RS-232 to LVTTL

parallel a resistor with D1, to ensure the transistor stays off over temperature with the RS232 disconnected (Icbo has to go somewhere).

Cheers Terry

Doesn't the 22K (or is it 10K now?) resistor on the base take care of this? In a MC1489 this resistor is more like 3.8K.

Tim.

A BC847BPN is spec'd at Icbo = 5uA at Tj=150C. 15nA typ.

I recently picked up a copy of RF Shea's "Transistor Circuit Engineering" (its really a GE book), for the whopping price of NZ$0.50. The chapter on bias stability discusses using leaky diodes to compensate for leaky transistors over time, temperature and sock colour.

yep. I'd use a single 10k quad-pack, and have a 2V5 (cf 3V3) output.

It's even worse. Older versions allow for + / - 30 volts :-)

It would be nice to see it, though :-) IIRC you won't get far beyond - 5.5 volt :-)

I read in sci.electronics.design that John Larkin wrote (in ) about 'RS-232 to LVTTL', on Mon, 11 Apr 2005:

In another thread, we got from colloidal silver to blood, which is Gore, so that draws in the Florida election. Now, can we get from that to RS-232 or LVTTL?

FWIW, the mmbt2222a and the 2n3904 data sheets both list Vebo max at 6 Volts.

--Mac

If that is all you can say, then why join the fray at all?

I really only had one question, and it was answered satisfactorily by two people within just a short time.

Why did I ask? To benefit from the experience of others, of course. I've designed transistor switch circuits many times on circuit boards, but I've never done an RS-232 converter. Usually when interfacing RS-232 to logic, I have used a transceiver. But in this case, for various reasons, I don't want to.

As it turns out, I did receive a useful suggestion from John Larkin. He pointed out that I can't count on high voltage drive from some wimpy RS-232 sources (e.g., laptops), so I probably should be more conservative on my overdrive calculation.

After that, in most of my posts I was just trying to be polite, and in some cases to lead other people to enlightenment.

regards, Mac

It's 10k now. The diode is what protects against reverse biasing the base-emitter junction, not the resistor. When the RS-232 input swings into the negative range, very little current will flow through the 10 (or 22) k resistor, so the voltage drop across the resistor will be negligible. This means that the base will be at input Voltage.

However, when the input Voltage across the base-emitter junction reaches the breakdown (Zener) voltage, some current will start to flow. At that point, the resistor will limit the current. But by then it may well be too late for the transistor. I'm not an expert on the behavior of transistors after they have been abused. I try to be a good little boy and avoid exceeding absolute maximum ratings. ;-) In typical transistors, the Veb max is 6 Volts. So the diode stays.

By the way, the MC1489 also has a diode protecting the input transistor against reverse bias.

Oh, and since you asked, the rule of thumb I usually use for sizing base and collector resistors is that the worst case (i.e., lowest) base current should be 1/10 the minimum acceptable collector current. In other words, I use a design beta of 10. In practice, what this often means is that the collector resistor should be around 1/10 the value of the base resistor. In my original post, I was using an input voltage of 12 Volts, that's why I initially chose 22 k for the base resistor. But as was pointed out, some drivers barely make it to 5 Volts. So the 10 k base resistor is safer.

In the MC1489, the base resistor is 3.8k, but the pullup on the collector is 9k. At first glance this would seem to be a forced beta of less than three. However, there is also a 10 k pulldown resistor on the base, and there is also positive feedback from a later stage which goes to the base. I won't pretend that I understand all the details of the circuit, but it is clearly not an exactly analogous situation.

--Mac