Characterising cheap $1 ebay DC-DC buck converters eg MP1584 based D-SUN modules

[ornea]

I am curious what characterisation data for cheap DC-DC buck converter would be useful to others or even if the data has been previously provided. 

In my experience I have found a mix of results even among a batch of identical modules and therefore almost impossible to determine what is viable from an ebay listing.

As such I thought a repository of all things relating to these cheap modules maybe useful.

Things to consider may include:-

1. Pin pitch
2. Vpp Ripple at a given Vin, Vout Iout.  How much ripple is too much?
3. Vin, Iout where the module shuts down
4. Ripple Freq
5. Other

There are many cheap adjustable DC-DC converters available from ebay.

However, the casual buyer has no real data (apart from the very optimistic specs below) to base decisions on so I am curious if it be worth publishing the data and charts along with the python script etc. If so where would be the best forum, (here, hackaday, youtube).

Thus far I have automated a characterisation test.  The test is basically:-

FOR Iout = 0 to 2.5A in 50mA steps
    FOR Vin = 7V to 21V in 250mV steps
         Measure Vin, Iin, Pin, Vout, Iout, Pout, Efficiency, Vpp, Vripple-freq
         Terminate if Vout deviates more than 0.4V
     Loop Vin
Loop Iout

The setup uses SCPI commands via telnet to control DP832A (Rigol bench PSU), DL3021A (Rigol DC load) and SCPI commands via HTTP POST requests to control RTB2004 (R&S Scope) to measure ripple and freq.

I now have many tables and graphs see example attached but I would like to know what data is most important or useful.

I am hopeful for suggestions on ideal step sizes for Vin and Iout, Vpp and Vout delta limits etc.  What parameter is best to plot against which parameter.  i.e Current verses Ripple for a given fixed input voltage or Efficiency verses Vin for a given current (similar to the attachment)

An example of the data usefulness: It can be seen from the graphs etc that the optimistic efficiency spec of 96% is almost achievable if Vout is set to 5V, Vin is 7V and load current is about 200mA.

Below is ebay (very optimistic) specs for A DC-DC module identifiable by D-SUN silkscreen and MP1584 controller:-.
 
Input Voltage: 4.5-28V
Output Voltage: 0.8-20V
Output Current: Rate Current 3A(MAX)
Switching Frequency: 1MHz
Output Ripple: less than 30mV
Efficiency: 96%(max)

All thoughts, suggestions warmly welcome.

[splin]

It would be tricky to do, but monitoring the input and output capacitor temperature would be useful. I suspect that they are frequently seriously underated for the amount of ripple current they need to handle and consequently overheat. I had a '150W' boost convertor where the input capacitor exploded when testing. Calculating the ripple current, I seem to remember that the capacitor should have been at least twice the size and very low ESR into the bargain.

Even better would be to directly measure the ripple current of the input capacitor, but that would be even trickier.

You only need to test the cases where ripple current is at a maximum which, without looking it up, I think occurs when the duty cycle is 50%, at maximum load current.

For example, those Ebay LM2596 2A o/p convertors may have an i/p capacitor ripple current of over 1A rms when the input voltage is twice the o/p. Typical modules have a 100uF, 50V i/p cap. Looking on Digikey, of the 694 x 100uF, 50V aluminium electrolytics, only 4 have a ripple current rating of 1A or more (at 10kHz or 100kHz) and cost a lot more than the typical 400mA rated parts. What chance those $0.7 modules have the very low ESR caps they ought to have?