Is there a chip with multiple N & P Fets for fast 5V to 12V conversion?

  • Thread starter Thread starter Jmayes
  • Start date Start date
  • Tags Tags
    12v Cmos Ttl
AI Thread Summary
The discussion revolves around the need for a chip that can efficiently convert 5V signals to 12V with fast rise times (0.05us to 1us) for high-speed applications. The original poster has tried various solutions, including open collector buffers and high current drivers, but has faced issues with speed and power consumption. Suggestions include using the CD4504B-EP as a potential solution, which offers a suitable level shifting capability. The conversation highlights the importance of managing capacitance in long cable runs, as it significantly affects rise times. Ultimately, the goal is to find a single-chip solution that combines level shifting and buffering for multiple signals.
Jmayes
Messages
6
Reaction score
0
Hi, new to these forums and glad I found them!

I am working on a project that requires conversion of fairly hi-speed signals (10us pulses) from 5V to 12V. The usual open collector with a pull up is much too slow getting from 0 to 12V. I really need the ramp to happen between .05us to 1us. I can achieve this with N & P Fets (push-pull) but really want a chip with several of these buffers built in as I have many signals to convert. At slower speeds using a UDN2982 worked great (with pull-ups) and if there were another chip with a similar pin-out that would be ideal for me.

http://www.allegromicro.com/en/Products/Part_Numbers/2981/2981.pdf

Any help will be greatly appreciated!
Jmayes
 
Engineering news on Phys.org
That IC you linked is a relatively high power driver and is much slower than what you require for simple buffer/level shift.

The 50 to 100ns edges you require and not the least bit difficult to achieve with a simple open collector buffer as long as you keep the pull up resistor value reasonably low (1k to 2.2k). Use the linked open collector buffer (7407) with a 1.2k to 1.5k pull up resistor and you'll easily achieve rise/fall times around 50ns into a single 12V cmos input.

http://www.google.com.au/url?sa=t&s...nbizAw&usg=AFQjCNEP0FbmogM-5bZax5dPlOpYmWh0HA
 
In fact I tried that exact combination of 7407 & pull-up, I had to go down to 330 ohms and the edge was only just fair, I could use it but it was not ideal and power consumption was high (most of the time the signal is at Gnd level). I know the UDN part is a high current driver with no pull-down, I also worked with the ULN2803 (reverse part) and a pull up. Also with just NPN transistors but everything left me on the hairy edge. I really need something with a push-pull output and many drivers in one chip if one exists. Looking at perhaps trying a LM3900 or opamps but don't want the extra outboard parts.

Thankx for the help!
Jmayes
 
Jmayes said:
In fact I tried that exact combination of 7407 & pull-up, I had to go down to 330 ohms and the edge was only just fair, I could use it but it was not ideal and power consumption was high (most of the time the signal is at Gnd level). I know the UDN part is a high current driver with no pull-down, I also worked with the ULN2803 (reverse part) and a pull up. Also with just NPN transistors but everything left me on the hairy edge. I really need something with a push-pull output and many drivers in one chip if one exists. Looking at perhaps trying a LM3900 or opamps but don't want the extra outboard parts.

Thankx for the help!
Jmayes

What type of input where you driving Jmayes. Are you trying to drive multiple inputs or do you have a very long wires or some other source of excess capacitance? You understand that this all comes down to RC right. The rise time is approx 2.2 RC. The capacitance of a single 74Cxx (or CD4XX) cmos input is only about 5 to 10 pf. Add another 5 to 10 pf of stray capacitance and you've got your RC time constant and rise times.

If the input capacitance is 15pf and the pullup 1.5k then the time constant is approx 22.5ns and the rise time approx 50ns (that's how I arrived at the suggested pullup).

If you needed 330 ohms just to get it down to 100ns then your input capacitance is out of control (>100pF). This is the thing you need to address. Are you using a breadboard?
 
BTW. One other quick thought. You wouldn't be the first person to ruin their rise times by connecting a CRO probe! I've even seen noobs trying to measure this kind of thing with a piece of 50 ohm coax connected instead of even using proper probe (haha probably several hundred pF!)

If you're observing this on a scope make sure you have low capacitance probes (<15pF)! Even this will double your observed rise times (compared to what they really are when you're not looking at them). Otherwise only observe at the cmos output and just deduce the input rise time.
 
Last edited:
Yes, using a proper scope probe :) and Yes, I am driving a length of cable (about 3 feet), using standard ribbon with every other conductor grounded. So yes the capacitance is what is killing me so hence the need for a stronger active pull-up and pull-down. Sorry I did not think to add that info to my OP. (Yes , I have played with termination too)

Thankx again for your help,
Jmayes
 
Jmayes said:
Yes, using a proper scope probe :) and Yes, I am driving a length of cable (about 3 feet), using standard ribbon with every other conductor grounded. So yes the capacitance is what is killing me so hence the need for a stronger active pull-up and pull-down. Sorry I did not think to add that info to my OP. (Yes , I have played with termination too)

Thankx again for your help,
Jmayes

Yeah that's about an extra 100pF right there, just what I was expecting. Can you run 0,+5V over the 3' wires and place your open-collector buffers (plus bulk and bypass capacitor) on the cmos board. There should be some way you can alter the layout to avoid having to run the open collector outputs across the 3' wires.
 
Some of the circuits end up going around several boards after the entry point so I really just need a good tight 12V signal that can withstand probably 6 feet of wire. This is a retrofit to a 80's model device so most of the signals go to plain transistors and diodes. It is a MUX buss type of circuit that already puts 3 items on the buss and I am expanding it to go to four so the edge specs need to be better then original. I have seceded with external push-pull logic but really want to find a chip to make this easier over 60+ circuits. A drop into the 2983/2903 is the ideal device as I have already wired for those chips.

Thankx again,
Jmayes
 
OK you could just buffer it (not open collector) after you do the level translation but before you go routing it all around the place. But I see what you're looking for now, you want the level shifting and the buffering (push pull) all in the one package to keep it simple. Maybe someone else will suggest a suitable part.
 
Last edited:
  • #10
Which brings me to the OP, I need a solution that is in one chip, If I convert and buffer again it's two chips + glue.

Thankx again,
Jmayes
 
  • #11
How about just a simple level shifter?
 

Attachments

  • Level Shifter.jpg
    Level Shifter.jpg
    29.9 KB · Views: 1,100
  • #14
Thank you for the suggestions! I remember finding the 4504 but did not have one in my parts kits to try and forgot about it, it looks very promising. I have over 60 circuits to buffer so the hi-speed MD1810 would be more then I could afford and requires outboard inverse drivers.

I really like the CD4504, I need to get some in and test them. If anyone knows of something equivalent that the inputs are on one side and the outputs on the other (dip part) that would even be better as I have already wired a prototype for that type of part.

Thankx again all!
Jmayes
 
Back
Top