Good question about the winder.
Frankly, I wouldn't worry too much about vacuum impregnation at this point. Perhaps at 3kV it might be needed to prevent corona but not at 1.5kV. In fact, impregnation might not even be desired. It will always increase capacitance... Some of the best sounding audio transformers I have heard had no impregnation whatsoever. And some of those were output transformers loading big transmitting tubes at 1.5kV.
In the end, experimentation will be slowed way down if you need to spend a day impregnating and baking every bobbin before you can try it. You should count on experimenting. Your first wind will not be your last....
-- Dave
Frankly, I wouldn't worry too much about vacuum impregnation at this point. Perhaps at 3kV it might be needed to prevent corona but not at 1.5kV. In fact, impregnation might not even be desired. It will always increase capacitance... Some of the best sounding audio transformers I have heard had no impregnation whatsoever. And some of those were output transformers loading big transmitting tubes at 1.5kV.
In the end, experimentation will be slowed way down if you need to spend a day impregnating and baking every bobbin before you can try it. You should count on experimenting. Your first wind will not be your last....
-- Dave
There was a thread here a while back with a builder constructing a transformer of this size. The thread went cold before we heard the final results.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=70519&highlight=
A few years ago I paid a transformer winder to make me a custom transformer for an 833A amp. I had purchased several smaller (300B) sized transformers from him in the past with good results. The big transformer did not quite make it. Output transformers of magnitudes will stump even veteran winders.
http://www.tubelab.com/833SE.htm
http://www.diyaudio.com/forums/showthread.php?s=&threadid=70519&highlight=
A few years ago I paid a transformer winder to make me a custom transformer for an 833A amp. I had purchased several smaller (300B) sized transformers from him in the past with good results. The big transformer did not quite make it. Output transformers of magnitudes will stump even veteran winders.
http://www.tubelab.com/833SE.htm
Hi,
use as bigger laminations as You can,
because You will have lack of space from the harder isolation,
(high voltage...).
So if some extra space left this is good thing because all windings are closer to the center of magnetic feild...
But You have to take care about permeability of the lamination material...
I dont know what primary inductance You need for the desired tube...
Anyway as internal resistance of the tube is higher the more henry You will need for the proper design...
ofcourse, that will increase number of turns (for low permeabile laminations), and result will be higher capacitance, that will cause the loss of high frequency signal...
Also take car about induction idealy is 0.5-sqrt(2) Tesla
(number of turns, volatage, low frequency depending)
next thing is to design a correct air gap.
1/800 gap/lenght of magnetic lines is very linear,
and to prevent saturation too...
(this is higher gap then usual...)
But also decreasing Primary induction...
next thing is that square area of iron core is preferable...
So
Use very big laminations with larger window if posibile
but very thin
with magnetic permeability of about 800 to 1200
(it is not easy to find because, try german wasner)
cheers
use as bigger laminations as You can,
because You will have lack of space from the harder isolation,
(high voltage...).
So if some extra space left this is good thing because all windings are closer to the center of magnetic feild...
But You have to take care about permeability of the lamination material...
I dont know what primary inductance You need for the desired tube...
Anyway as internal resistance of the tube is higher the more henry You will need for the proper design...
ofcourse, that will increase number of turns (for low permeabile laminations), and result will be higher capacitance, that will cause the loss of high frequency signal...
Also take car about induction idealy is 0.5-sqrt(2) Tesla
(number of turns, volatage, low frequency depending)
next thing is to design a correct air gap.
1/800 gap/lenght of magnetic lines is very linear,
and to prevent saturation too...
(this is higher gap then usual...)
But also decreasing Primary induction...
next thing is that square area of iron core is preferable...
So
Use very big laminations with larger window if posibile
but very thin
with magnetic permeability of about 800 to 1200
(it is not easy to find because, try german wasner)
cheers
ak_47_boy said:
http://www.dissident-audio.com/OPT_da/Page.html
Unless you are Window's allergic.
Have fun.
Yves.
Zen,
You do not have to vacuum pot experiments, as Dave has correctly pointed out.
However, corona begins it's destructive work at 30 volts AC, not 3500 volts AC. By 350 VAC the expected MTBF of an unprotected coil is about 5,000 hours. By 1.5 kVA it is down to about 500 hours, by 3.5 kVA you are looking at tens of hours. You can use armored wire to extend this, but you will not find it on eBay.
Another point to consider is the thickness of the dielectric material. The 15 mil material shown earlier in this thread is fine for material type, but far too thick for your use. Calendered 410 Nomex can be counted on for 500 volts per 0.001 inch of material thickness, although the manufacturer rates it at 750 volts per 0.001 inch.
However, it is notorious for having pinholes that will allow flash over. So, you must use two layers, regardless of the thickness of each layer, of this material. This is a very strong material but it does not compress in a linear fashion, so winding on it can be a delicate balancing act, unless you have it trapped between two strong side plates.
SE OPT's are very prone to poor high frequency performance, as noted earlier, from too much total capacitance in the winding. Having these layer by layer insulation sheets will add capacitance. One of the ways to lower this capacitance is by using smaller diameter wire and more turns. If the physical volume of a winding remains the same and its width remains the same then adding turns actually cuts capacitance faster than the total length of the wire can add it. Another balancing act goes on here because too fine a wire guage will overheat and shorten the life of the winding.
Also, at these voltages, every dielectric edge, called a dielectric discontinuity, will be an area that corona attacks with greater enthusiasm than at any other place. This means you must use a minimum number and size of pieces of tape to hold wires and dielectric pieces in place with. This will just add to the difficulty of winding with Nomex.
When you start considering what thickness of material you must have per layer, the minimum you can have is 2 X 0.003 inches. This means that you will have 1500 volts of almost guaranteed dielectric strength per laye, so that should be adequate for layer to layer.
Then you must consider turn to turn for every turn in a layer. For in layer, turn to turn dielectric strength, safety agency calculations are based upon 150 volts of dielectric strength for every 0.001 inch through air or over a surface, like the next turn of wire. A more normal approximation is double that. With active corona and no isolation of wire insulation from oxygen, you should use that lower value. With a good vacuum potting the numbers rise dramatically but, you can still get failures due to repeated peak voltage stresses.
Please trust me when I say you are attempting the most difficult winding that can be done, when your voltages exceed about 800 VAC total, in one winding. The number of possible failure modes becomes amazing, usually smells very bad and can easily kill you. Be warned.
Bud
You do not have to vacuum pot experiments, as Dave has correctly pointed out.
However, corona begins it's destructive work at 30 volts AC, not 3500 volts AC. By 350 VAC the expected MTBF of an unprotected coil is about 5,000 hours. By 1.5 kVA it is down to about 500 hours, by 3.5 kVA you are looking at tens of hours. You can use armored wire to extend this, but you will not find it on eBay.
Another point to consider is the thickness of the dielectric material. The 15 mil material shown earlier in this thread is fine for material type, but far too thick for your use. Calendered 410 Nomex can be counted on for 500 volts per 0.001 inch of material thickness, although the manufacturer rates it at 750 volts per 0.001 inch.
However, it is notorious for having pinholes that will allow flash over. So, you must use two layers, regardless of the thickness of each layer, of this material. This is a very strong material but it does not compress in a linear fashion, so winding on it can be a delicate balancing act, unless you have it trapped between two strong side plates.
SE OPT's are very prone to poor high frequency performance, as noted earlier, from too much total capacitance in the winding. Having these layer by layer insulation sheets will add capacitance. One of the ways to lower this capacitance is by using smaller diameter wire and more turns. If the physical volume of a winding remains the same and its width remains the same then adding turns actually cuts capacitance faster than the total length of the wire can add it. Another balancing act goes on here because too fine a wire guage will overheat and shorten the life of the winding.
Also, at these voltages, every dielectric edge, called a dielectric discontinuity, will be an area that corona attacks with greater enthusiasm than at any other place. This means you must use a minimum number and size of pieces of tape to hold wires and dielectric pieces in place with. This will just add to the difficulty of winding with Nomex.
When you start considering what thickness of material you must have per layer, the minimum you can have is 2 X 0.003 inches. This means that you will have 1500 volts of almost guaranteed dielectric strength per laye, so that should be adequate for layer to layer.
Then you must consider turn to turn for every turn in a layer. For in layer, turn to turn dielectric strength, safety agency calculations are based upon 150 volts of dielectric strength for every 0.001 inch through air or over a surface, like the next turn of wire. A more normal approximation is double that. With active corona and no isolation of wire insulation from oxygen, you should use that lower value. With a good vacuum potting the numbers rise dramatically but, you can still get failures due to repeated peak voltage stresses.
Please trust me when I say you are attempting the most difficult winding that can be done, when your voltages exceed about 800 VAC total, in one winding. The number of possible failure modes becomes amazing, usually smells very bad and can easily kill you. Be warned.
Bud
Hi I forgot to say that one thing among others is crutial
for the sound and performance...
The total DC resistance of the secondary winding...
for the best performance and capability of driving all types of open buffle
loudspeakers cabinets,
has to be extremly small under 0.5 ohm
as lower as posibile...
I have home made OT for 2A3 that measures 0.3 ohm DC secondary
4 secondaries about 2mm wire in parallel
with 25H of primary, going extremly down...
laminations are not standard shape but that one with bigger window
weight abourt 5KG per chanel.
*
But for the 304TL I am not sure, it will be not easy to acheive good results,
because of the very high Ua, and probably higher internal resistance of the tube
*
By the way my friend have SE with 250TH tube, I do not know the chr of the his OT, but working not bad...
*
Anyway take extreme care if You go into building OT
You will need isolation vith dielectric constant for very high tensions.
Very big lamination core.
Take care and good luck
for the sound and performance...
The total DC resistance of the secondary winding...
for the best performance and capability of driving all types of open buffle
loudspeakers cabinets,
has to be extremly small under 0.5 ohm
as lower as posibile...
I have home made OT for 2A3 that measures 0.3 ohm DC secondary
4 secondaries about 2mm wire in parallel
with 25H of primary, going extremly down...
laminations are not standard shape but that one with bigger window
weight abourt 5KG per chanel.
*
But for the 304TL I am not sure, it will be not easy to acheive good results,
because of the very high Ua, and probably higher internal resistance of the tube
*
By the way my friend have SE with 250TH tube, I do not know the chr of the his OT, but working not bad...
*
Anyway take extreme care if You go into building OT
You will need isolation vith dielectric constant for very high tensions.
Very big lamination core.
Take care and good luck
BudP said:
Hey Bud,
The 304TL is easier to deal with than a 211. Its plate Z is about 1/3 the 211 for starters...
Crank in a tertiary coil to run its 25A of DC filament through to cancel some of the plate idle flux and that output is not going to be very large at all. 3-3k5 should do it. You also mention being near retirement age. I hope you have a good plan in place to deal with your library of designs. Have you an apprentice? Would you try one?
cheers,
Douglas
cheers
Looking at class AB1, with 1500 volts, I find a reactive load of 2.3k Z ohms needed from the primary. For 20 Hz performance of -0.5 dB this will require 70 Henry's of inductance.
A power throughput of 256 watts, maximum, translates into 45 vac at 5.7 amps ac for an idealized signal transform in the secondary. The turns ratio is 20.9:1 into 8 ohms for 3.5 k ohms of reflected impedance to the plate. The actual loaded secondary voltage we will see likely will be about 34 vac, from 1500 vac X 48% for max primary drive voltage. So, with a nominal max current of 7.5 amps times a 63% audio power factor, you are looking at 34 vac at 4.7 amps AC, equivalent constant load for heating, as a secondary load, and 730 vac at about 400 ma ac in the primary, with a 10% overall transformer efficiency. Plus, a max, AC signal, DC current of 572 ma and a min DC current of 270 ma, for primary heating.
A big, heavy wire transformer, even if you multifillar it and cheat like crazy on max power assumptions. For heating in a VA of this size, about 800 cma is going to be the minimum, so #21 on the primary and # 14 on the secondary, if we used only one wire for one winding in both primary and secondary. Since the secondary will be in three sectors it will need # 19 in each sector. Since the primary will need to be in four sectors, it will need to be #27.
Now, some of you can think about this, critique my assumptions and then we can do the the rest of the SE equations for a day or two and we will have a basic first cut design. Then we can argue about probable capacitances and what % is stray and what will be working capacitances and then predict the frequency response.
Bud
A power throughput of 256 watts, maximum, translates into 45 vac at 5.7 amps ac for an idealized signal transform in the secondary. The turns ratio is 20.9:1 into 8 ohms for 3.5 k ohms of reflected impedance to the plate. The actual loaded secondary voltage we will see likely will be about 34 vac, from 1500 vac X 48% for max primary drive voltage. So, with a nominal max current of 7.5 amps times a 63% audio power factor, you are looking at 34 vac at 4.7 amps AC, equivalent constant load for heating, as a secondary load, and 730 vac at about 400 ma ac in the primary, with a 10% overall transformer efficiency. Plus, a max, AC signal, DC current of 572 ma and a min DC current of 270 ma, for primary heating.
A big, heavy wire transformer, even if you multifillar it and cheat like crazy on max power assumptions. For heating in a VA of this size, about 800 cma is going to be the minimum, so #21 on the primary and # 14 on the secondary, if we used only one wire for one winding in both primary and secondary. Since the secondary will be in three sectors it will need # 19 in each sector. Since the primary will need to be in four sectors, it will need to be #27.
Now, some of you can think about this, critique my assumptions and then we can do the the rest of the SE equations for a day or two and we will have a basic first cut design. Then we can argue about probable capacitances and what % is stray and what will be working capacitances and then predict the frequency response.
Bud
agent.5 said:
Then you have two transformers with their capacitance and inductance in parallel....and you still need a choke that can do the DC idle and AC for signal simultaneously. The capacitance will sum, and the parallel L's will still need to have a minimum value that is going to be difficult.
cheers,
Douglas
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.