Power supply build quality pictorial. part 2

[Agent24]

Re: Power supply build quality pictorial. part 2

But it won't need a mains connection because it will run from the battery and they'll call it a "WiFi PSU"

[tom66]

Re: Power supply build quality pictorial. part 2

Why do they bother with placing the spaces for filters and caps and coils, when they almost never use them? Perhaps some customers want them - but then why would they approach L&C!! Oh, they're idiots...

[momaka]

Re: Another L&C…

Seen it before. Really funny... they actually bothered to put the missing parts in there but forgot that they changed the PCB so that they don't have to bother using wire jumpers in the units without pi filters.

Well, it's all good anyways. I'll be using those coils to rebuild some other power supplies. I have a not-so-bad ISO with very undersized caps and missing coils. Once I put that stuff in there, it would be a decent bench PSU.

Unsolder the transformer inthat last one. Often, they say ERL-35 on top and ERL-33 underneath

Yes, I saw that in a review a long time ago. I don't think I'm going to bother with that. Maybe if/when I'm pulling other parts from that PSU. But right now, that PSU is a few thousand miles away so I won't be doing that any time soon .

I think momaka's meaning is that the caps used were not safety-agency approved polypropylene caps.

Correct.

[tom66]

Re: Power supply build quality pictorial. part 2

"Replacement AC Adapter"

12V @ 5.0A

eBay: £5 including postage from Hong Kong.

How bad can it be?

Claims: FCC, CE certification. Also "exoteric interference" may cause undesired operation.

First look:
- No input fuse! This is already a CE fail.
- No line filter! No way it'll pass FCC/CE!
- No transient protection: MOV, X caps, TVS etc.
- Full EMI metal shield around PSU... fairly decent and definitely going to help with emissions
- Cheap but okay caps. 1000u x 2, reasonable for output rating.
- Input cap is a 100u 400V. No PFC, of course. Decent size for output power.
- Good size transformer for 60W output.
- Not a self-resonant converter: it actually has a controller IC!
- TL494 feedback
- Fair heatsinks for the output power
- Output common connected to earth (EMI reduction again)
- Wave soldered well. All components are properly seated. Where necessary a bit of glue is used.
- Easy to disassemble - just two screws. Why can't every manufacturer do this?

It's a shame they just didn't go the whole hog by actually putting in an EMI filter - after all the metal shield and earthed output add slightly to the cost, how much does a simple filter cost?!

I will be scoping the output of this PSU later and seeing how well it copes with just 1.5A of load. I use it to recharge my LiPoly batteries with my B606 battery charger.

[b700029]

Re: Power supply build quality pictorial. part 2

- Not a self-resonant converter: it actually has a controller IC!

The self-resonant ringing-choke-converter design isn't used much beyond ~10W, because then it becomes cheaper to use a controller since discrete control components would become more expensive.

[Scenic]

Re: Power supply build quality pictorial. part 2

That's not a proper CE mark.
The real one has the C and E spaced further apart.


If they're super close to each other, it stands for "China Export" instead, which is basically the replacement of the Hongkong Flyapart LLC logo

I'm curious about what it can actually do.. I could use a few of those 12V 5A bricks.
If they work allright near or at full load, they should be OK after adding a fuse and recapping I guess.
care to link the seller?

[tom66]

Re: Power supply build quality pictorial. part 2

The self-resonant ringing-choke-converter design isn't used much beyond ~10W, because then it becomes cheaper to use a controller since discrete control components would become more expensive.

Well, you'd be surprised... I've seen pics of 32" Sylvania LCD TVs using them for the inverter output (about 150W)... that was a three-transistor discrete oscillator.

---

Agree on CE. Looked suspect - but auction claims CE so they're lying anyway. Anyway, the FCC certification is definitely a lie, the FCC logo is legit but the certification is faked. Including the picture on the auction showing it is certified.

Here's the adapter: http://www.ebay.co.uk/itm/12V-5A-4A-...ht_2854wt_1189

£5.89 from UK... can't find it from Hong Kong any more.

Closest for load I have is some 10 ohm 5W resistors (x5), 4 of them in parallel would load it down with 4.8A but each resistor would dissipate 15W so they wouldn't last long. My load tester is on the back burner.

I do have a 3.3R 25W resistor, as long as I'm careful to not leave it on too long I could probably use that...

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

Why do they bother with placing the spaces for filters and caps and coils, when they almost never use them? Perhaps some customers want them - but then why would they approach L&C!! Oh, they're idiots...

So as to use the same PCB for multiple models/grades of P/S.

The self-resonant ringing-choke-converter design isn't used much beyond ~10W, because then it becomes cheaper to use a controller since discrete control components would become more expensive.

I haven't checked Digikey, but is a TO-92 package TL431 more expensive than a 16-pin DIP TL494 (or TL494 clone)? I think you also need more associated components to use a TL494 than a TL431. I suspect there are more "cookbook" circuits "out there" for the TL494 than for the TL431 as a flyback controller.

[tom66]

Re: Power supply build quality pictorial. part 2

The TL431 is just a voltage shunt reference. The TL494 is a secondary side device; typically a TL431 will be used on the secondary but it will communicate with the primary controller.

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

A TL431 is a precision reference with an amplifier that allows adjusting the reference voltage. OR you can connect it so that the TL431 is a comparator with a precision reference at one amplifier input (which it is is the TL431) and a divided sample of the P/S O/P voltage you want to regulate at the other amplifier input. That comparator then can control the "On" time of a self-oscillating flyback. By controlling the "On" time, you control the amount of energy stored in the transformer core gap and the duty cycle of the oscillator. Boschert Inc. (bought by Computer Products, which later merged with Zytec to form Artesyn) had a whole series of flyback models, ranging from 40W to 165W (185W with fan cooling) which used the TL431 exactly as I described it. I worked at Boschert when that series was in its final stages of engineering development.

I also used TL431s in a series of window detector comparator circuits that could detect whether the O/Ps of an ATX P/S were in regulation. Each O/P had a pair of TL431s and an LED (as well as some load resistors so a functional P/S would be in regulation). The amplifier in a TL431 is an amplifier, not a comparator, but as long as your design takes into account the fact the amplifier output won't go to the low rail, it works rather well as a comparator.

[tom66]

Re: Power supply build quality pictorial. part 2

I think I have a Boschert power supply in my HP 54501A oscilloscope. By the way, it had a cold joint on the power supply that was causing power cycling. It was some kind of self oscillating flyback using a power NPN.

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

ICs with building blocks inside - e.g. the 555 timer or the TL431 adjustable reference or the uA723 linear regulator - that the user can configure in multiple ways can have multiple uses.

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

The Boschert models from that series would have included XL45, XL5x, XL75, XL100, XL125 and XL165. They would have been sold in the 1980s and possibly into the 1990s. Their efficiency wasn't fantastic, by current standards, and they used bipolar junction transistors (Motorola MJE13007 and MJE13009) rather than MOSFETs. The engineer for the design was utterly brilliant, but his answers to simple questions, while correct, bordered on unintelligible - serious TMI!

[tom66]

Re: Power supply build quality pictorial. part 2

The Boschert models from that series would have included XL45, XL5x, XL75, XL100, XL125 and XL165. They would have been sold in the 1980s and possibly into the 1990s. Their efficiency wasn't fantastic, by current standards, and they used bipolar junction transistors (Motorola MJE13007 and MJE13009) rather than MOSFETs. The engineer for the design was utterly brilliant, but his answers to simple questions, while correct, bordered on unintelligible - serious TMI!

Yes, this is a ~165W output supply, but the rated input is 250W!

The design is fantastic... and the output caps! So... many...! Noise is sub 1mVp-p on most rails, which is pretty amazing. About 16,000uF on the 3.3V rail, 8,000uF on the 12V. (Just for CRT and fan.)

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

Wow! What a tank! Double-sided G10/FR4 PCB, all UCC O/P caps, lots of heatsinks. Boschert used yellow silkscreen for reference designators and such. For this context, beautiful. The actual input power will be less than 250W, but that calculates to 66% efficiency. With real conditions and in the UK, I'd guesstimate the actual efficiency to be between 70% and 75%. That is a flyback design - no large powdered iron O/P inductor(s). Discontinuous flybacks are very stressful for O/P caps, so the large numbers are to get the ESR way down, which means low ripple. Since this is in a test instrument, HP may have specified extremely low noise so the ripple wouldn't screw up sensitive measurements.

[mariushm]

Re: Power supply build quality pictorial. part 2

"Replacement AC Adapter"

12V @ 5.0A

[...]
- No input fuse! This is already a CE fail.
[..]
- Full EMI metal shield around PSU... fairly decent and definitely going to help with emissions
- Cheap but okay caps. 1000u x 2, reasonable for output rating.
- Input cap is a 100u 400V. No PFC, of course. Decent size for output power.

Isn't there a fuse by the input connector, under the heatshrink?

PFC as far as I know is not required for power output below 75w or somewhere around that value (IEC555).

For 6 pounds, it seems like a pretty good psu.

[tom66]

Re: Power supply build quality pictorial. part 2

Isn't there a fuse by the input connector, under the heatshrink?

PFC as far as I know is not required for power output below 75w or somewhere around that value (IEC555).

For 6 pounds, it seems like a pretty good psu.

Nope, no fuse... I think that's a Y-cap (just one?!) The AC L and N go straight to the bridge rectifier. (I forgot to mention it has a real bridge rectifier.)

Yeah, I know about PFC, I thought the limit was closer to 90W, but that applies after losses in the supply.

I agree, it's fairly decent especially for the price. Much better than most of the cruddy supplies that come out of China.

[tom66]

Re: Power supply build quality pictorial. part 2

I'm mistaken. It is a fuse. I was looking at the PCB wrong.

Some tests of the output ripple:

At little to no load (just the LED or my charger paused) the ripple consists of a high frequency ringing, but overall quite low amplitude. If I turn the 20 MHz bandwidth limiter on my scope on, then the ripple appears smaller, but the ringing is still there.

With a 1.5A charging load (approx) the ripple is aroundabout 200mVp-p. It's not going to win any awards, but it's acceptable. My charger has a very poorly implemented software controlled buck/boost converter, so it's not entirely surprising that I saw that kind of ripple. I'll try it with some resistors later.

Frequency is more or less constant, around 33kHz.

[tom66]

Re: Power supply build quality pictorial. part 2

Wow! What a tank! Double-sided G10/FR4 PCB, all UCC O/P caps, lots of heatsinks. Boschert used yellow silkscreen for reference designators and such. For this context, beautiful. The actual input power will be less than 250W, but that calculates to 66% efficiency. With real conditions and in the UK, I'd guesstimate the actual efficiency to be between 70% and 75%. That is a flyback design - no large powdered iron O/P inductor(s). Discontinuous flybacks are very stressful for O/P caps, so the large numbers are to get the ESR way down, which means low ripple. Since this is in a test instrument, HP may have specified extremely low noise so the ripple wouldn't screw up sensitive measurements.

The input caps are Rubycons, too. The transformer is massive. Indeed, the ripple is low enough that the +/-5.2V rails from the PSU go straight to the buffer amps for the four channels. The +12V goes to the CRT, fan and beeper, not much else. I'm not sure what uses 170W on this model... there are a few chips with small heatsinks but that's it... the CRT is a monochrome green colour, nothing that would seem demanding of power.

After owning the scope for a year (it is now 19 years old) it started to power cycle. Would just reset in the middle of doing something. Or it would not turn on properly. I reseated the connector, that fixed it for about a year, but then it gave problems again.

I traced the problem down to a cold solder joint on the -12V output connector. I'm guessing the CPU relied on this in some way because it's an ancient Motorola 68K, or it might have just been a low voltage lockout.

[tom66]

Re: Power supply build quality pictorial. part 2

With a 10 ohm load (~1.2A), ripple is about the same - around 200mVp-p. I may end up replacing the cap at a later date, but my charger doesn't seem to mind so I'll leave it for now.

Very strange ripple waveform, almost like it's overshooting and then settling fine... I wonder if the control loop is properly compensated? It would be a shame as the rest is fairly well designed.