Decrease font size
Increase font size
Topic Title: circuit help (voltage spike)
Topic Summary:
Created On: 08 September 2010 11:07 PM
Status: Read Only
Linear : Threading : Single : Branch
Search Topic Search Topic
Topic Tools Topic Tools
View similar topics View similar topics
View topic in raw text format. Print this topic.
 08 September 2010 11:07 PM
User is offline View Users Profile Print this message


Posts: 10
Joined: 23 February 2010

I designed H bridge but i didn't use H bridge driver or CPU, the circuit was simple as 555 timer and logic inverter to control the MOSFET's.

When i used high resolution scope i found more than 2Kv spike across the high side MOSFET's. when i connect 4 scope probes for each MOSFET gate i found that the driver signal cross each other for about 5us so i knew this was the cause of the spike. I didn't care that the signal cross each other because the current that is going through the H bridge is not more than 50mA, and i used: 1000V, 5A, 200mJ MOSFET.

the H bridge is converting 700 dc 5mA to 700v ac square wave.

I am just wondering will that 2kV harm the MOSFET?

I ran the circuit for a long time and it seemed fine. But i am just not sure if the 2Kv will damage the MOSFET and sometime the spike hits 4Kv for not more than 5us or 10us

I would appreciate if someone could explain if the MOSFET is ok with that spike or how will the MOSFET get damaged, because i ran the circuit for a long time and it seemed to work fine without a problem.

look forward to some responses!

Thanks in advance.

Edited: 09 September 2010 at 03:12 PM by zhassan11
 17 September 2010 11:24 AM
User is offline View Users Profile Print this message


Posts: 1
Joined: 14 October 2002

What are you driving with the H-Bridge. I assume you are driving an inductive load, which will cause spikes during switching because or back-emf, etc. You could reduce the spikes, and possibly improve efficiency by adding a snubber circuit to damp the spikes and speed up the transition time. You could use a fast varistor or a RC snubber.

Hope this helps.
 17 September 2010 12:24 PM
User is offline View Users Profile Print this message


Posts: 141
Joined: 25 July 2008

As was suggested back in February when you first posted about this circuit, have you tried simulating the circuit?

Put in the basic MOSFET bridge and driver circuit but take care in modelling the load and any significant parasitics such as long wires and decoupling capacitors. Then you should get some insight into what might be causing the spikes.

To start with try using simple Spice (or Gnucap or QUCS) switch elements (but make sure they are not the very basic snap on/snap off versions as in Pspice: you want the softer switching such as the switch in LTspice offers to minimise discontinuities and so help convergence).

This will avoid some of the complication of using the (sometimes dodgy) vendor MOSFET models and may give enough insight to get you started.

Don't worry about trying to model any feedback control loop: run it as an open loop switching circuit and let it settle into the steady state, then look at the waveforms at the switching intervals.

Free and unlimited simulation tools: (FOSS) (FOSS)

Excellent free demo version:

Modern electronic design should not be attempted without using simulation.

Andy Fierman

 17 September 2010 08:46 PM
User is offline View Users Profile Print this message


Posts: 10
Joined: 23 February 2010

Thank you very much for the reply,

The circuit is driving a resistive load (resistor ladder), and without any load the spike is active across the high side MOSFETS and across the output terminals of the bridge.

I have attached the circuit and the signals so it will be easy to understand.

The circuit is running at 700V, I could not find a H bridge driver that works up to 700 so I designed a simple circuit (555 timer and logic invertors as you can see in the attached circuit).

Because the circuit is so simple I could not set a dead time so I know the results will be overlapped between the gates (as you can see), I did not care about the overlap as the circuit running on 50mA maximum and the MOSFET I used is a 1KV, 5A, 250mJ.

The circuit is running fine but I am just confused how come the MOSFET did not blow up from the high voltage spike (btw the current spike is 0.7A only).

I tried snubber circuit but it did not work very well for some reason, I will appreciate if you can lead me to any good method for snubber design, I did not try fast varistor, I thought the varistor is a slow device.

Have a look at the driver signal you will see about 10uS overlap, am I right if I said this is the cause of all the problems? But even though how did the MOSFET manage to survive this spike as i was running the circuit for a few months now?

I did simulate the circuit and I decided to use PIC24 to control the circuit to stop the overlap however after I used the CPU the spike disappeared across the MOSFET but the spike across the H bridge still active.

I am wondering If I am wasting my time using the CPU to kill the high voltage spikes ( 2 - 4KV or more) as the 555 timer and logic gate seems to work ok with all the voltage spikes for few months now?

Below is the link to the image of the circuit:

Below is the links to the images of the signals:

 19 September 2010 05:46 PM
User is offline View Users Profile Print this message


Posts: 141
Joined: 25 July 2008

Have you tried the null experiment of connecting your scope probe exactly as if you were going to measure the output voltage or whatever but then moving the probe tip to connect to the ground nearest to the point you would normally have connected to, to make the desired measurement?

That way you will see what background noise your probe is picking up even when you are measuring what should be a zero signal.

If you are using a probe ground lead of even a few cm then you are probably picking up huge spikes in the loop created by this ground wire and the probe capacitance.

Even if you are using a "short earth" probe you may still see sizeable spikes because you are measuring in the presence of very fast edges, high voltages and so high dV/dt.

If you do have "shoot through current in the MOSFETs due to not having any dead time then you may also be picking up a lot due to the high di/dt of this current pulse.

Doing a null measurement will sanity check what the probe is picking up in its ground loop so you get some idea of how much of what you see on the scope is real and how much is noise. If you can get the ground probing point from very close to the output and not have to move the probe body and ground wire much between the measurements then the noise in both measurements should be very similar.

A dead giveaway is if the size of the spikes is sensitive to the position of the probe and the ground wire even if the actual connection points do not move.

Andy Fierman

 20 September 2010 11:31 AM
User is offline View Users Profile Print this message


Posts: 10
Joined: 23 February 2010

Thank you very much for the comment; I really appreciate your help

The links below has a pictures that illustrates how I measured the signals (sorry for the hand writing):

I used 3 x 20M Ohm resistors connected in series with a 390K Ohm resistor and I connected the scope probe across the 390K Ohm to reduce the voltage from 700V down to 4V and I can control the input to the scope by the 1:10 scope probe.

The point of using 20M Ohm x 3 is not to drain any current during the measurements so I wont kill any weak spike or signal.

The other picture (below ) has two signals, the green is the GND noise (the scope probe connected to the GND) and the red signal is the high side MOSFET signal across the 390K resistor which is in series with the 60M Ohm.

I powered the scope from a car battery (the scope am using is PICO 6000) and I used a Laptop so it has its own battery and I powered the circuit from 6AA battery so I wont get any noise from the main or leakage from the circuit to the other circuit.

Edited: 20 September 2010 at 11:41 AM by zhassan11

See Also:

FuseTalk Standard Edition v3.2 - © 1999-2016 FuseTalk Inc. All rights reserved.