Re: GDT's for low frequencies
Marko, Thu Sept 07 2006, 05:00PM

100 turns is way too much for flyback frequencies.

Just couple of trns around a ferrite toroid is more than enough, and you need to use thick, possibly multifilar wires for good peak current.

Cat 5 cable like steve ward's DRSSTC GDT is usually a good approach.

Don't use litz wire like me; it gives better coupling but isn't worth the danger of windings shorting together.

Re: GDT's for low frequencies
HV Enthusiast, Thu Sept 07 2006, 05:01PM

You can download the student version of PSPICE and actually model the transformers before building them. Much easier. I have PSPICE simulation gate drive transformers on my website.

Re: GDT's for low frequencies
Thomas, Thu Sept 07 2006, 07:19PM

I made another transformer, this time with less turns. I used around 18. This is the same amount of turns I use for my SSTCs. However, I don't get the same preformace as I do at 100kHz - 300kHz. I get a nastly looking waveform that looks like dead-time at around 5 - 10kHz and at around 20kHz I just get some noise and the voltage is only 8 volts. The input is 15V, so i'm lossing quite a bit of power here. EVR I will try the program you recomended, however, I must first look for either a linux version or mac. But should I try oh say maybe 30 turns.

Thanks,
Tom

Re: GDT's for low frequencies
Marko, Thu Sept 07 2006, 07:32PM

What driver are you using?
What wire thickness and type (enameled, PVC) and HOW did you wind the GDT?
You want to reduce leakage inductance to minimum, all windings must be wrapped tight together.
What frequency you generally intend to run at?
What device (MOSFET/IGBT) are you driving with the GDT?
How does the output look without load?
Have you checked for shorts in the GDT?

Saturation is unliketly to be a problem at 20-40kHz, and there really isn't need to use monster amounts of turns.

Re: GDT's for low frequencies
Avalanche, Thu Sept 07 2006, 07:43PM

I've used a GDT at 13Khz to drive a halfbridge for a flyback.

For the core, I just used one of those ferrites from a monitor cable, and the primary and secondaries on the transformer were wrapped tightly together, total about 15 turns for a 1:1 transformer. It performed ok. How are you driving the gate drive transformer? This could be where the problem lies. I think you'd need to drive it with AC for best results (i.e. a small fullbridge, or 2 driver chips - 1 inverting, 1 non-inverting).

Re: GDT's for low frequencies
HV Enthusiast, Fri Sept 08 2006, 12:55AM


A gate drive transformer designed for low frequency will not work well at high frequency. A gate drive transformer designed for high frequency will not work very well at low frequency.

There is a balance between leakage inductance and magnetizing inductance which must be made for proper operation at the desired frequency. Get the PSPICE. You'll save yourself lots of time.

Re: GDT's for low frequencies
Sulaiman, Fri Sept 08 2006, 07:54AM

I have found a few problems using transformer drive;
1) Primary Inductance
The change of current in an inductor dI = (1/Inductance)xintegral(V.dT)
for squarewave drive that's Ipk = V.T/L

Core saturation is determined by amp.turns, N.I

The nett result is, core size is INVERSELY proportional to frequency
so a very large core is required at low frequencies.

Changing to a laminated steel core for low frequencies (low audio) is good
but steel is very lossy at high frequencies (above audio)

2) "Flux walking" (google it)
As above, the change in the current in an inductor is (1/L) x integral(V.dT)
With squarewave drive
When the transformer is ON dI=Von.Ton/L
When the transformer is OFF dI=Voff.Toff/L
IF these two are not IDENTICAL
then each cycle will see a nett change in current in little 'steps'
until the current in the winding causes core saturation.
For push/pull at 50% duty this 'flux walking' can be compensated for
with a little primary resistance.

With a Fluback where Ton >> Toff
the only way to 'reset' the core flux to zero is to allow
Von.Ton/L <= Voff.Toff/L

Voff >= Von x (Ton/Toff)

e.g. A flyback driven from 12V with Ton=45 us, Toff=5us (20kHz, 90% duty)
Voff >= 12 x 45/5 = 108 Volts.
so to drive a transformer with a high duty, you must allow for a very large "flyback' spike
If you supress/catch/snub this spike then you will cause the transformer to saturate.
This 'flyback' spike also appears at the output of the transformer
This reverse voltage will damage bjt junctions or MOS barriers
so is unsuitable for ALL transistors !!!!

SO, for a drive other than 50/50 squarewave OR sinewave, DON'T use transformer drive

EDIT: If you simply MUST use transformer drive thare are a few 'tricks' for getting around these problems, but it's better to use either direct-drive or optical-drive.

Re: GDT's for low frequencies
HV Enthusiast, Fri Sept 08 2006, 11:28AM

Thomas,

If you provide me with the following, I'll simulate it for you and design your transformer:

1. Ferrite Core Part Number
2. Schematic of your driving circuit
3. Device you are driving (MOSFET, etc...)
4. Frequency

Re: GDT's for low frequencies
Thomas, Sun Sept 10 2006, 05:58PM

Thanks EVR, but I don't really know what type of material I'm using. The package never said. I only bought them because I heard that they are a good substitute for material 77 ferrite. I never used material 77 on any of my designs. Now that I think about it even my SSTC mosfet chip drivers run pretty hot and require heatsinking to run. Maybe these toroids are not a good choice for these applications after all. For my flyback driver, it's pretty much a sstc circuit but instead of driving the oscillator at 200kHz it's being driven at around 20kHz. I'm using a TL494 for the oscilator and a TC4422 to 10uF cap to primary to ground on my flyback driver. I'm now actually using a audio transformer and it works fine, but I just series a 10 ohm resistor to dissipate some heat. I'm also running at 15V. Back to my sstc I think I'll aquire some M77 toroids and rewind them at at 18 turns. Also try driving a M77 toroid at low frequencies typical to a flyback, since I'd imagine it being better then a lower current audio transformer.

EDIT: I've done some reseach and I like the fair-rite toroids part number: 5977006401 these are material 77 though. I want to bulk pack some for future sstcs and drsstcs plus flyback related things. EVR what do you think of material 77 over 78, the difference is just an extra 300ui does it make much of a difference? (I can't seem to find 5978006401) PS: i'm gonna get a win xp version from my school (MSDN) so I can run pspice.
THanks for the software suggestion

Tom

Re: GDT's for low frequencies
HV Enthusiast, Mon Sept 11 2006, 11:19AM

Type 77 and 78 material will both work fine.

What kind of devices are you running???

Re: GDT's for low frequencies
Dr. Shark, Mon Sept 11 2006, 04:08PM


Just in case you missed this, I'll emphasize that it is a VERY important fact to consider. Even for SSTC frequencys I found that a lot of the cores I tried simply did not have enough crosssection to work without saturating, and I had to use way more turns that people usually suggest here. What is the size of your cores?

Re: GDT's for low frequencies
Thomas, Mon Sept 11 2006, 08:58PM

Oh this is for a flyback h-bridge.

I never knew that the size of the core had a impact on the windings. I never really took this in to concideration. The outer hole is 2.5cm, the inner hole is 1.5cm and the width is 1cm. I have pspice, but I don't understand how to use it yet. I need to read some documentation, however, homework is first still my priority :() Too bad i'm not take any electrical classes this semester.

Re: GDT's for low frequencies
Sulaiman, Tue Sept 12 2006, 08:03AM

OK, so the core is 0.5cm wide by 1cm thick, cross-sectional area = 50 mm2
For most ferrites you can use 0.2T peak an the 5-40 kHz region
so the peak magnetic flux is 0.2T x 50mm2 = 10 x 10-6 Wb (10 uV.s)

At 5 kHz and 50% duty the ON time is 100 us
For 10 V gate drive that's 10V x 100 us = 1 mV.s

So you need 1 mV.s/10 uV.s = 100 turns.

No need for a simulator!

Re: GDT's for low frequencies
Steve Conner, Tue Sept 12 2006, 09:53AM

Technically I think 50 turns should be ok. (In the steady state the flux will swing from -10 to +10 uv.s per each half cycle, not 0 to 10)

I notice Tom said that he already tried a toroid with 100 turns and it didn't work, though. I'm going to guess this is because he just wound two separate windings instead of using the multifilar technique with its much lower leakage inductance. (good luck making a 100 turn 5-filar winding!)

Steve

Re: GDT's for low frequencies
Marko, Sat Sept 16 2006, 01:12PM

5Khz is pretty low frequency for flyback driving!

You'l be in toruble fitting 100 turns on a small core using thick enough wire, and mantaining low enough leakage inductance and resistance!

For 50kHz, steve ward used 10 turns and I frequently use this as experimentally-confirmed orientation point

40kHz would probably requirefew more turns on core of same size, but on very small core (5mm^2) you will have trouble fitting everything.


For 5kHz and around, I would use a high-perm iron powder toroid (I guess that from ATX supplies could serve) and hope it would work for up to some 10kHz. It would allow you to use much more fair number of turns. (I need someone to veryfy this as I didn't try it^^)

Re: GDT's for low frequencies
Wolfram, Fri Sept 22 2006, 03:39PM

How about a high-side FET gate driver chip, for these frequencies they should work fine.