Nanosecond transient from relay with already supressed coil

Thread Starter

Goxeman

Joined Feb 28, 2017
77
Hello everyone,

I have an ULN2003 transistor driver array to control different relay coils, the ULN2003 is controlled by a MCU. For coil supression I decided to use a 4148 diode with a 6.8V zener diode. The attached schematic is just a design, it is not a real PCB in which I could measure.

Initially, the PCB in which I am making tests doesnt have the LDO as far as the DCDC supplies 3.3VDC. In my new design that I attached I increased the output voltage from the DCDC to 5V willing to use a low high SRR LDO to power the most sensitive ICs like the MCU.

Schematic relay.JPG

The fact is that within the PCB where Im making tests (everything works at 3.3V) is working well but I am concerned about some transients I measured and that spreading through the power net (I guess). I want the circuit to be very reliable as far as I am going to implement a bluetooth IC also besides the MCU and I dont think that the measured transients are good any sensitive IC


Transient relay.png
This shows all transients in I/O pin of the MCU while de-energizing a relay. Note that this I/O pin of the MCU is not connected directly to the relay or the power rail, it is just a normal I/O pin


Transient relay 1.png
This shows a zoomed transient, how wide it is, it lasts 129nanoseconds


How would you supress this transients? I think that even if everything works well while testing, it will damage the circuit somehow anytime
 

ericgibbs

Joined Jan 29, 2010
9,078
hi G,
Does the actual circuit have decoupling capacitors on the power rails.?
Is the Zener diode in the relay coil circuit required to speed up the relay release.?
E

Corrected Error
 
Last edited:

bertus

Joined Apr 5, 2008
20,176
Hello,

Why do you need a zener in series with the diode?
Replace both with a UF4007.
The transient will not go higher than VCC+Vdrop(UF4007).

Bertus
 

Thread Starter

Goxeman

Joined Feb 28, 2017
77
hi G,
Does the actual circuit have decoupling capacitors on the power rails.?
Is the Zener diode in the relay coil circuit required to slow down the relay release.?
E
Of course, for example the MCU has two decoupling capacitors

The zener diode is not must have but as far as I have been reading it helps the speending up the relay release, it doesnt slow it down but the opposite
 

Thread Starter

Goxeman

Joined Feb 28, 2017
77
Hello,

Why do you need a zener in series with the diode?
Replace both with a UF4007.
The transient will not go higher than VCC+Vdrop(UF4007).

Bertus
Why is that? Because of the speed of recovery? I mean what would be the difference compared with an standard 4148 diode?

I am asking because I didnt considered the option of an ultrafast rectifier. Wouldnt a shottky diode do the job better then?
 

MrChips

Joined Oct 2, 2009
19,734
Transients appearing on an oscilloscope screen may be misleading. A lot depends on the oscilloscope, scope probes, ground connections, ground loops, shielding, EMI, etc. Take the scope readings with a grain of salt.
 

ericgibbs

Joined Jan 29, 2010
9,078
hi,
Have you carried out the same transient test when using only a diode [ no zener].
E

A photo of the project , showing the CRO connection points would help.
 

RamaD

Joined Dec 4, 2009
323
I suppose the 6.8V Zener is to speed up the relay action.
Is there a decoupling cap near the ULN?
Try adding a ferrite bead in series to the Vcc to the microcontroller, and also near the ULN. These are lossy, acting resistive at high frequencies.
 

Thread Starter

Goxeman

Joined Feb 28, 2017
77
hi,
Have you carried out the same transient test when using only a diode [ no zener].
E

A photo of the project , showing the CRO connection points would help.
I havent tried taking off the zener, that is a good idea, just for the test

The connection point I take are MCU VSS and random I/O pins. The measures vary always a bit but is very similar

Capture.JPG
Reviewing the block diagram of the transistor array IC I see that it already includes one flyback diode for each channel but I cant see the specifications of it, then does it make sense using an external diode?

The diode inside the device connects the output pin (which is connected to one side of the relay) and the power supply (which is connected to the other side of the relay). So the internal diode in the device would be in parallel with the diode and zener across the coil, basically conducting before they could switch on and then this would not take into account the micro-weld issue (reason of using the zener for fast de-energize)
 

bertus

Joined Apr 5, 2008
20,176
Hello,

Did you connect pin 9 of the ULN for the free wheeling diodes?
When that is connected to the powersupply of the relays, you will not need a diode accross the relays anymore.

Bertus
 

Thread Starter

Goxeman

Joined Feb 28, 2017
77
Hello,

Did you connect pin 9 of the ULN for the free wheeling diodes?
When that is connected to the powersupply of the relays, you will not need a diode accross the relays anymore.

Bertus
Yes, COM pin is connected to the VCC rail, same VCC rail of the relays

The Series Diode / Zener diode would not be used, as the internal diode in ULN is parallel to them!
Ok I will test taking of the diodes from my PCB and see what happens, in any case there will be a transient from the relays de-energize that I measured. Would you use a TVS diode in the VCC rail?

Would it make sense using an ultrafast diode paralelled with the coil? Do you think it would be faster than the internal diode? I couldnt find much information about the internal diode of the ULN
 

bertus

Joined Apr 5, 2008
20,176
Hello,

How much current does the relays need?
The ULN20013V12 datasheet tells me that the recommended maximum current per channel is 140 mA.

Bertus
 

Attachments

Thread Starter

Goxeman

Joined Feb 28, 2017
77
Hello,

How much current does the relays need?
The ULN20013V12 datasheet tells me that the recommended maximum current per channel is 140 mA.

Bertus
Indeed each current per channel is very low compared with the traditional ULN2003 which can sink 500mA per channel I think if I remember well

The relays would be drawing something between 70mA and 100mA depending on the relay that I pick at the end. Right I am making tests and they are drawing around 90mA, everything works normal with the ULN2003V12
 

MrChips

Joined Oct 2, 2009
19,734
Try this. Disconnect your scope probe from your circuit. Connect the scope probe grounding clip to the tip of the scope probe. Now capture some traces with no connection to your circuit while the relay is being switched on/off.
 

Thread Starter

Goxeman

Joined Feb 28, 2017
77
Try this. Disconnect your scope probe from your circuit. Connect the scope probe grounding clip to the tip of the scope probe. Now capture some traces with no connection to your circuit while the relay is being switched on/off.
What do you mean with "capture some traces with no connection to your circuit". Do you propose that I leave the scope probe over a track? I dont understand where are you proposing that I connect the probe
 

MrChips

Joined Oct 2, 2009
19,734
I am saying you do not need to connect your scope probe to any part of your circuit. Even if you ground the scope probe to its own ground clip you should be able to pick up transients when your relay switches ON and OFF. What would that tell you about your circuit and EMI?

It will be nice to see your screen shots when you have done this.
 

crutschow

Joined Mar 14, 2008
23,756
Would it make sense using an ultrafast diode paralelled with the coil?
No.
The "ultrafast" is the turn-off recovery time of the diode, which is not of concern in this application.
All diodes turn on rapidly, so an ultrafast or Schottky diode will have little effect on improving the transient suppression.

I agree that the transient you are seeing is likely from the current pulse through your grounds and supply leads when the relay is turned off.
Where did you connect the oscilloscope ground probe?
Try measuring the ground voltage as suggested, and post what you see.
 
Top