Beefing up a Won-Hung-Lo Boost Converter

[unit998x]

Hi All, I recently bought a DC-DC boost converter from eBay:
http://www.ebay.co.uk/itm/10A-DC-DC-600W-10-60V-to-12-80V-Boost-Converter-Step-up-Module-Power-Supply-UK-/350975904870?pt=UK_BOI_Electrical_Test_Measurement_Equipment_ET&hash=item51b7cb4c66
It claims to be able to switch 10A, however details on it are incredibly vague and although the page shows it to be UK based, its actually made in china, so I wouldn't trust it at 10A at all.

I need to make a boost converter that can switch 20A continuous 1hr+ with peak surge of 70A. The input voltage would be 22-26V and output maximum needs to be 36V. I have traced out the PCB for this 10A device and attached a schematic, I have no idea what inductance L1 is or the value of some ceramic caps as I have no LCR meter (but I have guessed some from the UC384X datasheet). For this reason I have attached some photos of the device and its heat sink (the inductor looks a lot shoddier than in it's eBay picture.

What kind of work would need to be done on this device to make it capable of the specs above?
I have so far thought of this:-

• Increase trace width on PCB
• Higher current inductor
• Replace the rectifier with a higher rating.
• Replace the caps with some of a reputable brand
• Larger input fuse

Also, would I be right in thinking that the control side of the circuitry is fine, and it is just the power side that needs beefing up?

EDIT: RV1 is the Current Adjust POT, RV2 is the Voltage Adjust POT

Thanks, if you need any more info, just ask.

[LukeW]

The current capacity of the actual switching device (eg. MOSFET, either integrated into the chip or external) obviously plays a large part in determining maximum power handling too.

It might be easier to just design and build something from scratch that meets your specs.

You mean to say you need 36V out at 20A continuous, or 720W? If that's honestly what you want you're basically wasting your time trying to hack that, you'll need to engineer something that is fit for purpose from scratch.

[mariush]

The MBR20100CTL diodes are 20A 100v ... for 20A continuous you'd need 30-40A diodes.  That's one of the first things you'd have to change.

The UC3843 is ancient, probably more than 15 years old now. There's much better driver chips out there.  Not worth the trouble imho..

[T3sl4co1l]

How about a top view photo?  Trace the circuit?  Any other identifiable or at least important-looking components besides chip resistors and capacitors?

Also, would I be right in thinking that the control side of the circuitry is fine, and it is just the power side that needs beefing up?

Not at all.  The whole thing might be shit. 

The UC3843 is ancient, probably more than 15 years old now. There's much better driver chips out there.  Not worth the trouble imho..

You must be f'n kidding... people still use the MC34063 for god's sakes.  Now there's a terrible chip.  Only complaints I have about the UC384x are, the driver output is bipolar (= weak saturation voltages; but, beefy enough for its applications) and there's no way to compensate the overall chaotic behavior of the system (which isn't so much the chip, as the fundamental concept it is based on: peak current mode operation).

Nothing wrong with UC384x in and of themselves, but you really need slope compensation and a current transformer here.  Shunt resistors and DCM aren't going to cut it, not if you want something better than, well, what's pictured above let's say.

Tim

[unit998x]

How about a top view photo?  Trace the circuit?  Any other identifiable or at least important-looking components besides chip resistors and capacitors?

Hi Tim, I attached a Schematic of the device to the original post called ShenzenConverter.pdf, hopefully I did trace it correctly, would the current transformer be used in place of R5 in this schematic then?

[T3sl4co1l]

Oh, didn't see the attachment    Yes, in place of R5, except it's usually placed in the drain instead of the source circuit.  Same idea.

Eww, LM358, I hope that's not what's actually there.  No way in hell a 358 is fast enough for a switching converter current sense, marginal at best at unity gain, no way in hell at a gain of 30!

Bizarre, they used a semi-discrete 10.83V regulator with current limiting instead of a 7812 (or similar).  It probably works worse than a 7812, and I don't see how it could be cheaper.  Why?!

Example:

http://seventransistorlabs.com/Images/Mag_Amp_PSU.png

I haven't built this, but it should work.  Main thing lacking is slope compensation, which I think for the values shown is actually a problem (it'll operate in the chaotic region -- nasty).

By the way, this was sized for 3A 30V output I think.  The snubber is already recommended, and for 70A capacity, you'll *need* not just one, but several, spread amongst the switching transistors (plural) and diodes.


If you really insist on building this thing... you'll need to clarify that "70A peak" demand.  For how long?  If microseconds, it doesn't matter -- it draws from the caps and doesn't affect anything else.  If miliseconds, enough cap can be added to accommodate it.  But more than that, and you must have the converter sized for at least that much current.  The only thing you can skimp on then is heatsinks, which isn't saying much.

Doing just 20A capacity, continuous, in just a few transistors and diodes (one or two each), is possible, but you'll have an easier time using 2-4.  You'll need some heatsinking, possibly a fan, but it won't be obnoxious.  A snubber on every other transistor is probably a good idea, so 1-2 of those.  And as many caps as you can stand, since skimping on those *will* release the magic smoke\\\\\steam.

Tim

[T3sl4co1l]

The values in the schematic make no sense. The CS trip threshold is 1V, so 1V/0.3 ohms is 3.33A which is impossible for a 600W converter. I would expect somewhere around 30A peak currents. Then they have a unknown gain block? The inductance seems excessive even takeing into account the material properties (inductance roll off). The switching frequency from the values should be about 80kHz or so. For 600W somewhere in the 4 to 12uH maybe but given the wide range in and output it could be more but I can't see 680uH.

I'm thinking he didn't measure either of them (or, measure correctly).  0.3 ohms sounds like contact resistance, and 680uH...might be a default value for the symbol?

Tim they maybe went with discrete regulator because of the wide input voltage range. Most linear regulators top out at 30V or so.

Possible.  7812 does more like 40 doesn't it?  There's also LM317HV and the like, but anything the slightest bit special sounds like dollarsigns to me...

There is nothing wrong with the UC384X series, current mode control is current mode control. Some controllers have some bells and whistles, mainly though the only drawback is its bipolar. They do make CMOS versions and they are cheap.

http://canada.newark.com/texas-instruments/ucc28c43dr/ic-current-mode-pwm-ctrl-18v-8/dp/77C2003?CMP=TREML008-004

Damn, why haven't I seen these before?

On a related subject, UC3808 is neat: think 3844 -- the one with the divider flip-flop -- with both output phases taken out to pins.  So you can do the same thing as usual, like, a single or two switch forward converter, but push-pull (or bridge) for more power / better transformer utilization / lower supply/output current ripple.  Two downsides: no Vref, and the current sense comparator has an awful tolerance on it.

Tim

[unit998x]

Thanks Tim and Achmed for looking at this.

This project would be solely for hobbyist purposes, and I don't know in great detail about SMPS to completely redesign it, so I think that I should probably just try to upgrade the components that are already in place, the MBR60100PT seems to be a suitable replacement for the rectifier and the current FET is capable of 80A continuous anyway. I have also touched up the traces on the back of the board with more solder and copper to increase the surface area as well as a higher rated inductor. At worst all that I see could go wrong is the FET/Rectifer/Traces break.

I will post the results of these upgrades, if not, eh it was only £13 and anything that blows up should be easily replaceable.

[ConnorGames]

Remember that the FET can only handle 80 amps if it has infinite cooling... Also, you might have considerable switching losses adding to the thermal problem. It will probably need an upgrade -- after all, that converter is seriously built down to a price -- do you really expect that thy used a fet that can handle twice what the circuit is rated for

[dmwahl]

I realize this is an old thread, but perhaps we can start it up again.

I've been attempting to get one of these boards to perform a bit better, mostly in regard to functional current limiting. I've replaced the R9 shunt with a 0.01ohm resistor, which allowed me to regulate the input current much better than before, but the LM358 driving the current sense seems to run out of bandwidth at around half the range of the adjustment pot, which is about what I expected given the GBW of the LM358 (0.7MHz) and the ~35kHz switching speed of the converter. I'd like to be able to use the full range of the 100k current adjustment pot, which should give a gain of ~35.

So to try to improve things a little, I swapped out the LM358 with a TL072. Offset voltage of the TL072 is slightly better and GBW is 3MHz vs 0.7MHz, so I figured it should give me back the full adjustment range. Instead I get an output voltage of the input minus a diode drop. First thought was I fried something, so I put the LM358 back on and it all worked again. Swapped the TL072 back onto the board, and noticed some interesting things. For one, the designer of the board left the second op-amp channel floating (despite the fact that they could have tied it off with just a few more traces and zero additional components ). The second channel output was actually floating a few volts above Vcc, which I don't quite know how to explain, but I don't believe is the primary cause of my troubles.

Any ideas why replacing the LM358 with a TL072 would prevent the board from working properly? The theory I have is that the board designer is relying on the slow slew rate of the LM358 (0.3V/uS) to effectively filter out the initial current spike when the switch turns on, and when I replaced it with a faster amp (13V/uS) that "filtering" was no longer present and the UC3843 shuts down right away each cycle. Any other ideas? These boards are really bare-bones, no slope compensation at all despite how cheap it would have been to add. I'll probably just end up designing my own, but wanted to play a bit with these first.

[dmwahl]

Here's a schematic of the circuit for reference:
http://cr4.globalspec.com/comment/1018737/Re-Power-Supply-DC-DC-Step-Up-Converter-Problem

[dmwahl]

Thanks for the reply, seems obvious now that I know my mistake. I realize it's only limiting input current, which is fine for me. My goal is to turn it into a cheap mppt boost controller to charge my lawnmower with. I'm going to use a digital pot to control the op-amp gain, and monitor input current and voltage to find mppt point.

[Seekonk]

I'm going to be doing this to charge my vintage boat at low speed which had a 6V generator.  I inly need 3-5A.

[dmwahl]

I'll post my results here, I made a quick PCB to hold the digital pot, current sensor, and an arduino pro mini to do the math. I'm planning on using a 30W solar panel intended for 12V batteries. Peak current is only around 2A. If it works well I'll post the PCB files somewhere so anyone else can use them.

[metallicus69]

"Any ideas why replacing the LM358 with a TL072 would prevent the board from working properly?" I think itś because LM358 is Rail to Rail and TL072 is not.
Has anyone succeeded in beefing up this circuit?