0
\$\begingroup\$

I have a relay board controlled by an STM32 to switch two AC loads (a compressor and an MD40 pump). The circuit seems to work fine but after about a day of operation, my BC847 got burned and the LM3940 LDO which supplies the STM32 with 3.3V was heating up too much. This happened again and again. I can't figure out the root cause of this issue. The load is a compressor which takes about 5A from the AC supply. A 12V supply feeds the relays and the voltage regulator (an LM2576 and LM3940) that powers the STM32.

Relay circuit

\$\endgroup\$
9
  • \$\begingroup\$ Please spend a little more time clarifying some details. What's the relay, exactly? You show Vcc tied to one side of each relay. But you say the relays use 12 V. Yet I'd want to guess that Vcc is really 3.3 V. But maybe it isn't. Because of my confusion there, I'm not willing to guess about what voltage is supplied to pins RLY1 and RLY2. Could you clarify that detail in more explicit ways? \$\endgroup\$
    – jonk
    Commented Dec 10, 2019 at 6:54
  • \$\begingroup\$ @jonk Vcc is +12V like I mentioned. I connect the two jumpers together using a cable as is evident from the circuit diagram. The relay is an HF115F rated 12VDC-12A/250V AC. What other info do you need? \$\endgroup\$
    – sixter
    Commented Dec 10, 2019 at 7:09
  • \$\begingroup\$ Cascaded BC847 without R shunt on Vbe of T4 may have excessive voltage gain to resonate from crosstalk, draw more current and overheat LDO choose R17 to give 3% of coil current and delete T3,T1 \$\endgroup\$
    – D.A.S.
    Commented Dec 10, 2019 at 7:10
  • 5
    \$\begingroup\$ @sixter It certainly was NOT clear to me that Vcc was 12 V. Even looking back I can't see that for sure. So I had to ask. Maybe it's just me. So the coil current is about 35 mA, I gather. I still don't want to assume anything about what you connected to RLY1 and RLY2. But since you are being relatively terse (and perhaps I deserve it, so I am not complaining but just admitting I feel more is needed and isn't coming yet) I'll bow out and let others offer their help. Best wishes and I expect you'll do fine. \$\endgroup\$
    – jonk
    Commented Dec 10, 2019 at 7:52
  • 3
    \$\begingroup\$ Which BC847 got burned? Be precise. \$\endgroup\$
    – Andy aka
    Commented Dec 10, 2019 at 8:11

1 Answer 1

Reset to default
1
\$\begingroup\$

The Relay specs of 12V 650 Ohms means 20 mA is expected so a Darlington is NOT NEEDED. The base drive of 2mA is more than adequate for Ic/Ib=10. 2.6V/1k = 2.6mA may be reduced depending on Vdd tolerances.

50mA coil was estimated originally without looking up datasheet. Just ignore that.

schematic

simulate this circuit – Schematic created using CircuitLab No problem using 330R instead of 1.5k

But changes for hFE of 100x100 =10k with EMI from relay wires turning relay on with spurious noise during shutoff can cause spurious load currents and VI power drain might exceed 300mW on transistor. i.e. Darlington may be overkill and too sensitive near relay coil EMI.

Normally a resistor across 2nd Vbe is all you need to shunt Q1 leakage.

This was just a SWAG.

update

@EJE reminded me to improve EMI so I have but filters on Vbe, RC filters to match coil resistor and Plastic Snubber for motor below to significantly improve relay open speed, reduce arc time, and shunt arc noise with a snubber. Use twisted pair, high surge currents on start are not as noisy as shutoff due to resonant arc currents and isolate at right angles with shielding if necessary.

SIM

enter image description here

\$\endgroup\$
9
  • \$\begingroup\$ To increase noise immunity and to better cut the conduction of Q1, wouldn’t be advised to use a R2: 1.5K ~ 4.7K between Q_base and Q_emitter? Besides an increase in parts count, would be any disadvantage doing so? \$\endgroup\$
    – EJE
    Commented May 17, 2022 at 23:51
  • \$\begingroup\$ There is no advantage to that as the source impedance is assumed lower than input. But if you expect open circuits Ok. If you have long cables then you must consider ESD protection for -5=Vbe max. Also one could add R in series with the D1 to reduce the arc duration in the contacts as this will speed it up using say R= 1/2 of Ic Vce max and that will reduce the duration of EMI emissions by increasing slew rate L/R @EJE FWIW But you are right on, this answer needs more EMI design, with attention to cable proximity/orientation, shielding, and snubbers. It smells of bigger EMI issues with LDO \$\endgroup\$
    – D.A.S.
    Commented May 18, 2022 at 0:10
  • \$\begingroup\$ the 5A compressor probably draws 30A on startup \$\endgroup\$
    – D.A.S.
    Commented May 18, 2022 at 0:16
  • \$\begingroup\$ Thank you, Tony. A very interesting “update” and discussions. I still need to learn more about Falstad’s tricks & tips to make it more useful as you did. That simple C made an RC at Vbase and may protect from erratic EMI pickup. (R + diode // C) makes a faster commutation to minimize Arcing in the contacts, but also increases the voltage spike in Vce of Q1 which is not galvanically insulated from input; I wonder how bad would this be? Nice set of EMI-related concerns! \$\endgroup\$
    – EJE
    Commented May 18, 2022 at 4:42
  • \$\begingroup\$ Why is the linear regulator getting hot here? I can't see a correlation path from the given schematics. Maybe there is an elephant in the room we can't see. Whatever the transistor stage may do here, this should not affect the regulator with this significance. \$\endgroup\$
    – Jens
    Commented May 18, 2022 at 5:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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