| Prune |
On an old microwave oven transformer I have the high voltage secondary has one lead going into a diode, and the other one is soldered to the transformer core. Can I disconnect this so I can use a bridge rectifier?
BTW, I'm very well aware of safety issues, lest someone should feel in a patronizing mood in replying. |
|
|
| Sch3mat1c |
Probably not, the insulation isn't very great against the core. If your circuit is isolated (no), you could.
Can use two in series to get a CT'd plus and minus with FWB's though. Then you have the 2kV or so difference in DC which may or may not be isolatable.
Tim |
|
|
| Prune |
| quote: | | If your circuit is isolated (no), you could. |
I'm not sure what you mean by whether my circuit is isolated. Do you mean from earth ground?
There is no way for me to get another transformer like this, as I found it on eBay. I paid more for shipping than for product -- it is a good deal larger than any other MOT I've ever seen, which is why I got it.
RMS under 1 A load is 2100 V (or approximately so; I had to use a resistor divider as my multimeter doesn't go that high). I was hoping to get close to 2 kV DC filtered...
BTW, I'm not sure whether to leave the 3V heater winding on or remove it. Three turns right on top of the HV winding. I don't know how to figure out how much current it can supply; could I feed into a full wave doubler to use for 6.3 V tube filaments? |
|
|
| Prune |
Oops, I just realized that I can wind my own filament winding.
I read on another board a post by someone that left the secondary connection to the core and still used a FWB. Does the core need to be grounded? |
|
|
| Sch3mat1c |
Hmm, that is indeed a monstrous transformer! Where'd you find loads for something like that? (125W light bulbs?)
If you use a FWB, the core (and the rest of the winding) will be at a DC potential of half B+. This can be seen easier by imagining what a center tap does. In addition to the bias, it will be at AC potential due to swinging full Vout. The result is that the minimum peak voltage (with respect to circuit ground, after the FWB) reached is about .8V and the maximum is .8V less than +V, simply because those very voltages are derived from those very sine wave peaks, minus a bit of drop.
As a result, you end up with full ACV placed across either the secondary-to-core or primary-to-core insulation.
If you ground the core, you get ACV swinging on the power supply instead. Think: ground loop up the...erm... words allowed here cannot describe the place it goes to. :devily:
What I mean by isolation is: if you can't have a power transformer isolating, you can have input and output transformers rated for the isolation instead. I drew up (and will eventually build) such an isolated line operated design.
In terms of DC isolation instead, just use coupling caps rated for the DCV. Ground the CT so you only have to worry about ripple on the negative rail.
Tim |
|
|
| Prune |
Last time I saw a microwave open I was left with the impression that the core was grounded to the chassis, and presumably earth. This schematic at least shows the first one:
http://stefanbinder.privat.t-online.de/pics/uw9.jpg
I don't need isolation. As I'm feeding a glow discharge, I just care about relative potentials between the electrodes. I'm trying to figure out a way to get about 2 kV (filtered for ripple; 300 mA will do). Is this possible with this transformer, and how?
BTW, there seems to be contradictory advice on the Internet whether the magnetic shunts should be left in or knocked out. As I don't need the full current capability of the transformer, current limiting should be a good thing. Are there any possible cons to leaving them in?
Also someone ground down the welding of the core sheets (runs cooler without it with less losses in the core) and rebuilt them, covering with laquer, adding more winding insulation, all without disturbing the windings. If there's no suggestion of how I can achieve what I said in the first paragraph, I'll go this extreme route, with the risk of wrecking the beast. |
|
|
| Sch3mat1c |
Hm rebuilt? That's neat... Can you contact the guy and see if it's rated?
Ok, since you only need potential, a FWB will work. But how do you interface that to your signal source?
At that level you can leave the shunts alone. Without, you'll have a bit better coupling and as such more voltage, better regulation and unlimited current (like a normal transformer). Hm, since you want a variable current output, that might be a good thing. You decide.
Tim (is eventually going to use a similar, though 800V 15A, transformer for a "small" induction furnace) |
|
|
| Prune |
So should I ground the core or not in a FWB arrangement?? I opened up my microwaves and the transformers are all earthed, both core and one lead of the secondaries. The other lead goes into a 1 uF capacitor with diode to earth on the other end of the cap -- half-wave multiplier biasing the heater coil.
The coils are very solidly packed with resin-impregnated cardboard or something like that. I don't think they'll budge if I take apart the core. If I find that isolation is a probem, I'll try to rebuild it. What would be a good replacement isolation?
For interfacing to the audio signal, I'm going to replace the ballast resistor with a tube that will be controlled by the audio signal. I don't mind using AC coupling if needed -- I already have a capacitor in my source's signal path. |
|
|
| Sch3mat1c |
Yeah sure, ground the core. That way if it does arc over, you have the safety ground handling fault current like it ought to, rather than tracing up the primary wiring, or whatever's connected to the chassis, etc...
In ovens it's both convienient and safe to ground everything.
Tim |
|
|
| Prune |
| What's the best approach for filtering the rectified DC? In Power Supply Designer, a CRC with 200 uF total gives a good result, but that capacitance at that voltage is still very expensive. Is there another arrangement that will be cheaper and not lower voltage too much? And for capacitors, what's the cheapest way to achieving this? I was thinking of series/parallel array of lower voltage caps, say 450 V electrolytics. |
|
|
| Sch3mat1c |
Ya, series is the deal.
I once e-bayed a surplus board of some 16 x 780uF 450V and 8 x 270uF 450V computer grade caps... plus a tonne of equalizing resistors, and a few chunky diodes. My guess was a three-phase power supply at around 2.5 and 5kVDC... :D
I'd do CLC with the inductor on the ground side, so it can mount to the chassis without needing absurd voltage ratings. Capacitors will have a bit of AC hum on them so metal cans chassis-mounted are no good. But most of them have voltage vs. chassis anyway.
Tim |
|
|
| Prune |
Are equalizing resistors necessary across series capacitors, and what are appropriate values?
I just bought from eBay 25 pieces of 330 uF rated at 450 V for $46. Couldn't find anything cheaper. How many would I need to put in series to avoid frying them? And what filtering configuration would be optimal for a load from 200 to 400 mA? You mentioned a choke, but power chokes I found for sale are all extremely expensive, $100 and up. |
|
|
| Sch3mat1c |
Well with so much power flowing, of course it's not going to be cheap. You said some 2kW???
Stack as many caps as you need. Figure 350-400V per cap, that allows surge overhead to be within ratings. You can go rated (450V/cap) but they might not live quite as long. (Who knows...)
Tim |
|
|
| Prune |
Not 2 kW. I said 2 kV at 200 to 400 mA, so 400 to 700 W.
What's the best configuration to use, given the capacitance I have? LC, CLC, CRC, or CLCRC, and how do I best distribute the capacitance to these Cs? I tried PSU Designer, but it and LTSpice give me different results. I'd wind my own chokes if I knew where to get appropriate cores and wire... |
|
|
| Sch3mat1c |
Ouch, big tin can. :D 720J total, eh?...
Depending on how much ripple can be tolerated, I'd take... 1 of those caps at the rectifier would give about 28Vrms of hum. That's all of 1%. Add on a 2H 300mA choke and another cap or two and you'll get about .6V (0.03%) hum.
Tim |
|
|
| Prune |
I haven't gotten the can caps yet, but even with just the 165 uF of the multi-mini-cap in a simple C filter, no CLC or CRC, I'm getting a very quiet glow discharge at 100 mA. Haven't tried more current yet as my ballast resistors overheat.
:hot: WARNING! :hot:
The following images shows a very dangerous setup and should not be attempted by any sane person. I was two meters from the thing with a dry chemical extinguisher besides me when I powered it up. The capacitors have equalizing resistors about ten times the leakage ratings, and wiring is with 5 kV insulation, except the jumper cables.
The discharge started on its own because I had a very thin wire between the electrodes that burned off. I had my finger on the power bar button. Two cathodes to one anode, about 50 mA each. Yes, this is a glow discharge and not an arc, though it's hard to tell from the poor quality photo. I also tried using a microhollow discharge as the cathode, but my copper/mica sandwitch broke before I could take a picture. That resulted in larger and more stable discharges. Even larger ones were achieved by forcing air through the hollow with an aquarium pump. After about five to ten seconds the ballast resistors overheat and melt the solder.
 |
|
|
| Sch3mat1c |
What resistance is that? Did it make any sound (hissing, buzz, etc)? Did you measure the voltage across the discharge?
Tim |
|
|
| Prune |
There is no noise that I can hear, so it's low enough to be drowned out by the transformer buzzing (I still haven't rebuilt the core as the welding proved too tough to grind out by hand). Occasionally the discharge would loose stability (probably due to horizontal placement and air drafts) and momentarily break down into an arc, extinguish as resistors heat and transformer magnetic shunts kick in, and restart as capacitor voltage rises again ; this caused a few crackling noises. It was stable with the MHCD cathode, however, and vertical placement also improved stability (and size). In the final setup I may have about five centimeter discharges.
Ballasts are 25K for each of the two chains. My measurements are not very good since the voltmeter doesn't like the high absolute voltage, and resistors heat a lot and that must change the resistance a good deal between short-circuit and discharge mode, but it seems very roughly that resistors drop on the order of 2/3 of the voltage, and the discharge the other 1/3 (an arc would drop very little voltage). In any case, since I'll be using an array of microhollow cathode discharges as the cathodes instead of this simple electrode configuration, this measurement doesn't really mean much.
Sch3mat1c, what would make a good power resistor of about 300 ohms for a CRC? A few incandecent bulbs in series? The problem with bulbs is that their resistance is very low when cold. Especially when I knock out the transformer's magnetic shunts, that would make the circuit breaker trip at power on (already the lights dim). I guess I need some sort of soft start.
I was thinking of using a CRC before a CLC so that the choke would see little ripple and I could use a lower voltage choke, such as one of the Hammonds.
BTW, how can I figure out an appropriate power factor correcting capacitor across the primary without a scope? Most people use 100 to 200 uF. |
|
|
| Prune |
| Is there any disadvantage to a choke on the ground side (other than the fact that I can't simulate this in PSU Designer)? |
|
|
| Prune |
| With a choke on the ground side, if I want to put a CRC before it, where does the R go? |
|
|
| Prune |
| One more question: how do I find out the output impedance of the transformer, so I can simulate it properly? DC resistance measures about 80 Ohms. |
|
|
| Sch3mat1c |
Let's see, the problem of turn-on surge is threefold:
1. You have a big system. That means big inrush currents, period. (Small stuff like a 6V6 amp can get away with this because...they're small).
2. The iron has to establish a field, which is a pretty drastic surge if turned on at the right point in the cycle (zero crossing I think).
3. You have a lot of energy in the caps.
A slow turn-on would help (series inductance to soften the blow, resistor, etc. with relay to short after a second or three). Only thing that will reduce the effect of C at the FWB is to use less there; you could use maybe 10-50uF. So what if the ripple is 300V, that's only 10%! Remove the rest with the choke. Use a lot of C after it if necessary.
Choke in ground is electrically the same as choke at +V, except the winding is at 0V as well as the frame, so there's no insane insulation issues. You just get those 300V of ripple on the minus side of the rectifier bridge instead.
PFC and output impedance go hand in hand, as PFC cancels the winding and leakage inductances the transformer creates. When open-circuited, it acts as a choke equal to the primary on the core; when secondary is shorted, low coupling manifests itself as leakage inductance. The second only shows up under a full short, any resistance adds vectorially creating a slightly inductive load. The primary inductance appears in parallel with this, but in most cases I'm sure pales in comparison to the lower impedance presented by the leakage inductance (Ltot = L1L2 / L1+L2, just like resistors) and transformed load resistance so can be ignored.
Qualitative, but not quantitative. I wouldn't be able to guess, but you could try taking some measurements depending on your available equipment. However, once you know the inductance seen through whichever winding you are going on, you can calculate it as impedance, and reactance to cancel on the primary side. I would guess you need no more than 100uF. Maybe 20uF.
Tim |
|
|
| Prune |
Ripple as percentage is misleading here, because of the following: SE class A output where the output varies only a fraction of the total bias; the average bias goes into powering the plasma. Since the audio frequency modulation is only a fraction of the total output, ripple needs to be lower. How much exactly I have no idea, and that's why I'm doing extra filtering.
BTW the tin can capacitors are only rated at 3 kV, and even though I'm using them as the second and third Cs (the first is the series capacitors which add up to 3.6 kV), I'm worried about running them so close to the maximum, especially when there is no load connected.
I ground out the welds and disassembled the MOT (it was a pain, that's for sure). Without the magnetic shunts, what should I do with the gap left between the primary and secondary windings? Do I leave it, or do I move the windings closer together? Also, the filament winding was only three turns and giving 3 V RMS, so I took it out. It was wound around the HV secondary (as in normal operation it is biased by the HV). I'm thinking of rewinding it with more turns (what AWG wire do I use?), perhaps in the gap left by the removed magnetic shunts. If I do this, do I wind it close to the central bar of the E, or in a wide circle?
The insulation of the HV secondary was quite heavy around the outside, but almost nil at the inside, as that end was grounded to the core. Since I don't want to ground it again, I need to add better insulation. However, I only have about 1.5 mm spacing between the inner bar of the core and inner diameter of the winding. What's the best insulation to use that will fit? I was thinking mica, which is supposed to take at least 20 kV/mm, held together by paper and shellac.
Since there is supposed to be a small gap between the E and the I (http://geek.scorpiorising.ca/windingtransformers.html), how is the core supposed to be grounded to the chassis? Do I ground the E, the I, or both? Also, should I add electrostatic shielding (aluminum foil)? |
|
|
| Sch3mat1c |
So you took the core *entirely* apart?
O_o
Um... ok. I'd interleave them (as at the bottom of Geek's page). Only carrying AC so no need of a gap. Mind that those transformers are wound very close to saturation and might need lower input voltage if you keep getting current spikes on the peaks of the primary voltage.
For the heater, I'd put it inbetween (doesn't matter really), close or far doesn't matter as the magnetic flux flows almost entirely in the core - as long as you have one turn going around the center (or two turns, one around each arm) you get the same voltage. Think of it this way, farther out from the center leg, you might get less flux, but you make up for it by using more wire (longer perimeter). Not really true but you can imagine it that way. As for turns, you can experiment or use an educated guess - count the turns on the primary (ought to be able to count layers and turns per layer, then multiply), divide by line voltage to get turns per volt and multiply by new winding volts to find turns.
Dunno about your insulation, I'd be afraid the mica wouldn't be complete (you've got some tight corners after all). I suppose you could test it with a low current 5kV source after and see if it holds.
Tim |
|
|
| Prune |
No, I didn't separate all the laminations, just enough to take out the windings. The laminations do not appear to be insulated from each other. I won't leave any gap other than the lacquer coating (I coated the E and the I separately).
BTW, I would like to take this opportunity to thank you for all your help, when no one else replied! :) |
|
|
| Prune |
Well, I separated them all apart. 216 laminations total -- that's going to be a lot of nail polish... I'm worried that the windings won't fit due to the increased width from the nail polish. Also, the parts where I grounded out the welds are rough and may make electrical contact... sanding 216 laminations is nuts.
I had read somewhere that transformer cores should be grounded. I guess this is impossible if the laminations are insulated from one another. |
|
|
| Sch3mat1c |
I've never re-varnished the lams... as long as they get back in the same order, the broken varnish should line up, no? Well okay, if you interleave them, it's different... so what.
In fact... as I recall, the black coating (either Parkerizing (iron phosphate) or just scale (iron (II) oxide) from heat treating) keeps them insulated adequately anyway.
It just matters that most of the lams are grounded, or if there is a faliure, there's a path for it.
Tim |
|
|
| Prune |
But that's exactly what I'm saying, how can they be grounded if they are insulated from each other? On the other hand, he says if they are electrically connected there will be more losses with heat and buzz (and let me tell you, did that MOT buzz...)
There is no oxide coating; they are clean silicon steel, grain aligned, with a very thin layer of shellac which has mostly broken up.
The interleaving I like since I don't have to worry about keeping the E and I nicely aligned. The problem is that there's no way I can slip the windings back in, so I have to interleave them with the windings in place... |
|
|
| Prune |
| BTW, the smallest mains-AC rated capacitor I found locally is about 135 uF. Is that way too big for PFC here? If it is no good, where can I source an appropriate cap, and what can I use the one I have for (I'd hate to throw it out after I spent some money on it). |
|
|
| Sch3mat1c |
Well, now that you've removed the inductance from the transformer (interleaving the core and removing the shunts, decreasing the leakage inductance), you don't need PFC any more than any other transformer does.
If you're worried about power factor, switch to choke input :devilr:
Tim |
|
|
| Prune |
Grinding out the welds was easy. I washed the laminations in lacquer thinner, but some shellac was still left, so I had to go over with steel wool, and this was the most time consuming part. The nail polish I'm using is labeled "With Nylon" and I hope that won't be a problem. Increased the insulation on the inner side of the secondary with four layers of mica, so that I don't have to ground that end as was the case in the original construction.
 |
|
|
| Prune |
| I got a couple of used microwave oven capacitors for $1 a piece. They are rated at 2000WVAC. WTF does the W stand for? Also, I'm wondering what they are made from. Are they electrolytic, or paper/oil? |
|
|
| Prune |
What does that mean?
Also, what's inside the metal cans? |
|
|
| imix500 |
| That's the safe working voltage for that cap, similar to a safe working load limit on a rope. For dc operation, you can multiply that by 2.88 as I remember to get a safe voltage limit, but don't quote me there. Those are probably film in oil type. |
|
|
| Sch3mat1c |
Crack one open, it's just one wasted dollar right?
WVDC = Working Voltage Direct Current. WV is usually omitted because in the usual understanding of capacitor ratings, it's redundant.
Tim |
|
|
| Prune |
All the capacitors I found at the repair shop were different, exept for this pair. As I wanted to have a pair (if I need to use it, I'll need one for each of two channels after all), I didn't want to take one apart. In retrospect, I should have gotten one more, but I didn't think of that at the time.
Interestingly, all the MOTs there were smaller than the one I have, which makes me wonder whether I can push it more than I thought originally, as I had assumed it was from a regular 1 kW microwave. Judging by the size difference, especially when rebuilt, I'd guess over 1.5 kW.
Sch3mat1c, perhaps you can answer my question here:
http://www.diyaudio.com/forums/show...&threadid=41165
as no one else has yet. |
|
|
| Prune |
How do I shield this? The large transformer is going to radiate like crazy. I don't think an aluminum enclosure will do much; I'll need steel. But that would be expensive. What if I just shield the transformer and choke, so less material is used -- how do I avoid the heat problem then?
I do plan to have a slow fan for cooling, but I'm wondering if it would be a problem to add a winding for it on the main transformer instead of using a separate one (in terms of EM noise the fan can send back through the transformer).
Finally, for for HV cabling from the supply to the speakers, I was thinking that coaxial cable with the shield ground and center conductor HV would be safest. But what about connectors? I'd like to use a single connector for everything, but the highest rated multi-pin connectors I found locally are only 600 V. |
|
|
| Prune |
| How much continuous bridge rectified, filtered current can a 1500 W, 2100 VAC RMS transformer supply? |
|
|
| Sch3mat1c |
1.5kVA @ 2.1kV = 714mA / 1.5-2 (RMS derating) = around 400mA. You can draw more at the expense of greater temp rise and voltage drop.
Tim |
|
|
| Prune |
Are you sure that's right? The diodes only conduct near the peaks of the voltage waveform, and so you can't just divide by the RMS. At 400 mA, LTSpice shows RMS about 2.4 kW!
In the image, the load has 100 mA sine superimposed on the 400 mA DC because I wanted to simulate how much voltage would vary; the voltage source sine amplitude is 3050 as that's peak voltage, giving close to 2100 RMS.
 |
|
|
| Sch3mat1c |
Well, whatever. Standard derating is 1.4 (equivalent to AC RMS > DC peak change) to 2 times (good SOA).
Tim |
|
|
| Prune |
SOA? I'm not familiar with that abbreviation.
OK, so 400 then. I'm just curious what's wrong with the simulator. Or is there nothing wrong and I should be looking at the Average number (1.25 kW) instead of the RMS?
BTW, the only mains IEC plugs with built in EMI/RFI filters I could get cheaply go up to 10 A. I wonder how conservatively rated they are. The only series device inside is a common mode choke, the rest are capacitors and in parallel. I'll have to measure the power factor somehow and correct it if it's not 1, otherwise I'll fry these mains filters. |
|
|
| Prune |
| What do you think about that, using a 10 A mains filter? I was hoping to run the transformer at 1200 W, but if the power factor is not 1, wouldn't it draw more than 10 A from the mains, and overheat the choke in the filter? BTW, would the use of a mains DC blocker as discussed in other threads change the power factor? |
|
|
| Prune |
The transformer is leaking. A lot. The steel chassis it's in sings loudly with magnetostriction.
Hmm, I just calculated that Bmax looks to be about 1.7 T, which is above saturation for the core material. Looks like I'll have to add more turns to the primary and loose some secondary voltage. Dropping secondary voltage to 2.6 kV brings Bmax just below 1.5 T, which is probably just under saturation, but I can't afford more loss in output voltage. |
|
|
| raypsi |
I have tonnes of these transformers. I tore the "I" section off of 2 of the same transformers, and stacked the "E" s together. Potted them in a coffee can. I get 2500VDC unloaded off just one secondary. Loaded up it goes down to 1750VDC. And is rock stable into a bank of 175WVDC caps 20 of them in series for 3500WVDC. One transformer is the same just half the power. The secondary's of these transformers make great chokes too.
The caveat is they have welds on the laminations to hold together the core. These welds act a shorted turns, and draw copious amounts of power up to 800 watts depending on the transformer. And that's with no load.
Lastly these suckers can kill you. I've been lucky I've only lost my 2 best DMM's to the ways of 1750VDC. |
|
|
|
