Faster BJT in levelshift

Well, I knew that suppressing parasitic oscillation was what they were for. My thought was that the cascode structure could possibly make the bandwidth _too wide_ as opposed to not wide enough. What happens if there's no signal?

Reply to
bitrex
Loading thread data ...

I made a thing back several months that while it didn't work quite right, even in that state looked pretty fast at driving square waves using jellybean transistors. Sort of like a diamond buffer:

I wonder what a "Specchio de Corrente" is, and where to get one

Reply to
bitrex

I really feel like there should be a solution using depletion mode FETs or JFETs here. What about like, driving two complimentary BJTs at their bases connected to the output transistors as current mirrors, with FETs as the mirror collector loads?

Reply to
bitrex

Random idea (there are several variations on this theme):

Reply to
bitrex

But, you are replacing one part with about 6 - 8, depending on the version of circuit that has been bounced around. That certainly narrows the difference. IR makes some good drivers that fit in an SO-8 package. That will drive BOTH the high-side and low-side FETs.

Jon

Reply to
Jon Elson

The PFET has higher RDSon than a NFET, so for the same price, about 30% reduction in conduction loss

Cheers

Klaus

Reply to
Klaus Kragelund

Your alternate has a p-fet "booster" (Q2) in the driver, but an n-fet in the output (Q3). That's decent.

Yep. We don't know what's driving us, what we're driving, how quickly, duty cycle, frequency, what voltage needs switching, nor the cost constraint. We don't know how fast the existing circuit is, nor how fast it needs to be.

Definitely underspecified.

Here's a minimal variation on Klaus' original that's a bit faster, no extra parts:

Vcc Vcc -+- -+- | | [R1] | | |/ +---+----| Q2 | | | |>. | | | | ||--' | | | ||. | | >---[R3]---'

Decrease R1 to make it go faster.

I've sometimes used a Darlington for Q2 when R1's dissipation started getting too high.

The MMBTA42 looks promising for Q1--it's the only one I glanced at that spec'd Miller capacitance. 2pF--not bad.

Cheers, James Arthur

Reply to
dagmargoodboat

Sure, but knowing Klaus $0.03 isn't going to fly :-)

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

Joerg, are YOU suggesting Klaus is cheap?

I'm shocked, shocked.

Grins, James Arthur

Reply to
dagmargoodboat

They usually cost the least when it comes to driving staunch loads.

It probably needs to be aged 10-12 years in the basement to reach top taste.

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

Oops, scratch that. It doesn't pull low enough on the negative edge. You could use a 2n7002 instead for Q1 though. That works.

Cheers, James Arthur

Reply to
dagmargoodboat

That's no good. The gate drive amplitude needs to be referenced to the source node. The OP circuit is also inappropriate, as many of the circuits turned up in google search are.

Reply to
bloggs.fredbloggs.fred

Since I didn't know that for sure, I intentionally didn't do that. If Klaus had told us the motor (or whatever) had a separate supply, I'd have thrown in a bootstrap cap and diode.

But what was posted was, I hoped, enough to illustrate techniques for speeding up that internal node, the weakest link.

Cheers, James Arthur

Reply to
dagmargoodboat

Signal is from 3V microcontroller (so FET output)

18kHz Duty cycle from 0 to 100% Cost is low, but I need ideas, so I will sort out the ideas that cost to much (to many parameters to convey) Output of circuit is driving a 74AC14 schmitt trigger, which drives a FET (so the input impedance is very high, so I just need a voltage that ramps fast) Voltage is 24V DC, boot strap is 5V

Existing circuit has 5-600 ns response time, too long. I would like to come down to 200ns max

Yes, sorry for that

Reply to
Klaus Kragelund

The illustrate techniques was very good :-) You have pointed me in a good direction

Just wonder if this was 20 years ago with no NG/internet, I would then be sipping through countless books to get insight

Cheers

Klaus

Reply to
Klaus Kragelund

I tried to simulate it, got around 250ns, so that is good (with 2N7002 in the bottom and BC846 as cascode transistor)

It is with an ideal current source instead of the top resistor, since the topresistor without current source was bad

So now I need to make a cheap current source

Cheers

Klaus

Reply to
Klaus Kragelund

Here you go:

formatting link

...

Reply to
glen walpert

He has a problem with OUT going "up and down", really can't see any moving parts there, so he must mean the fixed ground referenced pull-up fet gate drive is problematic.

Reply to
bloggs.fredbloggs.fred

I think Klaus and I are roughly on par when it comes to the cheapness scale during designs. Sometimes I get to design stuff where cost is not important but mostly it is, very.

Only in Casablanca :-)

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

This should work, with some low-threshold fets.

formatting link

That upper diode might be a small zener.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.