Ebay (Chinese) boost converter blown up. Uncontrolled output voltage??

[Chris Wilson]

I was running an Ebay sourced, Chinese made boost converter similar to this one:
https://www.ebay.co.uk/itm/224127945383?hash=item342f110aa7:g:ICEAAOSwF9FfQL30

It had 13.8V on the input and I initially had the output adjusted to 20V. It powered a 2.4GHz 20W RF power amplifier for amateur TV transmission. The PA, this one:

https://www.sg-lab.com/AMP2400/amp2400.html

can run up to 28V voltage input and has a maximum current draw of 1.5 amps. Having made my first ever transmission of video through the Es Hail amateur TV satellite today I was very pleased with my carefully made home brew set up, but I needed a bit more power, so I turned the boost converter output up to 26.5V. I made a test transmission, slowly increasing the drive to the amplifier, and had not even got to an output level as high as that given reliably on 20V when the transmission ended and DC power to outside stopped.

I checked the waterproof cabinet by the satellite dish outside where most of the gear lives (it keeps lossy coax runs at 2.4 GHz to a minimum, having the amp by the dish) and on opening it smelt a pungent odour and found the output 35V rated electrolytic on the boost converter had done a proper exploding job. The input to the converter was a dead short, and had blown the linear power supply 7.5 amp line output fuse in the house that feeds 13.8 V to the outside cabinet.

Getting worried now I metered the (expensive) power amp input voltage pin and found it also dead shorted. My elation at my first successful transmission of video via the satellite abruptly ended.

All this leads to the question as to whether such boost converters are liable to go bang with uncontrolled output voltages, and if so, are there better ones available, able to give say 25V out from 13V in, at say 2.5 amps current, continuously? Is there any way to clamp the output voltage of one, so if another one goes berserk it protects whatever it's powering, even if it dies a terrible death itself in the process?

Many thanks if you have got this far and can advise!

I am now looking to see if the amp is fixable without sending it back to Bulgaria. They are a great company but always mad busy, so I fear a long wait if it needs to go back, and they are willing to fix it. Nothing dramatic looks to have happpened inside the amp, so I have some small hope there...

[strawberry]

zener or TVS diode
https://en.wikipedia.org/wiki/Crowbar_(circuit)

diy boost converter LM2577

[Chris Wilson]

Thank you strawberry, appreciated. I looked in trepidation at the amp tonight, and my attention was drawn to what looked like a slightly bulged input electrolytic. 35V at 47uF. I un-soldered one leg and the short was gone. I then did a sanity check and removed the cap and checked it, itself. It was dead shorted.

I am wondering if the el cheapo boost converter was giving out a load of ripple current? Blowing its own, and the amp's capacitor. The amp cap is the 6.25mm round can one near the input +ve voltage pin. The 5V regulator still regulates to 5V and the main PA transistor isn't shorted. I powered the amp without the input cap and it seemed happy and drew modest quiescent current. Without the cap I don't want to put RF through it just in case it decides to oscillate. I have ordered some very low ESR, high ripple current Nichicon caps for the amp.

I will look into building my own HD boost converter, I did once make a seemingly successful SMPS board, to see how they worked. I will put a crowbar and fuse on the output of whatever I use next though, thanks again. Any other experiences of these boost converters and their reliability welcome.

[strawberry]

made of factory reject components, opened feedback loop, feedback loop design error(they copy each other including errors)
have repaired a lot of dc-ac inverters even brand new

[NiHaoMike]

Take a look at laptop power supplies for something that can supply a lot of power at 20V or above.

[Chris Wilson]

Thanks, but I need to UP convert from circa 13V DC to circa 28V DC, I have no mains where the amplifier is.

[T3sl4co1l]

I mean yeah, even the "better" looking ones I guess, aren't rated anywhere near what they're sold as.  Even this promotional one: https://youtu.be/WVz0jfgIqyM?t=398

Boost is all ripple at the output, and usually not much less at the input.  It takes big caps to handle that.  Aluminum polymer can be a good idea.  Phase interleave gets attractive at fairly modest power levels, at these voltages (though that's quite a bit more than what you were doing there -- "modest" as in hundreds of watts).

Like this compact 200W converter I made,
https://www.seventransistorlabs.com/Images/CukHack1.jpg
The capacitors handled intermittent use okay, for a while, but eventually...
https://www.seventransistorlabs.com/Images/CukHack4.jpg
...one of the leads fused open!  The subsequent uncontrolled voltages blew up a couple things in the circuit.  (This is a Cuk converter, with inverting voltage, and triangular current ripple at both input and output.  The inductor is about as big as can be afforded, but nowhere near enough to get the ripple low enough for this to be safe in continuous duty.)
https://www.seventransistorlabs.com/Images/CukHack6.jpg
I later rebuilt it in a slightly larger box, providing enough room for doubled up filter caps, and also some ceramics in there.  The leads are shorter, and also input and output come together on the the same board so EMI should actually be pretty modest (I haven't measured it, no point it's just a handy converter).

Might be able to do something similar with those modules, but there's so much to go wrong it's hard to say.  Improve the caps, improve the heatsinking, replace other underrated components -- after all of which you're better off buying a name brand module for $50, so... I'd recommend starting there.  (Downside is they may not be available in wide voltage ranges.  Some have a narrow say 10% trimming range, which is something, but not as wide as you were looking for.)

BTW, if the ripple was THAT bad... was it not obvious on the video output?  The modulation must've been terrifying... maybe not so bad class A/B, but..?

Tim

[NiHaoMike]

Thanks, but I need to UP convert from circa 13V DC to circa 28V DC, I have no mains where the amplifier is.

There are DC input laptop power supplies.

[amyk]

Thanks, but I need to UP convert from circa 13V DC to circa 28V DC, I have no mains where the amplifier is.

There are DC input laptop power supplies.

I don't know of any laptops that need a 28V input. 20-21V is usually as high as they go, given that they have 25V caps in the power circuitry.

[Chris Wilson]

Thanks for the helpful replies. I have the jitters now, so am torn between running an armoured mains cable to the dish and box, or buying a decent, branded step up converter.  Mains to the box adds the option of a serious amp later, that will need about 48V at 30 amps or so. Mains to the box and Meanwell SMPS units is looking favourite right now. Where's me spade... ?

Thanks again!

[Ian.M]

Take the cheap Chinese boost converter, replace the electrolytics with good quality low ESR high ripple current 105 deg C ones fitted on the other side of the board so they aren't being cooked by the heatsinks, rip out the dodgy Bochen 3296 Guosheng multi-turn preset (a Bourns 3296 clone), and fit precision fixed resistors  to set the output voltage, and go over any dodgy soldering and, derated for current to 50%, it should be as reliable as any other boost converter.

Equipment protection is another issue.  The classic TL431 crowbar circuit is probably the best option here, but crowbarring the boost converter input after a fuse, based on its output voltage, rather than simply crowbarring the output, would be advisable to ensure the fuse blows quickly.  The output voltage cant rise any higher if no more energy is available at the input.

Note that the UC3843 controller on your Chinese boost converter has a min. UVLO threshold of only 7V so you'll have to derate it additionally to prevent input current overload during any brownouts of your 13.8V suppl  or add more closely calibrated external UVLO protection, if brownouts other than transiently during powerup/powerdown are likely to be an issue.

[schmitt trigger]

The MC3423 crowbar sensing circuit also includes an adjustable minimum duration of the over voltage event to avoid false triggering.

The circuit is unfortunately obsolete now, but you may still find inventory.

[Ian.M]

Another option would be to switch over the 13.8V 7.5A DC feed to the outside box to low voltage line frequency AC, rectified and regulated to 13.8V in the box.  If you've got low voltage AC to the box, any oddball equipment is easily accommodated with a mains transformer of 10% to 20% lower secondary voltage than your low voltage AC supply, wired backwards to provide an isolated low current mains supply suitable for powering the equipment's OEM mains adaptor.

[Chris Wilson]

Again, thanks for the replies. in the interim I put into service an Ebay SMPS that gives an adjustable 18 to 28V  output at an (optimistic) 17 Amps. I also used a Meang Well  (yes, really, these lot are real comedians..) which I also have had for ages, unused, from Ebay. The 28V one lasted a few days before it went bang with a split output capacitor. No damage to my expensive amp, thankfully this time.

Now well fed up having laid armoured mains  cable to my outside box I bought two reputable and branded SMPS from RS Components and have enjoyed RF quiet, cool running reliability since.

In the meantime as a learning exercise I have stripped down the broken 28V SMPS and recapped ALL the electrolytics with Panasonic ones, replaced both output transistors (J13009), one having blown, the dual diode (TJ13AC) which had also blown,  and tested all other transistors, diodes and Zener diodes lifted from the board. These latter devices all checked good using Peak testers.

After all this I hoped the damned thing would work. It does not....

I removed the PWM IC, a TL494 with the intention of building a tester for it. Before its removal I noted a low level buzz from the supply and after only 20 seconds both output transistors became warm. The Vcc and ground pins on the TL484 showed only about 250mV on them. After removing the TL494 it still hums and the output transistors still get warm with no load.

There is zero output voltage in all cases. Without the TL494 fitted why might it hum and the J13009 output pair still warm up rapidly please?

Thanks.

[Zero999]

I don't know. It's difficult to tell, without a schematic. Did you check the J13009s again? Perhaps running it with the faulty TL494 damaged them?

[Chris Wilson]

Thanks Zero999, the circuit is extremely similar to this, a smaller cousin I guess:

http://danyk.cz/reverz45_en.html