Faster BJT in levelshift

The passive pullup, and capacitances, may be more of a speed limit than the Ft of the transistor. Q1 will turn on slow (R1) and turn off slow (Rg).

How much output voltage and current do you want? What speed?

Why not buy a serious half-bridge driver chip?

The MMBT5551 is faster but only by 50% or so. Could be too pricey.

The fast 50C02CH from ON Semi would require really tough negotiations for pricing. All the others that come to mind would be way out of league for cost or too low in max Vce.

Can you add a small capacitor in parallel to R2 as a "gooser" for a spike in base current during turn-on? Unfortunately that adds cost.

The problem I have is compliance (that the levelshift current source works) when OUT goes up and down, which means VS+12V also goes up and down

A speed up capacitor across the bottom resistor will let it turn the transistor on/off faster, but will impede the current source action when VS goes up and down

Cheers

Klaus

To maintain emitter and collector current matching, one could transformer-couple those. It only takes a ferrite bead for transformer; inductance won't be much, but this is about HF response.

is running and a low side FET is turned off, the current mirror (BJT, R1 a nd R2) cannot keep up due to limited GBW of the BJT

An old technique in high voltage systems would be to scrap the current sou rce and clamp the collector node with a voltage limiting zener, resulting i n rise times of 0.2RC or so. And then there are conventional techniques to speed up the transistor if needed like a Baker clamp.

Does it need to be DC translated? If this is for say, a PWM, and is always running square wave input, you could just use a coupling capacitor and a diode DC restore clamp

-- Kevin Aylward

Have you actually simulated it with 500MHz GBW BJT?

It's probably a problem with Miller capacitance... or simply a timing sequencing issue. ...Jim Thompson

I use that sometimes when it is a highside PMOS, but in this case it is a NFET, so the gate is jumping up and down with the output, so capacitive coupling does not work

Cheers

Klaus

Huh ?>:-} ...Jim Thompson

Lookie here:

I just tossed in the ferrite because I hate stray fields, but a printed circuit pair of inductive windings could work (unless you crowd the box with other circuitry).

The idea is a hybrid of the transistor level translator, and a transformer.

Why don't you use a commercial half-bridge gate driver? There are tons of them around, they have all sorts of options for dead time delay, and solve a lot of problems with home-brew circuits.

Jon

That's not a current mirror, it's a level-shift comprising a current sink and a load resistor. Its speed is determined by the Cob** of the BJT followers and to a lesser extent by high beta to isolate Ciss. Improve speed by increasing current (lower R1 R2), or better yet, add an npn emitter follower with high pulldown current to ground, to drive the Cob and isolate it from R1's high driving impedance.

How about an arrangement like this? Looks pretty speedy in the sim:

I have simulated it with a BC846, and the dV/dt on the output is so high that the BJT cannot keep up

I wanted to hear here about devices before simulating with higher BW BJT

Cheers

Klaus

This is for a high volume product, price is king

Cheers

Klaus

Looks interesting, will try it out

I have tried increasing the current, but did not help much

So the circuit you are suggesting, is that like Figure 16 in this link:?

Cheers

Klaus

Hmm, once upon a time I had made a discrete circuit that worked very nicely, don't remember if it was for a discrete buck converter or synchronous rectifier or what. Anyway, don't use a pullup resistor, use a current source. Even better, use a current mirror, and instead of a single transistor sending current, use two, in a diff pair. Now you also get better immunity from dV/dt because it's differential.

Tim

-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website:

Aha! Bitrex' famous works once/weeps mostly circuit ;-) ...Jim Thompson