AC "Fiddy" flybacks for sale / Sale and Trade / Forums

Forums

4hv.org :: Forums :: Sale and Trade

« Previous topic | Next topic »

AC "Fiddy" flybacks for sale

Move Thread

LAN_403

kiat

Thu Oct 15 2015, 07:15PM

Registered Member #2115 Joined: Fri May 08 2009, 01:17PM
Location: Singapore
Posts: 46

Hi,

Following some intense discussions with a chinese flyback manufacturer, I am proud to present THE optimized "flyback" transformer of all time (More correctly, a high voltage step up transformer).

I was inspired by Fiddy's flybacks to investigate how best to increase the performance of high voltage ferrite xfmrs -- how do we increase output voltage and self-resonant frequency, while minimizing Rac ? I was rather disappointed by the tiny V/turn Fiddly limited his flybacks to..

And so began my reading spree of the attempts of 4hv members in creating the segmented secondary bobbin, which allows high voltage while minimizing parasitic winding capacitance and winding voltage stress.

My intention: build a 100kv CW multiplier capable of providing reasonable power, as small as possible.
How? Maximize output voltage and frequency.

What limits output voltage? Almost exclusively the winding's internal insulation breakdown/operating lifetime limits. Above ~10kHz , core saturation or core temp. rise is not the limiting factor.

How do we optimize insulation? Minimize points with high E-field on the bobbin.
Previous attempts at building segmented bobbins consisted of a "2-step" transition between segments, with a small gap in the middle for the wire to slowly transition from the top of segment one to the bottom of segment 2, without bringing segment 1 and 2's tops in close proximity.
Good as this is, this doubles the number of slots the bobbin needs and is a waste of precious window space.

Furthermore, some destructive research showed that flyback transformers exclusively use a bobbin without "2-step" segment spacers -- segments transited directly from old to new, with little regard that the ENTIRE segment voltage is developed across the point where the wire crosses over/down.
Such as in the attachment.

Well , although that seems to work good enough for DC, as most of us know, once flyback diodes are broken and the output is AC, corona quickly damages the innards of the flyback and it mysteriously heats up and reduces in output, unloaded!.

And so some form of insulation of the wire, specifically only at the crossover point, was critical. How? Well heatshrink would be rather impractical due to the amount of winding required (imagine threading 10 bits of heatshrink through a wire for 200meters), and furthermore i had to convince the chinese guys to get all this done with my rather poor google translate assisted communications. So I decided: just use triple insulated wire.

so I specially requested (at the expense of complicated logistics arrangements arranging the shipment of the TEX wire from china to china, and winding setbacks due to broken sections in the wire), for the use of triple insulated wire to replace magnet wire. Besides having a much improved dielectric strength where it's most needed (between segments), the larger wall also means a slightly smaller parasitic capacitance, and fewer secondary turns (higher f0 and lower Rac).

Specifically, TEX-B has a 0.1mm insulation wall consisting of 3 layers (mylar/PPS/polyamide), supposedly withstands 3kvrms (50hz/1minute) and (according to the datasheet) has a DC breakdown voltage above 15kV, not bad for a 0.1mm wall!
Of course significant derating is needed if operating at high frequencies (20~80khz) and continuous voltage stress. Its still a matter of time before internal corona wrecks the insulation, and if anyone wants to stress test these (and maybe compare em with Fiddy's?), please let me know.

These cores would ideally be suited to a half- or full-bridge drive, hard switched at low 10's kHz or possibly reaching a quasi-resonant state as we approach the flyback's self resonant frequency (40 or 80kHz).

Sadly, I was tight on cash and wondering if they may screw up , which, thankfully , they did not. and so I only got a sample order of 3*their normal design (31AWG magnet wire*2800T, similar to fiddy's) and 3*(32AWG triple insulated*1600T) on hand.

Below are teh Tech Specs:

Core: UY20, material is TDK's PC40. Clip used has M3 threads. Default ungapped -- gapping spacers available but they aren't for this core.
Al ~ 3.2uH/N^2
Ve 60cc
For 200mW/cc, reasonable if forced air cooling or under oil,
f, B, fB
24khz 300mT 7200
35khz 250mT 8750
54khz 200mT 10800
91kHz 150mT 13650
For fB around 10000, 13.3V/turn

Bobbin: 8-segment, ~0.8mm insulation between segments, Segment depth ~7.5mm and width ~5?mm, OD~45mm and ID 21~23mm.

Winding: 2 options, 2800 Turns 0.23mm/31AWG double build magnet wire or 1600T 0.2mm(32AWG)/0.4mm OD triple insulated wire (TEX-B), good to many kv more than the magnet wire.
Both options fill the bobbin completely.

Ls/Ls,leakage=26/1.65H (2800T) or 8/0.5H(1600T)

High voltage winding self resonance: This is the limit of high frequency operation, which is important if driving a CW multiplier/rectifier/delon doubler.
1600T, SRF of the secondary's LEAKAGE inductance and parasitic capacitance is ~ 80kHz,
2800T, 40kHz.

Both are impressively high IMO. I hope the insulation stands up to such high frequency AC.

Current (winding heating) limit: Due to the many layers in the segmented winding, AC resistance due to proximity effect, of the secondary. far exceeds the DCR of about 100~300 ohms, especially with increasing frequency.
My own back-of-hand calculation produces:
1600T, full Ls: 120 ohm + 50ohm/kHz
1600T, leakage Ls: 185 ohm + 20ohm/kHz ??
2800T, forgot: 340 ohm + 100ohms/kHz.

We should be able to use, Pwinding= sum of Irms^2*Rac(f) for the different frquency components of the output signal, maybe some approximation like P=Irms^2*Rac(fundamental)*2 for when driving a ()-stage half wave CW

Secondaries are vacuum potted in white epoxy. 2 or 3 have been "topped up" with silicone as the epoxy didn't cure flat -- I have dug some up to verify.

Primary: Right now all come with a primary of ~130Turns, 14//34AWG "bunched" magnet wire, not quite a litz winding.
Lp= 54mH, k(pri-sec)~0.9685

Yes. The primary was likely suited for a low power off-line driver at about 2.5V/turn; perfect for the 2800T winding running all day long.
To get the most out of the 1600T winding though, the primary needs to be chucked and replaced with ~25T or more of wire. Keep the OD < 1.7mm for a single layer winding.

Anyways thats about 11 meters of 0.28mm^2 bunched wire, good to about 200kHz i guess, so go ahead and use it for ur other projects.

So basically, I have some "and settled on 2 models for sale, "Fiddy type" and my "optimal" TEX winding type. To attain a sufficiently high output voltage the primaries will have to be re-wound for the TEX flyback.
The price comes up to be 35usd for the magnet wire one and 51usd for the triple insulated one, before shipping from singapore.
Indicate interest here:

Wolfram

Fri Oct 16 2015, 02:33PM

Registered Member #33 Joined: Sat Feb 04 2006, 01:31PM
Location: Norway
Posts: 971

Excellent work. I suggest you also post a copy of this thread in the High Voltage forum as it contains some new and very useful information.

I have some thoughts on the driving method, I will move this to the other thread if you make it.

For the driving, I can recommend driving them above resonance with a half/full bridge, while controlling frequency to adjust (or even regulate) output voltage or current. This allows you to run the transformers at the maximum practical frequency, while also allowing zero voltage switching of the bridge MOSFETs. I've used this topology before in a 70 kV 4 mA supply, consisting of a transformer and a 4-stage multiplier, and it was very efficient.

There's possibly also a second benefit to this driving method, but I haven't gotten this confirmed, so far it's just a speculation from my side. When you're operating the transformer close to the resonance, you're resonating the leakage inductance with the reflected secondary capacitance. Since a significant fraction of the output voltage is developed across this "leakage inductance", the core doesn't really see all the "volts per turn" that the secondary does. This means that core losses are lower for a given output voltage when operating in this region. The backside of the coin is that copper losses are increased, as some of the current passing through the secondary is not going to the output but into the parasitic capacitance.

Do you have a rough estimate of shipping cost for a pair to Norway?

kiat

Fri Oct 16 2015, 04:23PM

Registered Member #2115 Joined: Fri May 08 2009, 01:17PM
Location: Singapore
Posts: 46

Sure. Link:

Replied you on that thread

shipping for 2 will be 2kg, 54usd by standard air mail (10~14days) with speedpost. I am trying to find a cheaper option -- suggestions?
Actual weight is about 1.45kg without packaging.

Can throw in some spare fets that I have, till the package is 2kg. Have some IPW60R045CP and C3D20060D and DSEI30-06A (all used, pulled from equipment)

Do you happen to have a Fiddy's flyback too?

************************************************ *************
Update: I have talked the manufacturer into doing smaller minimum orders of qty 3 and above.
Shipping from China is much cheaper at 45usd for 3 flybacks, to most of Europe and USA.
More details at

The prices are 35usd (magnet wire) and 51usd (triple insulated wire) before shipping
Shipping from Singapore is 46usd/flyback and directly from China is 45USD/3 flybacks.

kiat

Mon Nov 30 2015, 04:37PM

Registered Member #2115 Joined: Fri May 08 2009, 01:17PM
Location: Singapore
Posts: 46

If anyone is interested drop me a mail or fill in the form

(my email is in there), by next week. I'm getting another batch of samples done.

It's 35usd (magnet wire) and 51usd (triple insulated wire) each, (please get at least two), and 45usd shipping from china to most areas for three flybacks.

I will send your address to the manufacturer after confirmation.

The "magnet wire" version is roughly the same as "fiddy's flyback".

Anders and hen918 too -- haven't got any updates from you guys

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob