When is an op amp not an op amp?

Antique opamps tended to not swing to the rails very well.

The output from the op amp is either +12V or -12V. The op amp drives a 1 kohm resistor which is grounded through several different value resistors and a diode so the +12V through the 1 kohm resistor is pulled down to 9V, 6V, 3V or 0V.

No one is asking you to design a circuit. I'm asking if it is in any way reasonable to use a 820 ohm resistor in place of the 1 kohm resistor, so that the limited drive level of the op amp (not to the rails) is mitigated. This drive level is specified very crudely, and the circuit would not work if the op amp were remotely like the actual specification.

Of course it doesn't need an op amp. That's an entirely wrong part for the job. It's actually in the circuit like a comparator with the - input wired to a fixed voltage and the + input wired to a 5V MCU. As I said before, a much better part is an analog switch, 2:1 mux. Then the signal is ±12V, period and the 1 kohm resistor can be a 1 kohm resistor.

The real problem is that the guy who designed it, can't seem to understand that it's a bad idea to use this part in this way. He is relying on the part being much better than the data sheet specification.

Given a chip with a finite output impedance and a desired drive impedance, it's perfectly reasonable to add a resistor that accounts for the chip. I make 5-volt sub-ns 50 ohm drivers with a cmos chip known to be 5 ohms, in series with a 45 ohm 1206 resistor. Nobody makes a zero ohm driver IC and relays are too slow.

We know that an FPGA driver pin is 20 ohms, so we add a 30 ohm resistor to drive a 50 ohm line. We know the FPGA is 20 ohms because we measured it.

Not exactly. Different values are used to pull down the junction with the 1 kohm resistor. The resistors are on two opposite ends of the cable. The 1 kohm in the EVSE and the various values in the car. The resulting voltage on the cable indicate a state the car wishes to communicate to the EVSE. The diode is in series with the car resistor so it only impacts the positive part of the pulse waveform, which is PWM at 1 kHz.

??? Who said I don't know. It's in the spec. The spec gives values for the resulting voltages on the interface, which, much like RS-232 have a range for the actual voltage, as well as a range for the detection thresholds, leaving noise margin.

The issue I'm asking about is the misuse of the op amp. The only relevant specification in the LF353 datasheet I could find was when powered by ±15V the output is guaranteed to swing ±12V. This circuit uses ±12V power and to work correctly, needs to drive the 1 kohm resistor with something pretty close to ±12V.

I prefer to use parts according to the spec in the data sheet and not assume that any specific behavior beyond that, which might appear on the bench at the moment is also specified. Stuff changes. That's why they provide data sheets. The output drive details may change, but they are guaranteed to meet the data sheet.

Normally, the -12V is a power supply. Feel free to be creative.

The diode is there to prevent loading the -12V output.

Diagrams are better than words to describe circuits.

I have not looked at the lf353 but I would be very surprised it its output range is specified over 1k. Back then 2k used to be the norm for that. And of course it would get no closer than 1-2 (3?) volts to the power rails, this is certainly specified as well.

There is no "tapped" divider. The 1 kohm resistor is connected to the interface point. That is where voltages are measured. The car uses *different* resistors to load the interface point to bring it to different voltage levels to communicate information about the state, "Vehicle connected / not ready to accept energy", or "Vehicle connected / ready to accept energy / indoor charging area ventilation not required".

The op amp initially drives +12V. When the car pulls the signal down to 9V, the EVSE drives a 1 kHz PWM signal, ±12V, with the car loading the positive portion to indicate state. The diode prevents loading the -12V portion to maintain a significant voltage pulse for the PWM to be measured.

Yes, but no. The drive needs to be ±12V. Otherwise, you seem to be starting to understand. If you could put the diode on the correct side of the interface and accept that the "tapped divider" is actually different value resistance, selected by the car, then you would be home free.

But, again, I will explain that you don't need to understand any of this, to respond to the original issue. The op amp is not specified to supply close to ±12V to even a 10 kohm load, when powered from ±12V rails. So I can see no justification to use it in this circuit. That's why I recommend the mux.

Ok. I guess it's beyond you.

No, but it is specified in the same way with a range for the receiver to detect and a range for the driver to drive, with the difference being the noise margin. That's why I said "much *like* RS-232".

Is there some purpose to your statement? The interface voltage targets are +12V, +9V, +6V, +3V, 0V and -12V. The steady -12V is from the EVSE to indicate an error condition. The other voltages are produced as a result of the load the car puts on the +12V signal through the 1 kohm resistor in the EVSE.

So this escaped your attention, or you don't understand?

It is specified with ±15V rails, and a 10 kohm load, to be within ±12V. So clearly if you reduce the supplies to ±12V, it won't get anywhere near the rails. The spec allows ±5% tolerance when open circuit.

Strange. I have seen tables of the interface voltages with tolerances. I don't see that in the spec. Instead, they specify the resistor values and the open circuit voltage of the ±12V drive. Seems the voltage table I was looking at was from the EVSE maker, not the spec.

Other than -12V, yes.

Yes.

You really should learn to read more carefully. Go back and search for the word "state". I've used it several times.

Yes, this is not a new idea. No, current is not measured, the voltage at the interface is. No, the EVSE does not always output a DC signal, again, read, read, read.

Dear God! It is clear that you will never read what I write, no matter how clearly I try to explain it.

You mean designing so a design will actually work when something beyond your control (or specification) changes. Yup, I'm all about that.

Why would a statistical analysis be required? This is a data sheet specification on the output voltage. The part is not rated to drive anywhere near the rails. Using an op amp in general is a terrible idea, even if it were rated to drive near the rails.

I think your question should be how accurate is saturation voltage drop with various loads?

Although Op Amps make non-precision comparators, they do work for logic-level specs if you use less than Iout min. The LF353 has Ios min = 10 mA, when saturated which is independent of Vcc from +/-(4 to 15) V from the tables and Figure 7 , due to the Idss bias. Vout has additional fixed series R's 100+200 in figure 1

Using Kirchoff's Law , one can compute the worst-case Voh min and satisfy the analog properties to meet a certain logic state. Is it perfect ?

- that depends on all the specs: - for price , performance margin and availablity. - It satisfies all these, then yes, it's perfect.

Can it be improved? - Yes, but that depends on which criteria you need to meet.

Is it necessary? - not if it always meets spec by design.

Sorry, what do you mean by "logic state"?

Does the Ron of FPGA output pins when in open drain mode tend to be repeatable enough in general that you can voltage sense with a comparator for over-current protection?

I don't know, but semis from dependable suppliers tend to be very repeatable. But not everything is specified, so we measure and guardband.

You could certainly protect an open-drain driver by checking its voltage when it's known to be on. That's commonly done to protect mosfets too.

Ricky wrote: ==========

---------------------------------

See figure 6 of the ST data sheet - shows max p-p voltage into a wide range of resistances.

Yep. Figure 6 is for ±15V supplies and at 2 kohm load, it looks like the swing is reduced by 6V, or ±3V relative to the rail. Drop the rails to ±12V and you can only assume you still lose about ±3V giving you a ±9V output, far below what is required. Oh, also, this curve is not a guarantied value. It's probably "typical", and the output could be worse.

Thank you for that information.