# Transformer high voltage spikes

#### Burak HANÇERLİ

Joined Jan 4, 2016
9
Hi all,

I'm using a 220V - 24V transformer as a supply of 24V AC - 18V DC converter which is my own design. Since I cannot control the output of the transformer I put a crowbar circuit in the DC converter circuit as a precaution for the high voltages and it triggers around 42-43V DC. So the whole system is as described below.

220V AC -- Transformer -- 24V AC --> Full rectification -- Crowbar Circuit (42-43V) -- DC-DC Boost Converter -- 18VDC

However I'm observing high voltage spikes on the output of the transformer which sometimes triggers the crowbar circuit and blows up the protection fuse. Here is a scope analysis.

Green line is 24V AC input, and the yellow one is the input of the crowbar. These spikes are captured at the exact moment of turning off the transformer. There are also some other minor spikes captured when I turn it on.

As you can see from the measurements, crowbar input can jump up to 49-50V level, which causes it to be triggered and resulting in a blown-up fuse.

So my questions are;
1 - What is the root cause of those spikes? (actually I'm curious about this one more than the second question)
2- How to prevent them from happening?

I already read about zener transition suppression circuits and MOVs that can prevent such things, and will try to test them in this week. I wanted to get some ideas before ordering the components.

bh

#### MrChips

Joined Oct 2, 2009
25,920
Your oscilloscope HOR setting is 50μs/div.
Let us see the waveform at 5ms/div.

#### Ian0

Joined Aug 7, 2020
4,838
Let's have a look at your 24V to 18V converter circuit.

#### kubeek

Joined Sep 20, 2005
5,772
Please show the complete circuit, and how you measured the waveforms - try to have as short a ground lead as possible, while referencing the ground of the crowbar circuit, and far away from the magnetic field of the transformer.
Did you check whether the spikes are actually coming from the mains, or if they are coming from the secondary side?

#### Burak HANÇERLİ

Joined Jan 4, 2016
9
I just noticed that I mentioned the DC-DC converter as a boost instead of buck, sorry for that.

@MrChips;
Below are the 5ms/div turn-off capture and 50ms/div turn-on capture.

@lan0 & @kubeek Whole schematic is attached as a pdf.

@kubeek Spikes are coming from the secondary side. I have a variac which doesn't have such problem.

#### Attachments

• 51 KB Views: 12

#### MrChips

Joined Oct 2, 2009
25,920
You are trying to build a SMPS (switched mode power supply).
The spikes are occurring when the transistors switch.
Show us the full circuit schematic.

#### Ian0

Joined Aug 7, 2020
4,838
There's so much filtering/smoothing after the bridge, that I really think that the "spikes" are switching noise being picked up in the 'scope probe.
Unless your supply comes from a generator and you are sharing it with someone doing arc-welding, I can't imagine you'd ever get more than 36V on the secondary.
Design the switcher so that it will take 60V and leave out the crowbar!

#### Burak HANÇERLİ

Joined Jan 4, 2016
9
Thanks for all responses.

The thing is those spikes are there even if my SMPS is not connected to the transformer. I know that I have a couple of noise/spike sources in the DC-DC conversion part, but the ones that I keep mentioning are coming directly from the secondary side of the 220V-24V transformer.

@bertus thanks for the suggestion. Actually I've chosen that chip for that high input voltage tolerance, but after observing more than 60V input in the field (and ended up with a couple of burned PCBs and the devices after that) I decided to put that crowbar to at least protect the devices connected to this SMPS.

@lan0 Me neither expected higher than 35V-36V output on the secondary side of the transformer, but that's the current situation And what's more surprising is I'm working in the lab which has a smooth and low-noise 220V AC source.

#### Deleted member 115935

Joined Dec 31, 1969
0
What the green scope trace volts per division ?

On the green, I count about 20 cycles in 50 us, which is around 400 Khz,
and why is the green disappearing ?
how are you turning this off ?
can we see the full schematic please as to what you are doing

#### Ian0

Joined Aug 7, 2020
4,838
Does it still happen without L2? I don't know how much good L2 will be doing:- there will be far more than 10uH leakage inductance in the transformer.
Also, does the converter IC have a UVLO? Could that be used to turn the converter off as the voltage decays before anything unwanted happens?

#### MrChips

Joined Oct 2, 2009
25,920
And what's more surprising is I'm working in the lab which has a smooth and low-noise 220V AC source.
It should not come as a surprise. Turn off your circuit and the noise goes away. The noise is generated by your switching circuit. Some of what you see on the oscilloscope is from pickup back through the AC power line and the oscilloscope probe.

#### Burak HANÇERLİ

Joined Jan 4, 2016
9
@andrewmm it is 40V/div on the green scope. I put 2 pics, one is captured when turning on the transformer as the other one captured when turning it off. I suppose I don't get your question about how I am turning the transformer off. I mean, by just turning off the switch? )

I've already put the schematic of the PCB as pdf, but if you're expecting the schematic of my whole test setup here it is.

I was testing the design and first prototypes with a variac instead of the transformer in the image and there were no spikes and the crowbar was not being triggered. However, spikes become visible just after switching the variac with the transformer.

And again, spikes on the green scope happens even if the PCB is disconnected from the transformer. Actually, this is why I asked if the Zener Suppression circuits may help to solve the issue in my very first question.

@lan0 If the information above does not clarify your question I'd be glad to provide more detail.

@MrChips After your comment, I felt like my previous explanations might be a bit deficient, so I hope the above information would fill the gap.

#### Deleted member 115935

Joined Dec 31, 1969
0
what is your DCDC convertor ? whats its switching frequency ?
what reference are you using for the scope probe grounds ?

#### Burak HANÇERLİ

Joined Jan 4, 2016
9
@andrewmm I don't quite get the relation of the DC-DC converter with the spikes coming from transformer but it is LT3844 as described in the schematic and the switching frequency is around 100Khz. And the probe ground for the yellow (DC) scope is ground of the smoothing and filtering circuit.

For green scope, there's no ground reference since it's AC.

#### Ian0

Joined Aug 7, 2020
4,838
I was also concerned about the green probe Earth. I’d be much rather inclined to believe the voltage if it were directly across the transformer (but disconnect the other earth clip first! - remembering a former colleague who connected one earth clip to mains neutral, thought of something else that he needed on screen and connected the second clip to mains live - then we phoned RS to order some more probes)

#### Deleted member 115935

Joined Dec 31, 1969
0
My first worry would be the measurement set up,
your scope has a high impedance input, and its not grounded for the AC measurements,
QED, its a very good antenna,

I'd suggest that you instead look at the input and output of the filter circuit, with a good ground connection for the scope.
see what the signals look like there.

my reason for the questions, is a basic one that mains transformers are just not that good at transferring such a high frequency of 400 KHz you are seeing as you remove the mains input,

It looks to me as if your DCDC is "gulping" large and larger lumps of current as its DC input disappears, causing great big di/dt spikes radiating through the system / air.

One other comment, its unusual to put a crow bar at the input to a DCDC,
the environment is horrible, and a crow bar is a final HARD level of protection.

A crow bar is normally on the output of the DC as a final protection of the DCDC going wrong, such as the switcher staying on, connecting input to the output. You then also have a fuse on the output of the DCDC, normally a reset able PTC type.

On the inputs, one would normal have things like a MOV for real big over volt such a lightning strikes effects on near by mains,

#### Deleted member 115935

Joined Dec 31, 1969
0
BTW: if you want to measure mains voltage , either use a differential probe,
or get another transformer, to provide an isolated measurement,

#### Burak HANÇERLİ

Joined Jan 4, 2016
9
@andrewmm Your response (#17) was a source of enlightenment.

Let me work on it a bit more, gather some more information, and share them with you again.

#### MrChips

Joined Oct 2, 2009
25,920
@andrewmm I don't quite get the relation of the DC-DC converter with the spikes coming from transformer but it is LT3844 as described in the schematic and the switching frequency is around 100Khz. And the probe ground for the yellow (DC) scope is ground of the smoothing and filtering circuit.

For green scope, there's no ground reference since it's AC.
Your thinking is flawed in many ways.

1) Your oscilloscope needs a ground reference, DC or AC.

2) Your oscilloscope is already grounded in two ways, via the AC mains plug and via the ground clip on the yellow channel.

3) Your oscilloscope probe on the green channel needs to be grounded. As I already pointed out in post #12, what you observe on the screen is partially due to stray pickup and interference, especially when the probe is not properly grounded.

You need to learn how to use your oscilloscope properly.