I have a bank of relays (5) that I'm looking to run from my Pi. I'm using 5V TE Connectivity relays, powered by Sparkfun's high power relay kits, which use a NPN transistor to trigger the relay.

So far the problem I'm having is I can't get the relay to switch. Do the GPIO output pins not supply 5V? I thought the Pi did on the GPIO.

  • 4
    No the Gpio pins are 3.3 volt. – Steve Robillard Feb 21 '15 at 4:18
  • Then do you know of a way I could modify my circuit? – hightekjonathan Feb 21 '15 at 4:19
  • 1
    It will require a 5 volt supply and remeber to connect all of the grounds together - I am doing this to switch a tower signal light like this one ebay.com/itm/… – Steve Robillard Feb 21 '15 at 4:21
  • 2
    I would suggest getting it working with a separate power supply first then tackle the complexity of the power needs of the Pi and portability. Like they say the best way to eat an elephant is one bite at a time. – Steve Robillard Feb 21 '15 at 4:47
  • 1
    All grounds are tied together. And the 5v positive is connected to common on the darlington. I followed the diagram. – hightekjonathan Feb 27 '15 at 5:27

Why not simply like this?

Raspberry Pi switches between 0 and 3V3, more than enough to saturate Q1, which takes over the "heavy" work: switching the +5V relay on/off. Depending on the relays you're using, small modifications for D1 and Q1 might apply.

Raspberry Pi controlling a relay

| improve this answer | |
  • Excellent answer! And a very safe way of doing this. – Piotr Kula Mar 1 '15 at 21:50
  • 3
    Nice answer, that should solve "relay-questions" once and for all. Maybe it would be good to add that the fact that supply voltages of the Pi and the relay should either be the same or at least share a common GND. While this might seem obvious to the electrical engineer it could be a pitfall for beginners. – Ghanima Mar 4 '15 at 14:57
  • 1
    What's the purpose of R2? – erikH Nov 5 '15 at 7:28
  • 2
    @erikH: this resistor will make sure the base of Q1 has a fixed level (in this case, connected to ground) in case the input at R1 would float. If you're sure the input will never float, then R2 is not needed, but it's a kind of "safeguard". If the input can float and you don't have R2 connected, then the transistor can start switching in an uncontrolled way. – GeertVc Nov 5 '15 at 16:33
  • 1
    When the relay is activated, it builds up energy in its coil. When the relay is all of a sudden de-activated (that is, switch Q1 opens) the energy needs to be released. If D1 would not be there, then Q1 has to absorb all that energy. Since we're talking about huge voltage spikes (much more than the allowed C-E voltages of Q1), it might almost for sure damage Q1. D1 prevents this: it kind of 'short circuits' the energy and as such, the energy will be dissipated in the relay itself in stead of the transistor. – GeertVc May 10 '19 at 15:02

# Introduction #

The OP would like to use Rpi to safely control a bank of 5 Sparkfun's Beefcake relay module. He had a problem because Rpi GPIO logic level is 3.3V, but his relay uses 5V logic control. He wants to know how to modify Rpi to get around the logical level disparity problem. His choices including the following: using the transistor BC5468 to drive the relay coil; getting an opto isolation relay and drive it using ULN2803; using a source driver such as UDN2981, ...

After investigation, I now suggest a couple of solutions, with their respective pros and cons. The OP can choose a solution after trading off risk, reliability, cost etc.

# Contents #

Solution 1 - Modifying NPN transistor's biasing resistor

Solution 2 - Using UDN2981 to shift up Rpi's 3.3V GPIO signal to 5V

Solution 3 - Using 74HC03 and 74HC04 to shift up Rpi's 3.3V GPIO signal to 5V

Solution 4 - Using 74HCT125 to do logical level convering

Solution 5 - Using TXS0102 to do logical level converting

Soution 6 - Using 2N2222 to do logical level converting

Solution 7 - Using 2N7000 to do logical level converting

FAQ1 - How to power Rpi and relay module and tie grounds together

FAQ2 - How to avoid floating input problem

FAQ3 - My relay is always on, whether High or Low input, is it because the Rpi Low signal is not low enough?

FAQ3 - My Rpi GPIO Low signal cannot turn relay off, but setting GPIO as input would do. Will I hurt my Rpi if I do so?

Hardware Troubleshooting Suggestion

Software Troubleshooting Suggestion


# Solution 1. Modifying NPN transistor biasing to make it 3.3V compatible #

The are two general types of solutions:

(1) modify the module's 5V logic level input circuit to adapt to 3.3V signals,

(2) use a 3.3V to 5V logical level converter to shift up Rpi's 3.3V signals to 5V.

I now start with (1).


The Sparkfun's Beefcake relay module has an NPN transistor 2N3904 (Q2) driving the coil (U1). It is designed for Arduino's 5V logic signals.

SparkFun relay control kit schematic

I have a similar NPN transistor module KY019 which can be driven by Rpi's 3.3V signals. So I checked out its input signal requirements to find why KY019 can entertain 3.3V signals but Beecake can't.

KY019 spec

I found that KY-019 has a trigger level of 2.5V and 0.1mA. This signal is amplified by the NPN transistor to 50mA, high enough to engerize the coil to activate the relay.

Tongling coil spec

Rpi GPIO (with High level above 2.8V, and max current limit of 16mA), can comfortably source 4mA, there should be no problem directly driving the module.

The coil has a response time of 10mS. I programmed Rpi GPIO pin 17 to toggle the relay module at 40mS period (25cps) and found the relay clicking happily as expected. (I was using 2 meter long connecting wires for the GPIO signals, so the signal at at relay input end is a bit noisy.)

KY019 toggle test

How to modify the Beefcake module to make it compatible to 3.3V logic

The Beefcake NPN transistor has a current limiting resistor R2 of value 1K. This resistor limits the base current at Arduino 5V logic high level. The base current within limit, after amplification (usually hFE > 100), is big enough to engerize the coil.

Calculation of Arduino 5V GPIO current into Beefcake relay module:

Arduino current i ~ (4V [Arduino High] - 1V [Vce(sat)]) / 1K [R]) = 3V / 1K = 3mA

However, the Rpi's logic High signal is lower than Arduino, so the corresponding limited current is smaller and after amplification is not big enough to drive the coil.

Rpi current i ~ ((3V [Rpi High] - 1V) / 1K = 2mA

The modification is simple - just replace 1K R2 with a smaller resistor, say 510R.

Rpi current i (after modification) = (3V - 1V) / 501R = 4mA

I am making education guess based on circuit analysis and experimentation. I think my guess is 90% likely correct.

Risk Analysis

Though the small signal NPN transistor 2N3094 can be used for small load switching, it is not that reliable. For relay switching, it is safer to use power transistors such as SS8050, UDN2981, specially designed for inductive loads.

The OP wants a safe method which would not fry his Pi, so for reliability, a source driver such as UDN2981 is the way to go.

/ ...

# Solution 2 - Using UDN2981 to drive the Beefcake relay module #

The comments point out that the OP's Sparkfun Beefcake relay module is high level trigger, therefore the commonly used sink driver ULN2803 cannot be used. A driver similar to ULN2803, but current sourcing, rather than current sinking, should be used instead

I think UDN2981 is a suitable driver for the OP's relay module.

I verified successfully UDN2981 driving a high level tirgger relay module similar to the Beefcak, and ULN2803 a low trigger one. Below is a summary.

UDN2981 controlling High trigger, NPN transistor input type relay modules

I first manually tested UDN2981, without connecting to Rpi, to blink 4 LEDs, to make sure the circuit is working OK.

uln2803 udn2981 photo

Then I setup 4 NPN transistor input type relay modules (KY019), and connected the 4 relay module inputs to 4 UDN2981 channel outputs.

KY019 x 4

Then I connected 4 Rpi 3.3V GPIO pins directly to the 4 UDN2981 channel inputs. I used the following python function to toggle 4 relay modules at 25 cps.

python function to toggle 4 relays

The result was good. The 4 relay modules click and LEDs blink at 25cps as expected. The Rpi GPIO output signals stayed near 3.3V, and UDN output signals around 4.0V, implying that no input was overloaded.

Toggle module waveform

UDN2981 controlling High trigger, opto-isolated relay modules

The OP also considered using opto isolated relay modules because they are safer. I have successfully used the same UDN2981 to control 4 High level trigger, opto isolated relay modules (MK055).

MK055 relay modules

Actually UDN2981 can be used to control any kind of High trigger modules, no matter NPN transistor or opto isolated types.

However, for Low trigger modules, whether PNP transistor or opto isolated, source driver UDN2981 does not work, ULN2803 or other sink driver should be used.

ULN2803 controlling Low trigger PNP transistor input or opto isolated relay modules

I successfully verified ULN2083 sink driver can control 4 Low trigger opto isolated relay modules. I first tested manually blinking 4 LEDs, then used the same python function above to test the 4 modules. The results was also good.

MK101 relay module


ULN2803 and UDN2981 Pros and Cons


  1. ULN2803 and UDN2981 can be directly driven by TTL or CMOS logic signal with 3.3V or 5V supply voltages.

  2. Their rated 500mA outputs with clamp diodes are suitable to switch relays and stepping motors.


  1. ULN2803 and especially UDN2981 are not so common.

  2. They have 8 channels and therefore have a bigger 18 pin DIP package size. For less channels, more common 74HC03/04 or 74HCT125 with 14 pin DIP package are more common and easier to handle..

# Solution 3 - Using 74HC03 and 74HC04 to shift up RPi's 3.3V GPIO signal #

Using UDN2981 to drive a relay module is a big over kill, because they are designed with built in fly back diodes to directly energize relay.

The UDN2981 is not common and not for beginners to experiment. For beginners, the very common and cheap logic gate ICs, 74HC03 Quad NAND gates, and HC04 Hex Inverters can do the same job as UDN2981, shifiting up 3.3V logic signals.

I have successfully verified HC03 and HC04 shifting up 3.3V logic to 5V and found it working for both transistor input and opto isolated high level trigger modules.

hc03 04 converter schematic

hc03 hc04 ky019 mk047

hc03 hc04 ky019 ky047

# References #

R1. How does an Electric Relay work? - TechyDIY

R2. Relay Switch Circuit - Electronics Tutorials

R3. Beefcake Relay Control Hookup Guide - SparkFun

R4. Digital Buffers and Tri-state Buffers - Electronics Tutorials

R5. Pull-up Resistors - Electronics Tutorials

R6. Logic Levels Tutorial - SparkFun

Arduino Voh 4.2V, Vol 0.9V

R7. Rpi GPIO pin voltage and current specification

Rpi Voh 2.4V, Vol 0.7V

R8. Bipolar Transistor - Electronics Tutorials


# A.3 Logical Level Converter using HCT125 #

So I tested another up converter, HCT125. I was happy to find that it works well. The HCT125 converted 5V0 signal did not drop when connected to the NPN transistor driven relay module.

HCT125 level converter

/ ...

End of Appendices

** * Long Answer To be deleted * **

This long answer is too long winded and messy. I am now trying to remove the irrelevant paragraphs and perhaps replace them by asking relevant question and answering myself.

How to check Photo Coupler / Opto Isolated Relay Module

  1. Get a jumper wire.
  2. Connect one end to signal/input pin of relay module.
  3. Hold the other end and touch the Vcc(+) and Gnd(-) pin and check the results below.

Relay Module Test Results Table

2.1 Transistor input type

For the popular bipolar NPN transistor input type, the sourcing driver signal (Rpi GPIO signal or RPi GPIO signal after 3.3V to 5V logical level conversion) goes to the base of the transistor through a series LED and biasing resistor.

Example of transistor input type (BJT NPN) relay module

There are other not so popular relay switching circuit as described in this relay switching tutorial

2.2 Photocoupler input type

The photocoupler input type relay has a phtocoupler as the input. The photocoupler drives another transistor which in turn drives the relay coil.

Appendix C - Logical level converter using TXS0102

Now I know that Rpi GPIO can directly drive the relay module, but there are two problems. First, the GPIO signal with a long connecting wire is noisy, therefore not that reliable. Second, the flywheel diode 1N4148 might not completely suppress the coil back EMF, and if unluckily the 1N4148 breaks down, or not properly connected (poor contact, dry soldering joint etc), the back EMF might damage the Rpi.

So I decided to use a logical level converter to shift up the Rpi GPIO signal from 3V3 to 5V. I first tried TXS102 converter and found it working well. Besides shifting up the GPIO siganl, the noise at the high level is also greatly reduced.

TXs0102 photo and waveform

However, I found a big problem when feeding the converted 5V GPIO singal to the relay module. The relay still turned on and off as before, with the 3V3 signal, but when I used the scope to check out the waveform, I found the very surprisingly that 5V signal dropped by half, to 2.2V.

I suspected the reason was that TXS0102 can sink current much better than sourcing current to the relay module. To verify my guess, I fed the 5V signal to another relay module, a pull down photocoupler type, model MK01.

This time I found the 5V signal did not drop any noticeable amount.

So I quickly concluded that the NPN transistor type relay module is a bad choice. I would stop testing this kind of relays from now, and move on the photocoupler type of relays.

I also tested another photocoupler driver module MK101. This module has a jumper to select Higher trigger or Low trigger. I found that for Low trigger, the TSX0102 converted 5V signal level is not affected. But when Low trigger is selected, the converted 5V signal level dropped to around 2.5V, though the relay is still working.

txs0102 test high low trigger modules

Appendix E - Logical level converter using HC04

HCT125 is not so common. So I tried one more converter circuit, using HC03 quad open drain NAND gate and HC04 hex inverter. When I tested the HC04 output, I found it very noisy. I guessed one reason was that I was using dirrerent power supplies, one for rpi, another for the converter. Even I connected the ground points of the power supplies to make a common point, the noise did not go away. I then used one power supply for both rpi and converter, and the noise disappeared.

Noisy converter output

enter image description here

I tried the HC04 output signal for the relay module in Low trigger mode (which requires sinking current, but not in High trigger mode (which requires sourcing current), . So I am going to add the HC04 hex NOT gate which can source current to the relay module.

Appendix F - HC04 Level Converter Floating Input Problem

Last time when I first tried the HC03 based level converter, on a photocoupler relay module, I found that if I left the input floating, the module picked up the noise and the relay switched on and off crazily. I thought the frequency was perhaps 1kHz. I was not sure if it was some sort of positive feedback oscillation. But when I used the scope to check out, I surprisingly found it was 50Hz! I guess it is some sort of resonance. But I don't know what is the difference between resonance and oscillation. Perhaps I should goggle again. Anyway, I think I need to add a pull up/down resistor somewhere.

HC04 Level converter floating input problem

Below to be shortened or deleted

# Appendices #

# A1. Opto isolated / Photo coupler relay module board and schematic #

Opto isolated relay module has a photo coupler which is 4 pin IC. The picture below shows a photoCoupler PC1 (with its 4 pins labelled 1, 2, 3, 4 in green) and a transistor Q1. The ICs are not always marked. In this picture, PC1 is EL354, and Q1 8050.

Photocoupler input type relay module photo

Photocoupler input relay module schematic

Diagram Links

 35: https://i.stack.imgur.com/cWkRi.jpg

| improve this answer | |
  • 1
    tlfong01: "So Rpi's 3V3 GPIO signal can directly drive the relay.": You shouldn't say that because reading this line might give the impression to non-technical people they can directly connect a GPIO pin of the RPi to the coil of a relay. With devastating consequences... There's a reason for the addition of the 2N3904... You should say "So RPi's 3V3 GPIO signal can directly drive the relay module" instead... – GeertVc May 11 '18 at 8:27
  • Thank you for pointing out my careless mistake of not being thoughtful of the non-technical people. And when I made the correction, I found I made another mistake of inserting two identical pictures of the KY019. Actually the first one should be SparkFun module's schematic. – tlfong01 May 11 '18 at 13:40
  • 1
    No problem, we're here to help each other... :-) – GeertVc May 11 '18 at 15:45
  • The reason there is a discrepancy using a GPIO directly vs. a GPIO with a shifter is the shifter is intended to deliver a specific voltage (5V). A 3.3V logic pin, on the other hand, is spec'd with thresholds indicating the difference between low and high. For the pi this means anything over 1.3V is high. If you are sinking through resistance to ground (i.e., pulling current), you should get 3.3V, but that is probably not what the input on the relay does -- it floats. – goldilocks May 11 '18 at 16:03
  • @goldilocks: Yes, it is confusing. OA says his relay kit uses a NPN transistor to trigger the relay [coil], One answer seems to suggest to modify the transistor [to BC5468]. Another suggestion is to use darlington ULN2803, .. I need to read the comments one more time to clear my mind. – tlfong01 May 13 '18 at 1:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.