Needs some pieces of advice on how to diagnose an electric fence

[gg64]

Hey,

So I am trying to figure out the problem onto an electric fence (for cattle, advertised 3J 10kV).

(The main power here is 230v, 50Hz)

Peoples already hepled me figuring what a component was.

So here is a schematic:

(schematics on QUCS: regular_sch.PNG)

So my understanding is:

the left side is supposed to provide the power to the circuit (left of the transistor here, which actually a TRIAC).

So the C3 capacitor (25µF) get charges and could reach up to 650v~(230*sqrt(2)*2)

I know it used to reach about 420v before triggering though.

Here is a simplified version of the triggiring side:

(schematics w/ grouped resistors: simplified_sch.PNG)

So the pulse tranformer and the C3 capicitor (25µF) would provide the voltage divider (1.48M/(1.48M+2.21M)).

(I know right, the voltage divider formula works if the output current is zero but there, I think the final voltage is relevant)

Therefore multiplying the voltage by 0.401 to gives charges to the C2 capacitor (4.7µF).

That would yield a maximum voltage of 260.82v (=650.44v*0.401)

So I replaced the original DIAC by a DB3 which should be about 30v-32v (threshold).

Let's take 30v for the maximum best case.

So the charging ratio would be of 11.5%~(30v/260.82v)

Which according to this website (rc circuit basically charges up to 11.5%) (:http://sim.okawa-denshi.jp/en/CRlowkeisan.htm)

Which would be about 2.1sec to reach the 30v (2.21M+1.48M 4.7µF) (R4+R8 w/ C2)

Then the discharge circuit would be triggered (allowing a current pulse through the pulse tranformer).

Another R-C would be formed when the DB3 diac would conduct (221ohm w/ 4.7µF) (R8 w/ C2)

The datasheet of the triac says (nxp bt139-600e) 1.6v to trigger it.

The discharge ratio would be 96.6%~95%~(=(30-1.6)/30).

The discharge time would be 3.1msec.

Making a cycle 2103msec.

Question:

Did I made an error ? (I think I got a PEKBAC problem, aka I am probably making a very obvious mistake)

Because I measered a 901ms cycle so I suppose I am wrong somehow, and I can't figure why my TRIACs and DIAC keep being fried for some reason...

If you guys want to know what I already did:

https://www.eevblog.com/forum/repair/electric-fence-with-an-unknnown-component/

I did some more testing but I am pretty much out of ideas (maybe I need to suppose it is a faulty diode)

Basicaly, I can diagnose the symptom but not the cause, therefore I keep frying components.


Thanks for reading.

[Armadillo]

I think there are 2 obvious errors.

1st error is obvious, 2nd error is dangerous and may trip the earth leakage breaker.

The frequency is 50 hertz which is 0.02 seconds and your cycle time is nearly ~ 1 sec.... do you think anyone will design such circuit?

Maybe you  should retrace the circuit or post front and back of the pcb clearly, so that helpful members can participate to point you in the right direction despite your 2 postings already on this site.

[gg64]

Hey, thanks for the reply.

The ground on the left would not be here, it is just for QUCS simulation (basicaly sometimes nothing beats paper and a brain).

The ground on the transformer exist IRL, you can't escape it.

In practice it would sometimes trigger the differential breaking. I don't want to fix it, because it allows a feature: know when a steady defect on the fence line-> it does not trigger it most of the time though since the differential is a 500mA. But when it does the user would go search for a fence defect. Which would occur every ~two weeks or so.

Here are some photos from the modified original board (on imgur: https://imgur.com/a/a1lfhSK ).

FYI: I already did a copy of this board (components wise and PCB wise (w/ jlcpcb and KiCad), at least I thought so ?!?), and I managed to reproduce exactly the same symptom.
And yes this circuit ran for years (this thing was made in the 80/90s I think, last service was in 2004)
Name of the original product:
PICFOR RB80

Here is a typical look of it (in French):
http://www.clotures-electriques.pyrat.net/Cloture-secteur-PICFOR-RB-80-1-247.html
http://www.clotures-electriques.pyrat.net/Picfor-RB-80-Secteur-230-Volts-184.html

So basicaly I think I measured and reproduce the problem somehow.

To answer your point about the 1st problem: the resistor are 5watts on the main side, I used 7Watts resistor on my copy.

Should I provided a Bill Of Material ? (I'll try to provide something)

[Dacke]

I'm just curious but, you said you ordered these triacs from Ebay,  which is notorious for selling counterfeit components.  I see a multiple 5 packs of BT139's there for $1.79 with free shipping and the genuine component is around 90 cents each from Digikey,  which would be around $5 plus shipping.  I only say this as I've had many issues with Ebay parts,  my last experience being with counterfeit IRF510s which destroyed themselves as soon as power was applied.  I no longer order components from Ebay unless I have no other choice.  I'm just pointing this out,  you could be chasing your tail here.

[gg64]

Well, I think the problem is in front of the computer (me):

I did change the pulse transformer but my own and the coil winding are a bigger section. Therefore allowing a bigger Ampere at once.

That being said, that would not explain why it did not burn right away (after I fixed the tranformer issue on it) before I fried the board (when I did not plug the tranformer correctly....).

Question: Could a non-lamented tranformer core impact a lot the capacity of transmission to the secondary that badly ? (a signal generator would help me, and even then, I don't that is the main problem)

Speaking of the TRIACs, the very first ones I fried (from Ebay) were (it seemed legit, I got 5 for 7.2eur):
https://www.ebay.fr/itm/322308031962?ViewItem=&item=322308031962&var=511208998756&ppid=PPX000608&cnac=FR&rsta=fr_FR(fr_FR)&cust=2KK89337BG366891X&unptid=a17d3d96-b79d-11e8-86af-441ea1472d58&t=&cal=63a2a5576fd82&calc=63a2a5576fd82&calf=63a2a5576fd82&unp_tpcid=email-receipt-auction-payment&page=main:email&pgrp=main:email&e=op&mchn=em&s=ci&mail=sys

So since the coil winding was bigger than before, could it just be a pulse way to high ? Because when I think about it:
(see simplified_sch_25u_short.PNG)

Those triac are rated 16A, I am thinking I am way over that (I don't have an oscilloscope :/)

For Your Information, there was enough energy to either blow up the triac or the track (probably bad soldering after several attempt).

During my last attempt (for science): I also managed to blew up a part of the track by shorting the 1 and 2 on the triac. I am not really astonished by the energy (0.5*420*420*25*10^-6=2.2 Joules) but mostly bother by the fact that did not got transfered to the secondary somehow.

Do I need thin wire to help getting voltage at the tranformer primary ?

Reason: Limiting the current-> by resisting, because the orignal tranformer got a an isolating tape that look like it is heat resistant.
I know that is also because of likelyhood of arcing at 10kV

I might tried with a 1000 resistor in serie, to see if the triac survive.

[Armadillo]

OK, I got a bit of time to study this circuit.

C1, C3 and associated Diodes act as a Voltage Multiplier.
The device on the Sim is not a triac but a mosfet. I don't know why you select that, but my comment is;
Since the actual firing [when c2 charges up on alternate half of the power, is enough to fire the Diac], current is only 1 direction only, never both ways,
hence a SCR thyristor can be used instead.

The device used must be able to withstand at least Vdrm of 650V and with the current limit resistor as you have used, I recommend BT258-800R SCR, for example. Your selection of a mosfet is incorrect in your sim.

The output transformer in your sim, I suppose is preliminary with a ratio of only 0.03 step down.... but incase of actual step up, its going to be very dangerous without proper limit HV resistors in circuit. Just a word of caution there.

Cheers; 

[EDIT: C3 must be able to withstand at least 600V similarly, assuming very long firing time. Hence this is a lethal circuit to be handled with care. The mosfet in your sim is only rated at 100V. Of course all this depend on the firing time or time for charging in the trigger circuit selection].

[gg64]

Hey,

I tried to use a TRIAC inside QUCS but I got poor (the 25µ was charging, the triac was not triggerred, the diac was nice IIRC)/no results.

By the way the transformer is actually define (in QUCS) by the inverse of the factor for some reason, hence the 0.03 instead of 30. Unless I switched primary and secondary.

I find out about the voltage multiplier circuit recently actually.

The 25µ is rated at 450v maximum voltage, before frying the board I remember it used to charge up around 420v and trigger.

For the caution part, yes, after my first try, I am using an extenion cord with a switch, since the main is really too easy to reach.

I might give the BT258-800R SCR a try but the previous (and factory) component was a NXP BT139-600E or NXP BT139-600, (I am not sure about the E part but I am sure about the voltage).

In some technical data onto the French website (http://www.clotures-electriques.pyrat.net/), the pulse time was 120µs (it is a PICFOR RB80):

http://www.clotures-electriques.pyrat.net/Picfor-RB-80-Secteur-230-Volts-184.html

And the Itsm is 155A on the bt139-600e (but I am not sure If I am reading this right) (I think I am way above that).

http://biakom.com/pdf/BT139-600E_NXP.pdf

On fig.4: this is full sin wave cycle, I have no idea how it will behave for my case.

On fig.5: there is a dIt/dt limit, since the the pulse is supposed to be in the µs kind of order, and it is out the graph.

This confirm the current limiting resistor I was willing to give a try. Because I am not sure if it is a maximum derivative which would destroy it if I exceed it.

[Armadillo]

The limit is set by R8 and I don't know the value of 221 ohms is how you get it, but that's limit the current to less than 3A. So that R8 also plays a vital role in Components destruction if not size correctly.

The QUCS, I pronounce it as Quack Quack..... if you know what I mean.

[gg64]

Thanks for the reply, no the capacitor (25µF) is shorted by the triac.

You can look at the schematic, R8 is on the command part, not for the power side.

Follow the current.

[Armadillo]

Oh yes I can see it now.

Then the triac is going to blow because no current limit, you need to put something there to limit the current, though the transformer reactance does limit the current, so you better calculate it. what did you simulator says the peak current without the resistor?

Also where did you find a 25uF capacitor that is rated at >600 V Interesting Value but practically I have not seen one so big in electronic. In electrical yes, they call it capacitor bank house in big cabinet.

 

[gg64]

Yeah it is a big capacitor but on regular operation it would not go above ~450v.

I blew up one triac with a 221ohm (1/4watt one).

I tried again but with a 1000ohm 7watt, nothing (I need to replace the diac, it should be dead, it is not the first I blew one up).

I will keep you updated.

Hopefully the voltage don't go too high during the pulse (due to the discharge time which might overlap with the main power).

[gg64]

Ok so I tried with a triac and new diac. I could measure 618v to 612v in dc mode (at 25µF) with a 1k 7watt resistor onto the 1 pin of the triac.

It seems to oscillate with about a 1Hz frequency, I measure at the diac as well, I get 26-31v oscillation. (I suspect that the diac is polarized ?!?, I Will check this eventually)

So it is not blewing up but not working either. The command seem to trigger but the current is probably not enough. No spark on the secondary...

For measuring the inductance, I only have multimeter and capacitor, and I am not really hype on having to mess to much with the main...

[Armadillo]

My Sim Model attached;

It shows the problem;

The Peak voltage can hit as high as 763V due to some back emf by just a 1uh primary side transformer. Either you design a snubber or select a SCR high enough for it. and be weary of the transformer inductance.
The Peak current with a limit of 100 ohm Resistor and the 1uH transformer is 3.26A
The Peak Spike Current with ZERO OHM limit is 1830A !!!
The trigger happen at 0.53 seconds with the value as shown.

[gg64]

Thanks for the working simulation . I might give another try after work, I need to go to work.

What did you use for simulation ? I tried several software but this one looks familiar but I can't remember the name, LTSpice ? (I already got familiar with several, orcad, qucs, altium, and so on).

[Armadillo]

Yes;

There is a mistake in previous model. The circuit fires One Shot and didn't recharge because the triac remains On, I noticed.

Attached, revised with the missing resistor at the gate and the revised waves. The resistor is necessary.

Regards; 

[Armadillo]

Hi;

If you are still having problems, probably its time to check your HV transformer for open or shorts from last known unlimited pulse, for example.
If you have a DSO, you can ring test it or if you have a ring tester better still.

 

[fzabkar]

ISTM that the value of L1 can be chosen by assuming that all of the energy of the capacitor is dumped into the inductor/transformer.

So 0.5 x C x V squared = 0.5 x L x I squared

L = C x V^2 / I^2 = C x (V/I) squared

[Armadillo]

You are right, actually.

But that will mean that the transformer has got to be changed to suit the cal.

It would be easier to measure the self inductance of the coil to adjust the circuit to suit it since physical transformer is unknown.

Also how much juice will not kill the cattle.....  no ideal.     

ISTM that the value of L1 can be chosen by assuming that all of the energy of the capacitor is dumped into the inductor/transformer.

So 0.5 x C x V squared = 0.5 x L x I squared

L = C x V^2 / I^2 = C x (V/I) squared

[Armadillo]

Oops ! in the first post;

[for cattle, advertised 3J 10kV]

LOL.....

[Edit; the time constant 0.53 sec is still not long enough. The charge must be at least minimum >500V]

I think is advertisement BLOW BIG!.

[gg64]

I wish I had one Digital storage oscilloscope as well.

A 2Msamples/sec dual channels would be enough for most of what I need to do.

(Maybe checking fuel injection as well with it ? )

Anyway I got a schematic working in LTspice XVII.

I did a copy of yours to check if I can get it working.

Did you used stock components or some found on the internet for the diac and triac in LTspice ?

Could the 100ohm resistor need to be bigger than a 1/4watt ?

I already tried with a 220ohm 1/4Watt it blew up the triac. I think it made an arc therefore shorting the resistor, and thus shorting the capacitor completly again (I saw a big flash.... )

[gg64]

Just so you know, I got video footage with me playing with the new tranformer before I fried the board.

I used to do spark about 8mm long (about a third of inch). And no it won't kill you because the pulse is too short to be an issue (like the french website would say, about 120µs).

http://www.clotures-electriques.pyrat.net/Picfor-RB-80-Secteur-230-Volts-184.html

[Armadillo]

Great your model works....

Better be sure than be sorry;

Its how much in how short the time.

[Edit: Oops yah the time is really short. Always have been trained to be coward when it comes to high voltage...... hahahhaha the french is right.]

[gg64]

Well I do agree that you should use the moto: better safe than sorry. But I actually like the modification you did. I have ideas how to modify it so it fits my 'requirement'.  

At least thank to you guys I am moving foward. Not doing this anymore.

[Armadillo]

Could the 100ohm resistor need to be bigger than a 1/4watt ?

I already tried with a 220ohm 1/4Watt it blew up the triac. I think it made an arc therefore shorting the resistor, and thus shorting the capacitor completly again (I saw a big flash.... )

Yoo, you need to be WARY of your existing transformer inductance. It is the inductive reactance + the limit resistor that limits the current and affect the back emf.
I suspect the transformer is shorted some turns in the primary. The inductive reactance is way too low already. It would be better to eliminate the limit resistor come to think of it, otherwise it would be really nonsense too big in size. I think, the primary inductance should be much much more than 1uH.

Why don't you try take away the limit resistor and insert increase the size of the primary transformer inductance to from 10uH ~ 1mH in the primary thus eliminate the limit resistor and also find out what is reasonable inductance to be used to check the HV transformer whether it is defective. I hope you know what I am saying. I am not at my station now, so can't SIM it.

[Armadillo]

With the above learnt knowledge, I am confident to say;

Your HV transformer primary is shorted resulting in a lack of inductive reactance causing the current to overshoot the limit of the triac.

Test Method: LCR measure the primary self inductance of the Primary coil of the transformer, it should NOT be less than 200uH.
Preferred Method: Compare a know good transformer with the DUT.

Attached the components value selection that will produce the 10KV pulse and close to 3 joules of energy without the limiting resistor.
With this selection the original BT139-600E "must" be used.

 

[Armadillo]

Can't find the video?

Just so you know, I got video footage with me playing with the new tranformer before I fried the board.

I used to do spark about 8mm long (about a third of inch). And no it won't kill you because the pulse is too short to be an issue (like the french website would say, about 120µs).

http://www.clotures-electriques.pyrat.net/Picfor-RB-80-Secteur-230-Volts-184.html

[gg64]

I went to sleep. So I found the files mp4 but those are corrupted , could not get it fixed with vlc for some reason...

Anyway, for the inductance of the tranformer, the new transformer is a custom one, the black wire goes around just once, the other one should be about 10 times as much. You can see it WIN_20180903_00_05_26_Pro.jpg.

Anyway I still got the former one, it is written (he is busted he was 20kohm when I took apart and it goes infinite from time to time):

RB80
RG 03/30

on it but I am not sure of the meaning of it (3 at primary and 30 secondary ?). I am pretty sure the RB80 refer to the model number though.

So to guess what kind of inductance I should get with my current setup according to this:
https://www.allaboutcircuits.com/tools/coil-inductance-calculator/

So I need modify the area made by my coils to get a lot more of inductance.

I have idea, I goes some shapable plastic around I could make something with my hot air station.

To isolate core from the wires and make something which can hold on his own.

[Armadillo]

Take the former busted HV transformer, note every details example the size of the air-gap etc... open it up and unwind the coil, counting the numbers of turns, right hand turn or left hand turns must be noted, note the separators sheet all such details very important to follow every details.

Throw the old winding  away and rewind it back exactly the same and you have a new HV coil. The pain will be the secondary turns because I imagine thin wires and many turns.
I have a coil winders, bobbins and jig holders for such repair.  You should also build one.

The one holding on your hand is a ferrite core, I don't think is the original, the way you wound it, you must need alot of spares traics ready... 

If you play the sparks for too long time, the wire gets very hot and melt and will short circuit.

This one from ebay definitely is better. It can spark 15mm wide gap. And its cheap.

https://www.ebay.com.sg/itm/Ultra-high-1000-kV-Pulsed-High-Voltage-Inverter-Arc-Generator-Ignition-Coil-NEW/292667655617?hash=item44245a05c1:m:mFlHQRgYmcyvcfb5gCHDBow:rk:27:pf:0

[gg64]

I am starting to unwind the previous coil, this:

https://www.ebay.com.sg/itm/Ultra-high-1000-kV-Pulsed-High-Voltage-Inverter-Arc-Generator-Ignition-Coil-NEW/292667655617?hash=item44245a05c1:m:mFlHQRgYmcyvcfb5gCHDBow:rk:27:pf:0

is a 3v input.... and 1Mv output ? And if they meant 1kv, I don't think that would do it either.

I don't plan to do a tesla coil anytime soon (1Mv)....

I will see how the unwinding goes and decide from there.

[Armadillo]

To be honest, I am a bit confused if you will be using it for cattle range fencing or something else?

Anyway;

Waiting to celebrate your final finished product put to good use... 

[gg64]

It is for cattle. I am not north korea or something.

Anyway resin/epoxy is holding the tranformer way to hard for me to not destroy it in the process. I am making a plastic holder for the winding but I am not 100% confident about it we will see how it goes.

[fzabkar]

I have a partially assembled electric fence controller kit (Silicon Chip magazine, April 1999). The transformer was hand wound using these components:

http://www.farnell.com/datasheets/1469874.pdf
https://au.element14.com/search?st=etd29

http://archive.siliconchip.com.au/cms/A_102476/article.html

Alternatively the ETD29 core assembly could be used but it will not fit readily into the PC board holes. It is a little larger but the cores are similar and so you can expect similar results. These cores, former and clips are available from Farnell Electronic Components Pty Ltd. ... Their catalog numbers are 178-505 for the cores (2 required per transformer assembly), 178-506 for the bobbin (1 per transformer assembly) and 178-507 for the clips (two required per transformer assembly).

Secondary -- 250 turns 0.25mm dia enamelled copper wire (ENCU) wound in 10 layers with insulation tape between each layer and a 2mm gap between each end of the windings and the former cheeks.

Primary -- 7 turns 0.4mm ENCU wound over secondary

The dump capacitor's rating was 6.8uF or 7uF 250VAC. It was charged to 340VDC. The time between pulses was 1.5s.

The kit was designed to be used with fence runs up to 5km. The relevant Australian standard was AS/NZS 3129.

Suitable Triacs were BT10-600, BT12-600, or BT138X600G.

[Armadillo]

This post is getting more interesting;

Allow me to show you my HV CORES in the photo;

The left side is how the core is assembled and potted from bottom.

On the right you can see the serrated bobbin. Its easy to imagine the windings as a constant increase of voltage along the wires, hence as the length increases, the voltage gets larger hence it must be separated by the serrated separator bobbin to avoid crossover flashing. As you can see the arrowed bobbin. Winding is an Art especiall when you deal with 0.02mm wire size!... you know what I am saying when you have handled it, really!

And then the potting, another piece of Art. 

Glad you guys are interested in it.   

[tautech]

Old medium size NZ made Gallagher unit at my bench, ~5 J output.

40uF main pulse grade caps switched via BTW69-1000 into 0.1 ohm primary with dual MOV's for back EMF protection and ~50t air cored inductor in series with the transformer primary. 10 ohm secondary.

Output is never direct to a fence, there is some pulse management circuitry (passives) on the secondary side.

From the test sheet:
4/8/89
Output test load 500 ohm
5.33 KV 5.18 J
Interval 1.27s Width 4.76 ms
No load peak 6.33 KV

Pics available if required.

[Armadillo]

@ tautech; Thanks for sharing,

It would be great to see the inner workings of how Gallagher implement their circuit and those transformers.

Some pics would be appreciated. thanks.

Old medium size NZ made Gallagher unit at my bench, ~5 J output.

40uF main pulse grade caps switched via BTW69-1000 into 0.1 ohm primary with dual MOV's for back EMF protection and ~50t air cored inductor in series with the transformer primary. 10 ohm secondary.

Output is never direct to a fence, there is some pulse management circuitry (passives) on the secondary side.

From the test sheet:
4/8/89
Output test load 500 ohm
5.33 KV 5.18 J
Interval 1.27s Width 4.76 ms
No load peak 6.33 KV

Pics available if required.

[tautech]

Pics attached in no particular order.
Main and output PCB's should be able to be aligned with the connectors as a origin.
Main switch to dump the caps into the primary is on the end of the PCB by 2 big MOV's.
Most connectors are spade terminals of some sort or another to allow for easy module replacement.
Anything else you can't work out, just ask.

[Armadillo]

Wow, solid. It makes "Picfor RB 80" pale in comparison.   
Yeap!, this is the real stuff.
Aside from HV generation, there are more circuit boards there? Seems like doing some protection or something functions?
Interesting.. thanks.

[Armadillo]

I think we should Sim this circuit, what say you gg64?

Can't believe I am reading cattle now..... hahahahahaha..

The schematic don't quite appear to be the same, though.

[gg64]

Hey, thanks for sharing

Well I fried some TRIAC again. This is pissing me off really bad, I spend so much hours trying to figure out what is wrong with mine.

(basicaly I put 2 1000ohm resistor in parallel (making it 500ohm measured) (in serie onto the 1 of the triac) and it still managed to kill (tick sound) the bt139-600e w/ a single turn on the primary and 10 turn on the secondary, I don't get it, it worked really nice at first -> before not connecting the load properly just once).

Anyway you gave me some ideas I might have a ignition coil lying around (might make/take a power supply and make a command for it).

On this: gallagher_villanypasztor_electrical_fence_sch.pdf

What is the name of the NE1 component ? (It is for watching if the fence working (properly))

Three others option:
- as Armadillo suggested, I might have been chasing my own tail the entire time, maybe I got counterfeit triac (I mixed some from ebay and some aliexpress). Ikr, I ain't proud.
Thus I need to buy from a thrustworthy seller BT139-800.

-or buy a big beafy: BTW69-1000 (that would finally explain why the silk footprint on the board was so big in the first place)

-or provide it to someone to service it (I spend way to much time on it to give just now, even if I am pissed. Money wise it is still ok)

3 Joules is already enough for our use (we got 3 25µF in working condition), 5 Joules that would do some serious muscle contraction.

I am going to finish that problem one day. 

[tautech]

3 Joules is already enough for our use (we got 3 25µF in working condition), 5 Joules that would do some serious muscle contraction.

These single digit Joule fence units are just toys by comparison to what we use today, my current unit is 40 J !
You really don't want to tangle with it.



I strongly suspect your problems are related to transformer and switching and a scope will be required to sort it out.

[Armadillo]

@gg64

You need to use the LCR meter to check your primary inductance, its needs to be > 200uH . Below 200uH, your Triac will be fried. In fact, we should category that anywhere 300uH and below should be avoided.
Please read Reply #24. The nominal design Primary inductance should be 500uH to eliminate *ANY series Resistor.
 
1 turn, is far far from 200uH. The air-gap of the Ferrite Core plays a vital role in the inductance number of turns, you have got to tape the 2 half if they are of split half.

The NE1 is a neon light bulb.

[tautech]

Aside from HV generation, there are more circuit boards there? Seems like doing some protection or something functions?

2. One, vertical attached to the MB is for an LCD display and the other seems to get a magnetic coupling pickup from the transformer and give some input to the display board but traces go into comparators so maybe load sensing. 

[Armadillo]

That brings up pertinent questions and thoughts;

Those 3 joules or 40 joules, despite what the French says non fatal at 120us, the question is what does the joules have any impact on human being?. I mean, there is no published science on this. The fatality chart start off at 10ms. So this is quite a grey area now AFAIK. The next question for the French is "WHAT IF" instead of a short pulse, the rise time exceeded due to some circuit problem, resulting in 10ms HV within AC 4 category?

So, that brings us to the inductive pickup coil;
IMO;

1. Protection in ensuring short pulse.
2. Detection of malfunction in the field e.g. shorted by iron bar in the field.
3. Detection and joules limitation when the cattle gets shorted [no idea of joules control here]

So between the old circuit and the new design with safety feature, I'll go for the safety one.     

Aside from HV generation, there are more circuit boards there? Seems like doing some protection or something functions?

2. One, vertical attached to the MB is for an LCD display and the other seems to get a magnetic coupling pickup from the transformer and give some input to the display board but traces go into comparators so maybe load sensing. 

[tautech]

Electric fence technology is quite advanced these days where there is distinction between battery and mains powered units maximum specifications.
IIRC it goes something like 10KV for battery and ~6KV for mains with pulse discharge times less that 5ms.
Mains units are normally restricted to ~1Hz while battery units sometimes have a fast mode where the pulse rate is doubled to ~2Hz for training stock to.....keep away !

Modern mains units sense the load and while they might be rated at 40J+, not all that energy is available into a single pulse in normal operation. When grass growth touches the fence wire the load increases and extra energy is released to kill the growth pulling down fence voltage.

Also if we consider the normal basic electric fence design, rectified mains charges the pulse caps for them to be discharged into the transformer primary @ ~1Hz rate.
The charge rate of the pulse caps is normally managed via a series resistor and pulse repetition rate traditionally by a simple RC network.
In their simplest form there's not much that can go wrong and they always seem to fail safe.

For the many I have seen, used, installed and a few repaired never have I seen one speed up to any noticeable degree or fail to anything but a safe mode.....normally dead......no output.

[Armadillo]

Oh yeah, the field environmental leakage conditions like the moist grass etc.... there I miss.
Sounds logically correct. thanks.

[gg64]

Does DUT stand for Device Under Test ?

I don't have an LCR meter.

Basically I just took apart the previous tranformer, I found some burnt stuff across 4 layers including the primary coil (as I expected), they are 9 layers (8 secondary and 1 primary).

I really got astonished to see the coiling inside to be honest. This did not looked that crowded from the outside look.

Amardillo was right for the inductance. I am going to provide the stats from it when I am done counting. I recorded it the tear down on video (6GB) and on another camera.

Anyway, my next logical stop would be to get supplies to make a new one, any advice where to get it ?

In case you wonder the previous one I used (and it worked somehow) was this:
https://www.ebay.com/itm/1set-PC40-E65-EE65-6-6pins-Ferrite-Cores-bobbin-transformer-ferrite-core/153032005558?hash=item23a16b17b6:g:8T0AAOSwo9VbA9~g:rk:1:pf:0

But I plan to use the previous one, the one I just tear down.

Anyway specific place to get coils and insulation tape for the transformer ?

[tautech]

While this site isn't the bible on transformer construction it give excellent advice on the basics:
https://ludens.cl/Electron/trafos/trafos.html

[Armadillo]

@gg64

The PC40 Ferrite core of the size E65, the flux density saturates at 17A ~ 20A only.
It simply means, the size E65 cannot be used [too small] for your purpose for 3 joules.

With that said, I wonder what the Marketing Gimmick is? Is the 3 joules on the Charging/Discharging capacitor Or the output is "3 joules" or "40 joules"? Have anyone verified or tested it before?

Have you seen the rat's electrocution coil before? Is just a long Ferrite rod, easy to mount on the lathe and do the windings there. The size of the Ferrite Rod determines the Power Transfer.
Hand winding the primary is OK [40 ~ 50 turns]. Hand winding the secondary is going to be a bit difficult [around ~ 1200 turns] [for 3C94 Core].

In those previous discussion before, I was having the vision that the transformer you used is the French Guy Pyrat-RB-80 type.

Can you photo the Core used for the dismantled coil? it will be interesting to learn how they did it. thanks. Would China sell the replacement coil cheaply? Any photos?

[Armadillo]

DUT is device under test, Yes, correct.

Can you post the video to Youtube. But 6g is a bit too big I think. Maybe you need do some editing.
I have no experience posting in youtube and I think they must have some forms of limit in sizes.
I will put a thumb's Up for you whether it comes with audio or not. If you introduce in French, I'll give you 2 thumb's up.

Please post up, don't just learn yourself. 

thanks.

[gg64]

Here is the core measurement video:

https://youtu.be/vdtzyWQPcYI

At the end I meant thirteen not thirty.

I tried to get the tear down video ASAP, I need to clean the audio background.

[Armadillo]

 
Great Video, Good Start;

Can you next time put the microphone closure, too soft, can't hear clearly.

Looks like a E I laminated core, Yes?

You miss the I block ?

[gg64]

It is just a E block, I am working on the new video (tear down).

[Armadillo]

Hi,

Ok, I'll eagerly wait for the tear down VIDEO.

By the Way;

The magnetic path must be closed, either EI core or EE core or simple rod ferrite.
See attached;

Then better use the tear down core. More reliable, surely.   

[gg64]

Ok, I am rendering the video.

I spend 3 hours figuring out why I got a crispy sound/pop that was mainly because my usb sound card is getting satured.

I clean the sound with a sound gate/low pass and high pass.

It is only in French for now. I might do a voice over in a bit.

EDIT: I am currently uploading it would available here:

https://youtu.be/fNSP_znZyLs

[Armadillo]

Hey; thanks for the video, seems to be freezed for very long time there. 

A look at the size of the wire gauge, it's NOT going to do 3 joules. Is it from the original 3 joules pryat unit?
But nevertheless even 1 joules can play a good thick spark.
If the magnetic flux path is not closed with a correct energy air gap, its going to be very difficult. 

[gg64]

hey, so I scanned one full layer to count the number of turns.

Per layer 7*9 + 3 turns for this layer, times 8 layers for the secondary (this layer is labeled 3 of out 8 BTW).

528 turns on the secondary.

edit is coming for the primary.

EDIT: 5*9+5 times 1 for the primary, 50 turns on the primary.

I need to check if the turns are somewhat consistant on the secondary former layers.

EDIT2: On the secondary the wire diameter is about 0.27mm (I took about 5wires side by side). (With one I can measure 0.30mm down to 0.22mm)

It is probably 0.25mm (tarnish included).

Now for the primary, in the next edit.

EDIT3: the wire diameter for the primary is 0.68mm (0.62mm when I press hard on the caliper with 4 wires)

For the primary, it is probably 0.3mm for the wire diameter.
And 0.68mm for the secondary (it seems that my digital caliper is 0.04mm loose when I am pressing hard)

[gg64]

Hey, so using this:
https://www.allaboutcircuits.com/tools/coil-inductance-calculator/

with a Relative Permeability of 640 (assume ferroxcube)

I am getting 0.113H for the primary (50; 29.3 'equivalent' circle diameter; 0.68; 640)

I am getting 15.2H for the secondary (528;29.3 'equivalent' circle diameter; 0.3; 640)

But those numbers are through the roof to my opinion, but since I am novice.

And the simulation argh_too_long.PNG is showing a dangerous behavior (too long for a pulse) to my opinion.

I would like to know what kind of value I could assume for a laminated core.

But if I am using 1 for the Relative Permeability, i am getting:

176µH, 23.8mH

And the simulation look like looks_ok.PNG


Look likes I need to do a lot of reading.

[Armadillo]

Just too many cores and I am getting a bit of confused over the period of this subject really. 

I thought you are unwinding and rebuilding the transformer? Normally just follow exactly.
If you want to deviate, you will have to impulse test the core to verify the knee and saturation point, to extract the core properties of the "X" core you will use.
So, the best is don't deviate.

Alternatively if you plan to get new cores from ferrocube for example, they have many types of ferrite cores, size and forms and published technical datas available for you to use in your calculation.
But, Academic and Practical are 2 different stories altogether, but ultimately you will need test instruments to verify your build. This is a deep science you are delving into. Only the courageous dare venture.
But I think it will take times as you build up your equipment to verify and experiment with this deep but certainly rewarding subject.

I am not verse with the magnetic software and can't be of much help in there about.
But your sim looks wonderful though.   

[gg64]

Hey, I will rewind the former stock pulse tranformer.

I might tried to do another spare with my ee65 core I bought on ebay.

Here is the math (for me, to remember the day I need it again):


Measurement of the former stock pulse tranformer:
35.1 thickness of the E + primary layer
34.27 minimal thickness of the E + primary layer

about 47.19mm maximum

7*9+3=66 turns per layer

Since the
(35.1*4)*66=9266.4mm for the first layer

Low estimate:
If we approximate for a layer with 35mm to 42mm for the last
( 42(42+1)/2 - 34(34+1)/2 ) * 4(width of a square) * 66(turns per layer)

0.5*(42*43-34*35) * 4 * 66
2* (1806 - 1190) * 66 = 78540mm

Medium estimate:
(35+37+39+41 +43+45+47+49) * 4 * 66

(152+184) * 4 * 66 = 88704mm


I still need to work out the taping material (I lost the info...   ).

EDIT:
22AWG 0.63mm for the primary
and
29AWG 0.28mm for the secondary
should do.

EDIT2:
For the primary:

50turns *35(width) * 4(square) = 7000mm

Meaning those would do the job just fine:
https://fr.rs-online.com/web/p/products/7790700/
https://fr.rs-online.com/web/p/products/7790716/