High Side Pulse Transmitter

100k is a pretty wimpy pulldown for the PNP emitter follower, and 22k is pretty wimpy for a high-capacitance MOSFET pullup. I can't make out the rest of the gobbled PNP pulldown circuit, is that a pulldown current limit? All three point to trouble with the falling edge.

--
 Thanks,
    - Win

It actually looks like positive feedback to me, pseudo-SCR sort of style. Looks tricky.

I'd vote for complementary mosfets, drains to the output.

We did one version that used some cmos gates as the driver: DC-coupled directly into the nfet and AC-coupled into the highside pmos gate, and then added a slow photovoltaic coupler on the high side to make it DC there, too.

John

Hello Jeff,

The 200 Ohm resistor to ground doesn't allow it to pull hard to ground. Can't you do it with a simple push-pull driver?

1.5uF is a pretty stiff load, depending on the transition time you are trying to achieve.

Regards, Joerg

It appears to me the source of your falling edge "bounce" (or rather undershoot at the output) is caused by the output PNP, positive feedback NPN, the 200R resistors, and the non idealities of the zener diode. If a zener diode were ideal, IE: zero internal resistance, perfectly fixed voltage drop from cathode to anode, then your circuit would probably function as you want it to.

As it is, regular sized zener diodes have quite a bit of internal effective series resistance. Depending upon it's size and whatnot, your "6V" zener is probably more like a 5.7V ideal zener along with along with some series dynamic resistance (and a parallel capacitance, which isn't too important in this application). During the turn off transition a significant amount of peak current must flow through this zener diode in order to halt the output fall. With the zener diode internal resistance, the anode voltage of the zener may fall well below Vcc-6V during the high peak zener current turn off halt event. I think this is the source of your output undershoot.

So how does one improve this?

Well ideally we would know more about exactly what you are driving, because there may be a better way altogether, and it is possible I don't fully understand your requirements... However... As it is, I would suggest the following:

Make the middle 200R resistor much larger. 2k or so would be more appropriate, this should dramatically decrease the peak current through the zener at the halting of output turn off, and thus decrease the output undershoot. Unfortunately, if you grow that 200R resistor to near 2k, that means the effective pull down strength of your output driver will only be around one half of what it is now. Right now, you essentially have the 200R bottom resistor in parallel with the 200R middle resistor for a pull down resistance of 100R.

So to compensate for this, I would replace the bottom 200R resistor with two or maybe three silicon diodes in series wired anodes towards the collector of the output PNP and cathodes towards ground. Then I would shrink the 10k resistor to something more like 100 ohms. Then I would place a 2k ohm resistor between base and emitter of that bottom NPN device to improve turn off performance and insure minimal leakage.

Note that these modifications should increase the peak dissipation requirements of the output PNP device quite a bit from the way it is now. As it is now, it appears most of the transient power gets dissipated in the

200R resistors, but my modifications would dump practically all of it into the PNP device. Therefore, make sure to select a PNP with big enough safe operating area for this kind of use. Note that the peak pull down capability is now roughly: (output voltage/2k)*beta of output PNP device.

I put the 200 Ohm resistor there to limit the current in the output PNP to about than 250mA. Without that resistor, current throught the output PNP in some versions of the circuit shoot right pass 2A.

Jeff Stout

You have the current-limit _phasing_ wrong, it's presently positive feedback ;-)

...Jim Thompson

-- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona Voice:(480)460-2350 | | | E-mail Address at Website Fax:(480)460-2142 | Brass Rat | |

| 1962 | I love to cook with wine. Sometimes I even put it in the food.

Yes, 100K is pretty wimpy. But if the load is purely resistive, then the load will pull down the output push-pull and the 100K doesn't do much anyway. If the output is capacitive, then the 100K gets it started and bottom npn takes over to complete the job. To really be effective, the 100K resistor would have to be less than 5K and the static power starts getting too high (> 1/2W) at 50V. I don't like chewing up that much power in one component in a standby state.

I don't know if 22K is too wimpy. These two resistor switch the high side input PNP and they perform at an acceptable level (i.e. that PNP turns off pretty fast). There is room for optimization here, but this part of the circuit is not the problem.

The problem is the falling edge of the output. I'm trading off speed, static power, and bounce.

Jeff Stout

Jeff Stout

Mark another one up to not enought information. The circuit spends most of its time in the high state. It transmits a signal between 1200 BAUD, and 9600 BAUD with a duty cycle of less than 10%. But your right, the current throught the zener is off the scale in some versions of the circuit.

Jeff Stout

A fast prototype would look like so. If you want current limit on that PNP pulldown then add a resistor at its collector: View in a fixed-width font such as Courier.

. . 2V-48V . VCC . + . | . +---+-------+-----------+ . | | | | . z | | | . 6V A .-. .-. | . | | | | | | . | | |1K | |200R |/ . | '-' '-' +---| . | | | | |>

. | | | | | . | | | | | . | +--|>|--+-----+ +---->

. - | | | | . v | | | | . | |< | | |< . +-| | +---| . | |\\ | |\\ . | | | | . +------------+-------|----+ | . | | | | | . | | | | | . | | \\| | | . | | |--+ | . | |

Not to mention he will either fry the zener or require a big one that takes 1.5W. Otherwise positive feedback is conducive to high speed.

Hello Fred,

and to ringing or oscillation ...

Regards, Joerg

Isn't this getting a little crazy and component-rich? You can do the whole thing with 5 parts. Or maybe less.

John

Serial interfaces like RS-232 are slew rate limited to 30V/us or less for EMI reduction and to avoid problems with cable reflections, the signal is not transmitted with edge rates "as fast as possible." Your circuit can be boiled down to something like this: View in a fixed-width font such as Courier.

. . 24V-48V . VCC . + . | . +----+--------+-----------+-----------+ . | | | | | . | | | | .-. . | | | | 4.7R| | . | | z | | | . | | 6.2V A .-. '-' . | | | | | | . | .-. | | |220R |/ . 47p| | | | '-' +---| T1 . === | |2.2K | | | |>

. | '-' | | | | . | | | | | | . | | +-|>|---+---+-----+ +------>

. | | | | | | . | - z | | | . | v 6.2V A | | |< . | | | | +---| T2 . | | | |< |\\ . +----+--------|-----| T3 | . | | |\\ | . | | | .-. . | | | __ 12R| | . | - +--|__|---+ | | . 5V |/ v 100R | '-' . CMOS o----| T4 | | | . |> | |

Nah- I need to add even more parts to charge that BFC at turn-on: View in a fixed-width font such as Courier.

. . 24V-48V . VCC . + . | . +----+--------+-------+------+-----------+ . | | | | | | . | | | | | .-. . | | | | | 4.7R| | . | | z | | | | . | | 6.2V A .-. .-. '-' . | | | | | | | | . | .-. | 100K| | | |220R |/ . 47p| | | | '-' '-' +---| T1 . === | |2.2K | | | | |>

. | '-' | | | | | . | | +-|>|----------+-----+ +------>

. | | | | | | | . | - z | | | | . | v 6.2V A | | | |< . | | | | |< +---| T2 . +----+--------|------------| T5 |\\ . | | | |\\ | . | - | | | . | v | .-. | . | | | 100R| | .-. . | | | | | 12R| | . | | | '-' | | . 5V |/ | | | '-' . CMOS o----| T4 | +------+ | . |> | - | | . | | ___ ^ |

I simulated your circuit and I can get it to produce perfect square waves. I had to balance the top 200R and the bottom 200R resistors and make a few other gratuitous changes. But God, it produces some beautiful output.

However, if I change the load or the source voltage the output is not so perfect. I tried to find a flat middle ground which would give acceptable performace; but I gave up.

Thanks anyway for the circuit.

Jeff Stout

VCC + | .-------o------o-----------' | | | | .-. | | | | |2K | | | | | | 6V | | '-' | z |/ | |< A .---| o-----| | | |>

| |\\ | | | .-. | | | | | | o------o--o--o o------ | |20K | | | | '-' .-)----. .-. | | | | | | | | | |< o-----' | .-. | | '---| | | | | '-' |\\ ||-+ .-. | | | 50R | ||

I DID see other circuits in the thread, but I did not have time to look and play with them. I had to ship the project last Friday and did not have time to play around. But I assure you I will eventually examine all replies for their technical merits.

As for my maturity, you may be right. But if one were to compare your rant and this reply, I don't think you would win the maturity prize. But I have sometimes observed that maturty and technical ability have little to do with each other.

Thank you for your help.

Jeff Stout

Pulldown nfet, directly driven from the input. Pullup pfet, ac coupled (that should be ok here, with low duty cycle). Pfet drive is just a cap, a zener s-g, and a resistor g-gnd. OK, maybe add a resistor in the pfet drain for powerup shoot-thru and short-circuit protection.

For one more part, you can dc-couple the whole thing.

You could almost do it with one spdt opto-solid-state relay, but they're a tad slow.

John

You're not grown up enough to get that circuit to work. Other simpler circuits absent feedback were posted in the same thread, but apparently you're not conscious enough to see them.