RE: [TCML] SISG Repair

["caspina" <caspina@xxxxxxxxx>]

Hi Steve,

Thanks for the info.

What you say doesn't surprise me, 800 A is indeed a lot, however in practice
I think it will be lower, but still...
It's just that I read a post once (by Marc Dunn I believe) that his SISG
works so well and so long (years).
If the IGBT can take say 'n' cycles before it wears out then with the
modulation I use I'm gonna be much faster at this number
then one who is stuck to mains frequency of course, apart from the thing
getting hotter too.

Well, my depression is almost over and then I start taking things apart and
do some repair ;)
Curious what I will find...

Michael




-----Original Message-----
From: Steve Ward [mailto:steve.ward@xxxxxxxxx]
Sent: vrijdag 8 juli 2011 0:23
To: Tesla Coil Mailing List
Subject: Re: [TCML] SISG Repair


Micheal,

It is possible that the pulse lifetime of the IGBT is a limiting factor.  I
believe the main failure mode with this style of IGBT (versus the module
style) is fatigue of the bonding wires to the die.  Thermal cycling on short
time scales causes stress between the silicon and the bonding wires, and i
guess eventually the bond wires start to let go.  This can lead to reduced
contact area, or just more contact resistance, which eventually causes
localized overheating and failure.

Im not sure what the voltage drop of your IGBT is at the 800A, but given
that the device is really only rated for 240A pulse, i think the 800A is
really pushing it pretty far.  I personally would try to estimate the
junction temperature increase on a per-pulse basis... luckily your switching
losses are basically zero, so you should be able to estimate the conduction
loss pretty accurately.  Using the transient thermal impedance data, you can
then estimate the delta Tj.  Im not sure if there are any good "rules" about
this, but a powerex engineer told me if delta Tj is less than 30*C, it
should last "a really long time", but then he also seemed to indicate that 2
million cycles was also "a really long time", in which case 30*C delta is
probably wayyy too much for us!

In any case, if you cannot find any evidence of power supply failure causing
large follow-through current on your SISG, then i would seriously look into
the thermal cycling issue.  I suspect it might actually be the problem, and
it could indeed take many many many pulses before it shows up.  Yes, these
things *do* wear out over time, especially in pulsed operation!

By the way, excellent coil ;-).

Steve

On Thu, Jul 7, 2011 at 2:15 PM, caspina <caspina@xxxxxxxxx> wrote:

> Ok, so here's some more info on my coil:
>
> It's a 6" DC-Resonant Charging coil.
> Power supply = 2 MOT secondaries in series bridge-rectified, and another
> identical setup like this in parallel for more current (hence 4 MOT's).
> Smoothing cap = 30uF (28 elco's 220uF / 450V), so I have clean 6kV dc.
> Charging choke = 8 fluorescent ballasts wich are wired in such a way that
I
> can change from 4 in series (2.5H) to 8 in series (5H).
> When I put 4 in series I can put the other string of 4 in parallel with
the
> first one (1.25 H).
> This is the configuration I use most, it allows higher BPS.
>
> >From here it goes to the SISG (wich is a triggered SISG).
> IGBT = IRGPS60B120KD
> SIDAC = K2401GRP
> TVS = 1.5KE24CA (24V) and P6KE33CA (33V)
> Because I absolutely wanted a triggerable version I designed the PCB's
> myself: 1 Module has 4 250V sidacs, hence 1kV / Module.
> To build a triggerable SISG I studied many posts on TCML by Finn Hammer
and
> Terry Fritz.
>
> Tank Cap = 100nF (8x8 strings of 100nF / 1600V caps)
> Resonant Frequency = 125kHz with smallest toroid (6" x 23.6") and 104kHz
> with biggest (5" x 40").
>
> In ideal case the primary peak current should be around 800A... maybe too
> close to the limit of the IGBT's :0
> The IGBT's get good warm if BPS is between 400 - 800, above that they will
> go hot but I try not to do so too much/too long.
> Since I can trigger the SISG I use the coil frequently with audio
> modulation
> and it has been working fine for months, though I must say
> I don't apply longer runs then 2 minutes.
>
> The SISG may have been destroyed due to internal arcing of the fluorescent
> ballasts: a lot of energy then can find a clean path to the SISG...
> If the IGBT's are broke because of too much stress (current) during
> operation then I would have expected them to have died earlier...
>
> So in a way I am curious of other 'SISG'-users that have encountered
> problems with theirs...
>
> Cheers,
>
> Michael.
>
>
>
> -----Original Message-----
> From: Quarkster [mailto:quarkster@xxxxxxx]
> Sent: woensdag 6 juli 2011 21:09
> To: Tesla Coil Mailing List
> Subject: Re: [TCML] SISG Repair
>
>
> Michael -
>
> There hasn't been much information circulated about failure modes with
SISG
> coils, so would it be possible for you to provide some additional
> information on your coil?
>
> 1. Are your SISG modules from Mark Dunn (4 modules per PCBA), copies of
> Terry Fritz 2-module PCBA, or some other board design?
> 2. What IGBT P/N did you use?
> 3. What SIDAC P/N did you use?
> 4. What value capacitors are used in the charging/tank circuit?
> 5. Are you using a 1-MOT or a 2-MOT power supply, or something else?
> 6. Do you know the approximate resonant frequency of your coil?
>
> It should be fairly easy to desolder the IGBT and determine if it has
> failed, or if the SIDACs have failed.
>
> Quarkster
>
> --- On Wed, 7/6/11, caspina <caspina@xxxxxxxxx> wrote:
>
>
> From: caspina <caspina@xxxxxxxxx>
> Subject: [TCML] SISG Repair
> To: "TCML pupman" <tesla@xxxxxxxxxx>
> Date: Wednesday, July 6, 2011, 4:33 AM
>
>
> Hi list
>
>
> I have a SISG coil, SISG = 15 modules (15kV)
>
> Now 6 Modules are shorted and need to be replaced.
> I'm not sure what to do with the others...
> Could they be 'partly damaged' but still work for a certain time ? Best to
> replace all modules, maybe even everything?
> Has anyone experience with this?
>
> Michael.
> _______________________________________________
> Tesla mailing list
> Tesla@xxxxxxxxxx
> http://www.pupman.com/mailman/listinfo/tesla
>
>
> _______________________________________________
> Tesla mailing list
> Tesla@xxxxxxxxxx
> http://www.pupman.com/mailman/listinfo/tesla
>


_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla