Logic level converter?

A logic level converter based on a MOSFET transistor has probably been seen by all of us.
Professional integrated converters are probably based on this principle as well.
For slow events with a frequency of kHz and units of MHz OK.
Transistors like the BS170 have a Turn−On Time of 10ns and a Turn−Off Time of 10ns, so I am definitely not suitable,
Are there any available discrete transistors in cases like SOT23 capable of operation in a similar converter?

 

[KlausST]

Hi,

Why "not suitable"?
There are many MOSFETs out there. Just do a search.
And there are dedicated level converters for alomst every application.

****

no voltages, no load current, no timing requirements, nothing about capacitive || resistive load, one direction or bidirectional ....

So how can we help? What reply do you expect?

Klaus

 

[asrock70]

Sorry, I missed some of the questions.
I have an output in 3.3V logic and an input in 5V logic, frequency 80 MHz.
These are the pins of two different ICs. It is not possible to connect them directly because of the logic level 1 on the 5V side. Capacity. I don't know, if I say 10pF I probably won't be confused by more than +-5.

 

[KlausST]

Hi,

Again: One direction only or bidirectional?
I ask because you talk about IN and OUT, which sounds unidirectional, but your circuit shows a bidirectional solution.

****

I have an output in 3.3V logic and an input in 5V logic, frequency 80 MHz.

Still not clear if the "output in 3.3V" means the Converter output or the "one of the IC" outputs. Please clarify.
As always: A sketch could be so helpful ... (Not only in your case I ask myself why it is hard to use a pencil and a sheet of paper)

Also why hide the "two different IC" as a secret? Giving a link to the datasheets gives us all informations we need to know. At once, low effort.

****
Possible solutions unidirectional:
* 3.3V levels --> 5.0V levels: use a (fast) HCT type gate supplied with 5V. There are many different families supplied with 5V and TTL input levels.
* fast CMOS switch. 5V supplied with TTL level control input.
* Maybe no converter needed at all if the 5V decvice has TTL input thresholds.

Klaus

 

[dick_freebird]

This is not a level shifter. Looks more like what is done for "5V
tolerant" inputs to 2.5 / 3.3V logic. If you want speed then a
designed cross-coupled level shifter with output buffers is the
usual thing.

 

[FvM]

The "1-transistor level shifters" are used for birectional signals, e.g. I2C and depend on pull-up resistors. They are surely not suited for high speed. Dedicated levelshifter ICs with push-pull drivers are available from different manufacturers, e.g TI and NXP.

 

[crutschow]

Perhaps one of these.

 

[D.A.(Tony)Stewart]

To design anything you must specify all important parameters for input and output.

Polar, unidirectional, push-pull??

e.g. Vcc1, Vcc2, Z1, Z2, Tr,Tf in/out or max transition frequency with logic thresholds for ”1&0”.
Z2 may include R,C,L.

- for the fastest speed with voltage gain you will probably use CML or ECL as it was called for BJT devices. These devices have many GHz gain x bandwidth.

For linear devices, you might want microwave devices but I suspect you want a push-pull driver where speed is limited by current rating.

Please define your purpose and more details.

 

[asrock70]

Again: One direction only or bidirectional?

One direction from 3.3V to 5,0V

 

[D.A.(Tony)Stewart]

How fast is enuf?

 

[asrock70]

I'll try to gather what I know and want.
Output is +3,3V Push Pull, impedance maybe about 50 Ohm
from this output I would need to convert to an IC input with 5V logic as a 74HC clock signal with a frequency of 80MHz.
ideally, if it could be done by discrete switching with a transistor, i.e. no integrated voltage converter
Why? Availability and repairability even years later

 

[D.A.(Tony)Stewart]

I'll try to gather what I know and want.
Output is +3,3V Push Pull, impedance maybe about 50 Ohm
from this output I would need to convert to an IC input with 5V logic as a 74HC clock signal with a frequency of 80MHz.
ideally, if it could be done by discrete switching with a transistor, i.e. no integrated voltage converter
Why? Availability and repairability even years later

I'm not sure I agree with your assumptions on why. Are you familiar with the 100 MHz 74AHCT family? Your 74HC family will go obsolete before this family. It has Advanced prop delays with TTL hysteresis input thresholds hence the pre & suffix letters A...T

It is 5.5V output capable while having 3V Schmitt trigger input specifications with ~ 50 Ohm max output impedance using 4.5V at 85'C. useful for 100 MHz clocks.

I don't know your clock distribution requirements but for synchronous operations at 80 MHz target threshold variations must be verified with longer Tt transition times for skewed operations.

Otherwise if you wanted to use a CB common base config consider this ft= 700 MHz common part, MMBT6429LT1G gain should be 10 a decade lower than ft. Unfortunately the hFE*ft = GBW of 700 MHz was the maximum and min is only 100 MHz.

 

[KlausST]

I fully agree with Tony.

for 80MHz you need push-pull drivers. If you want to build it with discrets you need a bunch of devices.

I vote for a 74AHCT1G125 .. and a power supply decoupling capacitor.

Why repair? Then it needs to fail first. Why fail? Design it correctly.
And it won´t hurt to put a couple of the AHCT devices on stock.

Klaus