Silicon chip - everyday practical electronics capacitor reforming/leakage tester

[viorel]

I have a problem with my newly finished capacitor tester/reformer built from an Altronics kit bought a while ago. This was published in Silicon Chip
 and Everyday practical electronics in August 2012.

So everything is normal until I set it up to measure leakage over 100V. It simply does not generate more than 100-120V. I have tried a few power supplies, no difference. Less capable power supplies do stop earlier (maybe 90V or so), more capable supplies go up to 120V but no matter what, I can not get it to generate more than that.

I have checked everything with the exception of the mosfet (IRF540N), it is from Altronics, new, I presume good if it works up to 100V. All soldering is nice and neat, everything else works fine, displays correctly everything, it is adjusted as per the instructions procedure, voltages in the schematic are what they should be, can't think of anything else.

Anyone else built the kit and have some thoughts?

I can post a schematic if it is allowed, just not sure about the copyright problem, I can mail it to anyone who would like to take a look, maybe they can help.

It is basically a part that generates the high voltage (up to 630V) from a supply of 12V using a MC34063 then through a push pull BC 327-337 pair into a mosfet driving an autotransformer. The second part has to do with the digital measuring and displaying of voltage, time and leakage current on a dot matrix display.

The voltage selection is done by daisy chaining appropriate resistor values that divide the output voltage (after the autotransformer) and feed it into the MC reference pin. Hope this makes some sense.

Nothing fancy really, yet it doesn't work as intended.

From the brief description above, where should I look?

[timeandfrequency]

Hello viorel,

Can you provide the schematic of your tester ?

[coromonadalix]

https://archive.org/details/EPEMagazine/EPEnonlinear.ir2012-08/page/n1/mode/2up

pdf

https://dn790004.ca.archive.org/0/items/EPEMagazine/EPEnonlinear.ir2012-08.pdf

as for the pic fw,  need to go at EPE to check availability

[floobydust]

Don't wreck it for everyone.
SC August 2010 (part 2 sept.) to 630VDC uses MC34063/PIC16F88. Also one in Dec. 2009. Seems to have been reprinted in EPE 2012.

OP check the feedback resistors if they are the right value, that the kit was assembled correctly.
Next I would suspect the handmade transformer and of course no details given about the core p/n, surely it has an airgap... make sure it looks physically OK not crooked.

One errata in 2006 but SC is pretty good about them

[timeandfrequency]

Hello viorel,

Probe the waveforms on Q3 :
- Vgs
- Vds

And show also the waveform on TP3 (GND = TPG)

Finally found the part 2 of the article
I hadn't considered that you had to wind the transformer yourself.
Can you show how you connected the transformer wires to the PCB ?
Did you follow the transformer manufacturing guidelines?  All five layers should be covered with insulation tape.
Did you include the plastic washer ?
As  floobydust mentionned, there is a possibility that your transformer has an insulation / voltage breakdown problem between turns or between primary and secondary.

[viorel]

Hi guys, thank you.

Yes, the schematics are the correct ones.

All resistors rechecked when I found the problem (I also measured them when I was soldering). Will check again.

The transformer is actually an autotransformer (primary end connects to sec start), and it is built as per instructions in the article, each layer insulated with Kapton tape, airgap set with a few layers of kapton to spec. Wires soldered in the correct place, isolation scraped, they are nicely tinned and soldered correctly (primary to primary pads etc). Yes, plastic washers all in place. I will try to take some pictures tonight when I get home. The core is the correct part number specified in the kit instructions, bought from the supplier they recommend. I will look up the specs and post here. The wires are again bought from the same supplier as per kit instructions, enamelled copper wire of the correct diameter and I wound it very nicely by hand, no messy crossed wires going all over the place. I checked it before I soldered together the primary and secondary wires and there is no internal short, good continuity across both primary and secondary.

I will try to scope the mosfet as per suggestion above and post the piccies.

ONe thing I may not have made clear enough is that the output voltage goes up if I use a power supply with a higher voltage (I have used different supplies in the specified range). So if I supply exactly 12V I get a max of about 100V, if I supply close to 15V, I get about 120V max.

One thing I can not figure out for sure (and the article doesn't say anything about), what is the point of the 4.7 Zenner (ZD2) to ground in the feedback loop? Drain too much voltage in case of spikes, etc?

[floobydust]

Does the power supply regulate at all? Or it it just having problems putting out high voltage?
If you selected say 25V is it a steady 25VDC out regardless of 12V input voltage moving around a little?

The zener I think protects IC1 in the case the rotary switch is break-before-make which would nail the feedback input pin.

[timeandfrequency]

I checked it before I soldered together the primary and secondary wires and there is no internal short, good continuity across both primary and secondary.

I was rather talking about a partial insulation breakdown. You can't measure that with a multimeter because the voltage applied(*), usually 1 to 9 VDC, depending on meter, is just too low.
The low impedance path only appears if a significant voltage is applied (about 50 to 200 VDC usually) and vanishes when a vanilla multimeter is used. To measure a partial insulation breakdown, you need a megohmmeter.

In the BOM, I could not find any information about the core material used by the author.
I don't really know if it's absolutely decisive (**), but there's a small chance that using a wrong magnetic material (3H1, 43) with very high permeance might drive the core into saturation when the output voltage raises.



(*) for purists : it is actually the compliance voltage of the current source activated while in Ohms mode
(**) ferrite core materials is a realm by itself (here and there) and I'm not totally fluent with it

[viorel]

 LF1060_datasheetMain_95160.pdf (115.66 kB - downloaded 23 times.) LF1060_datasheetMain_95160.pdf (115.66 kB - downloaded 48 times.) LF1060_datasheetMain_95160.pdf (115.66 kB - downloaded 23 times.)floobydust, the power source does not show significant sag over the entire range. I am pretty sure it can handle much more than the 1.5A recommended in the instructions. I resorted in the end to a huge transformer out of a 100W per channel 5.1 channel receiver, just rectified and with a piddly capacitor on the output just to test things and the same result. The voltage sags say .3V or something like that when I crank the tester up to 630V. I have a regulated power supply that can deliver easily 5A but it shouldn't need that.


timeandfrequency, I see your point. Here is a link to the core used:

https://www.jaycar.com.au/transformer-pot-core-pair/p/LF1060

And here's the datasheet:

https://media.jaycar.com.au/product/images/LF1060_datasheetMain_95160.pdf?_gl=1*gtomaf*_gcl_au*MzQ0Mzc4MDQ1LjE3NDQ4MDEzNTY.

Sorry, can't manage to upload the .pdf but the link above should work.

Hmmm. Megohmmeter, eh? I have an old style vacuum tube capacitor checker with leakage function that goes into the realm of hundreds of volts (I think the highest scale is 650V) and can check leakage on mica capacitors (what is that, a few microamps?), would that catch a leak at high voltage in the insulation? You are tempting me. I am thinking if I disconnect the primary from the secondary and apply 650V between them a small leak would show immediately (that thing is crazy sensitive). What I am not so sure about is whether the primary of the autotransformer would take 650V bu if the insulation on the secondary can withstand 630V, I wonder why would the primary insulation not? It's the same enamel as far as I know. Temptation, thou art a heartless bitch.

[timeandfrequency]

Interesting core datasheet : many informations provided but I do not see the 'Al value' (permeance, whose usual unit is nH/t²) (*) nor the material mix. Actually, the two informations I needed are just missing(**)
Could not find any information either on the 'Philips FX 2240' core that was mentionned by Jaycar as an equivalent. Bummer.

So, I won't dive into the theorical flux density calculation (max 500 mT) because of the air gap (plastic washer) that significantely alters the part caracteristics.


Just putting 650 VDC onto the secondary will not lead to a valuable measurement because the megohmmeter outputs a DC voltage. So you will just short the voltage supply by the DC resistance of the secondary wire.
 It would need a pulsed DC high voltage source to perform the mesurement you suggest.

To check the insulation of your windings, you can try to put about 250 volts between any transformer connexion and the ferrite core wrapped in aluminium foil. If there's an isolation problem from the winding to the ferrite core, it will show up.

The isolation of an enameled wire against a conductive surface can withstand about 100 to 200 VDC.

If we really narrow down to the transformer as the culprit, in-deep mesurements of it's behaviour is not trivial at all especially if it comes to leakage and shorted turns. Maybe just rewinding it with much care and new wire could lead to a positive issue.


I would at first measure some waveforms around the switcher to see if there's not something else that creates the issue.
Core saturation can be guessed by measuring the instantaneous current through the winding.

Indeed, as we don't have the core Al value, inductance can not be calculated, so no relationship with the switcher frequency can be made.
Mesuring the core inductance (LCR meter) would not lead to a trustable result  either : a potentially damaged winding + low measurement voltage of the LCR meter cannot provide reliable values for high voltage transformers.



(*) From a top tier manufacturer, one can see the different Al values (here in mH/1000 turns) of the 'PC26/16' pot core vs 6 different magnetic materials mixes (R, P, F, ...).
(**) Very few manufacturers provide comprehensive data of the magnetic parts they sell.

[viorel]

I didn't suggest putting 650 in the secondary. I suggested disconnecting the primary from the secondary and applying 650V across from the primary to the secondary. Not having any electrical connection between the two leaves only leakage as a possible path and that should show. A similar test can be carried out on the core itself. The only problem it seems would be what would the result mean?

Not sure what to say about the material. Apparently, the SC/EPE staff made a reformer with the same core and theirs works, so I take it there is nothing inherently wrong with the core itself.

Again, I am pretty sure there is nothing wrong with the transformer, it is wound very nicely (or I don't think I can do it any better, if you will).

The only thing I can question is what would happen if I squashed the airgap too much when I tightened the screw that bolts the core to the PCB?

[timeandfrequency]

I didn't suggest putting 650 in the secondary. I suggested disconnecting the primary from the secondary and applying 650V across from the primary to the secondary. Not having any electrical connection between the two leaves only leakage as a possible path and that should show.

OK, if you modify the part so that it becomes an isolated transformer, the test you suggest makes sense.

A similar test can be carried out on the core itself. The only problem it seems would be what would the result mean?

It means that your part has leakage, is dangerous, and should be rewinded.

Again, I am pretty sure there is nothing wrong with the transformer, it is wound very nicely (or I don't think I can do it any better, if you will).

You may be right. But something is not OK on your board. So we are still waiting for the waveforms...

The only thing I can question is what would happen if I squashed the airgap too much when I tightened the screw that bolts the core to the PCB?

Flux density and Al vs airgap dimension needs very comprehensive data from the manufacturer to provide a serious answer to your question.
Put a 0.5 to 1 mm plastic washer to increase the airgap : it will significantely decrease the likeness of a core saturation.
If saturation is the culprit, it should solve the malfunction.

[viorel]

Yeah, sorry, we are on easter holidays here, nothing much will happen for a few days.

[Martin72]

Hi,
Does anyone have any idea how I can get hold of the hex file for the Pic?

[floobydust]

I find it on the SC website here {seach leakage} and Jim Rowe was nice enough to also post .asm source code.
"Capacitor Leakage Meter Firmware (0411209A).zip"

[Martin72]

Thanks!
I ordered the circuit board earlier in the shop.
For a little over €10, I'm not going to sit down and make my own.