I had the idea to make my own current limited transformer, and I am looking for any comments or direction that anyone may have. I have adequate experience in areas of building and working with such things and believe that I would be able to do this. a am, however, unfamiliar with working with any voltages over about 3 kilovolts, and may be mistaken in trying to do this. Attached below is a rough schematic (sorry for any problems in advanced, I am not the best at drawing them and it is not the highest quality) to try to provide further information as needed.
Any help and advice is thanked in advanced,
Nelson
HomeMade eTranny
HomeMade eTranny
- Attachments
-
- transformer.tif (736.87 KiB) Viewed 2906 times
-
Chris Bradley
- Posts: 2930
- Joined: Fri May 02, 2008 7:05 am
- Real name:
Re: HomeMade eTranny
That's the way to rectify a transformer output, of course - are you saying you want to make the transformer?
Re: HomeMade eTranny
Yes, I want to build all of it.
-
Tyler Christensen
- Site Admin
- Posts: 551
- Joined: Mon Feb 23, 2009 9:08 pm
- Real name:
Re: HomeMade eTranny
If you want to wind your own transformer, I'd suggest making a lower voltage high frequency core and then use a multiplier to increase the voltage up to fusion levels. This prevents issues of needing excellent insulation (which is hard to do yourself) in the windings since often it can be only one layer this way. Also it requires fewer turns generally which prevents a lot of spent time on coiling.
Re: HomeMade eTranny
I am fine spending more time coiling, and would probably be able to find enough material such as teflon or some polycorbonate to be able to insulate it, but a voltage multiplier is always an option.
-
Doug Coulter
- Posts: 1312
- Joined: Sun May 27, 2007 3:18 pm
- Real name: Doug Coulter
- Location: Floyd, VA, USA
- Contact:
Re: HomeMade eTranny
I'd say listen up to Tyler on this one. I'll add my inflated 2c worth.
Except for very rare instances, the pros do it that way, and unless you believe you're somehow smarter than a few generations of smart engineers, you might want to learn why they do it the way they do. If after that, you can improve on them, by all means do, but first understand what you're trying to improve on.
All high voltage solutions have their problems, and some of the solutions only work well on large scales (power companies, for example). There are X ray supplies around, but these are almost universally designed to be as inexpensive to make as possible for the job at hand, which does NOT include hour long runs at full power -- they won't take it, because making them that way would be several times more expensive than the way they are (really, were) made. There are current posts on this forum in just the past few weeks about burning these things up easily. Not where you probably want to go. Current X ray supplies, by the way, all use HF switching and multipliers. Now that we have decent fast switching devices, it's the best way to go, there's no argument among pros on that.
And that's why the old 60hz one stage ones are out there surplus -- they stink compared to newer tech.
All (that's 100%, no 99.9's) of the pro HV DC power supply companies use multipliers and fairly high switching frequencies for a good set of reasons, some of which Tyler mentioned. Some of the others are, less ripple on the output with less stored energy in the filter capacitors (which in turn get cheaper because they are capacitively smaller for the same impedance at the higher frequencies), which is hazardous to all of the user, the diodes, and the driven gear -- and even the driver when an arc blows all that stored energy back into the switching devices through the transformer. Working with switched square waves is nearly always much more efficient than with sine waves, more controllable with PWM, and any heat you make with inefficiency has to go someplace, usually at some cost to the maker and certainly the user.
One disadvantage to trying to do it all on one transformer other than insulation issues (which are a big one) is that when you increase the inductance without limit (more turns), and add capacity (from the wires to the next wires in the winding) you're making a tuned circuit - you may not be able to get the self resonance high enough for your drive frequency, even at 50 or 60 hz and still get the turns ratio you want without using such a massive core (too keep that out of magnetic saturation) you can't afford it, lift it, and so on. X ray machine transformers are usually designed to run only a full load, for short times (maybe a second or less then cool-down while someone changes filme etc) and only at full load, and are very inefficient for that reason -- since the manufacturer isn't paying your power and wiring bills, so what if it's only 20% or less efficient? Even the user doesn't care, as the thing is only on a couple of minutes total to take a couple hundred X rays. So they all use inadequate core area for the job we want to do, just enough for the job they were meant to do instead.
Trying to drive a transformer above it's self resonance frequency doesn't work out well at all, as you'll find when you try to do it. It's like trying to force AC across a capacitor, and any losses at all (series R) get hot fast, and so do the drivers.
Of course, I've been reading up on Pelletrons lately, and they point out that their way is even better in a lot of ways than solid state supplies, or other things that get killed by arcing, something you're going to encounter making a fusor. And for what they're good for (very HV, low current) they are right, too.
But for most fusors, a pelletron would not be the thing -- too high volts, and not enough current. You'd need too many "pellet belts" to get the current up for it to be real practical, or too much belt speed for any belt to handle (it's charge per second past the pulleys). Now I suppose I can expect someone to prove me wrong (even me! I might try one.) and it'd be fun to see indeed.
Engineering is not about brute force, it's about doing trade-offs right, which you cannot do unless you map out the trade-off space.
Dumb example:
An engineer wants to improve efficiency of a given switching supply. He sees I^2R losses in the FET switches as well as losses during the finite switching time, so he changes them to bigger ones with lower "on" resistance. Efficiency goes down. Why?
The bigger FETs are harder to drive (more gate capacity -- can't force an infinitely fast risetime across a good capacitor) and now the main loss is during the now-slower switching time. So he drives the gates harder -- which takes more power, probably from a component that can't stand it as easily, and now he still has somewhat lower switching time, and also more power loss in gate drives and driving the greater intrinsic capacity of the FET D-S loop. He goes round and round until he figures out the thing to do is drive the transformer more wisely (either make the driver match the xfrmr, or vice versa) as regards when in the waveform it's going to draw current (always a tuned circuit with some sort of spectrum, and probably not a pure resistive load at any frequency) and goes back to the original driver design, which is now efficient again, and has fewer I^2R losses because the transformer isn't drawing current in a direction that matters during the time it matters but rather is helping the switching take place instead.
Maybe not a dumb example after all. That little bit of complexity was considered a black art by all who hadn't mastered it (nearly everyone during my long career, and still....), and that's only a couple of variables in a rather big trade-off space. ( I could add hysterisis loss in the transformer core and skin effect in the windings, diode recovery time losses -- there are a few more) The point is, brute force often is not only expensive, but also plain doesn't work!
Try brute subtlety with extreme malice aforethought, and be prepared to have to try a few variations before you have something that is solid. Or just buy one....you may find that cheaper if you've not had experience in power supply design and don't want to spend the time to become good enough at it.
(which itself costs money -- you're eating all that time, right?).
OK, maybe a nickel's worth -- I type real fast.
Except for very rare instances, the pros do it that way, and unless you believe you're somehow smarter than a few generations of smart engineers, you might want to learn why they do it the way they do. If after that, you can improve on them, by all means do, but first understand what you're trying to improve on.
All high voltage solutions have their problems, and some of the solutions only work well on large scales (power companies, for example). There are X ray supplies around, but these are almost universally designed to be as inexpensive to make as possible for the job at hand, which does NOT include hour long runs at full power -- they won't take it, because making them that way would be several times more expensive than the way they are (really, were) made. There are current posts on this forum in just the past few weeks about burning these things up easily. Not where you probably want to go. Current X ray supplies, by the way, all use HF switching and multipliers. Now that we have decent fast switching devices, it's the best way to go, there's no argument among pros on that.
And that's why the old 60hz one stage ones are out there surplus -- they stink compared to newer tech.
All (that's 100%, no 99.9's) of the pro HV DC power supply companies use multipliers and fairly high switching frequencies for a good set of reasons, some of which Tyler mentioned. Some of the others are, less ripple on the output with less stored energy in the filter capacitors (which in turn get cheaper because they are capacitively smaller for the same impedance at the higher frequencies), which is hazardous to all of the user, the diodes, and the driven gear -- and even the driver when an arc blows all that stored energy back into the switching devices through the transformer. Working with switched square waves is nearly always much more efficient than with sine waves, more controllable with PWM, and any heat you make with inefficiency has to go someplace, usually at some cost to the maker and certainly the user.
One disadvantage to trying to do it all on one transformer other than insulation issues (which are a big one) is that when you increase the inductance without limit (more turns), and add capacity (from the wires to the next wires in the winding) you're making a tuned circuit - you may not be able to get the self resonance high enough for your drive frequency, even at 50 or 60 hz and still get the turns ratio you want without using such a massive core (too keep that out of magnetic saturation) you can't afford it, lift it, and so on. X ray machine transformers are usually designed to run only a full load, for short times (maybe a second or less then cool-down while someone changes filme etc) and only at full load, and are very inefficient for that reason -- since the manufacturer isn't paying your power and wiring bills, so what if it's only 20% or less efficient? Even the user doesn't care, as the thing is only on a couple of minutes total to take a couple hundred X rays. So they all use inadequate core area for the job we want to do, just enough for the job they were meant to do instead.
Trying to drive a transformer above it's self resonance frequency doesn't work out well at all, as you'll find when you try to do it. It's like trying to force AC across a capacitor, and any losses at all (series R) get hot fast, and so do the drivers.
Of course, I've been reading up on Pelletrons lately, and they point out that their way is even better in a lot of ways than solid state supplies, or other things that get killed by arcing, something you're going to encounter making a fusor. And for what they're good for (very HV, low current) they are right, too.
But for most fusors, a pelletron would not be the thing -- too high volts, and not enough current. You'd need too many "pellet belts" to get the current up for it to be real practical, or too much belt speed for any belt to handle (it's charge per second past the pulleys). Now I suppose I can expect someone to prove me wrong (even me! I might try one.) and it'd be fun to see indeed.
Engineering is not about brute force, it's about doing trade-offs right, which you cannot do unless you map out the trade-off space.
Dumb example:
An engineer wants to improve efficiency of a given switching supply. He sees I^2R losses in the FET switches as well as losses during the finite switching time, so he changes them to bigger ones with lower "on" resistance. Efficiency goes down. Why?
The bigger FETs are harder to drive (more gate capacity -- can't force an infinitely fast risetime across a good capacitor) and now the main loss is during the now-slower switching time. So he drives the gates harder -- which takes more power, probably from a component that can't stand it as easily, and now he still has somewhat lower switching time, and also more power loss in gate drives and driving the greater intrinsic capacity of the FET D-S loop. He goes round and round until he figures out the thing to do is drive the transformer more wisely (either make the driver match the xfrmr, or vice versa) as regards when in the waveform it's going to draw current (always a tuned circuit with some sort of spectrum, and probably not a pure resistive load at any frequency) and goes back to the original driver design, which is now efficient again, and has fewer I^2R losses because the transformer isn't drawing current in a direction that matters during the time it matters but rather is helping the switching take place instead.
Maybe not a dumb example after all. That little bit of complexity was considered a black art by all who hadn't mastered it (nearly everyone during my long career, and still....), and that's only a couple of variables in a rather big trade-off space. ( I could add hysterisis loss in the transformer core and skin effect in the windings, diode recovery time losses -- there are a few more) The point is, brute force often is not only expensive, but also plain doesn't work!
Try brute subtlety with extreme malice aforethought, and be prepared to have to try a few variations before you have something that is solid. Or just buy one....you may find that cheaper if you've not had experience in power supply design and don't want to spend the time to become good enough at it.
(which itself costs money -- you're eating all that time, right?).
OK, maybe a nickel's worth -- I type real fast.
Why guess when you can know? Measure!
Re: HomeMade eTranny
The best insulator is polythene (polyethylene), but not so good if it gets hot.
It must be 'virgin', not recycled, and the thicker the better. I can't remember the figures off the top of my head, but go for the type they use for good quality rubble sacks, builder's sheets (the type they put down under concrete floors) etc.
The pro's use mylar mostly.
Go for high frequency for the reasons given by Doug and Tyler.
A few of us are working on this type of supply, but I've had to postpone my project due to work commitments. (see recent threads on the subject).
There are pro's and con's with multipliers.
Do a bit of research into ferrite cores and 'switch mode power supplies'. As Doug says, this will mean huge savings on capacitors.
Read the threads in the High Voltage forum.
Above all, remember, THESE THINGS ARE LETHAL.
It must be 'virgin', not recycled, and the thicker the better. I can't remember the figures off the top of my head, but go for the type they use for good quality rubble sacks, builder's sheets (the type they put down under concrete floors) etc.
The pro's use mylar mostly.
Go for high frequency for the reasons given by Doug and Tyler.
A few of us are working on this type of supply, but I've had to postpone my project due to work commitments. (see recent threads on the subject).
There are pro's and con's with multipliers.
Do a bit of research into ferrite cores and 'switch mode power supplies'. As Doug says, this will mean huge savings on capacitors.
Read the threads in the High Voltage forum.
Above all, remember, THESE THINGS ARE LETHAL.