Sunday 10 November 2019

Why everyone is building their bench power supplies incorrectly (fix = cut the thin black wire)!

I ordered some  cheap Volt\Amp meters from Banggood and also a DC-DC buck-boost LTC3780 converter board. Many people have made YouTube videos and written Instructables about how to make a variable PSU from this combination (plus a DC source such as 12V from an ATX power supply) but nearly all of them are WRONG - but the solution is simple!

Usually described as "LTC3780 Automatic lifting pressure constant voltage step up step down 10A 130W"
Digital VA meter (note they seem to only have one decimal place for voltage, not two as shown in this stock picture!)

Note: This type of VA meter has thick Black+Red wire and thin Black+Red+Yellow wires. Do not confuse it with the other types which may have a thick blue wire or a thick yellow wire!

Many people who have built a power supply using these two modules have complained that the over-current pot on the Buck\Boost board did not seem to work. Other people have complained that the Ammeter reading was incorrect. A few have blown up their boards.

The diagram below shows the typical wiring arrangement that is used by most of these designs, but I have added in some extra purple lines to indicate the 0V traces which are already inside the DC-DC Buck\Boost board and already inside the VA meter.
 If you add the rotary pots you must remove the two blue 30-turn trim pots.
The purple lines indicate INTERNALLY CONNECTED circuitry which is always present.

So can you see the problem?

Problem #1

The two purple lines inside the Buck\Boost converter board show the 0V tracks on this board. Notice that the 0V line from the 24V 5A PSU on the left goes through a 7 mOhm shunt to the 0V terminal on the PCB. The PCB circuit measures the voltage drop across the shunt and the centre pot is used to set a maximum current limit.

However, note that the thin black wire connects together the 0V input AND the 0V output of the Buck\Boost converter because of the internal connection in the VA meter. This means that the Buck\Boost shunt has been shorted out (though the thin black wire will have a small resistance too, so it is not a 0R short).

Problem #2

The purple line next to the VA meter in the diagram above indicates an internal connection and the purple resistor indicates an internal 0.023Ohm shunt which is inside the VA meter. The VA meter will interpret the current used by the load from the voltage across the internal shunt, but the internal shunt is connected to the 0V rail of the 24V power supply, so when a load is added, it could be measuring the current used by the Buck\Boost PCB as well as the load.


I powered this circuit using my new PSU to supply the input voltage (see a previous blog). The thin black wire to the VA meter is connected.
Thin black AV meter wire is connected (bottom-right of PCB next to the blue-tack).
The rusty pliers + elastic band makes a handy vice (very useful to hold components when soldering)!
I connected a DVM (set to measure Amps) and a load resistor of 30 Ohms across the Positive and Negative 'posts' and adjusted the voltage of the Buck\Boost converter to give 30 Volts at its output. The DVM display showed 1 Amp as expected (I=V/R = 30V/30 Ohms=1Amp).

However the reading on the VA meter was showing 1.53 Amps!

I also noticed that the centre over-current  pot had very little effect (if any).

Next, I disconnected the thin black wire going into the VA meter, and it immediately displayed 1.0 Amps. Also, the over-current pot on the Buck\Boost board now worked correctly...
Thin black wire is now disconnected - and now we get the correct reading!


If you look at the Banggood product listing, you will see this diagram if you scroll down:

Notice that, unless you are using two independent power sources (with separate floating 0V such as a battery) then you do NOT connect the thin black wire. i.e. If your project uses a common ground then do not connect the thin black wire.

The LTC3780 Boost/Buck module

After playing with this module for a few days, I cannot recommend the type pictured below. It is unstable at certain voltages and under certain constant-current conditions, it can be unpredictable and draw heavier currents than it should, it can overheat and the capacitors appear to be of poor quality and under-rated (they can actually get hot!). Watch this YouTube video to see how it performs.

I am currently testing other modules...

Extending the pots on the LTC3780 board

Many people have attempted to use this in a bench power supply and have extended the adjustment pots so that they can be mounted on the front panel of the PSU.
Diagram courtesy of BeyondLogic.
Note the three holes for an On-Off switch at the top-right.

You can unsolder a pot by turning the board upside-down and using a soldering iron on all three pins at the same time - it should just drop out.

You can get an idea of how these pots are connected onto the PCB from this excellent page here. If you have broken the PCB through-hole plating, you can solder on small trace wires to replace the missing/broken tracks.

The 500K Voltage pot is used to control the output voltage according to the formula: 0.8V * (1 + VR/14K):

VR (K)   Output (V)
500           29.37
450           26.51
400           23.66
350           20.80
300           17.94
250           15.09
200           12.23
150           9.37
100           6.51
50             3.66
0               0.80

Note: if your board will not go up to the full 30V, it may be because your 500K pots are actually 460K pots (as I found with one board I bought on eBay!).
10-turn locking knob
For the output Voltage control, 10-turn 500K pots are hard to find and expensive and then you may also fit a multi-turn knob. An alternative is to wire another 10K pot in series for a fine tune control (approx +0 to +0.6V). I prefer having a coarse and fine control because it is much quicker to adjust the output from min. to max. Note that the total possible resistance should not exceed 500K because too high an output voltage can damage the components on the board.

Using a higher value than 500K can cause an over-voltage on the capacitors which are only rated to 35V. If you exceed the design  limits (30V) the capacitors can be damaged. I would suggest that unless you really need 30V,  then 26V max would be a better maximum voltage (wire a 2M resistor across the 500K pot and add a 10K pot in series for a fine control).

Electrical noise
Note that the wires going to the pots will pick up noise from your hands and the switch mode PSU, etc. because 500K and 200K is quite a high impedance and is easily affected by radiated EMI. I found that the normal current used by the unmodified DC-DC board was approx 20mA with no load, but when I removed the Volt-adjust 500K pot and connected three wires to an external 500K pot, the board drew over 100mA with no load and when I put my hand near the wires, it drew 1A with no load! I replaced the wires with a shielded cable (I used on old USB cable) and connected one end of the braid to 0V\Gnd - this cured the problem.

Be careful not to touch any of the pot wires whilst it is on, because this can inject lots of noise into the circuit (you may hear it buzz!) and can destroy it.

You could replace the 500K pot with a 100k one (a multiturn one if you like), but to keep the voltage divider in the right range replace the 14k SMD (1402 0805 near shunt resistor) resistor with a 2k8 one. By calculation, this should yield a maximum output voltage of 29.3V, which turned out to be the case. This would improve the noise immunity.

The 500K low-voltage input cutoff pot adjustment range is 2.5V-37V. This is useful if you are running the converter board from a battery and you don't want to discharge the source battery too much. For a mains operated supply, this pot is not really useful and so can be set to a few volts above the expected output voltage from your source PSU or just wind it right up (anticlockwise).

The Current limit 200K pot range is approx. 160mA up to 13A. 13A at 30V is 390W which is too high for the VA meter and too much Wattage for the LTC3780 board which will probably overheat and melt or blow up without extra heatsinks and cooling! I have yet to experiment with resistor changes here, but a range of 0-6A would probably be safer which means using a 50K pot. To get below the 160mA lower limit, I may need to adjust one of the resistors (R6 1K) on the board (perhaps add another 1K resistor in parallel with it to get 80mA - TBC)? This pot is connected to a 1.6M resistor on the board and so the current limit pot wires should also be connected using a well shielded cable to avoid picking up noise. This is another reason why using this board as an adjustable bench PSU is not advisable.

On-Off Switch

There are three holes for an ON-OFF switch on the LTC3780 board - see picture above (one hole is not connected to anything). This can be wired to a SPST toggle switch on the front panel instead. When the two wires are shorted, the board will be disabled.

A Normally-Open (NO) thermostat could also be wired in parallel too, so that if the inside of the box gets too hot, it would simply disable the board.

The Fault and OK lights can also be wired up to the front panel by removing the LEDs on the board and connecting wires to two other LEDs (red and green?) on the front panel.

Note that the over-current circuit is very primitive and can cause the output to oscillate. When it detects an over-current demand it turns off or reduces the output voltage, the over-current trip is therefore reset and the output is turned on again which trips the over-current limit and turns off the output, which resets the over-current trip, etc., etc., etc. Not very suitable for any load unless it is a motor/heater/bulb/battery and certainly not suitable as a bench power supply for your electronics projects.

DPH5005 boost/buck module

In my experience, many of the boost/buck modules have an inferior design and are made with poor quality components. They are not very forgiving if you exceed their ratings or change any component values even when they do work correctly (which is rare!) and you still have to add a VA meter and switches/pots/LEDs too.

The Riden DPH5005 will convert  a DC input to 0-50V 5Amps and is designed for the purpose
If you are planning to use an old laptop power adaptor or perhaps an ATX power supply as a DC source, I would suggest that you simply use a Riden DPH boost/buck module such as the DPH5005 50V 5Amp 250W module (£32). 

It has a programmable memory and you can set a constant current and a total power limit too. You can optionally add bluetooth and/or USB modules for control from a PC or android app. You can also buy a nice case (Type B) for it too but you won't be able to fit an AC-DC board inside the case as well as the boost/buck board.

This unit is specifically designed to be used as a bench power supply. If you add up the cost of a LTC3780 boost/buck converter plus VA meter plus extra pots/switches/LEDs it will probably come to almost the same price but with this unit it should just work!

Or for £100 (inc. case+AC-DC board+front panel+all wires, terminals and fittings = complete PSU) you can build one of these to get 0-60V @ 6Amp (and no or very little soldering is required)...

No comments:

Post a Comment