Hello, please let me know if this is a repost.

I would like to use a microcontroller running at 3v to send data to an LCD that runs at 5v. To do the level conversion I plan to use NPN transistors. The data should be sent at around 8mhz. I have looked at different transistors but the datasheets aren't exactly clear about what the maximum frequency the transistor can work at would be. I tried adding up the various rise and fall times for the 2N2222s I have and came out with about 2MHz as the max frequency. The logic levels for high and low of the LCD are listed as 0.8*Vdd and 0.2*Vdd, though, so the signal may not have to completely rise or fall to be valid.

Is there any straightforward way to calculate how fast I could send data by looking at a transistor's datasheet?

 

2SC2229
2SC1815

Some high frequency transistors in the tens of Mhz range

 

Looking at the datasheet the test conditions are at 100Mhz, so 8 Mhz should work for you.

2n2222

 

Thanks for the quick replies, guys.

@Shagas, how did you figure out the frequency of the 2SC2229? It lists the transition frequency as 120MHz. Is it that divided by about 10?

 

Thanks for the quick replies, guys.

@Shagas, how did you figure out the frequency of the 2SC2229? It lists the transition frequency as 120MHz. Is it that divided by about 10?

http://www.edaboard.com/thread18748.html

transistion frequency is the frequency at which the gain is equal to 1 .
So basically if you are operating in at about 8Mhz then you should be able to get decent gain . Sorry I don't know the exact math used for this but you need a gain of 5/3 if I understand correctly so it's more than enough

 

Also , as Dodgydave said
the 2N2222 has a transitional frequency of about 300Mhz so you can use it

 

The small signal frequency of a BJT is much higher than the switching frequency, principally due to saturation storage delay. The 2N2222 has a combined rise-time, fall-time, and storage-time of 310ns, thus the maximum switching frequency is about 3MHz so it's not appropriate for this application.

You could use a high speed level shifter such as this or this.

 

The small signal frequency of a BJT is much higher than the switching frequency, principally due to saturation storage delay. The 2N2222 has a combined rise-time, fall-time, and storage-time of 310ns, thus the maximum switching frequency is about 3MHz so it's not appropriate for this application.

You could use a high speed level shifter such as this or this.

hmm thx for the info

 

2N2369 is one of the best saturating switch NPNs you can buy.

 

2N2369 is one of the best saturating switch NPNs you can buy.

Good suggestion. That should work as an 8MHz switch with a sufficiently low collector resistor.

 

Good suggestion. That should work as an 8MHz switch with a sufficiently low collector resistor.

And Baker clamps can help. The Schottky diodes MUST be low capacitance.

 

The small signal frequency of a BJT is much higher than the switching frequency, principally due to saturation storage delay. The 2N2222 has a combined rise-time, fall-time, and storage-time of 310ns, thus the maximum switching frequency is about 3MHz so it's not appropriate for this application.

You could use a high speed level shifter such as this or this.

These parts are very useful in mixed voltage systems because the inputs are 5V tolerant AND they have TTL thresholds of [0.8, 2.0] Volts

 

2N2369 is one of the best saturating switch NPNs you can buy.

The data sheet lists Turn-On + Turn-Off + Storage Time as 43ns. Is this enough information to consider 23mhz the max you could could signal at?

Good suggestion. That should work as an 8MHz switch with a sufficiently low collector resistor.

Would around 4k work or something lower?

 

The data sheet lists Turn-On + Turn-Off + Storage Time as 43ns. Is this enough information to consider 23mhz the max you could could signal at?



Would around 4k work or something lower?

The problem with high resistances (e.g., 4k) is with the RC time constant. The capacitance consists of load (LCD) capacitance, collector-base capacitance, and wiring capacitance.
A total capacitance of only 10pF will yield a 40nS time constant. Even if the transistor turns on and off instantly, the rise time (10-90%) will be about 90nS, which is incompatible with a pulse width of 62.5nS (50% of 8MHz period).

You really need a push-pull driver for 8MHz.
What is the part number of the LCD you are driving? A link to the datasheet would be even better.

 

The quoted 310ns time for a 2N2222A is at Ic=150mA. Lower the collector current and that time decreases. The majority of that time is because of the collector resistor, and the value of the resistor counts as well as the voltage swing and receiver thresholds, so it is hard to give a concrete answer given the known information. But, for a few cents more the 2N2369 is faster, but it is specified at 10mA.

 

You really need a push-pull driver for 8MHz.
What is the part number of the LCD you are driving? A link to the datasheet would be even better.

Here is the LCD:
http://www.hmangas.com/Electronica/Datasheets/LCD/EW32F50YLY.pdf

It lists 8mhz as the maximum clock frequency although data can be clocked in at a much slower rate. Hopefully I will have some time left over to prepare the next frame if I can transfer the data in quickly. (Never mind the requirement for -27 volts to bias the LCD. This is more for fun than utility.)

 

The quoted 310ns time for a 2N2222A is at Ic=150mA. Lower the collector current and that time decreases. The majority of that time is because of the collector resistor, and the value of the resistor counts as well as the voltage swing and receiver thresholds, so it is hard to give a concrete answer given the known information. But, for a few cents more the 2N2369 is faster, but it is specified at 10mA.

The PN2222A spec sheet (plasitic case version of 2N2222A) shows that the switching times versus collector current increase, not decrease, as the collector current decreases. And that switching time is unrelated to the collector resistor, which can add its own delay due to stray collector capacitance. The times shown are due just to various internal transistor delay mechanisms.

 

The LCD needs 6 drivers. They need to be push-pull in order to operate at 8MHz. I would use 74ACT04, which is a high speed hex inverter with TTL-compatible inputs, which means you can drive it from a microcontroller running off 3V. It's in a SMD package.

 

The 2N7000 series switching FET was designed for this sort of thing.

You can make Ron's push pull with these as well.

http://www.fairchildsemi.com/ds/2N/2N7000.pdf

 

The 2N7000 series switching FET was designed for this sort of thing.

You can make Ron's push pull with these as well.

http://www.fairchildsemi.com/ds/2N/2N7000.pdf

That is a poor choice, IMHO.
2N7000 needs 5V Vgs to guarantee Rds(on). Maximum Vgs(th) is 3V. The OP has a 3v source.
Push-pull is not possible without a lot of extra circuitry.
The load resistor has to be low (i.e., ≈100Ω with a 5V supply) to get decent waveforms at 8MHz.