Low Voltage Extra

I’ve had a bench top power supply for a while - it’s invaluable to be able to dial in the required voltage and also see what sort of power draw your circuit has. My old supply was linear - so it was heavy and bulky! 

I’d seen a few projects where people built their own bench top power supplies using the RuiDeng DPS 5005 - so I set out to do the same. After spending ages scouring AliExpress, I found a suitable enclosure (220*195*80mm Iron metal enclosures for electronics junction box diy electrical instruments box) and some very compact 300w switch mode power supplies (Dual Output Power Supply 48V 12V Led Stage Lights Driver Led Moving Head Power Supply 300W SMPS).

The power supplies were well specced, but the reason they are so compact is because they have a fairly high ripple voltage (advertised at 480mV p-p). With that in mind, I set out to build my PSU.

Cutting holes for the fan and to mount the two SMPS:

Both SMPS in place (with thermal grease underneath)

Fan clearance - its pulling air in from the bottom and it goes out through the top past both SMPS.

Top down with wiring

Close up of wiring

Cutting holes for the two DPS 5005

All up and running - pushes 500w without getting warm!

What about ripple?

At full power - 50v ~250w - we get 314mV p-p and 21mV Rms:

At 25v ~60w we get 178mV p-p and 25mV Rms. 

At 12v ~15w we get 112mV p-p and 15.6mV Rms

At 5v ~2.5w we get 82 mV p-p and <10mV Rms

Adding two large capacitors on the output doesn’t help at all - here we have 50v again with two 1500uF caps:

These are hugely popular, and cheap buck adjustable power supplies.

They claim 250w output, but they are tiny!

I wanted to run a quick test to see how they handled a little bit of that power.

Using my trusty electronic load (rated to 150w) I dialed up as much power as my bench top power supply could muster.

How hot do you think it got?

At ~90w the electronic load was up at 100+ degrees C.

The DPS5005 was hardly hot at all!

Looking at it from the top, the components were

Less than 50 degrees C.

I recently ordered a couple of these from a cheap Chinese website thinking they would be useful to power a Raspberry Pi (or other 5v devices) from a variable power source. They were only a couple of dollars each, so I shouldn’t have expected too much from them.

The specs on the website show the following details:

A pretty broad input range (5.5v - 26v) and up to 3A. The packaging even says max 5A! That’s a fair amount of power for such a small device.

I thought I would run a quick load test at different input voltages to see whether they can really do the claimed 3A buck conversion.

I setup a small testing rig using a (cheap) digital load and a bench power supply.

For each run I set the load to draw a fixed amount of current and then varied the input voltage on the bench power supply. For example, the first run I set the load to 200mA and the output of the bench top power supply to 5.5v (the minimum claimed input on the spec sheet).

Even at a very low load, the UBEC could not provide 5v output with only 5.5v input. At 200 mA the minimum voltage required at the input to get a stable 5v out was 6.5v. 

At higher loads, it was even greater.

So the specs aren’t quite accurate. While testing the 3A load, I noticed that the UBEC had heated up significantly (too hot to touch!).

Wow. The bottom of the device was 130 degrees C! Ouch! After a little while the red light went off and it stopped providing any output.

So, clearly the device shouldn’t be used for applications that require 3A current draw!

After removing the metal shield we can see why the device isn’t suitable for 3A. Infact testing the steady state heat generated at 1A and 2A draw showed that the inductor and schottky diode next to it were getting up to 52C (at 1A) and 93C (at 2A).

At 3A the device cut out once the inductor reached around 174 degrees...

I’m pretty sure this isn’t a case of a cheap knockoff or anything like that. Notice the HW212V1.2 silk screen on the PCB (which was underneath the shield).

Looking at the components on the device, we can see that the main IC is a MP1593DN - which is rated for 3A. 

The inductor, however, appears to be well under-rated given its size (which, in addition to the heat, could also cause problems with RF noise?).

So. I would suggest the following “corrected” spec sheet:

Output Voltage: [email protected] Continuous output current: 1.5 Amps Input: 7.5V (@1.5A), 6.5V(@500mA) to 26v

Three ESP32 boards but they are not equal!

Price

The GeekCreit DevKit-C is between USD12.50 and USD6.50 from Banggood 

The Wemos Lolin 32 Lite is USD4.90 from AliExpress 

The “Wemos” (it’s a fake!) TTGO is USD8.99 from Banggood

Power consumption 

All running esp-idf hello world

DevKit-C: 41mA@5v or [email protected] (regulator bypassed)

Lolin32-Lite: 29mA@5v or [email protected] (regulator bypassed)

TTGO: 34mA@5v or [email protected] (regulator bypassed)

ESP32 - Silicon Revision

DevKit-C reports Rev0

Lolin32-Lite and TTGO report Rev1

Battery support

DevKit-C is missing any LiPo support. Both the TTGO and Lolin32-Lite have connectors and support USB charging. Great!

Size

DevKit-C is the largest at 58mm x 30mm

Lolin32-Lite is much smaller at 50mm x 25mm

TTGO is just slightly longer and thinner at 51mm x 23mm

Conclusion

I really like the Lolin32-Lite. It’s the cheapest and the build quality is fantastic. The power consumption is very low too.