Home | Giveaways | Rules and About
Admin | Edit
published: Sunday 23 April 2017
modified: Sunday 23 April 2017
author: not set
markup: textile

Making a bench PSU out of a cheap buck/boost module off the web

For the past few years I have thrown together all sorts of random things when I've needed a power supply:

From left to right:

The ATX power supply is an old group-regulated design that expects a 5V load. Without this load the 12V rail sits around 10 to 11V, especially under load, which tends to be a bit lower than I want. Its 5V rail is also akin to a spot welder, with little resembling over-current protection. The sharp nailbed and fake LiPo sticker makes up for some of these failings.

The little phone charger and the computer PSU are both fixed voltage, which means I rarely use them. But they do at least have functioning over-current protection.

Overall: something with an adjustable current limit would be nicer. And as of today, I need not dream anymore.

The synchronous current recycler from hell

I had been reading a few things about the LTC3780 boards that you can buy off eBay:

... in other words, almost a complete bench power supply for a fraction of the price of a normal one.

Alas, we'll also need some nice front-panel controls:

Wonderful!

Modifications for the pots

After receiving and testing my 3780 board I immediately went onto modifying it to use the new pots.

This turned out to be harder than expected. My new pots were 5Kohm max and the old ones were many hundreds of Kohm max. I presumed that the circuit would use these pots as voltage dividers, so this would not make much difference, but that didn't turn out to be true.

The pots were connected via three wires to the PCB, as shown by the three circles at the bottom of this diagram:

On the right we have the voltage coming out of the power supply and on the left we have it going back to the main LTC chip. You can see that two of the pins for the pot are commoned together, effectively turning it into a two-terminal variable resistor.

When I switched out the original multi-hundred Kohm pot for my measly 5Kohm one, the usable voltage range I had out of the board was greatly reduced. This was because of the fixed 14K resistor shown in the diagram above. After some quick calculations I decided I needed to replace it with a 120 ohm resistor instead.

Going from 14Kohm to 120ohm is a massive jump, and you should always be wary when you do this. Luckily however this 120 ohms would (almost) always be guaranteed to only have 1.25 volts across it (due to how the feedback system worked), and this would always be coming from the main power supply output, not the poor LTC chip directly. All good.

As per usual: I didn't have any replacement SMD resistors at hand, so I had to bodge a dinosaur-size 1/4W resistor in instead. It looked quite neat and pretty. Sadly I don't have any pictures of this.

Instead I have photos of my horrific failed attempts to rescue the situation after accidentally snapping this resistor off the board:

"Ooh, that resistor does not look straight. I'll just bend it a litt--" SNAP!. There goes the pad. Cue many hours of reverse engineering this board and trying to bodge replacement parts in.

And please don't mention that little purple wire. Trying to solder it to a specific pin on the LTC, without making solder bridges, was a nightmare. Those pins are spaced every 0.65mm.

Unfortunately my repairs did not work. It seems my constant attempts to re-solder a wire to the LTC killed it, and it refused to spit out much more than a volt after that without over-heating.

Let's buy another one!

I was nowhere as enthusiastic as this title would suggest. This is money I could have spent on several packets of chocolate biscuits.

Let's compare new and old:

Note that I've removed the heatsinks from the old (right) board, and attempted several other repairs (look for blobbly solder bits).

Whilst these two boards use the same layout, there are a few differences. Sadly the new one used narrow holes for the pot pins, so I wasn't able to solder 2.54mm headers in their spots like I did in the old one. And the screw-holes for mounting were slightly smaller/different.

Stupidbloodyboard.

I'm not sure why they went for a different number of turns on the main inductors. Different core material? Different parts cheaply available? It's likely a wide range of inductances will work well with this chip.

Primary power supply

This LTC3780 board can't accept mains power directly. I needed a PSU between it and mains.

Luckily this PSU can be almost anything I want. I have a stack of old laptop chargers that would fit the bill, but I decided that I wanted to try and use one of the toroidal transformers I had lying around instead. A bonus of this is that my power supply would become completely isolated, which is handy when using an oscilloscope.

Unfortunately none of the toroids I have scavenged over the years were suitable. Let's take Bertha, for instance:

She outputs a nice 12VAC, but weighs a few kilos and is absolutely massive. One day I might make a spot welder out of her several jiggaamps.

I did have a few other toroids, but they all outputted voltages that were a bit low for my liking (eg 7VAC) or too high for the board (30VAC).

Rewinding a toroid

This wasn't anywhere as bad as what I had expected. The outside layer of wires was the secondary, so I just cut it off and installed a new one.

I wound two wires together at the same time. I didn't know how much wire this would take, so I guestimated a length and cut it. This didn't end up being enough so I had to do this twice and put a couple of joints halfway around my toroid, but this didn't bother me.

Testing

It worked remarkably well:

Note the custom-made primary board with a fuse holder (bridged by my multimeter) and the secondary board with a bridge rectifier.

Boxing up

I looked through my collection and found a nice aluminium box, previously used by the power supply for a CCTV camera. This old analogue camera was fun whilst it lasted, but then something in it disintegrated into cat piss.

I was however slightly annoyed that this box already had an input fuse and switch:

Bastard. I threw out my earlier boards (including the rectifrier one because it was too big).

This is what the final insides looked like:

Mains power starts on the right, goes through the toroid in and ends up on the left. Also shown in this picture is the lid where I mounted my pots and banana jacks.

And here is the final product:

Final notes

Problems still to be addressed:

Other things not mentioned:


Tiago - Saturday 10 February 2018

Hi,

I have now to burned ltc3780...
one i think i killed it myself. the other one it came dead. Well, none of them is enterely dead, just the current adjustment doesnt work.
You know if replacing LM358 would do the trick or might something else be burnt?

thank you.

Hales - (site author) - Saturday 10 February 2018

Hey Tiago,

It completely depends on what has died on your board, I can't guess from here. Make sure you're not doing something silly first (eg turning the current pot the wrong way, using a replacement pot of the wrong value, etc). In this particular design the current pot's resistance is unfortunately extremely important.

Indu - Thursday 5 July 2018

Hi,
My LTC does not give anything, none of the LEDs light up. The white fuse on left side near the input (I think thats what it is) gets quite hot. Any ideas how to proceed?
Thanks,
Indu

Hales - (site author) - Thursday 5 July 2018

Hey Indu,

The fuse getting hot suggests something is shorted. Check for other hot parts and make sure you're not powering the things backwards.

Tip: pour a small amount of methylated spirits onto the board and then run it. The metho will evaporate fastest near the hot components.

Otherwise: there's not much advice I can give from here, these boards have dozens of parts and many things can go wrong. Probe around, look for voltages on the LTC chip itself, see if you can find anything suspicious. If you want more help from lots of people: hobbyist forums like eevblog.com are also good starting points.

Kevin - Thursday 26 July 2018

Hi Hales,

I'm considering getting one of these boards to build a battery-powered portable CV CC power supply. Is the current pot acting in the same fashion as the voltage pot? (variable resistor instead of as a pot)

The reason I ask is because I'm considering using digital potentiometer ICs to control the CV and CC using a microcontroller with a feedback loop.

Thanks,
Kevin

Kevin - Thursday 26 July 2018

Just found the answer to my question. Maybe this will be helpful to you: https://wiki.beyondlogic.org/index.php?title=WD2002SJ_LTC3780_Synchronous_Buck_Boost_Converter_Ebay_Automatic_lifting_pressure

Hales - (site author) - Thursday 26 July 2018

Hey Kevin,

Thanks for the link.

By memory: both are used as voltage dividers, but one of them has the "other half" of the voltage divider in the form of a separate resistor. So in practice one is used as a 3-pin voltage divider and the other a 2-pin variable resistor.

Big warning: I've seen a few variants of this board now. Some of them may do this differently

I don't like how the pots in this design are intrinsic parts of the feedback loops. Buffered voltage control would be better, as that means adding a capacitor to the wiper on the pots would not effect the device behavior. But this would require more parts and complexity.

Sidenote: those big 10-turn pots I'm using were cheap jobs off ebay. To begin with they were great, but now they're very nasty and dirty. I can't keep the output voltage stable, even a tiny bump of the desk takes me from 5.1V to 4.9V. This effect appears to be worst in the ranges I use the device the most (so I suspect the insides are turning to powder

Kevin - Monday 6 August 2018

Hi Hales,

Just thought I'd share what I've done regarding this converter in case you're interested.

I only wanted a max current of ~3A so I was able to get a 20K 10-turn pot not too expensively and put a 180K fixed resistor on the other end of that. This gives about a ~0.36A-3.7A range.

I used a 40K 10-turn pot I already had for the voltage adjustment, and a combination of resistors to get 1.201K for a voltage range of ~0.8V-27.6V.

I had, due to my needless-purchasing habit, a USB power monitor whose input is rated for 0-24V (when externally powered) and a current of 0-3A. I studied the circuit and figured it would be safe for ~30V and probably no more than 5A. But I'd rather not push it.

Then I threw that in a 3D-printed enclosure along with 3x 18650 batteries and a BMS.

Here are some pictures:

https://pcbx.us/cfkv.jpg

https://pcbx.us/cfkw.jpg (it has a compartment to hold the charger and some short alligator to banana plug leads)

https://pcbx.us/cfkx.jpg
https://pcbx.us/cfky.jpg (both bottom sides have rubber footies :P)

And a video:

https://www.youtube.com/watch?v=4PNyn3nbTZE

Thanks,
Kevin

Hales - (site author) - Thursday 9 August 2018

(Apologies Kevin, spotty internet access. Having a look :)

Hales - (site author) - Sunday 19 August 2018

Graphing... graphing! What is this opulence? Consider me jealous :)

Not sure how long the 3D case will hold up. Does it feel like it might shatter if you drop it?

I had a transistor tester I modified to be powered by two 18650's, but I thought it wise to revert to a standard 9V alkaline instead. 18650's are amazing batteries, especially since they cost $0 most of the time, but I needed to go overseas and I didn't feel comfortably carrying an already hollywood scary device in my luggage with the added appearance of two big unlabelled cylindrical cells. Fear is the mind killer.

On a related note: make sure to remove your 18650's _before_ attempting to desolder battery holder leads. I discovered that I didn't have to bother desoldering the other ends of the wires. Or the diode I used to join the two cells.


Add your own comment:

Name:
Email (optional):
URL (optional):
Enter the word 'irrlicht' (antispam):
Leave this box blank (antispam):

Comment (plaintext only):

If you provide an email address: it will only be used for the site admin to contact you, it will not be made public.

If you provide a URL: your name will hyperlink to it.