EEVblog #659 - Medical Plugpack Teardown

[EEVblog]

What's inside an IEC60601-1 medical class 5V mains DC-DC plugpack?
What is the difference between PCB clearance and creepage?
A German FRIWO 5V FW7555M/05 brand medical grade isolation unit vs a cheap Wun Hung Lo consumer brand.
Also a comparative look at transformer construction.

Datasheets & Links:
http://www.farnell.com/datasheets/30889.pdf
http://en.wikipedia.org/wiki/IEC_60601
http://www.rubycon.co.jp/en/catalog/e_pdfs/old/aluminum/e_yxa.pdf
http://xcnet.de/projekte/xnetcreate.de_v2/pics/kunden/dateimanager/5221/JE8_CD_263_BK.pdf
http://www.vishay.com/docs/83515/tclt1000.pdf

[Vgkid]

Thanks for the teardown. On a related note it is rather interesting to note build quality practices among various ebay/ oem(real/fake) quality laptop psu's.

[Dave]

I think they didn't route under those optos because they would risk having the board fall apart. Not to mention, there is glue under those optos, so the creepage distance would effectively be increased, as there is probably no direct path under those packages.

I didn't like the flimsy heatsinks. The only thing that held them attached to the board were the leads of the components themselves. Mechanical stress on the solder joints should be avoided, especially if you are trying to make a very reliable product. The same goes for pop rivets, the "pop" gives a mechanical shock to the component, which can cause damage to it and significantly reduce its life.

[coppice]

The transformer construction goes a bit farther than most, but the rest of that medical supply is just following standard approvable design - 8mm creepage, air gaps, and so on.

I have never understood why you can get away with such ropey transformer construction as most approved supplies have. The opto couplers, capacitors, and board are well controlled by the approvals requirements. They have to not just fail, but fail badly to present a serious risk. Most transformer's, on the other hand, have to be working perfectly to provide safety. The tiniest flaws in the insulation somewhere in those windings, and there's pretty much no isolation at all.

[Phoenix]

A few thoughs on some of the components (nothing definitive). Would the signal diode you identified in the DaveCAD be a zener? Then that circuit provides some sort of overvoltage/peaking voltage level feedback.

The weird external winding may be providing an electrical connection to the transformer cores since it neatly wraps around both E cores. This might have an EMI/shielding effect. Where does the pin go to on the primary side?

The orange wrap looks to be Kapton, rather than Mylar. Though the terms seem to get used interchangibly.

[firewalker]

I believe that the winding above the secondary is not a feedback coil. Just a winding to power the controller.

Alexander.

[mux]

I'm not sure if I said this before on the forum, I remember saying this on the internet once... The reason they are using Samxon (crap garbage bin caps) on the secondary and name brand on the primary is that the secondary caps are much more expensive, simple as that. High voltage aluminum electrolytics are peanuts to build, you can even literally do this yourself and get great stability with some oil paper and aluminum foil. The difference between a name brand and cheap chinese brand cap is less than a cent for these manufacturers, so the choice is easy.

The secondary side is a very different story. Low voltage caps that can supply a reasonable amount of ripple current have electrolytes and oxides that have to endure much higher field strengths without breaking down. It's an art to make these caps and they are subsequently usually the most expensive component after the transformer and non-BOM costs. That is why you almost always see crappy secondary caps and good primaries.

As for the general construction: Dave, you are giving them way too much credit. This is a crappy power supply. It's not $4 QA reject crappy, certainly, but it's not even near high quality in any respect. It's a phenolic, wave soldered, one-tier-up-from-the-shenzhen-market-brand-components, just-meets-requirements design. I don't necessarily blame them, but I've seen medical supplies from Edac that have proper (4 layer even) FR-4 boards, slot isolated transformers (not this 3M foil garbage), name brand controllers, 125C caps, great output filters, etc..

If you ever get a hold of an EEG power supply, I highly recommend tearing those down. They are the absolute best PSUs in the world.

[Dr. Frank]

Some comments on the components:

The TH20594 DC/DC converter (?) may originate from former company THESYS, a small ASIC maker (design only), as far as I remember.
Beneath MELEXIS (mixed signal ASICs fab), it was also a company (VEB) from the former Eastern Germany / Thüringen, nowadays it's XFAB: http://en.wikipedia.org/wiki/X-Fab

The two resistors on the 400V cap seems to laid out ok, they have 2M both, (200* 1e4) and are across the cap, intended to discharge it. So they should carry about half of the 400V each.

There's also no strange thing about the secondary windings and the MBR1660 Schottky (?) power rectifier diode.

Both windings and the diode are in series , therefore you may place the diode everywhere, also "in between" the windings, wouldn't make any difference compared to  placing it "outside" the windings, as it's commonly done.


The PCB is made from CEM-3, a cheaper, but higher grade composite material: http://en.wikipedia.org/wiki/Composite_epoxy_material

Everything is chosen and designed "on budget", but quality seems to be ok, very similar to Automotive Industry.

Nice instructive video about these safety techniques!

Frank

[mux]

Fuses on both line and neutral are only specified for earth-connected (exposed conductor) devices. For 2-pin devices all you need is one fuse for all approval agencies.

That's also why the input circuitry is heavily reduced.

[FrankBuss]

The secondary side is a very different story. Low voltage caps that can supply a reasonable amount of ripple current have electrolytes and oxides that have to endure much higher field strengths without breaking down. It's an art to make these caps and they are subsequently usually the most expensive component after the transformer and non-BOM costs. That is why you almost always see crappy secondary caps and good primaries.

A Rubycon 680 uF / 25 V / 105°C, low ESR capacitor costs EUR 0.58 at Digikey, in single quantities. This power supply costs EUR 64 (at Conrad). Even if you calculate times four the BOM costs to get the retail price, and add a bit for the development costs and profit, it should be possible to use this output capacitor instead of the Samxon. Maybe they think if you can get the medical certification, then you can just double the price, it doesn't matter, the target group will pay it.

[SNGLinks]

Why is there a green LED on the PCB?
The bare wire on the outside of the transformer might be some form of shielding.

[mux]

(...)

A Rubycon 680 uF / 25 V / 105°C, low ESR capacitor costs EUR 0.58 at Digikey, in single quantities. This power supply costs EUR 64 (at Conrad). Even if you calculate times four the BOM costs to get the retail price, and add a bit for the development costs and profit, it should be possible to use this output capacitor instead of the Samxon. Maybe they think if you can get the medical certification, then you can just double the price, it doesn't matter, the target group will pay it.

I don't think you appreciate the markets we're talking about here. Power supply manufacturers don't shop at distributors, those prices don't mean anything. Distributor pricing is almost entirely a function of stock, overhead and supplier deals. I'd be surprised if those power supplies go for more than $10-15 a pop in volume.

Power supplies, yes even medical grade ones, are an extremely cutthroat business where volumes are in the tens of millions *per model* and margins are sub-10%. A one-cent difference between components is already considered large. And those samxons are definitely more than one cent cheaper than name brands. And yes, certification is definitely a big part of the price, but even if certification is responsible for a $20 increase in price, people are still going to choose for the $34 PSU instead of the $34.50 with slightly better caps.

[TheRevva]

Some time ago I was looking into making some significant 'mods' to a PC PSU...
The most informative article I found on the net was this one:
http://www.users.on.net/~endsodds/smps.htm

On that page is a link to his rewind of the main transformer (with pics)
http://www.users.on.net/~endsodds/tx.htm

It describes in fairly good detail a lot of the 'safety' aspects of SMPS transformers.
You'll even see how ONE of the pictured transformer bobbins 'forces' the presence of a 3mm gap to produce an acceptable 'creepage' distance (when coupled with the teflon insulated lead-in / lead-out of the windings).

Anyway, just thought I would 'share'.

[delmadord]

I was just wondering, where is the controller on the second unit. It has a one-layer board, as most PSUs have. But there are no SMDs on it, and I could not find any pins or traces that would resemble  any THT based SMPS controller. Is it possible to build it without one?

[VK3DRB]

That odd-ball outer winding of a few turns may have something to do with reducing eddy currents.

Rubycon... they were notorious years ago in TV's and audio amplifiers where their dielectric would dry out. The non polarised types were often used in cheapo radios and amplifiers as DC blocking caps between the output of an AF amplifier and the speaker. They may have improved over the years, but I tend to go for prefer Panasonic or EPCOS electrolytics if I want quality.

[ryanmoore]

Why is there a green LED on the PCB?

There appears to be a plastic window on the front of the device near where the cable exits. You can see it near the start of the video.

[diyaudio]

I'm not sure if I said this before on the forum, I remember saying this on the internet once... The reason they are using Samxon (crap garbage bin caps) on the secondary and name brand on the primary is that the secondary caps are much more expensive, simple as that. High voltage aluminum electrolytics are peanuts to build, you can even literally do this yourself and get great stability with some oil paper and aluminum foil. The difference between a name brand and cheap chinese brand cap is less than a cent for these manufacturers, so the choice is easy.

The secondary side is a very different story. Low voltage caps that can supply a reasonable amount of ripple current have electrolytes and oxides that have to endure much higher field strengths without breaking down. It's an art to make these caps and they are subsequently usually the most expensive component after the transformer and non-BOM costs. That is why you almost always see crappy secondary caps and good primaries.

As for the general construction: Dave, you are giving them way too much credit. This is a crappy power supply. It's not $4 QA reject crappy, certainly, but it's not even near high quality in any respect. It's a phenolic, wave soldered, one-tier-up-from-the-shenzhen-market-brand-components, just-meets-requirements design. I don't necessarily blame them, but I've seen medical supplies from Edac that have proper (4 layer even) FR-4 boards, slot isolated transformers (not this 3M foil garbage), name brand controllers, 125C caps, great output filters, etc..

If you ever get a hold of an EEG power supply, I highly recommend tearing those down. They are the absolute best PSUs in the world.

Did you even consider the power supply is old?

[diyaudio]

I believe that the winding above the secondary is not a feedback coil. Just a winding to power the controller.
Alexander.

Switching controllers are normally powered from a secondary power source, directly from mains and rectified or a smaller transformer (cost more though), I suspect it might be auxiliary tapping or some type of technique used to minimize leakage inductance.

[Switching Power]

The IEC60601-1 standard itself will cost you around €200 but that don't include all the other standards its referring to.

[SNGLinks]

Why is there a green LED on the PCB?

There appears to be a plastic window on the front of the device near where the cable exits. You can see it near the start of the video.

Thanks, I missed that.

[Flump]

What struck me is how bright and shiney the soldering is.

[wigman27]

Hi Dave!

It would be really awesome if you could do a fundamentals Friday on the theory of the power supply design. For example, what purpose the opto couplers have, what the purpose of the y rated cap has and the basics of operation! It would be really helpful :-)

Thanks

Lee

[patrik]

Here some Pictures of the same type of Power Supply (FW75550/05, build date October 2003), but not Medical graded.

It looks almost the same - the differences seem to be the PCB material,  the chosen Optocouplers (PC123) and the Fuse (T630 instead of the T800mA on the medical version).
As the Optos are different the routing is altered a bit, but interestingly the landing for the resistors used on the Vishay type are on this PCB as well.
The Housing was not welded, is seems to be glued on - nothing what a trusty Swiss army knife couldn't overcome (no need for an Dremel or a sword as Dave is using these days  )

The other components are similar to the -M type, even the transformer has the same type number (15 1430 OO RH), the caps are JH branded (all of them) and rated for 105°C
The similarity goes so far that the green LED is contained, but you can't see it on the outside 

[electronic_eel]

Thanks for the teardown video of this power supply. And thanks patrik for the photos of it's non-medical brother.

Seems like both units survived the EMC tests without the use of Y-caps. This is something you'll seldom find today and something you need for some applications: the Y-cap floats the output to about half of mains voltage in reference to earth. If you now plug the live power supply into a sensitive device which happens to sit at earth voltage, it could go boom. So if you can't get a psu without Y-caps, always connect the consuming device first and plug the psu into mains afterwards.

What I've never seen anywhere else is this dual optocoupler feedback design. Usually you have just a TL431 and a small compensation network with a few caps and resistors on the secondary side, driving one opto. Does anyone have further insight what this dual opto setup is all about and how it works?

[ddebeer37]

I design power supplies for a living and it keeps surprising me how obscure a  subject it seems to be.  Very few people seem to understand isolation, creepage distance and clearance distance and the techniques used to achieve compliance.

The secondary winding is normally made with triple insulated wire to avoid using a bobbin with a partition in the middle of the bobbin.  The secondary in the Chinese supply appears to be wound with triple insulated wire.  The medical grade supply uses silicone isolated wire.  See here for quality wire http://www.rubadue.com/products/reinforced-3-layers-tca3-triple-insulated-wire .

The flexible tape between the layers is self vulcanizing tape that forms an isolation barrier between the primary and secondary windings.  http://en.wikipedia.org/wiki/Self-amalgamating_tape

The tape to cover the windings is normally kapton tape for high temperature transformers and cheaper materials for lower specification transformers.

The 'winding' that goes around the outside of the transformer is not a secondary or primary winding in the ordinary sense of transformers since the net flux through the winding is zero.  To be a normal winding it has to enclose one of the legs of the transformer, either the center leg or one of the outer legs (for half the flux coupling).  Since the wire is not isolated with a coating its purpose is to 'connect' to the core pieces.  The core pieces are made of ferrite material that has poor conductivity (for low eddy current losses) and you can induce voltages on the surface of the material through capacitive coupling from the windings.  That bare wire provides a discharge path for those charges.  It is particularly helpful to keeps voltage spikes on the primary from coupling capacitively to the secondary winding.

The white spacing tape is to keep the secondary winding well away from the core since it is not a proper isolating material.

On low frequency mains transformers the primary and secondary windings are effectively on two separate bobbins and the plastic provide the creepage distance.  For high frequency transformers the coupling between the primary and secondary winding is more important and designers prefer to have the secondary on top of the primary.  Sometimes the primary winding is split in two with one section wound on the interior of the bobbin followed by the secondary followed by the second half of the primary winding.

I hope his helps.
Daniel