Author Topic: A fast and powerful bipolar drive stage  (Read 1296 times)

0 Members and 1 Guest are viewing this topic.

Offline T3sl4co1lTopic starter

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
A fast and powerful bipolar drive stage
« on: August 01, 2019, 10:53:55 am »
A freebie -- I'll probably never use it due to the sheer transistor count (at least until monolithic protos become viable?..).  Also, a discussion of its synthesis:



Core concept: use open collectors to drive an output, e.g., motor, switching supply, gate drive, etc.
- We can use low-Vce(sat) transistors, with high hFE, to get good performance.
- Even with the gain, we still need strong base drive, and strong turnoff for speed.
- For efficiency, we may steal some of the base current from the collector circuit when Vce is higher, i.e., make a Darlington.  We still want to drive the main transistor's base directly, to ensure low Vce(sat) at lower Ic (so that, for example, a gate driver will tend towards nearly zero saturation, rather than pulling to ~0.6V inside the rails and then just sitting there forever).

Putting these features together gives us Q113, Q114, and their turn-off drivers Q111, Q112.

We then have four bases to drive, which should be arranged from a CCS for level shifting purposes.  Current-steering logic (a diff pair) seems a good candidate, but a variety of possible drivers can be used.

- We can pair together Q111/Q112 bases, as long as we have some "squish" under their emitters.  (A monolithic circuit would not need this, as the transistors can be electrically and thermally matched.)
- We can provide the drive currents to Q113/Q114 simultaneously, in much the same way, i.e., pairing up Q115/Q116 with some squishiness under their emitters.  Note this reduces the diff pair gain some.  (R116/R121 should probably be smaller, so that somewhat more current is diverted towards the output transistors.  The resistor values in general are very preliminary.)
- R118/R119 could of course be a CCS of some sort, but this is rather overkill, at least for THR as shown.  A TTL-compatible version (THR ~ 1.5V) might be better served with CCSs.

THR is set by a voltage divider, or can be used as a complementary input.  It's also the positive input, so we can implement hysteresis by simply adding a large resistor from OUT to THR.

Reproduce everything on the bottom side, in complementary type, to get a full wave driver.

Similarly, a tri-state driver can had by unlinking the two inputs, driving them separately instead.  Logic must be added to prevent a shoot-through condition.

High voltage supplies, bootstrapping: the differential drive should afford good PSRR.  Probably it would be desirable to add another diff pair in front of this, so that level shifting can be done with less current, while the "core" diff pair can run at higher current, giving better on/off drive for the output transistors.  (You could of course expand this even further with trigger pulses and flip-flops, which is how the commercial bootstrap gate driver ICs do it.  I... would strongly suggest caution, to implement a pulse-triggered circuit yourself.  Unless carefully balanced, it will be prone to false triggering, and you'll get all kinds of nasty waveforms in a gate driver.)

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline Sylvi

  • Regular Contributor
  • *
  • Posts: 85
  • Country: eg
Re: A fast and powerful bipolar drive stage
« Reply #1 on: August 01, 2019, 05:32:54 pm »
Hey

Those surface mount output BJTs are interesting, but just as your circuit is, they are only optimised for switching.

It is a little optimistic to expect 2N3904/6 to survive as drivers at the 6A extreme of the output stage, and they limit Vcc to <40V. I suppose if you are normally designing for battery operation then 40V would seem like "high voltage".

No data for speed - so how "fast" do you think this circuit is?

No data for output voltage and current - so how "powerful" do you think this is?

It is ONLY for pulsing output NOT linear, which seems like an important point to highlight.

LT had an app note for their LT1010 buffer, back in the 1980s, where they showed how to improve a video line driver using it, but then taking the driver further by replacing the 1010 with diamonds of 2N2904/6. The crazy part was biasing them at 80mA with 12Va cross them - 960mW from a half-watt TP-98 - not going to last very long! The 3kV/us was impressive.
 

Offline T3sl4co1lTopic starter

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: A fast and powerful bipolar drive stage
« Reply #2 on: August 01, 2019, 09:23:42 pm »
Quite; it would take some effort to bias everything in a reasonable linear range to give linear operation without burning up.

Those output BJTs are quite good for anything you might use them on, really the only thing that's missing is power dissipation.  I suppose that's really all that you're saying.  (But worth noting, some related types have been found to have impressively low noise, a side effect of the wide junction area and clean processing.)

Right, easily modified for higher voltages.  The higher dissipation of Q115, Q120 would make that bias path unfavorable (or at least something you'd need to dial back), and the higher voltage makes the Darlingtons' saturation not as unpalatable, so I'd recommend dropping Q115 and Q120.  The pair of turn-off transistors I think would still be a good idea.  Candidates would be FMMT459/FMMT560 and STN83003/STN93003, perhaps?

On that note, an option would be merging Q111/Q112 together, and using schottky diodes to pull down the two bases, instead of using two collectors.  This will turn off Q114 first, and then Q113, which is fine.

Speed -- rise/fall time should be in the 10 to 100ns range, depending on how hot the drivers are biased, and optimization of resistor values.  And maybe some speed-up caps across some emitter resistors, that sort of thing.  Propagation delay and minimum pulse width should be larger, due to the number of stages being propagated through; I would imagine it's good to a few MHz.

Oh, a note on shoot-through -- given that the thresholds can be adjusted independently and high and low sides triggered independently, we might make the THR divider into a chain of three resistors, so that as the input transitions, the switches can be timed to interleave just right.  Given some known input dV/dt, of course!

Tim
« Last Edit: August 01, 2019, 09:32:16 pm by T3sl4co1l »
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf