High Side Pulse Transmitter

Yup. Might could use fets, or maybe darlingtons, depending on the swing requirements.

A depletion fet might be fun for the upper one.

John

It's a bunch of conceptual garbage: View in a fixed-width font such as Courier.

. . . V+ . | . +-----------+ . | | . z [Rs] . a | . | ||- . +--------||>

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

. | | . | ||- . CMOS IN>---------|--------||< . | ||- . | | . [Rb] [Rs/2] . | | . --- --- . /// /// . .

Mt latest "parts rich" el cheapo looked something like this- it can drive indefinitely long lines if you lower the baud rate and easily does

1200 baud into 1.5u- you can make it noninverting by putting 5V on T4 base and driving the bottom of the 1K emitter resistor with the CMOS signal: View in a fixed-width font such as Courier.

. 24V-48V . VCC . | . +----+---------+----+-------+-----------+-------->

. | | | | | | . .-. | | .-. .-. .-. . 100K| | | |47p| | | | 4.7R| | . | | z === | | | | metal| | . '-' A6.2V | '-' '-' oxide'-' . | | |2.2K| 220R| | . | | +----+ | |/ . | | | | +---| T1 . | | | | | |>

. | | 56R | | | | _ . | +-|>|-|>|-|__|-+ +-----+ +--|_|--->

. | | | | | | F1 . | | | v | | 1/2A . | | | - | |< . | | | | +---| T2 . | | | | |\\ . | | | |< | . +------------------------+-----| T5 | . | | | |\\ | . | | z | | . .-. | A6.2V .-. .-. . | | | | 100R| | 12R| | . | | | - | | metal| | . '-' | v '-' oxide'-' . |1.8K| | | | . | +----|--------------+-------+-----------+ . 5V | | - | | . CMOS |/ | ___ ^ |< | . o----| T4 +-------|___|--+-----| T3 | . |> | 470R |\\ |+ . | | | === . .-. .-. .-. |100U . | |1K | |1.2K | |100 | . | | | | | | | . '-' '-' '-' | . | | | | . +---------+----------------------+-----------+-------->

. === . GND

What's unusable about it? It's simple, it satisfies all the stated requirements and, more important, it should work in the application.

John

I think he was planning on use complementary MOSFETs.

Right- we told him to drop that "fast as possible" requirement.

Now you know that circuit is unusable- so you haven't "done" anything with those five parts.

I've uploaded one final circuit and simulation to alt.binaries.schematics.electronics; Win's circuit actually.

Jeff Stout

Hi John. I surely d>

Fred, one comment. Small is beautiful. :>) Oh, a second comment, wasn't the circuit supposed to be non-inverting?

--
 Thanks,
    - Win

That's a good point, but, if your circuit is modified by making T4 a common base, then you'll need +5 as well, right?

--
 Thanks,
    - Win

Jeff, did you ever tell us what you're working on? It looks like power with a signal on the same wire. What's it for?

--
 Thanks,
    - Win

Yup, he always does.

John

Yep- I told him how to do that by making T4 a common base, but one thing everyone overlooks with that idea is that he might not have 5V available at the driver, just the signal, so that's just more parts to derive the base voltage from the high voltage.

Talk is cheap- build the damned thing and then come back with your results:-)

[snip]

Yes, this circuit provides the power and transmits data down a "coax" cable which can be over 6 miles long. Since their can be many instruments at the other end of the cable, this circuit is used to transmit an address for each instrument. The instrument(s) on the other end of the cable make measurements and transmit the data back using current pulses in a variety of formats (AM, FSK, baseboand serial, etc.). The high voltage found in the design is to there to overcome IR voltage drop along the many miles of cable.

Jeff Stout

[snip]

No.

I'll review all the design ideas and recommend a solution to my boss in a couple of weeks.

Thank you, Jeff Stout

Oh, one more thing. The reason it has to have true logic is because we already have about a thousand units in the field and my boss doesn't want to change the software. Yea, I know software is easy to change. But it's his name on the shingle outside the door so he gets to make these kind of decissions.

Thanks much, Jeff Stout

--
 Thanks,
    - Win

Did you notice I posted some additional suggestions under the same SUBJECT: name, but under a different thread (due to a mixup)?

Date: 17 Oct 2005 09:51:48 -0700 Message-ID:

In particular, I felt the issue of surviving possible short circuits needed more attention. Unless the source power supply has its own average current limit at a fairly-low level, e.g. 250mA, the output driver we've been discussing needs to be able to handle a continuous short, which would be at a fairly high power dissipation for many of the suggested configurations. As I stated there, using a current- limit BJT rather than a series current-limiting resistor is a start in improving the circuit's performance and robustness. A foldback C-L provides further improvement. A thermal limit in the pass element is better yet, in many cases, hence my suggestion of an LM317 or LDO.

One workaround is to transfer the problem to the source power supply, which may already be prepared to handle it. If its current limit is higher than the driver's capability (750mA in my example circuit), then adding a power zener (AKA transient suppressor), or an active say 10V zener, may be another solution, e.g., limiting the pullup- element's dissipation to say 750mA * 10V = 7.5W. But 7.5W is a serious bit of power, hence my suggestion to use a thermal-limiting device, with a smaller heatsink, cutting-off at a lower power level.

If the source supply is powering multiple sets of data cables, then the short-circuit problem must be handled by each driver entirely on its own. In this case a thermally-limiting driver is very attractive. One simple solution is to use an LM395T active transistor, which can be dropped right into the NPN slot in most of the designs you've been considering.

But the LM395 is an old design, costs $2.64 each (qty 25), and isn't second sourced, hence the suggestion of using an LM317HV, etc., instead. If cost is not an issue, and this is a onetime buy-and-build situation, the LM395T may be an ideal solution. DigiKey has 1202 in stock.

. rail -----+-----/\\/\\--, . 24 to 50V | | . 6.8k | . | |/ Q3, LM395T . +---------| active power transistor . | |\\e . | | 1.0 noninverting 6V swing . +--|>|-|>|--+---/\\/\\----> output to 1.5 uF load . | | dv/dt = i/C = 6V in < 12us . | |/e (assuming +/-Imax = 750mA) . +---------| . | |\\ Q2 . Q1 | | 1A darlington . |/ | Zetex ZTX603 . +5 ----| | 22 . |\\e | '--/\\/\\--- gnd . 4.3k | V 1.0mA . o--/\\/\\---' for logic LO . 5V cmos logic

The two diodes protect Q2 in the event of an instantaneous short (the LM395 has a 500-ohm resistor in series with its base).

--
 Thanks,
    - Win

Sounds fun- just be sure to TVS the hell out of the thing...at least to the degree of that lightning induced 1500V standard. I would also substitute bulk ceramic for the power resistors, remove the output fuse, and **add even more parts** for a loop back capability of actual serial output and fault detection.

Let's just face- it is impossible to make this circuit.