BJT Pair for Push-Pull, 1W, 33V, 200mA, 1MHz (& maybe no heatsink)?

Yup, fuel-injected Vortec V8.

Interesting-looking beast, compared to past vehicles I've owned. It has a separate ignition coil for each spark plug.

I've identified my "air door" and sensor location, which is very easy to get to, at least. I'll go look for the pressure regulator, later today.

I, too, had a 77 LTD (Crown Victoria or somesuch). I liked that car. But its computer failed once, when my wife was about 8.9 months pregnant, on an unusually-hot Saturday afternoon, in a tiny little town out in the "corn desert" of central Illinois. The owner of a nearby store gave me his whole toolbox to take back to the car, to see if I could find the trouble. But, of course, I had to have it towed. Cost me something like $450.

Thanks, Jim.

- Tom Gootee

then

Quite

Graham,

OK, sorry, never mind that question about the maximum output level. I was doing it in Spice and happened to be trying it with an opamp model that wasn't able to supply enough current to the BJTs' bases.

I have modified the circuit to be a more-typical "opamp with push-pull booster" (inverting). So, instead of using the Howland topology thuat I showed in the schematic, I removed the feedback loop to the opamp's + input, changed the R for feedback to the - input to 10K, changed the input (to opamp -) R to 10K, and grounded the opamp's + input. I also changed the emitter resistors to 3 Ohms.

Once I had an opamp with sufficient output-current capability to comfortably get all the way up to 15v/150mA at the push-pull's output, I juggled the bias resistors' values and the boost capacitors' values, to lower the required maximum opamp current, and also to lower the size of the boost capacitors. That was done with the emitter resistors still at 3 Ohms each.

With 2.7K for each of the four split bias resistors, the required opamp current was 22 mA max, and about 4.5mA average over one staircase cycle (at 22kHz step-transistion rate). I may use an OP275 opamp, there, for now (after I check the offset and input bias specs again), which can push 40 mA max, since I already have lots of those on hand. The boost capacitors were able to be lowered to only 0.22 uF, which will be a whole lot easier to fit onto the pcb layout.

If I used larger boost caps, I would probably be able to lower the required max opamp output current, and/or enable the use of more-than-3-Ohms emitter resistors. I guess I'll just test it as-is, first, and see how hot everything gets.

Any suggestions or comments would be quite welcome.

Thanks again, Graham.

Regards,

Tom

somewhere, then

away. Quite

Uh Oh.

Unlike the Howland-based topology I had tried, this plain "opamp with push-pull booster" doesn't seem to work very well as a precision current source, at least not the way I now have it configured. I'll play with it for a while and report back.

- Tom

Just for an option, this constant current arrangement turns up now and again. Works OK from DC to a few kHz but there's opamp/instrumentation amp speed, slewrate issues beyond that. 15V current source compliance is also a pig with those 17.5V supplies. Higher voltage opamps start running thin on the ground at more designer friendly 20V rails. .--------. |\\| | Power | 0-15V/0-150ma o->--------|-\\ | Buffer | ___ Output Input | >------------o X1 o---o-|___|-o----o--->-o 0-15V .----|+/ | | |Isense | | o | |/| | | | 100 | | | | | | | | \\ o o | | | | | | \\ | GND | ___ '--------' | | \\. | o----|___|--. | | o | | 100k | | | | GND | /| | ___ 100k | | | | /-|---o--|___|--------------' | .-. o-

Thanks very much, John! I'll check it out.

- Tom

So give it a small LCD and four joystick buttons. You have digital control of everything, so if you need to use it while disconnected, you can wander through the menus... slow, but feasible. Plus if it worked in the automatic mode while disconnected, so you could just hook up, say, a BJT, and have it tell you "NPN BJT, Hfe 126" and the *pinout*, preferably while the device is still *in-circuit*, you'd have a winner.

Noise. Difficulty of mounting. Can't use a laptop. What interface - IDE? Various problems anyway. I don't think it's such a wonderful idea.

USB isn't going away any time soon, and you could include a flash chip as mass storage so you have the UI software anywhere the device goes, maybe even for multiple platforms.

Clifford Heath.

I agree, with so many people using laptops these days USB or Ethernet is the way to go unless one actually *needs* the bandwidth provided by PCI or similar.

Nice idea, although for something low volume like Tom's curve tracer, adding a built-in flash chip would undoubtely add a *lot* more cost than just buying a cheap 128MB or larger USB memory stick at retail (

Here's how I'd attack it...

Cautions:

(1) Headroom forces one into multiple supplies to accommodate the OpAmp requirements.

(2) The Power MOS devices selected are *only* because of what I had "laying loose" in my libraries. They're too leaky to accommodate the low current end, {R} sets the range. Might even be better to use Darlingtons.

...Jim Thompson

Joel,

If you see this, please email me at tomg at fullnet.com, again, so I can respond. (I am usually impossible to reach, by phone, without an appointment or something. Sorry.)

I finally had to call my ISP, to ask them to delete all of the huge emails from the attacker. But, since the POP file had grown to over 2 GB by then, they said they "just had to remove the whole thing". So, for now, I've lost all of the emails that were still on my ISP's server (at least a couple of weeks' worth)!

Just FYI: Luckily, they did keep a copy of the 2 GB POP file containing all of my stranded emails. So, they're either going to burn it onto a DVD for me, so I can try to sort through it (it's in ASCII), or, they said, they "MIGHT" be able to do something to get it connected to a separate mail client, somehow, and then be able to delete just the attacker's emails, or, they "MIGHT" be able to use something like the Unix "tail" command to get the messages from after he ceased his attack, about a week or so ago. It depends on how difficult each of those options is, for them. Even just getting it onto a DVD sounds like it will take some time, for them. I'm just happy that I have a local ISP who is willing to work on it, for me.

Regards,

Tom Gootee

Jim,

Thanks!!

I'll play with it in Spice and report back.

- Tom Gootee

Hello Tom,

That depends on the chosen bias current. I usually shoot for around 50mV drop but you can go lower. The better the thermal tracking the lower you can go. I have an RF amp that runs without any emitter resistors but with an extremely good thermal coupling where everything is bolted to a big heat sink. Don't forget an overtemp switch (very important).

-- Regards, Joerg

Some of us know how to do temperature stable A-B bias... but I won't divulge how, I need to file a patent application first ;-)

...Jim Thompson

Hello Jim,

Mine were ok so far. Never really had one fail except (almost) on that RF amp I mentioned. Someone had thought that lockwashers are optional. Arrrggh...

Maybe you can post your trick once it's filed.

Updated with staircase drive.

(Figured out a clever way to make a staircase simulation tool ;-)

...Jim Thompson

Joerg,

Thanks for the info.

With 1 Ohm for each Re, I get about 1 mv Re drop per mA output. So, for my max output current of 150 mA, I have about 150 mV Re drop. If "more is better", I can leave them that way.

"Overtemp switch"...?? Uh oh!

Is that something like a PTC "self-resetting fuse"? In series with the load?

- Tom

Love the circuit!.

Any chance of sharing that staircase gen?. Nearest I got needed 3 sources. john

I needed more than that...

I could merge all the behavioral sources into one, but it's hardly worth the effort... gets confusing keeping track of the SPAN and SVOS ;-)

...Jim Thompson

Hello Tom,

I meant 50mV at the selected quiescent current, not at max operating current. But if things don't get too hot you might be ok.

I use what is sometimes referred to as "thermo disks". Little thermo switches that are screwed onto the heat sink and are (typically) normally closed. When exceeding their rated temperature they open but then reset themselves after things have cooled off.

A temp variable resistor works as well but then you need some more parts to pull down the bias source.

One of my guys, on Monday, derived a closed-form equation to predict thermal runaway in a class AB amp. I did a quick numerical scribble to demonstrate that our circuit was OK, but he got interested in the general solution...

But just offhand, 50 mV sounds a little small. If the power transistor temp increases 20K, the Iq doubles, and that may run the power dissipation up enough to give you another 20K, and away you go. It depends on heatsinking and stuff. I'll ask Rob for his equation. I think he assumed no thermal coupling from the transisrot back to the bias diode, a prudent assumption for surface-mount stuff.

I saw a formal analysis of A-B runaway in an old GE manual, 1960's maybe, and it included leakage!

John