FR9888 or a synchronous step-down converter IC

This review continues my experiments with step-down voltage converters. Along the way, I’ll also explain a little about what a synchronous rectifier converter is.

I started looking into various converters quite a while ago, but the reason I bought these ICs was a recent review of a small step-down converter IC. In the comments, readers suggested a few more options.
First, here are links to the two previous reviews.
TRI1461
MP2359DJ

And here are a few links to my reviews of ICs and converters.
PT4115E
AMC7135
Inductor assortment CDRH104R
LTC4054
Adjustable DC-DC converter
SC6038
MC34063
ICL7660
DC-DC converter
A pair of step-up converters and a small upgrade to SEPIC
Universal voltage converter or a few words about what SEPIC is
Multichannel DC-DC converter
ZXY6020S
ZXY6005S
Converter based on LTC3780
DC-DC step-down module rated for 10 A

The ICs are sold in lots of 10. They cost 1.67 when I bought them and now cost 1.71, though there was probably a discount through the mobile app.
By the way, I only noticed today that the AliExpress order page no longer shows how much time is left before order protection expires.

The ICs arrived in a small yellow envelope, with a simple zip-lock bag inside. The quantity matches the order.


A brief description in English.
In short:
Input voltage — 4.5–23 V
Output voltage — 0.925–20 V
Output current — up to 3.5 A
Switching frequency — fixed at 340 kHz.
It also has various useful features, such as thermal protection, overcurrent protection, and so on.

Link to the full datasheet.


The IC comes in an SOP-8 package. There is a pad on the underside for soldering to the circuit board, which helps dissipate heat from the IC die.


The IC block diagram. It's worth noting the rather unusual 0.925 V reference voltage.
Someone asked me in the comments why you can't set a converter's output voltage below a certain minimum.
The reference voltage is what determines this limit. The IC compares the voltage at the feedback input with its reference voltage. If the input voltage is higher, it tries to shorten the pulses to bring the voltage down. Likewise, if the input voltage is lower, it tries to lengthen the pulses to raise the voltage.
So there’s no simple way to set the output voltage below the reference voltage: the IC will always try to raise it until the feedback input reaches 0.925 V (or the corresponding value for other ICs). Raising the output voltage is no problem; you just add a divider made from a pair of resistors.

The second unusual thing about this IC is that it has two MOSFETs on the output, not one.
This IC has a built-in synchronous rectifier. You’ve probably noticed in previous reviews that there’s always a small (or large) diode “hanging around” the IC. This IC doesn’t need an external diode.

The third handy feature is the SS (soft-start) input. When power is applied, the IC doesn’t start instantly; it gradually increases the pulse width until it enters regulation. This is useful with “heavy” loads, which can trigger the IC’s overcurrent protection if voltage is applied abruptly.


Let me digress a little on the subject of synchronous rectification.
For a long time, voltage regulation was handled with discrete linear regulators. Then came integrated versions, in multi-pin packages with external components and an external power transistor.
After that came the familiar KR142EN series of ICs, often simply called KREN.
They were convenient: low output ripple and very easy to connect. The only drawback was that they dissipated a lot of heat. But heatsinks were cheap, and nobody was particularly worried about the heat.
Although, in one of my reviews, I showed how to make a high-efficiency converter from a KREN5 (7805).
Meanwhile, switching regulators were also being developed. At first, they were discrete (without dedicated ICs) and quite complex, which held back their adoption to some extent. But then integrated PWM voltage regulators started to appear (often copied from foreign designs).

A simple switching voltage regulator consists of several main components (let’s look at a step-down converter):
The controller
The switching transistor
The diode
The inductor
The output capacitor.

We’re interested in the first three items.
Everyone knows that bipolar transistors and diodes have a fixed voltage drop across their junctions. It’s higher for ordinary diodes and lower for Schottky diodes.
This is a simple, inexpensive solution in principle. But then MOSFETs came into use, making it possible to improve converter efficiency because they have a lower voltage drop across the switching element.
Efficiency improved, but the diode remained. Replacing it with a transistor is much more difficult. That problem was eventually solved too, and this converter is an example of that approach.

The block diagram on the left shows a conventional converter with an external diode; the one on the right is the converter from this review.
At the start of the conversion cycle, the transistor turns on and energy builds up in the inductor. After a while, the transistor turns off, the voltage across the inductor reverses, and current starts flowing through the diode to the load.
In this IC, the diode has been replaced with a MOSFET built into the IC. This improves efficiency and means the converter no longer needs an external diode. There is still a diode, of course—the MOSFET’s body diode—but that doesn’t matter here.


That’s enough theory; let’s move on to the practical part.
The datasheet showed two converter circuit options: one with ceramic capacitors and one with electrolytic capacitors.
I decided to go with the first option, using ceramic capacitors, since it makes the design a little more compact.


I used the PCB layout example from the IC’s datasheet.


In my version, the PCB looks like this.


I found some of the parts I needed at home, and while I was at the market I bought a 5.2 µH inductor as well—but it turned out I didn't need it. It also cost almost a dollar, which is very expensive even for a brick-and-mortar shop.


I shouldn't have bought the inductor because I wanted a 5 V output, and according to the table, that requires a 10 µH inductor.
By the way, the table also lists divider values for specific fixed output voltages.
I used slightly different values from the standard E24 series: 9.1 kΩ for the lower resistor and 39 kΩ for the upper one. The calculations called for 40 kΩ on top, but I didn't need an exact voltage, so this was good enough.


After that, I moved on to assembly.
Since the IC's bottom pad has to be soldered to a copper pour on the board, I started by doing it as shown in the photo:
1. Applied a little solder under the IC.
2. Heated the copper pour and placed the IC, then heated it again from both sides while holding it down with tweezers.
3. Checked that the IC's underside had actually soldered by trying to pry it off, then soldered the remaining pins.
4. Cleaned the board and moved on to installing the other components.
Of course, soldering the IC with a hot-air gun would have been easier—and that's what I ended up doing. But not everyone has one.


This is the little board I ended up with.
The inductor isn't soldered in; it's just shown to demonstrate the original design. The board can be used this way, though, but only at low output voltages.


I had an identical inductor rated at 10 µH, but it wasn't suitable because it was rated for a lower current. The thing is, as inductance increases, the inductor's rated current decreases (for the same dimensions). The 5.2 µH one was still suitable for the current, but the 10 µH one wasn't—at least not from the series I designed the PCB for.

I had to look through the inductors I had at home, then rewind one of them to get the inductance I needed.
The inductor is wound with two strands of 0.63 mm wire.


A photo to give you an idea of the size, and a comparison with the boards from the two previous reviews.


A closer look.


For the test setup, I used the same equipment as last time
Power supply
Electronic load
Multimeter
Another multimeter
Oscilloscope
Non-contact thermometer

But when I powered it up, things didn't go as planned.
The converter started up, but at a load current of 0.7 A it went into protection mode and dropped the output voltage almost to zero.
The output voltage was also slightly lower than calculated. After a few experiments, the IC started putting out 6 V, then stopped working altogether :(
I removed the IC with a hot-air gun and soldered in a new one (also using the hot-air gun), but it still didn't work: there was no output voltage, and the current draw was 90 mA.
In the end, I removed the second IC too and soldered in a third. With that one, the output voltage was as intended, and the IC continued to work correctly.
The first photo shows the first IC, the second shows the third.


First test: measuring the current draw with no load on the output.
I wouldn't call that low; I expected the IC to draw less.
23 mA at 10 V and 28 mA at 20 V


Test procedure:
1. Output current: 1 A; input voltage: 10, 15, and 20 V
2. Output current: 2 A; input voltage: 10, 15, and 20 V
3. Output current: 3 A; input voltage: 10, 15, and 20 V


I won't include all the oscilloscope traces; I'll just show the no-load and maximum-load conditions with a 20 V input.
There is practically no ripple. I even checked whether the probe's 1:1 setting was selected.


Checking the minimum input voltage at different load currents: 1, 2, and 3 A.
The output voltage is about 4.75 V.


I also tested the output short-circuit protection.
The protection works very well, but it doesn't enter hiccup mode; instead, it stays in a mode similar to constant-current regulation.


The output current is where things get a little tricky.
At a load current of 3 A, the IC shuts down after a while because its case temperature gets too high. If I blow a little air on the IC, everything works fine.
The likely reason is that there should be vias under the IC connecting to the other side of the board, and the board itself should have been double-sided. I used a thin, single-sided board, and it simply couldn't dissipate the heat.


The IC itself can handle load currents up to 4.5 A; beyond that, the output current limit kicks in.
Of course, the IC can only deliver 4–4.5 A briefly, but even so, that's good.
The photo shows the input current at output currents of 3.5, 4, and 4.5 A.


And of course, I'll check the efficiency.
The manufacturer provides this graph for an output voltage of 5 V. I tested it at input voltages of 10, 15, and 20 V, rather than 12 and 23 V as in the datasheet, but I don't think that makes a critical difference.


Well, what can I say? I never got the claimed efficiency, although the efficiency is certainly not bad given these output currents.
For the 3 A measurements, the IC temperature shown is the temperature at which the thermal protection kicks in.
It's also clear that efficiency is higher with a 10 V input than with 20 V.
By the way, after the tests I decided to give the IC some more abuse. I heated it thoroughly with a hot-air gun and pressed it firmly against the board with tweezers. After that, the thermal protection kicked in noticeably less often, but it still couldn't handle 3 A. It worked properly at 2.5 A.


So, what can we conclude?
Pros
Price
Short-circuit, overload, and overtemperature protection that works properly.
Good overload capability
Soft start.
Very low output voltage ripple.

Cons
Efficiency below the claimed figure
Uncertainty about reliability when soldering the ICs onto the board.

My take: the IC fell a little short of my expectations. I was expecting more. Of course, I didn't follow all the manufacturer's recommendations and soldered the IC without providing a thermal path to the other side of the board. Still, the efficiency was the bigger disappointment. Although it's higher than with the previous samples in the 10–15 V range, here's a quick comparison at 2 A:
The one under review is on the left; TRI1461 is on the right.
10 V — 89.7/86
15 V — 87.4/86.1
20 V — 84.9/86.1

Overall, it can be used at currents up to 2.5 A, and at up to 2 A you can even get away without soldering on a heatsink. But the glitch with the first sample is still bothering me. I still can't figure out what happened. It was disappointing; I'm kind of used to putting something together and having it work :)
By the way, ICs with a flange are much easier to work with in this respect: heat the flange, and the IC solders into place. Doing the same with the one under review is very awkward.

I hope this information is useful. I think I've covered everything I could, and as always, I'll be glad to hear your questions and suggestions in the comments.