By Steve Bush 4th January 2022
The holiday gave me a chance to play with bipolar transistor mosfet drivers again.
And things did not go quite as planned.
I had been using the circuit on the right quite happily from a signal generator.
Rise time was quick enough with 470Ω for R3, and is did not have to resort to my improved version of the driver.
By ‘resort’, I mean ‘bother to build it’.
However, it was working well with a signal generator driving Q3 base through R2, but output pulses were suddenly too short when I connected a microcontroller there instead.
I had been using a +/-5V drive waveform, and now my 0-5V drive was insufficient to quickly pull stored charge out of Q3’s base, leading to a long delay (hundreds of ns) before the transistor turned off.
This made me look at the turn-off specs for the npn (a 2N2222A) and, for the first time I can remember, I noticed that the turn-off test circuit in the data sheet used a negative drive to the base. Looking at data sheets for other transistors revealed similar drives.
At this point, I concluded that many similar circuits advocated on the web have never been tried, or are at least have not being used at 300kHz.
So how to pull out that charge?
I tried the circuit on the left to speed-up switch-off, where: the capacitor allows charge to be tugged out of the base, R4 limits current flow for the preceding circuit, and R5 is enough to hold the transistor on or off after the switching transition.
It worked well – nice crisp turn off – with the signal generator set to provide 0V or 5V but, with the microcontroller connected, R4 could not be set low enough to get crisp switching without the high current pulse through R4 and the MCU upsetting other parts of the circuit (this was on a breadboard with wires up to 150mm long).
Stuck for something to try quickly (rather than make a pcb to tame the long connections) I remembered Jim Williams using ‘Baker clamps’ in some Linear Tech application notes – in his case adding a Schottky diode to keep transistors out of saturation so that the storage time problem goes away (right).
And indeed it did go away.
Even using a 1A Schottky that I had lying around, the transistor turned off quickly and the on-voltage of the transistor remained well below 1V – plenty low enough for the circuit to work: it might even have been <400mV, I can’t remember right now.
Looking at the Williams app notes bought to my attention the gold-doped low-storage-time 2N2369, which seems to be a bit of a legend for such applications. I wish I had one (or one of its many derivatives) to try – although its max collector-emitter voltage is a bit marginal for my application.
The 2N7000 small enhancement-mode mosfet also looked tempting as minority crrier devices do not have storage time, and I have a few of those.
As I have to make a circuit board to tidy up the prototype, I might as well forget the whole discrete driver idea and move over to using the integrated TC4428A drivers that Microchip generously sent me a few years ago, which I didn’t use up to now as the SO-8 package is just a bit too fiddly for me.
On the other hand, I might just add my proposed enhanced discrete high-side and seemingly-unique Cuk synchronous drivers alongside the IC drivers just to try them with real components.
However, right now, creating a pcb is turning out to be another can of worms….
UPDATE: Conversation with Steve Kurt and Mike Bryant below caused me to come across this brilliant application report by TI, that covers just about every technique that can be used to speed up bipolar transistors (eg, take a look at figure 13) – hats off to which ever TI employee was given the time to write this information-crammed article (it also caused me to see how glib the term ‘miller-killer’ is when applied to later TTL families).
Top diagram created in LTspice – so also hats off to ADI, for keeping LTspice free to use
Tagged with: bipolar transistor driver EinW Engineer in Wonderland mosfet
> eg, take a look at figure 13
Oh I remember that paper. I never understood why they left Q4 in the AS version as Q10 removes any charge from Q5 base. And looking at it again now, I still don’t 🙂
Morning Mr Bryant I am not very good at this but: is it possible that, if Q4 is deleted, there is nothing to keep Q5 off after Q10 has done its transient stuff in the face of system noise getting onto the output and through Q5’s collector base capacitance? (particularly as R11 is quite high so D9 can’t do much) Do forgive me if this is rubbish.
No you’re correct but I should have been more clear that I’d have put a resistor across D9 to keep Q10 slightly on. In fact I’m not convinced D9 does that much anyway as Q10 is mostly switched on through the capactive D10, but obviously all these things add up. It’s much easier nowadays with CMOS 🙂
Phew, thanks for that. I see your point: use Q10 for both instead of retaining Q4. Maybe they used the same mask set and just altered the metal layers to pick up the transistors they needed, and thought ‘Q4? Why not!’.
The issue of getting charge out of a base-emitter junction has touched my life too. One approach is to use a diode across the resistor where you show a cap. The anode is connected to the base… meaning that there is less resistance in the discharge path. However, I do like the idea of using the schottky diode to avoid saturation. Wasn’t this what was used in LS TTL logic? I have vague recollections of the class where this was explained (many decades ago). 🙂
Morning Mr Kurt. Does that diode pull out the charge that causes switch-off delay, even with its 700mV drop? (I wish I had listened harder at my semiconductor physics classes*) It will not be hard to try 🙂
On the subject of LSTTL – I did indeed read this once I started searching ‘Baker clamps’. The same was true of STTL too, I also read. This is a truly excellent document, which goes on to say why AS TTL was even better: https://www.ti.com/lit/an/sdaa010/sdaa010.pdf?ts=1641443359100 TI really went the extra mile on that one!!
*ok, I admit: I wish I was bright enough to understand my semiconductor physics classes – run that Fermi level maths past me again….
I think one of the main benefits of a solid state physics class was the reinforcement that the minority charge carriers had to be pulled out of a base-emitter junction before it would turn off. The other part was the education about metal-semiconductor junctions, aka schottky diodes, and how they were so much faster.
I’ve used the diode in the base network mostly as a way to control rise and fall time separately, but speeding up turn-off time is just a variation of that. I was doing it for EMC emissions reasons, though!
I noticed that the T.I. document was from 1985, about the time when I was in college. That was the era when I had a hard cover T.I. TTL data book, and it had a ton of info on each IC. However, it’s nice to not have to store dozens of databooks on my bookshelves. 🙂
Good morning Mr Kurt Sound like you understood your solid-state physics. If I get time before using the cmos driver, I will try swapping a diode in for my capacitor. There are some wonderful data books from those times – the Nat Semi linear data books, for example – I learned my wonder of electronics from them 🙂 (I recently notice some of the applications circuits in later Nat Semi linear books have the finger prints of Jim Williams on them – I miss those long challenging app notes)
I had a fine collection of ancient databooks acquired from various skps.
However once the cracks in the front room ceiling became a bit alarming I figured out there were about 3 tons of books in the box room.
The useless ones rapidly met the recycling.
Much to my amusement a chap at the dump asked if I should be putting the books into a different skip rather than the paper rec.
My reply was that I hadn’t read them in decades so why should anyone else be interested in 1985 era Mitel databooks.
Kept most of the NatSemi, Texas, and Moto databooks, though the same thing applies to them.
Having approached the local charity second hand book shop and discovered that it was full, I too took books to the dump over the festive season.
As I left, I heard a horrified voice: ‘Someone has thrown books into the recycling’.
I was that book burner…. 🙁
Could you output a complementary pulsed signal from the MCU to drive another transistor with it’s C-E connected across the B-E of the first one to turn it off quickly ? Once the first transistor is off turn the second one off as well so it doesn’t mess up the next turn on signal.
Thanks Mike, I hadn’t though of that. I will have a look at the resources – AVR PWM outputs are not quite as flexible as the PIC PWM outputs – no individual output inversion in a dead-banded pair, only the option to invert both, so I am not quite sure I have another output of the correct polarity to spare (that is not on the wrong side of a dead band gap).
Having found that document in my reply to Steve Kurt, I now realise ALS TTL used active turn-off similar to the idea you suggested Mr Bryant.
Another solution would be if the current handling requirement isn’t too onerous then a single gate NC7SZ05P5X or NXP equivalent in SC70 5 pin package. Or if more drive is needed to parallel up the outputs of a 74HC05 open-drain inverter.
To be honest Mike, the bipolar approach has become something of an academic exercise as the TC44xx range of drivers are available and are essentially CMOS inverters (or buffers) with gigantic output mosfets which do the whole job in one package (except if you need to work under 4.5V). That said, I am dying to try the bipolar approach on a Cuk synchronous rectifier mosfet, as I can’t see another way of doing it. BTW, I looked up the NC7SZ05P5X – what a find – impressively fast chip and a 24mA output.
Hadn’t seen the TC44xx before. Looks like it could be used as a powerful line driver in its own right.
Microchip has an amazing variety pf products these days. I wondered if those CMOS-like drivers would make a nice robust output stage for a pulse generator – although easily-adjusted slew-rate control might be tricky.
That’s because they’ve bought up almost every small semiconductor company going 🙂 Microchip and Diodes Inc are soon going to be broadline suppliers like TI and ST used to be.
Slew rate control is a bit hard. You could try variable constant currents feeding the positive and ground supply pins and a large capacitor on the output to slug any external capacitance. This would mean the input reference ground would be moving so you’d either need to overdrive/underdrive the input, as it says it can take -5V underdrive, or just use a transformer on the input.
I can see how that would work. After I got started pondering pulse generator output circuits, I was searching around and discovered a couple of clever ADI ideas which I have been saving up for a blog. Thanks to your prompting Mike, I will stop being lazy and do it NOW. Expect a link here later…….
Later: https://www.electronicsweekly.com/blogs/engineer-in-wonderland/circuits-make-nice-pulse-generator-output-2022-01/
Good to see you’ve been keeping busy.
Nice to hear from you zeitghost I felt a bit of a chump staring at the scope thinking ‘why transistor no work properly…’ Hope you had a successful festive season (which I define as: ‘not catching covid’ these days)
So far so good, thanks.
I ascribe my success to the Sampson option of not having a haircut in 2 years.
My success comes from being a natural hermit* A friend* of mine said of of covid-avoiding isolation: ‘at last, something introverts can enjoy’.
*ok, I admit the contradiction – but it was really an acquaintance….
Your email address will not be published. Required fields are marked *
Get our news, blogs and comments straight to your inbox! Sign up for the Electronics Weekly newsletters: Mannerisms, Gadget Master and the Daily and Weekly roundups.
Read our special supplement celebrating 60 years of Electronics Weekly and looking ahead to the future of the industry.
Read the Electronics Weekly @ 60 supplement »
Read the first ever Electronics Weekly online: 7th September 1960. We've scanned the very first edition so you can enjoy it.
Read the very first edition »
Electronics Weekly teams up with RS Grass Roots to highlight the brightest young electronic engineers in the UK today.
Read our special supplement celebrating 60 years of Electronics Weekly and looking ahead to the future of the industry.
Read the Electronics Weekly @ 60 supplement »
Read the first ever Electronics Weekly online: 7th September 1960. We've scanned the very first edition so you can enjoy it.
Read the very first edition »
Tune into this Xilinx interview: Responding to platform-based embedded design
Tune into this podcast to hear from Chetan Khona (Director Industrial, Vision, Healthcare & Sciences at Xilinx) about how Xilinx and the semiconductor industry is responding to customer demands.
By using this website you are consenting to the use of cookies. Electronics Weekly is owned by Metropolis International Group Limited, a member of the Metropolis Group; you can view our privacy and cookies policy here.