EEVblog Electronics Community Forum
Electronics => Repair => Topic started by: dorkshoei on December 12, 2022, 07:19:01 pm
-
I have 4 that are broken. I've read various threads on gluing them back together (https://www.eevblog.com/forum/repair/how-to-fix-a-broken-ferrite-core/) but alas they came to me as pictured so I lack any remnants to glue.
I asked the manufacturer of the board for the component ID but they consider it "top secret" info 8)
I'm not sure if based on the "893 9022" marking (that most but not all have) plus perhaps some induction measurements (I have an LC103) I can figure out a suitable replacement?
Or maybe I can fudge some kind of repair. My testing of basic functionality of the 4 that are broken didn't show any difference from the ones that are complete but I did only limited 10,000' testing.
Thanks!
-
Hi,
I asked the manufacturer of the board for the component ID but they consider it "top secret" info
What kind of board, in which unit is it assembled, model ? Schematics avaible?
Have you ever desolder one to look on it´s backside, maybe the name of the core is there...
Broken cores shouldn´t be glued, their performance will never be the same as before.
To find an alternative core/new transformer:
What´s the switching frequency, the turns ratio ?
Both you can determine by measure with scope.
Primary winding is the coil which are in the switching circuit, measure its amplitude, then measure the amplitudes of the secondary windings for determining the ratio.
Using more channels of the scope let you see the phase alignment between primary/secondary windings.
Finally the size is important.
There are lots of smd transformers on the market, knowing the parameters of the original one by measuring helps to find a new one.
-
No schematics. I'll desolder one and look at the rear.
Other threads disagree on no gluing with the proviso that you achieve a tight fit and the understanding there will be a difference but likely not massively significant. Regardless, I don't have any pieces to glue.
-
EER shape, TDK, NiCera
legnth, HT, with in mm for size eg EER 9.5, EER14
core matériel and gap needed.
Impossible to fix or glue, will create airgap
j
-
EER shape, TDK, NiCera
legnth, HT, with in mm for size eg EER 9.5, EER14
core matériel and gap needed.
I have no clue what any of the above means. Obviously I know what mm are but what part I am measuring I do not know.
I have no idea how to measure gap or what the core material is.
Please assume I know nothing about ferrite cores.
Here are some pics of it removed. No obvious manufacturer markings. The red dot is mine so I can remember orientation on the pcb.
-
Rebonjour
you asked what type of ferrite core.
EER = industry standard core shape, like EI, EE, PQ, pot cores etc.
TDK, NiCera = Japanese ferrites core brand
mm is core dimension of center, legs, etc. Largest dim is across the largest dimension, which identified the core size eg EER 9.5, EER 11.5
Suggest that you scrap the device, you cannot repair or replace the transformers
Very Sorry
Jon
-
You've got the wound bobbin, but need a new ferrite core. You won't tell us a make/model of what you are repairing?
Do you know what IC is driving it? i.e. for power supply or data isolation etc.
Coilcraft (https://www.coilcraft.com/en-us/products/transformers/) has many offerings, if you know the IC/purpose. Critical to know is if the old ferrite has an air gap in the center pole.
If it was a really important vintage repair, I'd get another new transformer and take it apart, a heat gun or heat up to around 80-100°C and the glue comes apart, swap in the old bobbin.
-
You've got the wound bobbin, but need a new ferrite core. You won't tell us a make/model of what you are repairing?
Correct. The coil seems fine. I just need a new ferrite.
Sorry, i thought I'd answered this. It's an old Noritake VFD. It still works even with the broken pieces but I'd like to fix it properly.
Do you know what IC is driving it? i.e. for power supply or data isolation etc.
I've not poked around (yet) to determine what is connected to it but in proximity there is an MB3800 (IC6) and a K2782 (TR2)
Critical to know is if the old ferrite has an air gap in the center pole.
Got it. I'm gluing the bottom and sides back on so I can drop top on and see if there is an air gap. I'm trying to glue carefully for a tight fit.
-
Since this works maybe just check with a FLIR if any parts around the coils get hot. As these are certainly custom made, I would just leave it alone - it won't get any better than "working".
(Good thing is - if it works with several damaged cores, there seems to be a huge tolerance allowance in the part.)
-
Looks like this is probably a flyback converter, in which case there would be an intentional gap in the core already. This is probably why it "miraculously" works with the cracked core and gluing the pieces back together (if you had all of them) would work just fine as a tiny bit of extra gap usually won't hurt a flyback type of power supply.
-
Looks like this is probably a flyback converter
Would you expect to find some shimming paper in the gap?
-
Ferrite cores come in standard sizes. It looks like a TDK ER11/5 size core. (https://www.tdk-electronics.tdk.com/inf/80/db/fer/er_11_5.pdf) Check the dimensions with the datasheet. The air gap is on the center poles and small at around 0.1mm, Mouser only stocks one (https://www.mouser.com/c/passive-components/emi-filters-emi-suppression/ferrites/ferrite-cores-accessories/?core=ER%20Core&core%20size=11%20x%205&type=Gapped) N87 AL=160 but you can also try getting samples.
I'd put that on the bobbin and give it a go. The circuit is a flyback converter (not current mode) and not knowing the design details this is a ballpark guessing game at best.
You could contact Noritake about the part, although they have done a huge purge with VFD being overtaken by OLED and might think the part is specific to that one model display.
Looking for surplus Noritake displays using the same transformer, pricing is crazy high on eBay. Maybe check surplus houses.
-
Hi,
Is this top-secret or why you don´t show the full pcb and the model name ?
-
Looks like this is probably a flyback converter
Would you expect to find some shimming paper in the gap?
No, the gap would be on the center pole only, the ferrite is manufactured with the air gap on the pole piece already in there.
At this point I'd be tempted to just find whatever gapped ferrite core of the correct physical dimensions I could find and give it a try. Since it works with the ferrite all cracked apart without releasing magic smoke, pretty safe to say finding the exact same ferrite type and gap is not critical.
-
The circuit appears to be a flyback converter. MB3800 is a switching controller: http://www.icbase.com/File/PDF/FJS/FJS00590106.pdf (http://www.icbase.com/File/PDF/FJS/FJS00590106.pdf)
There appears to be a MOSFET driving the primary, which is obviously the primary from the connections and the multi-strand (litz) wire used. The secondary is fine wire, probably around 10-40 times the primary turns, for the 80V output. VFDs need modest voltages at low currents (a few mA?).
The other secondary (three middle pins on the primary side) must be filament. VFDs are actually very small vacuum tubes; a microscopic filament spans across the front, emitting electrons over the electrodes. Oh, and this winding connects to the midpoint of the HV secondary, they must need some negative cathode bias -- hence the two diodes on the HV side. That explains that.
The transformer is probably massively oversized for what it needs to do (ER11 can probably do 5W or more, in an optimized design; this is probably less than 1W?), which may mean extra cost, but at least means few turns are needed, saving wire length and core gap.
A flyback requires a gapped ferrite core to operate (or other core types), so there should be some gap. I don't think the core is so big they could get away with none. But it may indeed be small, like 0.1mm.
Shimming the core is an acceptable substitute, but mind that the air gap counts twice: once for the middle peg (normally only this is ground down), once for the outer limbs. If that means 0.05mm spacing, well, maybe you can find some plastic film or tape to get there? A micrometer will be helpful.
A note, use insulators only. Any conductor in the gap will be blasted by the field. So like... just in case you had the clever idea to use the feeler gauges that way or something. :P
Tim
-
Thinking more. If I measure the height of the center part of the coil and then measure the height of the circles on the upper/lower part of the ferrite relative to the flat part, I should be able to figure out exactly what the air gap it. Picture paints a thousand words but this description seems sufficient.
Hmmn.
With my calipers I measure:
thickness of center section of coil (channel that ferrite upper/lower contacts) = 3.02mm
thickness of ferrite wall: 1.28mm
height of ferrite wall and circle: 2.96mm
[both sides seem symmetrical/equal]
which gets me 3.02 - ((2.98 - 1.28) * 2) = 3.02 - 3.28
so that is obviously not correct.
obviously it's possible that the ferrite doesn't fit as tightly around the coil (as my assumption above) but I'm not seeing room for much of an airgap.
-
doubt if these had any airgap.
More likely this is a push pull, or half bridge inverter, with AC output rather than a flyback that required a gap.
See the many old app notes on VFD inverter design.
Jon
-
Ferrite cores come in standard sizes. It looks like a TDK ER11/5 size core. (https://www.tdk-electronics.tdk.com/inf/80/db/fer/er_11_5.pdf) Check the dimensions with the datasheet.
I don't think it is. I'll have to look at other datasheets (TDK and other).
I'm not entirely clear what "11-0.35" means? 10.65 - 11.0? I'm more used to seeing tolerance expressed as +-
The outer dimensions are 15.03mm x 6.7mm plus the inside of the ends is flat, the TDK has a slight inner bow (red arrow - to match the bobbin)
-
Look at the next standard size up in TDK ER shapes (https://www.tdk-electronics.tdk.com/en/529424/products/product-catalog/ferrites-and-accessories/er-etd-eq-cores-and-accessories-) which seems to be ER14.5 or 14.5mm across. The ERxx is a nominal size category rounded off but older datasheets do not give tolerances, so (ER11) 11-0.35mm must be 10.65mm; the exact core size is not critical, as long as the bobbin fits. The airgap is the only touchy dimension. See what the center pole diameter is as well.
Your DC-DC is good for a several watts, depending on the VFD size.
The exact air gap is very difficult to know and I see the ER14.5 (https://www.tdk-electronics.tdk.com/inf/80/db/fer/er_14_5_6.pdf) only offers one airgap of 0.21mm so that's your only option.
-
Also look for an "ER15"
-
doubt if these had any airgap.
More likely this is a push pull, or half bridge inverter, with AC output rather than a flyback that required a gap.
See the many old app notes on VFD inverter design.
Jon
You can see the one MOSFET, and single diodes on the HV ends?
Tim
-
ER14.5
Measurement wise ER 14.5/6 (https://www.tdk-electronics.tdk.com/inf/80/db/fer/er_14_5_6.pdf) looks spot on.
They show as available ungapped and gapped 0.21
The data sheet says exterior height is 5.9+-0.1 (so 2.95 for each half) and the interior height is 3.3+-0.2 which means the upper/lower wall thickness is 1.3mm.
I just measured:
- undamaged one on bobbin: 14.5mm outer length, 11.8mm inner length, 6.7mm outer width and 5.9mm outer height.
- damaged one removed from bobbin: 4.7mm core diameter, 3.3mm inner height, 1.28mm wall thickness and 2.95mm wall+core thickness
So precise match and no air gap.
5 different possible material values:
N49 1100 +30/–20% 800 B65513J0000R049
N92 1100 +30/–20% 970 B65513J0000R092
N87 1500 +30/–20% 1250 B65513J0000R087
N97 1500 +30/–20% 1290 B65513J0000R097
T38 6600 +40/–30% 5670 B65513J0000Y038
TDK came back to me with "This marking "893 S022" is not related to any TDK ferrite core or transformer part number, It is possible that such marking is more related to the transformer than the core"
-
Those all have high AL values, so they have no air gap. That leaves the B65513J0100A087 which Mouser (https://www.mouser.com/ProductDetail/EPCOS-TDK/B65513J0100A087?qs=xhbEVWpZdWe25brPDB%2Fzsg%3D%3D) stocks. I'd go with that. Ferrites always have lavish catalogue offerings but what you can procure is another story.
N87 100 +/–3% 86 B65513J0000R087
N87 is a common power ferrite material, they're explained here. (https://www.tdk-electronics.tdk.com/en/529404/products/product-catalog/ferrites-and-accessories/ferrite-materials)
-
Those all have high AL values, so they have no air gap.
Neither does the one on my board.
Ferrites always have lavish catalogue offerings but what you can procure is another story.
N49 1100 +30/–20% 800 B65513J0000R049 -- $2.16 @DigiKey
N97 1500 +30/–20% 1290 B65513J0000R097 -- $1.76 @DigiKey @Mouser
I asked TDK if they had an opinion. Maybe the rest of the peanut gallery here does too ......
-
Neither does the one on my board.
I don't see that you've established that, unless I'm misunderstanding your measurements? (Could you show a diagram..?) The dimensions are all out by 0.1mm or more, so the drawing doesn't tell you anything, nor can the parts be relied on to be that consistent.
An accurate measurement would be taking a whole core piece, clean and deburred (mind any glue residue!), placed flat (facing side) on a surface plate, then measuring the gap between center peg and plate. Test both halves, because often just one is ground down.
Put another way: the broken pieces are basically useless because there's no datum from the side limbs' faces.
Alternately, measure the inductance of one winding, then count its number of turns (may be a destructive test). Possibly, turns can be measured by looping one extra turn around and measuring ratios to it, but this does have some error (due to leakage; it's usually within a few percent, so, not bad for measuring something like a hundred turns). Or, more to the point, the single turn has the inductance of A_L of course, though this can be tricky to measure (it's a fairly small amount of course). (Also, try to measure at a related frequency like 10-200kHz.)
Anyway, it is still possible it's ungapped -- like I said, it's a fairly big part for what it should be doing. In which case, most likely choosing one of the lower mu materials (N97 probably closest?) would be perfectly fine.
Tim
-
Neither does the one on my board.
I don't see that you've established that, unless I'm misunderstanding your measurements? (Could you show a diagram..?) The dimensions are all out by 0.1mm or more, so the drawing doesn't tell you anything, nor can the parts be relied on to be that consistent.
The diagram is in the datasheet (copied below). For most of the outside shell w/h measurements I used an undamaged core. For the wall thickness and wall+core measurement (the later is the most important I think, red arrow in the diagram) I used 3 different broken pieces and got the same result on all 3, 2.95mm.
Would this pass scientific muster, no. Do I think there is an air gap, no.
0.21mm doesn't sound like a lot but I'm getting reproducible results to +-0.02mm with calipers.
As has already been said, the fact that with a broken ferrite the VFD still seems to work (note: I only ran the all-pixels test - shorting a test jumper when powered on - rather than programming selected pixels ) probably indicates that the actual value perhaps isn't that critical.
My orderable options seem to be: gapped: N87 ungapped:N49/N97
https://www.tdk-electronics.tdk.com/en/529404/products/product-catalog/ferrites-and-accessories/ferrite-materials (https://www.tdk-electronics.tdk.com/en/529404/products/product-catalog/ferrites-and-accessories/ferrite-materials); N87/N97 are both listed as f<500khz; N49 >500khz
I'm a digital (firmware) person, not analog so I'm reliant on the opinions of others here. I'm thankful for yours and floobydust's help (and PMd him to this effect) and am not trying to be argumentative but I'm not in a rush to order, new year is fine so I figured I'd see what the consensus of thought was.
If I had to pick something today I'd probably go with this: https://www.digikey.com/en/products/detail/epcos-tdk-electronics/B65513J0000R097/3913694 (https://www.digikey.com/en/products/detail/epcos-tdk-electronics/B65513J0000R097/3913694)
Thank you!
-
Something else would be measuring the MB3800 output pulses, they should be full tilt, at max. PW. That is, running the display with no transformer. I think it all runs off 5V?
It might help to estimate the core's inductance. AL is 10-15X higher for an ungapped core.
-
I think it all runs off 5V?
Correct. 5V is the sole input.
-
If you measure the bobbin's (primary winding) inductance in air, and know PW I think you can do math and infer the ballpark core AL.
-
If you measure the bobbin's (primary winding) inductance in air, and know PW I think you can do math and infer the ballpark core AL.
unfortunately as I said, I'm a digital type, not analog. given specify directions "measure x across pins y and z" I'm sure I could get it done, else .....
-
If you have a signal generator, you can do something like this,
https://www.seventransistorlabs.com/Calc/RLC.html#frq (https://www.seventransistorlabs.com/Calc/RLC.html#frq)
The numbers are exact for sine wave; you can use a square wave still, but the source voltage will be somewhat in error (if it's a 50% duty, bipolar (symmetrical +/- V) square wave, the fundamental peak is 4/pi times the square wave peak (not p-p!) voltage).
Typical values might be 0.1uF and 1kohm. Don't forget to include source resistance (of the generator) -- or, measure the voltage on both sides of the resistor, when at the resonant peak frequency.
You'll usually need an oscilloscope to do this, but you can still get the frequency right by, well, anything that can measure/infer the frequency, and using a peak detector to qualitatively measure the tank voltage (you just need to find the peak). (A DMM will not have high enough frequency response to measure the voltage here.)
So, the most basic case could be an Arduino generating a square wave, and you have a console or debug interface to tweak the period up or down as you find the resonance. A peak detector can be any signal diode, into a parallel RC network to GND, say 10nF and 100k, then measure the DC voltage on it with a DMM.
Tim
-
Some manufacturers put the air gap only on one half of the core's two pieces, others half on each half - you measure the center pole at 2.98mm on one, does that match the other?
I was thinking you could measure primary inductance bobbin's pins 1-5 in air with your Sencore LC103, and then estimate that resulting inductance in the various cores.
We don't know # turns but that could cancel in the two equations. Knowing the pulse-width IC6 is making at R18, you could ballpark if a gapped core is reasonable or not.
If you don't want the hassle, then take your best guess...
-
Some manufacturers put the air gap only on one half of the core's two pieces, others half on each half - you measure the center pole at 2.98mm on one, does that match the other?
I'd managed to lose the lower side of a core but I found it today. So I have 4 cores, from 2 bobbins.
Of these 2 are definitely tops (as I can see part of the serial# marking) and they measure 2.95 and 2.98. One is definitely a bottom as it's the one I glued back together and it measures 2.85. I'd discounted this measurement difference thinking I'd messed it up during gluing/clamping but as I said I found another today, small piece, no markings, which is also 2.85 and I suspect it's the missing lower.
So possible I was wrong and there is a 0.1mm airgap. Hey it's half way to 0.21 :-)
I have a third VFD with a broken ferrite (upper and lower cores are still intact). Tomorrow I could de-solder the bobbin from the pcb if it would be helpful.
If you don't want the hassle, then take your best guess...
To be honest I'm starting to get a headache. These inductance and frequency measurements, while well intentioned, all sound a bit vague. I have this nagging feeling that after taking them and doing some math I may still be no better off.
I have a friend who's a smart EE. I'll ask him for his thoughts and go with what he suggests. At this point my preference is whatever is least likely to make the board go poof and emit white smoke. Maybe I should flip a coin LOL.
-
Ah, seeing a correlation between parts is quite encouraging! Yes, that makes 0.1mm quite plausible, despite the tolerances on the drawing.
Tim
-
I ended up repairing the boards successfully.
I'd ordered the gapped N87 and ungapped: N97 based on the picture here (https://www.eevblog.com/forum/repair/can-anyone-id-these-ferrite-cores/msg4583191/#msg4583191)indicating the air-gap was formed by both halves. I'd hoped I could therefore use one half of an N87 and one half of an N97 to create the exact air gap
However in the parts I received the bottom half of both the N87 and N97 were identical and unmarked and the air gap was all in the upper half.
I tried both types (I had 4 boards with cracked ferrites) and they seem to work fine with either. I kind of assumed this may be the case as even the boards with worst damaged ferrite was still working.
Anyways, thanks for everyones help!