27 September 2011

Maplin N27GG - 80W Switching Bench Supply - Teardown!

I've been using this bench supply for about a year now and I absolutely love it! It's slim, easy to use and seems very well built. It boasts a maximum power output of 80W with three voltage range options. The interface is simple with rotary encoders for setting voltage and current as well as a separate button for applying output. Let's skip the review and get right down to seeing what's inside!!




First looking around the unit, there are seven black machine screws that hold the grey cover on. They all it in metal inserts or hold straight into the metal chassis. None of these self tapping screws.


Initially the inside looks quite spacious yet well designed. The soldering is well done, no cold solder joints and everything seem well secured and appropriately assembled.

There are three boards in here. The SMPS, some kind of controller board and also a front panel display/button board.


This little wire immediately caught my attention... a last minute fix that not only ruins the look of this board, but also makes this board very delicate to lift up.

On the board there is an ATMEGA48V as well as some what I presume could be op-amps but can't seem to find any datasheets for the parts.


Not an awful lot on the other side of the board, more op-amps and some pots.


There really is a mess of cables under this board. 5 - 6 wire headers and even a ribbon cable! So much so that they have had to cram them under the board and cable tie them to some of the board stand offs. Slotting headers or thin ribbon cables would have been more worthwhile not to mention easier to manufacture. There are around 30 wires running between all three boards, some with headers, some without... a lot of inconsistency in this design.

These boards all seem to have been well revised, the above image shows the main SMPS board to be Rev. 5 and the front panel board to be Rev. 3. Why do they still have such a mess in the interconnections?...

Above can be seen some floating voltage regulators in TO220 packages only secured to the board by their own leads, a big problem if they vibrated loose...


Again showing the mess of wires. Granted they have used a ferrite ring on the main output/sense feedback as well as extra insulation on the mains cabling, they've still cut corners on a lot of the simple minor details.


This is the front panel that holds two rotary encoders, some 7seg displays and various buttons and LEDs


This is the back of that front panel, again an Atmel ATMEGA can be seen. Note now the ribbon cable connector above is well placed, yet there are wires under it that go to the other boards. It looks as if there was due to be a pin header placed there but they've got rid of that and just soldered the wires on, leaving some holes empty. This board is held to the plastic housing shown below with four self tapping screws.


The mains switch (thank goodness it is an actual mains switch and not one of those standby switches) and output sockets are shown well attached with glue and resin. They've even gone through the effort of making a PCB for the output sockets rather than using wires soldered onto some screw tabs.

The mains side of the system is well laid out, if a little widely spaced. There is though a lack of glue or resin holding the tall free standing components to the board. They've seemed only to have added it to the transformers, presumably to aid soldering.


A similar story on the low voltage side of the board, again they've got a lot of free standing capacitors and resistors when I'd have expected them to be glued down to stop vibration damaging their joints..

The designers have seemed to have used specially made heat sinks for all the power devices apart from those floating voltage regulators...

I don't understand why so little regard has been given in areas such as reducing risk of failure from vibration when the designers have put so much effort into a relatively well designed system.

Ohh well, this makes me feel like making a version of my own. Perhaps a future project where I can build on their ideas of a nice and compact switching bench supply.

Lets put it back together and see if it still works okay.


No problems at all!

15 comments:

  1. Hi,
    My unit just failed this morning. I checked the plug fuse and internal fuse. Both okay.
    I was charging 10 AA batteries in a pack at 14.1V with 150mA input. Have been doing this for a year with no problems.
    Any suggestions where to start with repair?
    Thanks
    Brian.

    ReplyDelete
  2. Bit of a tricky one as the inside is such a mess of cables.

    I'd start out by checking if anything has gone up in smoke inside and depending what it is, that could point out what went wrong.

    Otherwise I think the best place to start would be to check the voltages coming out the main supply. (Very very carefully as there are high DC voltages!!)

    Does the display show anything?

    ReplyDelete
  3. Hi Raj,
    The Display is totally dead. There is nothing obvious internally. I checked continuity between power in and the circuit board. I'm guessing a component failure but,it is beyond my ability to determine which.
    I was hoping that someone might have had a similar problem that was repairable with basic skills.

    ReplyDelete
  4. Switching supplies are hard to fault find as especially the cheap non branded ones are seen as cheaper to replace than fix if they break and therefore have no repair documentation. The thing with switching supplies is that if something has gone horribly wrong, there will be a catastrophic failure of many components, not just one, think bangs and smoke...

    This leads me to believe something else could be wrong maybe with the control circuitry. Perhapse a loose cable? Give them all a wiggle and see if it works again, maybe you have bad continuity in your power cord? Check the mains voltage going into the unit. (Carefully of course, don't go round probing with bare hands on mains voltages, use correct equipment)

    If there's still a problem, I'd say it would be better to give the supply to someone who knows about switching supplies as they are tricky to debug not to mention very dangerous if you don't know what's going on in there.

    These supplies, especially the cheap no brand ones can hold over 300VDC in the capacitors even when you remove the power cord. So if you don't know what you're doing, it's best not to poke around in there without studying about SMPS technologies first.

    ReplyDelete
  5. Thanks Raj,
    I think I'll take your advise and bring it to someone who might understand circuitry better than me.
    I did check all the basic things like power lead, internal fuse and general connections etc... All okay.
    It is disappointing as it worked great for just over a year.
    I let you know if it gets repaired.
    Thanks again.
    Brian

    ReplyDelete
  6. Did you have a userguide/manual with your unit

    ReplyDelete
    Replies
    1. Yes I did but I don't think I have it any more. It contained some specs and fairly basic overview on the functions. Not more than a few pages.

      Delete
  7. I've had two of these Maplin/Manson PSUs go up in smoke (yes with a bang). Charging batteries over 12V is something they don't like. There's no back-feed protection on the outputs and when the output is "off" (and even when the AC is off) the output terminals are still connected to the output capacitors (and maybe more). If you connect a high current battery of more than 12V (mine was a 400Ah 24V battery), you get a massive back-feed impulse current to charge the caps and something is damaged. Then when you turn on the PSU and enable its output (with the soft button), the PSU self-destructs with some electrolytic caps boiling up and then exploding inside.

    You can prevent this problem by inserting a series 6A diode in the positive lead output so that the battery can't back-feed the PSU but then the Voltage at the battery is off by a diode drop (unless you use the remote sense wires to compensate for the diode drop).

    ReplyDelete
    Replies
    1. Thanks for the insight. I've personally never had this problem as I've never needed to charge cells above 12V. The use of a diode may be a good permanent safety solution. Looking at the above picture of the output terminals, you can see there are two sets of wires going to the output. The thin wire must be the internal feedback line. Putting a diode in series with the much thicker power wire next to it may solve the problem all together.

      Delete
    2. Hmmm... Number 3 went KABOOM! yesterday.

      This time I was charging a 12V lead acid battery and was using the remote sense wires to ensure the charge Voltage was right at the battery terminals. I made the fatal mistake of disconnecting the positive power lead from the battery before turning the output off or disconnecting the sense wire. The PSU immediately detonated in a series of explosions and a cloud of smoke.

      If you have one of these PSUs with the hardware version number "R1.0" on the serial number sticker (mine have all been version "R1.0"), then return it to Maplins for a refund or be VERY CAREFUL how you use this thing. It's literally a time-bomb.

      Yep, the thin wires are indeed the feedback sense point. I tried adding a power diode to the positive lead at the point where the wires emerged from the main PCB. This put the diode before the sense point on the front panel. It caused the PSU to become unstable with it unable to hold a set Voltage (it oscillated slowly around the set point by +- 0.1V). I don't think it would help the failure mode where the problem is probably that the sense input is still connected to the external battery (unprotected by the new diode) and so can be back-feed damaged by a high Voltage battery.

      I measured the reverse impedance of the output with the diode in circuit. When you connect an Ohm meter to the output, there is a current spike (it beeped continuity for a fraction of a second) that decays and then the Ohm meter reads about 35kOhms after a few seconds. There is a small capacitor in the sense input that has no current limiting resistor, so it charges with probably a few dozen Amps when connected to a big battery. This is what fries the sense input and then the feedback control on the SMPS PWM controller goes AWOL and the rest of the PSU goes thermo-nuclear in seconds. Quite why a Voltage sense input has such a low impedance input that is susceptible to over-current damage is a mystery of Chinese design.

      Probably bad software on the ATMEGA PWM controller too. Remote Voltage sense compensation should be limited to just compensating for the expected Voltage drop of the wires (no more than a couple of Volts max.). It shouldn't be able to cause the negative feedback loop to go out of range or even create positive feedback. Just printing a note in the back of the user guide that says, "don't disconnect the main wire before the sense wire" isn't a solution... If the sense input value is out of range it should be ignored or limited so that at worst it causes the wrong output Voltage. Better still, it should cause the output to shut down.

      Delete
  8. Ouch! I think my version could very well be the same revision as yours as it has the R1.0 sticker on it as well. I see what you mean about the sense line. An idea could be to use some sort of clamping diode arrangement between the sense and output so that they don't stray apart too much? Perhaps there could be a way to just put a diode on the external output and recalibrate the supply to compensate for it. Maybe something that could be adjusted with the various internal pots. I don't know how linear the voltage range will be afterwards though. Could just mess it up totally.

    I guess it's an important lesson to avoid these no brand supplies. Shame really as it's quite a nice compact unit that usually gives you a good bang for your buck..

    I'm hoping to put together my own bench supply soon and ditch this one. For me it's become a little noisy for sensitive circuits. I have a few high gain op amp circuits that really don't like this supply and just oscillate when powered. The same circuits remain stable and work fine when used with regulated supplies. My idea is to have an adjustable switched mode rail with constant voltage/current modes as well as a few low power regulated rails for your standard 3.3V, 5V, -5V, ect. Whole thing will run off a small form factor 12V ATX supply. Hoping to get the same if not a smaller bench footprint as this maplin supply.

    ReplyDelete
    Replies
    1. LoL... I'm getting plenty of BANG for my bucks.

      Linear PSUs with huge transformers are much less noisy (at least no RF noise) and more robust. I've got another very old Manson (Maplins) PSU that they still make today. It's a 16V 25A jobbie that weighs a ton but I've used it to charge batteries and even jump start cars. It's just not very efficient (<60%) and has a noisy fan.

      SMPSUs are prone to RF noise on the output (up in the 20-100kHz range especially) and can suffer from entrainment. This is where an oscillating load at a frequency that is close to the switching frequency can cause the SMPSU oscillator to entrain (resonate) with the load with unexpected Voltage surges and the like. Some SMPSUs sold for radio hams even have a tuner knob to dial out the RF noise if it happens to clash with the transmission frequency on the radio set by shifting the PSU switching frequency a bit.

      The output of this Maplins SMPSU has the power wires go through a big ferrite toroid to try and damp the residual ringing noise but it's possibly not enough, especially as they had an "oh f&ck" moment when assembling the unit. The main board has silk screens and holes for much bigger first line output caps after the power diode and then they realised that the brain board wouldn't fit over it on the stand-offs because the caps were too tall. So they just drilled some more holes in the PCB lands and jammed smaller caps in there :D You gotta laugh at that. Probably a cock-up in the box manufacturing. The PCB designer used taller stand-offs between boards but then it wouldn't fit in nice slim box someone else designed.

      On your op-amp circuits you may want to boost the local AC decoupling at the chips with small caps very close to them to prevent this kind of ill behaviour with PSUs. If the amps don't need high gain above a certain frequency, you could also roll off the gain with a first order low pass RC feedback network so that it doesn't try to amplify frequencies you aren't interested in that provoke oscillation.

      Delete
    2. I should have seen the pun coming :P

      That's the thing with some of these cheap Chinese products, they are probably rushed in development. Wouldn't be surprised if the board is an early revision that they ordered loads of then bodged just enough to get them to work first time.

      The chip I'm using is a DTMF decoder for part of a university project. It has an internal amp that I'm using to amplify a microphone input. I've tried an external op-amp and it worked a charm with plenty of gain. This built in one however struggles to do the same thing without going unstable. I've tried decoupling it but the noise from this supply is still getting through though not as much. With a linear regulator, which the final version will include anyway for other system components, I should be fine. The device has a bunch of internal band pass filters anyway to decode the signal so that helps too.

      Delete
  9. Do u have a user manual for this power supply

    ReplyDelete
  10. Working away on a 5V project, on /off switch it can spike to 36V max output. Now if it was just this one I would say okay but another one I use at work is also the same. This PSU is deadly for low voltage projects when it fails and took out my programmer connected too.

    The actual indicator of the 3 modes is not illuminated when this happens, just the power one and the PSU outputs its max voltage frying any circuitry.

    ReplyDelete