Need advice for TVS selection

Thread Starter

jrm

Joined Oct 11, 2011
38
Hi All,
I'm switching a large inductive load (350h) at low frequency (2-30hz) with 12vdc. The resistance across the inductor is 4ohms. The switching circuit is 5v logic (avr) and a pair of P channel mosfets in parallel. For a snubber, I have been using a byw29e (http://www.nxp.com/documents/data_sheet/BYW29E_SERIES.pdf) wired anti parallel across the inductor immediately proximal to where the inductor connects to the pcb. Looking at it on my low res scope I thought I had the transient issue under control but I'm not so sure now. I've had the voltage regulator for the logic fry twice now, this last time actually blew the side of the regulator off near the vin pin. The 5v side only demands 170ma from the Ti r104 regulator. The voltage regulator has a large copper area for heat dissipation and has internal thermal controls. I also have the same regulator with the same supporting components sans inductor in a couple of other projects and they work great so I'm lead back to the inductive spike.


The questions are: Does this BYW29e look like sufficient protection? If not, will a TVS diode be a better choice? When selecting the TVS for this circuit should I base the clamping voltage on the 5v side or the 12v through the coils?
Any help will be greatly appriciated.

Thanks.
 

Attachments

hgmjr

Joined Jan 28, 2005
9,027
You have drawn your mosfet as n-channel devices. The sub75p05 devices are p-channel mosfets. This makes the schematic a bit confusing.

Perhaps you can redraw and repost.

hgmjr
 

Thread Starter

jrm

Joined Oct 11, 2011
38
You have drawn your mosfet as n-channel devices. The sub75p05 devices are p-channel mosfets. This makes the schematic a bit confusing.

Perhaps you can redraw and repost.

hgmjr
Thanks for pointing that out. I put that symbol in the library a couple of months ago and never even noticed the error.
 

Attachments

hgmjr

Joined Jan 28, 2005
9,027
That is much better. I see you even corrected the connection of the sources so that they tie directly to +12V rather than being tied to +12 volts through a 10K resistor as shown in your original schematic. Now the 10K provides a pullup on the gates of the two parallel p-channel mosfets.

An inductor of 350 Henries is unusual. Are you absolutely certain of the inductance value?

hgmjr
 

Thread Starter

jrm

Joined Oct 11, 2011
38
That is much better. I see you even corrected the connection of the sources so that they tie directly to +12V rather than being tied to +12 volts through a 10K resistor as shown in your original schematic. Now the 10K provides a pullup on the gates of the two parallel p-channel mosfets.

An inductor of 350 Henries is unusual. Are you absolutely certain of the inductance value?

hgmjr
I was a bit hasty in my first attempt. Sorry about that. :)

The 350h is calculated only. I do not have any empirical measurements. The total inductor consists of 10, 12" diameter air core magnets with 230wraps of 26g copper each, wired in parallel. the sectional area of the coil is roughy .126^2"

re: floating pin2 - Pin 2 on this fixed output version is an optional noise reduction cap pin. If I'm reading the datasheet correctly the noise should be roughly .5uV rms without the additional cap.

The demands on the reg104 Vout are only an atmega164p, a 128x64 glcd, a small Piezo buzzer, and a resistive touchscreen. There are three color backlights on the glcd which draw 20,20,25ma max. In my code only one backlight is on at a time. I was using the smaller SOT package reg104 which should be able to dssipate 1.5 watts continuous with the cooling copper I have on the board. A fully charged battery pack can measure 14v for a brief period right after charging so I am right at the limits of the regulator at about 50 degrees C ambient. The only thing that leads me to beleive that heat is not the issue is because this chip has internal thermal shutdown circuitry and I've never seen it trip.
 

hgmjr

Joined Jan 28, 2005
9,027
That is interesting. What is the aim of this circuit?

I am not familiar with the term "air core magnets". Are you possibly referring to toroid cores?

hgmjr
 
Last edited:

Thread Starter

jrm

Joined Oct 11, 2011
38
That is interesting. What is the aim of this circuit?

I am not familiar with the term "air core magnets". Are you possibly referring to toriod cores?

hgmjr
They are just copper coils without any core, axial or toroidal.

It's heath related experimentation by a friend. I think he is trying to prove or disprove magnetic therapy. He just asked if I could make a control unit to work with his existing components.

I'm making a huge error in my calculations about the henries. I just revisited VolI and realized that I was adding inductance of the ten parallel coils instead of divding. L=3.5h which gives a time constant that makes a little more sense.

Nevertheless, I'm still a bit confused about picking the right TVS. The REG104 max voltage is -.3 to 15. The lowest clamping voltage I see is 7.5 volts which easily exceeds the detruction voltage of the regulator. Is adding a cap the simple answer?
 

hgmjr

Joined Jan 28, 2005
9,027
Air core inductors have typically low inductances since air is not an efficient storer of magnetic energy. On the plus side, air cores do not saturate.

I have no direct experience with flyback in conjunction with air core inductors but it would seem to me that any flyback produced would be easy to suppress using a reasonable fast diode such as the one you have chosen.

Might I suggest that you try putting one of your diodes (if you have an extra) in series with the positive leg of your power supply. That will drop your output by the forward voltage drop of your diode but I think that would be OK. Be sure to put an electrolytic filter capacitor (the larger the better) from the output of the diode to ground. That would prevent any significant spikes from getting back into your 12V power supply and from there into your reg104 regulator.

hgmjr
 

Thread Starter

jrm

Joined Oct 11, 2011
38
Thanks for your help.

Should I put the series diode on the drain side of the driving P channels to protect them as well, or would it be better to put it on the source side?

Example:
 

Attachments

hgmjr

Joined Jan 28, 2005
9,027
Your placement represents a good start. See what happens. If that saves your regulator without compromising the circuits performance then you are set.

hgmjr
 

Thread Starter

jrm

Joined Oct 11, 2011
38
You do need to remove the capacitor that you have in parallel with the inductor.


hgmjr
I guess I was a little confused about your cap suggestion. Were you refering to the flywheel diode when you said - "Be sure to put an electrolytic filter capacitor (the larger the better) from the output of the diode to ground. That would prevent any significant spikes from getting back into your 12V power supply and from there into your reg104 regulator."?
 

hgmjr

Joined Jan 28, 2005
9,027
I guess I was a little confused about your cap suggestion. Were you refering to the flywheel diode when you said - "Be sure to put an electrolytic filter capacitor (the larger the better) from the output of the diode to ground. That would prevent any significant spikes from getting back into your 12V power supply and from there into your reg104 regulator."?
Sorry for any confusion I caused. This comment was based on my earlier suggestion of placing the added series diode in series with the positive power supply feed.

hgmjr
 
Top