Bi-directional level shifting speed, active load and NMOS logic

[c--]

Hello,

From what I understand, using a N-channel enhancement mode MOSFET with a couple of pull up resistors to do level translation can be pretty slow, due to the RC load that is created between the pull up resistors and the input/output capacitance of the transistor, resulting in a delay when switching between high and low logic levels.

For instance, as seen in Fig 1, an AO3400 has a pretty big C_iss of 630pF and a C_oss of 75pF which would make it a pretty poor choice compared something like a BSS138 (C_iss 27pF and C_oss 13pF).

Now, this level shifting circuit looks pretty much like a discrete NMOS circuit with passive loads. So in theory it should be possible to replace the passive loads with N-channel depletion mode MOSFETs similarly to Fig 2.

In such a case, would an active load actually improve the speed of operation of the level shifter, regardless of the transistor used for the translation? And would such active load configuration also benefit discrete NMOS implementations compared to ones with passive loads?

[prints-f-darkness]

Hi,

Instead of worrying about the details, worry about the application. For 50 cents USD you can get a TXS0101DCKR, 24Mbps (push pull).

(datasheet attached)
https://www.digikey.com/en/products/detail/texas-instruments/TXS0101DCKR/1739893

There's no reason to use discrete components unless you absolutely have too (IE cost savings for a production design).
If speed is your primary concern, use a dedicated device with known specs.

That's just my advice, I hope my insight was useful.

[Zero999]

Figure 2 won't work, because Q3 & Q4 will always be off, due to their gates being shorted to their sources, leaving only reverse biased diodes to +V.

Why not use a small BJT? They have a much lower capacitance than MOSFETs, which are typically designed for power. A small capacitor across the base resistor can help to speed the circuit up a little.

^{Level shift bi-direc BJT 5V.asc}

Another option is a J-FET, but it's can be more tricky to find parts which will work. Choose a part which with which will cut-off when V_S-V_G≤3V, such as the J113 or PN4393.


^{Level shift bi-direc J-FET 3V3 5V.asc}

[langwadt]

Figure 2 won't work, because Q3 & Q4 will always be off, due to their gates being shorted to their sources, leaving only reverse biased diodes to +V.

depletion mosfets with zero Vgs  they are basically a constant current source 

[Zero999]

Figure 2 won't work, because Q3 & Q4 will always be off, due to their gates being shorted to their sources, leaving only reverse biased diodes to +V.

depletion mosfets with zero Vgs  they are basically a constant current source

I know, but thanks. I missed the depletion bit. I skimmed the text and paid too much attention to the schematic. Pictures speak more loudly than words to me.

The BSS126 will typically allow around 15mA to flow when V_GS = 0V and V_D = 5V which is quite a lot. A better use for a depletion MOSFET is in the circuit I posted above, but a small sinal J-FET is still a better choice, as it will have less capacitance.

[c--]

Why not use a small BJT? They have a much lower capacitance than MOSFETs, which are typically designed for power. A small capacitor across the base resistor can help to speed the circuit up a little.

But a BJT would only make such circuit unidirectional, right?

[Zero999]

Why not use a small BJT? They have a much lower capacitance than MOSFETs, which are typically designed for power. A small capacitor across the base resistor can help to speed the circuit up a little.

But a BJT would only make such circuit unidirectional, right?

A BJT will work in both directions. The reverse has a much lower gain, but it's still more than enough for this application.