dead vnd5n07

Testing and development of Megasquirt 3

Moderators: jsmcortina, muythaibxr

EFIDART512
Helpful MS/Extra'er
Posts: 46
Joined: Wed Mar 24, 2010 10:00 am

Re: dead vnd5n07

Post by EFIDART512 »

dontz125 wrote:What would the clamp consist of - simply a 1N4007 pointing from the source to the drain? Or combined with a 30-ish-volt Zener fitted opposite to the diode?
The most simple connection would be from the DRAIN to the INJECTOR_GND. Use a 33V TVS (they are fast - 1 pico second), cathode to DRAIN / anode to GND.
A 33V 5W part can be found at Digi-Key 5KP30ALFCT-ND. A 1.5W part might do the job but until it's tested/measured I'm going to guarantee anything. Try it at your own risk.
Jean is correct that clamping to a lower voltage will impact the closing times. At 33V, the fly-back load will be taken off of the VND part but still allow a rise to 33 which shouldn't affect the closing time much.
dontz125
Super MS/Extra'er
Posts: 4200
Joined: Mon May 11, 2009 7:14 pm
Location: York, ON
Contact:

Re: dead vnd5n07

Post by dontz125 »

SymTech Laboratories wrote:
dontz125 wrote:What would the clamp consist of - simply a 1N4007 pointing from the source to the drain? Or combined with a 30-ish-volt Zener fitted opposite to the diode?
To clamp the reverse EMF spike to ~12V, install a 1N400x diode between the VND5N07's drain and +12V, with the diode's cathode connected to +12V.
Understood, and understood why it's not great - a closing time measured in minutes (kidding) does bad things to your idle.
To clamp it to a higher voltage, install a 3W (or larger) Zener diode between the VND5N07's drain and ground, with the diode's anode connected to ground. You can add a forward biased 1N400x diode in series with the Zener diode to make a makeshift TVS and protect against negative voltages.
Vaguely understood, but confusing. The Zener and 1N400x diode are mounted parallel to the VND, with the 1N400x pointed south (assuming the injector is north of the VND and ground is south) and the Zener pointed north, correct? I've been sitting here scratching my head, doing some rummaging on Google, and even doing some scribbles with my daughter's crayons - given the north-bound body diode of the VND, are these diodes actually doing anything? A flyback pulse wants to go north, right? What's stopping the pulse from simply going through the body diode and bypassing the Zener and 1N400x?

I found this page showing snubber designs and mosfets. I realise we aren't interested in snubbers per se, but the Zener + 1N400x arrangement shown here across the coil made all kinds of sense to me.

Am I misunderstanding the flyback pulse? Is the pulse actually going "outward" from the injector / drain connection, not just "north" towards the +12v? If so, then the arrangement described by Symtech suddenly makes more sense ...

Edited to add the link mentioned in the 2nd-last para.
Last edited by dontz125 on Sat Mar 10, 2012 11:55 pm, edited 1 time in total.
QuadraMAP Sensor Module -- PWM-to-Stepper Controller -- Dual Coil Driver
Coming soon: OctoMAP Sensor Module
TTR Ignition Systems
dontz125
Super MS/Extra'er
Posts: 4200
Joined: Mon May 11, 2009 7:14 pm
Location: York, ON
Contact:

Re: dead vnd5n07

Post by dontz125 »

EFIDART512 wrote: The most simple connection would be from the DRAIN to the INJECTOR_GND. Use a 33V TVS (they are fast - 1 pico second), cathode to DRAIN / anode to GND.
A 33V 5W part can be found at Digi-Key 5KP30ALFCT-ND. A 1.5W part might do the job but until it's tested/measured I'm going to guarantee anything. Try it at your own risk.
Jean is correct that clamping to a lower voltage will impact the closing times. At 33V, the fly-back load will be taken off of the VND part but still allow a rise to 33 which shouldn't affect the closing time much.
Ummm - according to the data sheet this is a 5 THOUSAND watt device. Or were you looking at the 8W steady-state rating? Should the flyback pulses, coming as fast as they do, be considered a steady-state load?
QuadraMAP Sensor Module -- PWM-to-Stepper Controller -- Dual Coil Driver
Coming soon: OctoMAP Sensor Module
TTR Ignition Systems
EFIDART512
Helpful MS/Extra'er
Posts: 46
Joined: Wed Mar 24, 2010 10:00 am

Re: dead vnd5n07

Post by EFIDART512 »

dontz125 wrote:
EFIDART512 wrote: The most simple connection would be from the DRAIN to the INJECTOR_GND. Use a 33V TVS (they are fast - 1 pico second), cathode to DRAIN / anode to GND.
A 33V 5W part can be found at Digi-Key 5KP30ALFCT-ND. A 1.5W part might do the job but until it's tested/measured I'm going to guarantee anything. Try it at your own risk.
Jean is correct that clamping to a lower voltage will impact the closing times. At 33V, the fly-back load will be taken off of the VND part but still allow a rise to 33 which shouldn't affect the closing time much.
Ummm - according to the data sheet this is a 5 THOUSAND watt device. Or were you looking at the 8W steady-state rating? Should the flyback pulses, coming as fast as they do, be considered a steady-state load?
Hey sorry for my typo (K). Too many parametric tables open on a small laptop missing the units. Yeah it is a big device, like 20" brakes with 8 pistons. That big rating is for one single pulse. The problem is heat, pack everything in a small box, put it in a hot place and all this stuff gets hotter. Run the VNDs at -10 deg C, they will survive. The extra energy that they have to handle with the clamping of the fly-back might push some over the edge to failure when mounted in a hot place. Everyone's install is different. Adding the external diode takes that extra away from the device but that heat goes somewhere. Wrap it all up together again and you have the same heat just in slightly different places.
There are many ways to clamp the fly-back, zeners, tvs, and active clamps. They all have to put the heat somewhere.
These are just patch fixes, might work for some but not all.
SymTech Laboratories
Super MS/Extra'er
Posts: 2188
Joined: Sun Aug 31, 2008 4:02 pm
Location: South Florida, USA
Contact:

Re: dead vnd5n07

Post by SymTech Laboratories »

dontz125 wrote:Vaguely understood, but confusing. The Zener and 1N400x diode are mounted parallel to the VND, with the 1N400x pointed south (assuming the injector is north of the VND and ground is south) and the Zener pointed north, correct? I've been sitting here scratching my head, doing some rummaging on Google, and even doing some scribbles with my daughter's crayons - given the north-bound body diode of the VND, are these diodes actually doing anything? A flyback pulse wants to go north, right? What's stopping the pulse from simply going through the body diode and bypassing the Zener and 1N400x?

I found this page showing snubber designs and mosfets. I realise we aren't interested in snubbers per se, but the Zener + 1N400x arrangement shown here across the coil made all kinds of sense to me.

Am I misunderstanding the flyback pulse? Is the pulse actually going "outward" from the injector / drain connection, not just "north" towards the +12v? If so, then the arrangement described by Symtech suddenly makes more sense ...
A picture is worth a thousand words, right? The blue arrows signify the current direction (using conventional current) during a flyback pulse. This circuit is very similar to the TVS EFIDART512 is describing, but a proper TVS is much quicker.

The body diode in the MOSFET is usually drawn as a Zener diode, but its breakdown voltage is much higher than the VND5N07's clamping voltage. Due to the over-voltage mechanism in the VND5N07, the body diode shouldn't ever reach its breakdown voltage. We're not fully convinced the flyback pulses are destroying the VND5N07's anyway. We can't quite reproduce the problem in the lab just yet, but the VND5N07 in our test fixture is working as described: clamping the flyback pulses to ~60V. It doesn't get warm the touch driving a single high impedance fuel injector for extended periods of time, but the next step is the measure the temperature rise directly.

We still suspect the problem is due to the body diode conducting in the other direction. Until we can come to a definitive conclusion, we're going to recommend that our customers use 5A/100V Schottky diodes inline with each VND5N07. Of course, the diode's cathode should be connected to the VND5N07's source and the anode should be connected to the load.
tvs_current.jpg
SymTech Laboratories, LLC ----- MSQ File Repository: megasquirt.symtechlabs.com
Image
jsmcortina
Site Admin
Posts: 39585
Joined: Mon May 03, 2004 1:34 am
Location: Birmingham, UK
Contact:

Re: dead vnd5n07

Post by jsmcortina »

None of this would explain DOA parts which _appear_ to be the bulk of the problem?

James
I can repair or upgrade Megasquirts in UK. http://www.jamesmurrayengineering.co.uk

My Success story: http://www.msextra.com/forums/viewtopic ... 04&t=34277
MSEXTRA documentation at: http://www.msextra.com/doc/index.html
New users, please read the "Forum Help Page".
racingmini_mtl
Super MS/Extra'er
Posts: 9128
Joined: Sun May 02, 2004 6:51 am
Location: Quebec, Canada
Contact:

Re: dead vnd5n07

Post by racingmini_mtl »

Is there a difference in the assembly process between the MS3X and the uS? I don't see anything in the data sheet on the subject but are there any unusual soldering requirements? Has anyone contacted STM about this?

Jean
jbperf.com Main site . . . . . . . . . . . . . . . . . . . . . . jbperf.com Forum
Image
Peter Florance
Super MS/Extra'er
Posts: 3653
Joined: Fri Apr 02, 2004 8:40 pm
Location: Virginia Beach, VA
Contact:

Re: dead vnd5n07

Post by Peter Florance »

racingmini_mtl wrote:Is there a difference in the assembly process between the MS3X and the uS? I don't see anything in the data sheet on the subject but are there any unusual soldering requirements? Has anyone contacted STM about this?

Jean
They (MS3X and the uS) are made at the same factory, IIRC.
Peter Florance
PF Tuning
81 BMW Euro 528i ESP Car
60-2 Wheel LS2 Coils, Low Z Inj
Co-Driver 1999 BMW E46 DSP car.
SymTech Laboratories
Super MS/Extra'er
Posts: 2188
Joined: Sun Aug 31, 2008 4:02 pm
Location: South Florida, USA
Contact:

Re: dead vnd5n07

Post by SymTech Laboratories »

jsmcortina wrote:None of this would explain DOA parts which _appear_ to be the bulk of the problem?

James
The few that we've seen worked for a short while, then failed. If there are a significant number of DOA parts, however, we can compile the traceability codes in this thread and then contact ST for more information.
SymTech Laboratories, LLC ----- MSQ File Repository: megasquirt.symtechlabs.com
Image
racingmini_mtl
Super MS/Extra'er
Posts: 9128
Joined: Sun May 02, 2004 6:51 am
Location: Quebec, Canada
Contact:

Re: dead vnd5n07

Post by racingmini_mtl »

jsmcortina wrote:None of this would explain DOA parts which _appear_ to be the bulk of the problem?

James
The problem I see with this is that with SMD it is difficult to say whether they were DOA or the assembly process caused them to fail. I don't think anyone has tested the parts prior to assembly. Of course, there shouldn't be any reason a standard assembly process would be a problem.

Jean
jbperf.com Main site . . . . . . . . . . . . . . . . . . . . . . jbperf.com Forum
Image
EFIDART512
Helpful MS/Extra'er
Posts: 46
Joined: Wed Mar 24, 2010 10:00 am

Re: dead vnd5n07

Post by EFIDART512 »

jsmcortina wrote:None of this would explain DOA parts which _appear_ to be the bulk of the problem?

James
The DOA parts could be due to them getting over-cooked in the soldering process. I'm assuming that these are lead-free, higher temperatures are involved. The larger die pads are harder to solder with IR radiation (ask your manufacturer). The parts could be over-heated but then the rest of the board would be. Just a thought.
jsmcortina
Site Admin
Posts: 39585
Joined: Mon May 03, 2004 1:34 am
Location: Birmingham, UK
Contact:

Re: dead vnd5n07

Post by jsmcortina »

AEA use with-lead (as the Megasquirt parts a "control boards" and therefore exempt from the lead-free aspect of RoHS.)

James
I can repair or upgrade Megasquirts in UK. http://www.jamesmurrayengineering.co.uk

My Success story: http://www.msextra.com/forums/viewtopic ... 04&t=34277
MSEXTRA documentation at: http://www.msextra.com/doc/index.html
New users, please read the "Forum Help Page".
dontz125
Super MS/Extra'er
Posts: 4200
Joined: Mon May 11, 2009 7:14 pm
Location: York, ON
Contact:

Re: dead vnd5n07

Post by dontz125 »

I edited my post above to include the link that fell off somehow; they show the double-diode across the transformer / injector / whatever, rather than across the mosfet.

It's very much like the sketch SymTech posted, but instead of pointing down to GND it points up to +12v. Would this make a difference? Could one cause more problems than the other?

Image
QuadraMAP Sensor Module -- PWM-to-Stepper Controller -- Dual Coil Driver
Coming soon: OctoMAP Sensor Module
TTR Ignition Systems
SymTech Laboratories
Super MS/Extra'er
Posts: 2188
Joined: Sun Aug 31, 2008 4:02 pm
Location: South Florida, USA
Contact:

Re: dead vnd5n07

Post by SymTech Laboratories »

dontz125 wrote:I edited my post above to include the link that fell off somehow; they show the double-diode across the transformer / injector / whatever, rather than across the mosfet.

It's very much like the sketch SymTech posted, but instead of pointing down to GND it points up to +12v. Would this make a difference? Could one cause more problems than the other?
In that case, the clamping voltage is the Zener/TVS voltage plus +12V. For example, a 36V Zener diode configured that way would clamp the voltage to ~48V. Theoretically the circuits are the same, but it may be a little more difficult to avoid introducing noise to the +12V rail instead of a ground plane.
SymTech Laboratories, LLC ----- MSQ File Repository: megasquirt.symtechlabs.com
Image
Peter Florance
Super MS/Extra'er
Posts: 3653
Joined: Fri Apr 02, 2004 8:40 pm
Location: Virginia Beach, VA
Contact:

Re: dead vnd5n07

Post by Peter Florance »

SymTech Laboratories wrote: Theoretically the circuits are the same, but it may be a little more difficult to avoid introducing noise to the +12V rail instead of a ground plane.
True, but because the + 12V rail is not a reference like ground is, it may be less problematic.

Having said that, sticking to practice of returning all sensor grounds to MS/US seems to keep ground noise pretty minimal.
Peter Florance
PF Tuning
81 BMW Euro 528i ESP Car
60-2 Wheel LS2 Coils, Low Z Inj
Co-Driver 1999 BMW E46 DSP car.
Matt Cramer
Super MS/Extra'er
Posts: 17499
Joined: Thu Apr 16, 2009 8:08 pm

Re: dead vnd5n07

Post by Matt Cramer »

racingmini_mtl wrote:Is there a difference in the assembly process between the MS3X and the uS? I don't see anything in the data sheet on the subject but are there any unusual soldering requirements? Has anyone contacted STM about this?

Jean
I'm checking with them now.
Matt Cramer -1966 Dodge Dart slant six running on MS3X
SymTech Laboratories
Super MS/Extra'er
Posts: 2188
Joined: Sun Aug 31, 2008 4:02 pm
Location: South Florida, USA
Contact:

Re: dead vnd5n07

Post by SymTech Laboratories »

In case it's of any interest, the temperature rise for the VND5N07 part, operating a 14 ohm injector at 80% duty cycle over a thirty minute period is less than 5.5°C/10°F.

We contacted STMicroelectronics, explained how the part was implemented, and explained the troubleshooting measures we've taken so far. They confirmed that the VND5N07 is being used in accordance with its specifications, and they couldn't identify any potential causes for the failures. They ultimately suggested failure analysis through an ST distributor.
SymTech Laboratories, LLC ----- MSQ File Repository: megasquirt.symtechlabs.com
Image
racingmini_mtl
Super MS/Extra'er
Posts: 9128
Joined: Sun May 02, 2004 6:51 am
Location: Quebec, Canada
Contact:

Re: dead vnd5n07

Post by racingmini_mtl »

SymTech Laboratories wrote:In case it's of any interest, the temperature rise for the VND5N07 part, operating a 14 ohm injector at 80% duty cycle over a thirty minute period is less than 5.5°C/10°F.
Interesting results but the duty cycle may not be the most important issue. I would think frequency would also be something to check. Using the same 80% duty cycle but with a higher frequency to simulate higher RPM and number of squirts might be another interesting test. But that may not change your results.

Jean
jbperf.com Main site . . . . . . . . . . . . . . . . . . . . . . jbperf.com Forum
Image
dontz125
Super MS/Extra'er
Posts: 4200
Joined: Mon May 11, 2009 7:14 pm
Location: York, ON
Contact:

Re: dead vnd5n07

Post by dontz125 »

A high-revving bike would presumably make the heaviest demand - would you be able to try 50% and 85% DC at 310Hz?
QuadraMAP Sensor Module -- PWM-to-Stepper Controller -- Dual Coil Driver
Coming soon: OctoMAP Sensor Module
TTR Ignition Systems
SymTech Laboratories
Super MS/Extra'er
Posts: 2188
Joined: Sun Aug 31, 2008 4:02 pm
Location: South Florida, USA
Contact:

Re: dead vnd5n07

Post by SymTech Laboratories »

racingmini_mtl wrote:Interesting results but the duty cycle may not be the most important issue. I would think frequency would also be something to check. Using the same 80% duty cycle but with a higher frequency to simulate higher RPM and number of squirts might be another interesting test. But that may not change your results.
dontz125 wrote:A high-revving bike would presumably make the heaviest demand - would you be able to try 50% and 85% DC at 310Hz?
Those are both great points. With a 1k ohm gate resistor, these MOSFET's should switch in less than 2.5usec, which equates to just a fraction of a watt in switching losses at anything under 1kHz, but it's was still worth investigating further.
vnd_losses.png
At ~310Hz (18,600 RPM) and 50% DC, the temperature rise over a thirty minute period was approximately 10.0°C. At 85% DC, the temperature rise was about 14.5°C. The measured temperature rise is consistent with the DPAK part's junction-to-ambient thermal resistance and total power dissipation at that frequency and duty cycle.

A quick note about the test conditions: the readings were taken from bare parts on a paper substrate (no copper, no heatsinks). The VND5N07 was covered to minimize air flow over it.
SymTech Laboratories, LLC ----- MSQ File Repository: megasquirt.symtechlabs.com
Image
Post Reply