Continue to Site

# Voltage follower with amp-op

Status
Not open for further replies.

#### maxeen

##### Member level 1
Hello. I got two microcontrollers talking to each other (two launchpad kits) though SPI, but today I realized that one is powered with 3.3V and the other one is powered with 3.6V, which means that one's high logic level is 3.6V.

After reading the datasheet of the 3.3V microcontroller, it says that the maximum voltage applied to any pin is Vcc + 0.3V, which would be exactly 3.6V. I think that this is too risky, I am not sure if I already damaged one of the launchpads.

Anyway, to solve this problem, I was thinking of using a voltage follower amp-op to limite the 3.6V high logic level to 3.3V.

My question is: if I power the amp-op with 3.3V, can I input 3.6V on it withou damaging? I am thinking about the LM358, which is a single-supply. Its datasheet says that its maximum supply voltage is 32V, and its maximum input voltage is 32V, which does not seems to depend on supply level.

I am having trouble to upload an image of the datasheet. The maximum specs for the LM358 can be found at: https://www.ti.com/lit/ds/symlink/lm358.pdf, page 4.

Any help would be appreciated

Hi,

There is much simpler and easy way to connect 3.3V to higher voltage domain (3.6 or 5V) without affecting the speed - use Quick Witch IC

https://www.idt.com/document/dst/qs32x245-datasheet

You can get in one package single switches or multiple switches - select the proper part depend on your needs.

maxeen

### maxeen

Points: 2
Hi.

Is it possible to drop the 3.6V with a Schottky diode down to 3.3V?
Or lift the other to 3.6V

At he signal lines a diode and a pull down could also drop signal voltage.

Don't use an opamp for this.

Klaus

maxeen

### maxeen

Points: 2
no, that won't be acceptable.

maxeen

### maxeen

Points: 2
luben111, it seems that it won't be an easy task to get one of those were I live.

KlausST, I don't have a Schottky diode, but I think I can get one. I am starting to think that the op-amp is a bad ideia because of its slew rate..
How would i attach the diode between the two pins?

You need two diodes so that the signal fall-time is not seriously degraded.
Put the two Schottky in inverse parallel and connect the two in series from the 3.6V output to the 3.3V input.

maxeen

### maxeen

Points: 2
The maximum positive output voltage from a lousy old LM358 opamp is about 1.2V less than its positive power supply voltage when it has no load.
To reduce the signal level a little then why not use a voltage divider made with two resistors in series?

maxeen

### maxeen

Points: 2
SSI is based on RS-422 using using 24 AWG Standard, copper conductor, unshielded twisted-pair telephone cable with a shunt capacitance of 52.5 pF/meter (16 pF/foot) terminated in a 100 Ω resistive load.

Thus to limit voltage, examine the source resistance and load resistance to determine what series resistance is needed to limit the voltage. Many uC's now have 25 ~50 Ω sources.

Before designing a circuit, you should try to understand the +0.3V thing. It's a standard specification for analog and digital circuits with clamp diodes against Vcc. +0.3V gives a safety margin against forward biasing of clamp diodes to avoid either damage of the diodes by unlimited input current or possible latch-up. The fine print of most data sheets tell that it's permitted to feed the input with a higher input voltage if it's sufficiently current limited, to e.g. 1 - 10 mA.

This leads to a trivial solution: Use a series resistor (kohm range) at each input pin. A simple voltage divider can be used if you don't want to forward bias the clamp diodes.

LM358 has similar input specifications as the digital circuits and isn't suited for digital signals like SPI at all. It doesn't even achieve a sufficient high output level.

There are however logic families that allow to overdrive the inputs with a higher voltage, e.g. 74LVC family.

maxeen

### maxeen

Points: 2
Hi,

no, that won't be acceptable.

SSI is based on RS-422 using using 24 AWG Standard, copper conductor, unshielded twisted-pair telephone cable with a shunt capacitance of 52.5 pF/meter (16 pF/foot) terminated in a 100 Ω resistive load.
Thus to limit voltage, examine the source resistance and load resistance to determine what series resistance is needed to limit the voltage. Many uC's now have 25 ~50 Ω sources.

What are you referring to? Why do you talk about SSI and RS422? Maybe i missed something?

Klaus

Hello, thank you all for your answers and sorry for my delay.

You need two diodes so that the signal fall-time is not seriously degraded.
Put the two Schottky in inverse parallel and connect the two in series from the 3.6V output to the 3.3V input.
Even if the pin is one way communication? I thought that maybe a single diode in the pin's direction and a pull-up resistor would be enough.

To reduce the signal level a little then why not use a voltage divider made with two resistors in series?
For that I need to check the input impedance of each pin. Can't find it on the datasheet, or I am looking in the wrong section.

LM358 has similar input specifications as the digital circuits and isn't suited for digital signals like SPI at all. It doesn't even achieve a sufficient high output level.
Yes, the slew rate is too low, the clock period is higher than the time needed for the op-amp to present 3.3V on its output

I will test almost everything suggested, and I also came up with an ideia.
What if I use a scheme where I would connect the 3.6V pin in series with a Schottky diode and a 10k resistor. The voltage in the resistor would be 3.3V or less. How about that?

For that I need to check the input impedance of each pin. Can't find it on the datasheet, or I am looking in the wrong section.
I presume you are working with CMOS devices which have effectively infinite (high MOhm) input resistance. Voltage divider or series resistor values are more imposed by IC and wiring capacitance, at least for higher SPI data rates. A low kOhm range is usually appropriate.

...................
You need two diodes so that the signal fall-time is not seriously degraded.
Put the two Schottky in inverse parallel and connect the two in series from the 3.6V output to the 3.3V input.
Even if the pin is one way communication? I thought that maybe a single diode in the pin's direction and a pull-up resistor would be enough.
..........................
The communication may be one-way but the signal goes high and low. Two diodes will allow the driver to drive the signal high as well as drive it low, minimizing any rise/fall time degradation of the signal.
But one diode and a pull-down (not up) resistor may be sufficient

maxeen

### maxeen

Points: 2
Hi,
post#11:
What if I use a scheme where I would connect the 3.6V pin in series with a Schottky diode and a 10k resistor. The voltage in the resistor would be 3.3V or less. How about that?

good idea.

post#3:
At he signal lines a diode and a pull down could also drop signal voltage.

Klaus

I will test almost everything suggested, and I also came up with an ideia.
What if I use a scheme where I would connect the 3.6V pin in series with a Schottky diode and a 10k resistor. The voltage in the resistor would be 3.3V or less. How about that?

At lower data rates this will work, but at higher data rates the simplest solution would be to use 74LVC gates, at 3.3V supply they accept input signals up to 5V.

Hi,

What are you referring to? Why do you talk about SSI and RS422? Maybe i missed something?

Klaus

SPI is the single-ended unbalanced high speed version of SSI which is based on RS-422

In order to reduce overshoot, it must have low controlled impedances for source, cable and termination. I would expect the drivers to be 25 or 50 Ohm CMOS types.
Don't assume. Ask TI Forum from their website , if it is not documented. I would measure the input impedance and expect 100 Ohm either split between the rails with two R's or 100R to ground.

I suggest to attenuate the pulse height AND reduce overshoot by increasing the source impedance with something like 25~50R and check load impedance and ribbon interface.

Again check with ti's web community http://www.ti.com/lsds/ti/csc/support_Americas.page

https://e2e.ti.com/support/clocks/f...isearch=TI-e2e&keyMatch=spi level termination

We dont know which modules you have, it may be trivial or not.

maxeen

### maxeen

Points: 2
Hi,

Yes, now I see the relation...

Klaus

Status
Not open for further replies.