4 M Ohm resistor across the switch; that effectively still leaves the floating option, but it provides a path to prevent any chance of static build up (unlikely, but it would do no harm having it in there).

The one thing I would note about that particular variant of the transformer is that it really should have insulated guards to keep fingers separated from the dangerous bits. As can be seen in the photo, I have mitigated the risk by using heat shrink sleeves (and also stubs of small heat shrink on the exposed pins - which run at 90 deg to the main spade terminals - to which the transformer winding wires are terminated) but after ordering the above transformer, I noticed the below comment in the specification sheet:

Note: The VDE certified A41 series transformers with standard length and colour lead wires are readily available. See website for our AF1-L product series

If I'd spotted that the first time I'd read the specification sheet, that's the version that I'd have used (assuming that it is readily available, of course).

All the best
Briain

[smile]

So can anyone actually can take a scope and measure this NOISE, ripple increase on PSU output wires, before and after Y caps removed? I would do it myself but I do not have a scope.

thank you.

[T3sl4co1l]

So can anyone actually can take a scope and measure this NOISE, ripple increase on PSU output wires, before and after Y caps removed? I would do it myself but I do not have a scope.

thank you.

Obligatory,



What PSU and what Y cap(s)?  You will find the answer depends on everything.

You also need to specify measurement method.  A 20MHz differential ripple measurement might not show much if anything different.  The noise is largely common mode, much more pernicious, and requires a different measurement method.

Tim

[smile]

Well, then can somebody take a look at this schematic https://ab57.ru/schema/LiteOn-PA-1121-04.pdf
This if for Laptop PSU from Acer. It has no double square symbol, yet is it isoloated, no Y caps.

Can this be used as reference to modify Meanwell PSU? Like the RPS-300 27v or LRS-350 48V?