PopularFX
Home Help Search Login Register
Welcome,Guest. Please login or register.
2018-12-10, 09:23:34
News: Check out the Benches; a place for people to moderate their own thread and document their builds and data.
If you would like your own Bench, please PM an Admin.
Most Benches are visible only to members.

Pages: 1 2 3 4 5 6 7 8 9 10 11 12 [13] 14 15 16 17 18
Author Topic: Controller No5 With Protection - Looking for Explosions  (Read 90798 times)

Hero Member
*****

Posts: 1490

Increasing the drains (bifilar coils) voltage from 24V to 42V still does not return the crackling sounds.
A mild whispering is all i can detect together with some shown oscillations on the scope.
I tried all kinds of combinations of frequency, pulse width and delays.

Could it be that the present setup with 900V Cree MOSFETs and 14A drivers is to beefy for the effect to occure (like avalanching or so)?

Itsu
   

Group: Administrator
Hero Member
*****

Posts: 3603


Buy me some coffee
Hi Itsu
I never tried the Cree Sics they do have quiet a high on resistance but not sure if that is enough to stop the crackling, do you see the large pulse appear when you do a sweep on the delay.
   

Hero Member
*****

Posts: 1490

First screenshot shows the gate signals (yellow and Blue) of the 2 MOSFETs (220ns delay), and the drain signals (green and purple) of the same MOSFETs with attached bifilar coils.
To be noticed is that allthough the gates are switched 220ns apart, the drains show a timed switching.
I would have expected that we would see also a delayed switching on the drains (220ns apart).
(yellow and green are MOSFET 1, blue and purple MOSFET 2).
 
Second screenshot shows more delayed gate signals (700ns apart) and again the timed drain signals.

So the delayed gate signals do not translate directly to a delayed drain signal, is it the tramsformer function of this bifilar coil causing this?


Itsu
   

Hero Member
*****

Posts: 1490
Hi Itsu
I never tried the Cree Sics they do have quiet a high on resistance but not sure if that is enough to stop the crackling, do you see the large pulse appear when you do a sweep on the delay.


you think 0.065 Ohm is quite high?       I am sweeping (+15), but no pulse is seen up till now.

Itsu
   

Hero Member
*****

Posts: 1249
Could it be that the present setup with 900V Cree MOSFETs and 14A drivers is to beefy for the effect to occur (like avalanching or so)?
Yes it could be.
Also, the drain pulse shape can be different (and its frequency content)
   

Hero Member
*****

Posts: 1490

Hmmm,   i went back to using the 2x IRFP260N MOSFETs driven my a single dual MOSFET driver IXDD604 with 42V on the drains, but no more crackling is heard here too.
It looks like boxing up Peters box has tamed the crackling as that is the only change i did.

I will continue to do some sweeping, but up till now no crackling pulses / oscillations are noted.


Itsu
   

Hero Member
*****

Posts: 1490
Bump.


since i boxed up Peters box as mentioned above, i never was able to create the crackling sounds i had.
I will go over this setup again to see why that is.

For now i have again the earlier setup using Peters box (Freq. 200Khz, Pwidth 64ns (max.), phase 220ns).
I have 18V on the MOSFET driver (dual output IXDD604) and 25V (2x batteries) on the drain of ONE MOSFET.

I use only ONE MOSFET now to see how this ONE MOSFET behaves
The other MOSFET has no drain voltage nor is the MOSFET driver input connected.

I use the max Pwidth from Peters box to get the max gate signal.
The both MOSFETs are loaded with a 100 Ohm resistor (so no (bifilar) coil).

The screenshot below shows in yellow the drain voltage of the working MOSFET and in purple the gate signal of
this MOSFET.

We see that the MOSFET opens quite fast/good, but it closes very  very late.
We also see that the gate signal (purple) is nowhere near the 18V that the driver is running on.
This causes the MOSFET to linger in off/on state to long i guess.
The Pwidth of Peters box (64ns) has been shown by the vertical cursors.

I think i have to improve on this simple resistive setup before going into the inductive bifilar coil setup.

Why does the gate signal stay on longer then the driving pulse (108ns versus 64ns)?
When using the required 32ns Pwidth, the gate signal is only about 7V, how to improve this to 18V (which is the driver Vcc)?
How to improve the MOSFET close time (adding a schottky diode across the gate resistor?)?

MOSFETs are irfp460's

I added the schematic from Peters PCB i am using showing the IXDD604 dual driver and MOSFETs setup. (be aware that i do NOT use the PIC white noise chips as input, instead
i use Peters magic box)

Thanks,  Itsu
« Last Edit: 2018-10-20, 09:11:25 by Itsu »
   

Hero Member
*****

Posts: 1249
Try measuring what this type of MOSFET does with these kind of pulses, like this:

   
Sr. Member
****

Posts: 253
Hi Verpies,

It is good to see you around again, hope you are doing fine.

As an addition to your suggestion, I would like to add the followings to Itsu:

Hi Itsu,

I assume you checked the output pulse from the IXDD604 when the MOSFET gate is not connected, is this so? Did you find the output pulse ok? as its pulse width should be, coming from the box?

If the drive pulse coming from the IXDD is okay in itself, then perhaps the IRFP460 is to blame?

Otherwise, there is a method to improve MOSFET switch off time by using a pnp transistor and a diode between the gate-source, see this link:
http://www.imajeenyus.com/electronics/20111010_40-400V_supply/index.shtml

It shows a different driver IC which has an open emitter output hence the need for the 470 Ohm pull down resistor. Your IXDD should discharge the gate source capacitance without long delay, it is designed to sink current, so normally such speed up circuitry should not be needed for such IXDD type driver.
Anyway, such speed up circuit should help reduce the speed off time for the MOSFET.

Maybe the unconnected driver IC input ought to be connected to the ground if you left it floating, not to disturb the operation of the driver chip.

Gyula
   

Hero Member
*****

Posts: 1490

Hi verpies, 

nice to see you, and thanks for the diagram (Hmmm, where did i see that before?).
https://overunity.com/12736/kapanadze-cousin-dally-free-energy/msg342647/#msg342647

I thought that what we see in the first 60ns of the gate (purple) signal was due to the Miller effect.
Guess it will be hard to overcome with Peters small (32ns) pulses.

Let me see what i can do with your diagram, i still should have those source/sink output drivers (ucc27511) somewhere.


Regards itsu
   

Hero Member
*****

Posts: 1249
Let me see what i can do with your diagram, i still should have those source/sink output drivers (ucc27511) somewhere.
They are not mandatory. In that circuit, you can use your single output driver, too.
   

Hero Member
*****

Posts: 1490

Gyula,

i did check again the output of Peters box, and they are still the same as tested before in post #270 above, see here:
http://www.overunityresearch.com/index.php?topic=272.msg58828#msg58828

I think they are good for the job.

I did use the IRFP460 MOSFETs before and they did produce the crackling sounds, but i think Peter used IRF840's, so i could try them.

Thanks for the link, i will take a look tonight.

Good idea to tie the unused IXDD604 output to ground.

Itsu
   

Hero Member
*****

Posts: 1490
They are not mandatory. In that circuit, you can use your single output driver, too.

Ok, Roger that, even better.

Itsu
   
Sr. Member
****

Posts: 253
...

Good idea to tie the unused IXDD604 output to ground.

Itsu

Dear Itsu,

I wrote input, not the output.  Leave the unused output unconnected.

Gyula
   

Hero Member
*****

Posts: 1490

Input was i meant to write  O0
   

Hero Member
*****

Posts: 1490
Initial test done with verpies his diagram as shown below.

input pulse is 100ns wide
driver is ucc27425 (using non-inverted output only, running on 16V (as is max).
single R1/R2 of 3.3 Ohm
2x IRFP460 with 42V on their drains.
200Khz pulse repetition frequency.


Screenshot shows:
yellow: input pulse from FG
blue:   gate Q1 signal
purple: drains Q1/Q2 signal
green:  output across R4

Not sure where/what to look for.

 
I have some IXDD614PI drivers which can withstand up to 35V which i could use.

Itsu   

   

Hero Member
*****

Posts: 1490
Made some extra screenshots with varying input pulse widths (20, 50, 100 above, 200ns).
Last one is a zoomed out screenshot showing the massive ringing.

Itsu
   

Hero Member
*****

Posts: 1249
These MOSFETs have a large gate charge.
Could you make a video showing how these waveforms are changing as you are varying the R4 (make it a pot) ?
   

Hero Member
*****

Posts: 1490

Yeah,  like 4.2nF   (Ciss 4200pF) according to the datasheet.

I will change R4 for a 5K pot and vary it showing the waveforms.

Not sure i will get around it to do it today as i seem to have some other obligations  :)

Itsu
   

Hero Member
*****

Posts: 1490

I used the same setup as yesterday with a 100ns wide pulse.
38V on the drains, 16V on the driver.

Potmeter (5K) was initially set to 1K, and went down in 4 full turns from 1K to 800 Ohm, 580 Ohm, 360 Ohm, 200 Ohm.
Then up to 1K again and from there in 4 full turns up till about 2K.

No change was seen, so i turned it down even further untill a change was noticed which was when the pot went to 0 Ohm.

Video here:  https://www.youtube.com/watch?v=gSlTVd-gXgg


Regards itsu
   

Hero Member
*****

Posts: 1490

i realize that the video of the screen was not that clear, so i redid the tests with a lowered yellow amplitude
at the 4 positions of the pot and at 0 Ohm (1K, 800, 580, 360, 200 Ohm and 0 Ohm), see screenshots.

Itsu
   

Hero Member
*****

Posts: 1249
At 0Ohm the Ch4 should flatline.
Also, it looks like the impedance of the driver is too high to drive the gate stiffly enough.

There is a lot of ringing. How much inductance is between the driver and the MOSFET ?

...and why is the Ch1 signal so wobbly?  It is the input from the FG, isn't it.? What's affecting it?
   

Hero Member
*****

Posts: 1490

I can use another driver then the now used ucc27425, but then i need to remodify the used pcb i am using now
or build  a new setup.

The below layout is from the PCB used (modified), so there is some 1.5cm long link between the 3.3 Ohm
gate resistor and the Q1 gate.

The input to the ucc27425 driver is indeed from the FG, below 1st screenshot is that signal without power to the driver.
2th screenshot is when i put 16V to it.

Itsu
   

Hero Member
*****

Posts: 1249
Something is not right.  The power supply rails to the driver appear to be unstable.

I recommend your old ground-plane design in which you soldered to an unetched PCB and the driver was soldered directly to the gate terminal and separately choked and bypassed with a dedicated bank of different caps (each cap has a different self-resonance frequency)

The driver must be able to source and sink many amps, to provide stiff gate drive.  After the gate goes up, it should stay up despite the Miller current trying to pull it down.  The opposite goes for the case when the driver is pulling the gate down.  The gate resistors should be experimentally tailored for the MOSFET to be as low resistance as possible without allowing gate ringing and without overloading the driver.
   

Hero Member
*****

Posts: 1490

Hi verpies,

i understand and agree, but.......  Peter has seen these explosions / crackling effects using the PCB i am using
which has the layout i have added above, meaning without all these precausions you just mentioned.
(it off course verywell could be that thats the reason he saw them as after boxing up Peters box my crackling sounds stopped).

Anyway, this Miller effect test setup i have now, using the same PCB from Peter, obviously is not ideal so i will be
building one which does have these precausions you pointed out.

When successfull, i will incorperate these into a new PCB driven by Peters box and driving the Bifilar coil setup.

Thanks,   Itsu 
   
Pages: 1 2 3 4 5 6 7 8 9 10 11 12 [13] 14 15 16 17 18
« previous next »


 

Home Help Search Login Register
Theme © PopularFX | Based on PFX Ideas! | Scripts from iScript4u 2018-12-10, 09:23:34