Datasheet-a-Day W45 2020

More datasheets and application notes: starting to learn about DACs in the Analog Mini-Tutorials series, another DC/DC converter (in a slightly less silly package), plus a couple of randoms.

Optical transmission transmitter (SFH757) and receiver (SFH250) (Avago)

These are obsolete parts, but I know nothing about fibre applications, so I was interested to see how usable these kinds of things might be.

These are a pretty standard LED and photodiode pair. The interesting part is the packaging. Both of them are in a plastic packages with aperture with lens at bottom to insert fibre directly, i.e. there’s no stripping or termination of the fibre needed. I don’t know if you need to treat the ends of the fibre at all. For high-end applications, I know fibre ends are often polished (I think: like I said, I know next to nothing about this), but I don’t think these are intended for high-end applications.

There’s an additional mounting with an axial threaded cover to secure the fibre: you solder down the transmitter and receiver housings, insert the fibre and tighten the screw and you’re good to go, apparently.

They have a maximum claimed data rate of 100 Mbps, so I’m not sure what the intended application for them really is.

LM555 Timer (TI)

The venerable 555 timer...

I’ve recently been looking at more analogue IC datasheets that include schematics for the chips, and I’m starting to recognise building blocks in these, at least, even if I can’t explain how everything is connected and set up. This one has a couple of long-tail differential amplifiers, a couple of current mirrors, and something that looks like an output stage (being connected to the OUTPUT pin is a good clue!). There’s also something in between the comparators and the output stage which must be a flip-flop.

One thing (well, one of the many things) I don’t yet understand about these IC schematics is some of the weird-looking transistors in them: in this case, there are a couple with two collectors, and one with three collectors and no base! There’s a Stack Overflow answer that covers some of what’s going on here.

The applications of this thing are well-known, though I’ve seen some other applications in other datasheets for equivalent parts that are more interesting than the usual astable/monostable examples (PWM and PPM waveform generation, for instance).

TI datasheets are quite funny, because they include layout advice for every single part, including this one...

ISL91107 High Efficiency Buck-Boost Regulator with 3.6A Switches (Renesas)

This seems like a fairly normal buck-boost converter, except that it’s another one in a tiny WLCSP package. It comes in fixed 3.3V and variable output versions, from 1.8-5.5V input and claims to have smooth switching between buck and boost modes.

Seems ideal for Li-ion battery powered devices. There are a lot of these things out there, and I suppose it should be pretty easy to find one in a more usable package with the same kind of characteristics.

2.5 MHz switching frequency, so small inductors! 3.3V fixed version needs only input and output capacitors and inductor as external components.

Basic DAC Architectures I: String DACs and Thermometer (Fully Decoded) DACs (Analog MT-014)

The first few of the Analog Mini-Tutorials series were all about ADCs. Now there are a few about DACs.

They start with the basic observation that you can use switches as 1-bit DACs, then show a Kelvin divider (string DAC): this has $2^N$ switches and $2^N$ equal resistors for an N-bit DAC. The pros of this setup are that glitches are code-independent, small and at the DAC update frequency, so it’s good for low distortion applications. The cons are that you have lots of resistors, lots of switches, and trimming all those resistors to get linearity isn’t really practical...

String DACs are also used for digital potentiometers.

There’s a current-based equivalent to string DACs. These are called “thermometer” DACs. Some of these have complex switching schemes to improve DAC performance.

Basic DAC Architectures II: Binary DACs (Analog MT-015)

A simple binary-weighted DAC uses one switch per bit and resistors in binary ratios. It’s hard to manufacture at high resolutions and not monotonic though, so not often used. The equivalent current mode design is based on the same idea and has the same problems.

You can do something similar with capacitors in binary ratios. This isn’t generally useful for normal DACs because of capacitative decay, but it can be used as an internal component in successive approximation ADCs.

Then we get to R-2R ladders, whcih are something I’ve seen in a lot of circuits. They’re easy to trim and require relatively few resistors. A voltage-based DAC using an R-2R ladder has constant output impedance. The input impedance is tricky though, and the switches in the DAC need to be able to switch across the full input range, which can also make things difficult.

There are current-mode ladder DACs have different trade-offs, and these have different trade-offs.

Basic DAC Architectures III: Segmented DACs (Analog MT-016)

Segmented DACs combine multiple simpler DACs to meet whatever performance requirements you have. For example you might use one DAC architecture for the MSBs of your signal, and one for the LSBs, somehow...

For example, two string DACs, one $M$-bit, one $K$-bit, give you an $M+K$-bit DAC with $2^M + 2^K$ resistors, instead of $2^{M+K}$ resistors.

There are lots of possible architectures and it all gets complicated very quickly!

STBB1-AXX High efficiency single inductor dual mode buck-boost DC-DC converter with 2.3 A switches peak current (ST)

This is the same kind of thing as the other DC/DC converters I’ve been looking at recently, but in a more usable package (a 3 x 3 mm DFN10: still small, but easier to deal with than the tiny WLCSP packages).

What are the differentiators between these different devices? Presumably it’s mostly about the quality of the control algorithms, particularly for these buck-boost devices where you want smooth changeover between buck and boost operation, but those things are quite opaque in the datasheets (at least to me).

I suppose you need to do a side-by-side comparison of the performance to get some idea of what different parts do, but the graphs shown in different datasheets are hard to compare, and they’re presumably set up to show the parts in the best light!