One channel, both ways
Each channel is one BSS138 with its gate on 3V3 and a 10 kΩ pull-up either side. At rest both sides sit HIGH, each at its own voltage. When either side pulls LOW, the transistor pulls the other side LOW too. Nothing tells it which way the signal goes, and nothing has to.
Three situations explain it
The transistor's gate is tied to 3V3. It switches on when the gate is more than about a volt above the L side, and stays off when the two are equal. Each side also has a 10 kΩ to its own supply, holding it HIGH whenever nobody pulls it down.
- Nobody pulls. L sits at 3.3 V and H at 5 V. The gate and L are both at 3.3 V, so the transistor is off and the two sides are apart.
- The 3.3 V side pulls LOW. L drops to 0 V, the gate is now 3.3 V above it, and the transistor switches on and pulls H down with it.
- The 5 V side pulls LOW. The transistor is off, but the diode built into it lets L fall to about 0.6 V. That puts the gate far enough above L to switch it on, and L goes all the way to 0 V.
Step through the third one. It is why the same channel carries a signal either way with no direction pin, and why an I2C bus, where the two ends take turns, works through it.
What the design costs
Nothing on the board ever drives a line HIGH. A HIGH is a 10 kΩ resistor filling the wire, which takes time: the rising edge shows how much. A HIGH is also weak, about half a milliamp at most, which is plenty for a logic input and nothing like enough to light an LED.
A LOW is strong: whoever pulls it down, the switched-on transistor joins the other side to it. Each LOW also takes 0.33 mA from the 3.3 V side's resistor and 0.5 mA from the 5 V side's while it lasts, which is why a board with every line held LOW draws a few milliamps and one at rest draws almost nothing.
The 27-pack's loose 6-channel board is this circuit too, and the llc book's one transistor, two resistors walks the same three situations. What the TK97 adds is the supply for the gates, which the next page is about.
When it does not work
Then it is on an H pin, not an L pin: each H pin has 10 kΩ up to 5V, and that is what you are measuring. Unplug it and move it to the socket's row, L1 to L6, before anything else.
Not through the channel. A HIGH on L is 3.3 V from its own 10 kΩ, and the transistor switches off before the 5 V side can push L past about 3.3 V less its threshold. That is the point of the board.
Not for getting a signal through: both ways work, including I2C, where both ends take turns. It matters for speed, because each side's HIGH is made by a resistor. The rising edge article draws both directions.
The regulator that makes the 3.3 V side's supply, and what that means for your wiring.
The 3V3 pin is an output →Edit this page — content/books/logic-level-converter/one-channel-both-ways.mdx
Questions about this product
See what other owners have asked, and read their solutions.
Logic Level Converter
Loading discussions…
Discuss this article
Ask about this page. The answer stays here, on the page it belongs to, for whoever hits the same wall next.