Bulging cannot be fully prevented, but can be minimized by the following
1. Using a rounded core (R-core). Impossible for C-cores, EI cores, torroidals.
2. Using a finer wire for the first layers, where finer wire bends easier and gives a lower "bowing" radius.
3. Smash the bowing resulting from the first layers with a clamp after winding them. Lower gauges can be smashed using hand/fingers.
4. Wind using an "anti-bowing" wire positioning, which when layering the wire, bends it at a radius negative to the coil radius, hence the wire lays more flat. Check the picture.
5. Use a squarer geometry bobbin instead of a highly rectangular one. Very rectangular coils result into more bowing.
Hint No 4. plays a huge difference before intervention with Hint No.2
1. Using a rounded core (R-core). Impossible for C-cores, EI cores, torroidals.
2. Using a finer wire for the first layers, where finer wire bends easier and gives a lower "bowing" radius.
3. Smash the bowing resulting from the first layers with a clamp after winding them. Lower gauges can be smashed using hand/fingers.
4. Wind using an "anti-bowing" wire positioning, which when layering the wire, bends it at a radius negative to the coil radius, hence the wire lays more flat. Check the picture.
5. Use a squarer geometry bobbin instead of a highly rectangular one. Very rectangular coils result into more bowing.
Hint No 4. plays a huge difference before intervention with Hint No.2
Thanks Alexander. So, if I've understood correctly you put a pre-concave curve onto the wire to prevent a convex bulge. Do you have an issue with such an arrangement of rollers work hardening the copper wire? Is work hardening an issue?
I'm stopping every 10 turns of so to dress the turns by using wooden tools to tighten the space between turns & as you say smash ( very gently) , or flatten the turns being very carfeul not to damage the enamel coating.
Your using an automated winder so can hopefully achieve better turn dressing & minimise gaps between turns etc. Have you any tips for measuring ones work for calculating depth of layers/windings as a percentage for calculating fill. After 10T I used a used a digital vernier to measure the depth. Wire diameter is 1.08mm, I measured 1.2mm from the top of a turn, to the bobbin at the top of the bulge on the longest sides.
Thanks again for your IP, Andy.
I'm stopping every 10 turns of so to dress the turns by using wooden tools to tighten the space between turns & as you say smash ( very gently) , or flatten the turns being very carfeul not to damage the enamel coating.
Your using an automated winder so can hopefully achieve better turn dressing & minimise gaps between turns etc. Have you any tips for measuring ones work for calculating depth of layers/windings as a percentage for calculating fill. After 10T I used a used a digital vernier to measure the depth. Wire diameter is 1.08mm, I measured 1.2mm from the top of a turn, to the bobbin at the top of the bulge on the longest sides.
Thanks again for your IP, Andy.
A quick and dirty way of calculating fill is to add up the layers - measure wire, insulation, tape, window height etc with calipers - calculate turns per layer for each wire size and knowing number of turns calculate number of layers for each winding, rounding up to nearest whole number - use metric it is easier to work out.
An example - assume a winding window height of 9mm
Primary 5 layers of 0.5mm wire = 2.5mm
Insulation 2 layers of 0.15mm insulation = 0.3mm
Secondary 1 3 layers of 0.25mm wire = 0.75mm
Insulation 2 layers of 0.15mm insulation = 0.3mm
Secondary 2 1 layers of 0.85mm wire = 0.85mm
Insulation 2 layers of 0.15mm insulation = 0.3mm
Secondary3 1 layers of 1.25mm wire = 1.25mm
Insulation 3 layers of 0.15mm insulation = 0.45mm
Total build height = 6.7mm
Percentage fill = total build divided by window height multiply by 100
= 6.7/9*100 = 74.4%
If I was winding this then I would say it would be an easy fit in the window space
An example - assume a winding window height of 9mm
Primary 5 layers of 0.5mm wire = 2.5mm
Insulation 2 layers of 0.15mm insulation = 0.3mm
Secondary 1 3 layers of 0.25mm wire = 0.75mm
Insulation 2 layers of 0.15mm insulation = 0.3mm
Secondary 2 1 layers of 0.85mm wire = 0.85mm
Insulation 2 layers of 0.15mm insulation = 0.3mm
Secondary3 1 layers of 1.25mm wire = 1.25mm
Insulation 3 layers of 0.15mm insulation = 0.45mm
Total build height = 6.7mm
Percentage fill = total build divided by window height multiply by 100
= 6.7/9*100 = 74.4%
If I was winding this then I would say it would be an easy fit in the window space
Yes, exactly. I have no idea to what extent the roller harden the copper wire, keep in mind however, until this isn't objectively documented, it could be easily an overstatement.
Some copper wires are softer than others, for example Elektrisola wires are softer than my local class 2 enameled wire and more easy to wind. A softer wire needs less straightening. So the choice of your wire supplier is a bigger factor than the tensioner wheels configuration.
There are free access patents over annealing copper wires by electrical impulses, however it does partial annealing. Full annealing, I think, cannot be done without damaging the enamel, because it requires temperatures over 300C.
Smashing the windings does work, were the biggest difference is visible on the first layers, where the bobbin is still square. As windings progresses, turn radius increases and bowing becomes less of an issue.
An automated winder does not result automatically into gapless windings. A huge factor is also played by the winding tensioners, wire feeding towards the bobbin, precision of the tooling, etc. The only way to make a gapless winding is to verify and pack all windings by hand, which results into additional hand work and torn nails. My wife, which I taught how to wind output transformers, hates this part of the job. Using your nails for a long period of time does bring a certain "nail pain". For finer wires, one can pack windings using a "sticky" finger, by coating it with a layer of colophony for example, or use cardboard or plastic sheets.
To have near perfect windings by an automated machine, you need to ensure the following factors as perfect as possible.
1. Straight wire in the first place.
2. Straight feeding the wire towards the bobbin you're winding, which requires a long straight perpendicular wire path towards the bobbin.
3. Have a precise set of tooling as possible, mandrels and shafts. No backlash in your wire feeder. If there is backlash, I load all gears using weights or elastic bands.
4. Wire feeder as close to the bobbin as possible.
Some copper wires are softer than others, for example Elektrisola wires are softer than my local class 2 enameled wire and more easy to wind. A softer wire needs less straightening. So the choice of your wire supplier is a bigger factor than the tensioner wheels configuration.
There are free access patents over annealing copper wires by electrical impulses, however it does partial annealing. Full annealing, I think, cannot be done without damaging the enamel, because it requires temperatures over 300C.
Smashing the windings does work, were the biggest difference is visible on the first layers, where the bobbin is still square. As windings progresses, turn radius increases and bowing becomes less of an issue.
An automated winder does not result automatically into gapless windings. A huge factor is also played by the winding tensioners, wire feeding towards the bobbin, precision of the tooling, etc. The only way to make a gapless winding is to verify and pack all windings by hand, which results into additional hand work and torn nails. My wife, which I taught how to wind output transformers, hates this part of the job. Using your nails for a long period of time does bring a certain "nail pain". For finer wires, one can pack windings using a "sticky" finger, by coating it with a layer of colophony for example, or use cardboard or plastic sheets.
To have near perfect windings by an automated machine, you need to ensure the following factors as perfect as possible.
1. Straight wire in the first place.
2. Straight feeding the wire towards the bobbin you're winding, which requires a long straight perpendicular wire path towards the bobbin.
3. Have a precise set of tooling as possible, mandrels and shafts. No backlash in your wire feeder. If there is backlash, I load all gears using weights or elastic bands.
4. Wire feeder as close to the bobbin as possible.
It can be minimized with low RPM and high degree of tensioning. I use tensioner which have discrete values (level 1 and 2) and smooth with potentiometer.
For example, with 0.7 - 0.8 mm wire, very low RPM 25 - 30 rpm yields really good results.
CNC control software and photo of finished layer attached.
Attachments
An additional thought about the winding speed (RPM). The biggest factor which influences this IMHO is the response time of any wire tensioning system. With higher RPMs, the response time needed increases. Although the later isn't really necessary for round coils, it is detrimental for square coils, where the wire feeding speed spikingly changes due to the radius changing.
If the tensioner response time fails to cope with the wire deceleration, you will suddenly get a "loose" wire for a small timespan, which can be enough to break wire laying accuracy and potentially result into a crossed turn.
In past times, I would even change the bearing system of my feeding wheel to faster response, ceramic bearings, and add additional local tensioning using a fast spring. However in the end, I just blocked the feeder wheel from turning. Blocked wheel = no mass to move, hence only friction = no response time.
These sets of magnets basically clamp to the wheel and prevent it from moving. I only remove the magnets where a high gauge wire is wound with low RPM.
If the tensioner response time fails to cope with the wire deceleration, you will suddenly get a "loose" wire for a small timespan, which can be enough to break wire laying accuracy and potentially result into a crossed turn.
In past times, I would even change the bearing system of my feeding wheel to faster response, ceramic bearings, and add additional local tensioning using a fast spring. However in the end, I just blocked the feeder wheel from turning. Blocked wheel = no mass to move, hence only friction = no response time.
These sets of magnets basically clamp to the wheel and prevent it from moving. I only remove the magnets where a high gauge wire is wound with low RPM.
It doesn't take much to work harden copper, do an experiment - small diam copper tube illustrates it well but thick wire may also work ok, anneal the copper by heating to cherry red and dunk in cold water, then bend it back and forth, at each bend you can feel it getting more stiff/rigid, it only takes 4 or 5 bends to make it harder, my thoughts are that if you have a large spool from the maker then it will more than likely be reasonably annealed but if you have a smaller spool with wire that had been despooled 2 or 3 times already, much of the annealing has already been lost.
Even though my Aumann winder will do 1.5mm wire I often do wire that is over 0.7mm thick by hand to try and keep it flat, it depends on the amount of winding window fill I've calculated - if I think the windings will be a tight fit then I try to keep the windings flat as I go - rotate the bobbin and when you get to the bend hold the turn flat with one hand while holding the wire with the other as the wire bends over the bobbin, this needs a powered spindle with foot control speed which the Aumann has, the other thing I have done is to wind the complete layer and then use a clamp between 2 pieces of wood to squash the wire down flat.
Even though my Aumann winder will do 1.5mm wire I often do wire that is over 0.7mm thick by hand to try and keep it flat, it depends on the amount of winding window fill I've calculated - if I think the windings will be a tight fit then I try to keep the windings flat as I go - rotate the bobbin and when you get to the bend hold the turn flat with one hand while holding the wire with the other as the wire bends over the bobbin, this needs a powered spindle with foot control speed which the Aumann has, the other thing I have done is to wind the complete layer and then use a clamp between 2 pieces of wood to squash the wire down flat.
For already annealed copper it makes sense, however many winding wires aren't 100% annealed in the first place, they are quite hard. How harder they can go when wound? It also depends of the bending radius of the wheels. I'm a fan of low tension winding, so I tend to use the less wheels as possible.
Thanks all, some good observations there. going back to calculating window %, I've just sat down & had a go. Re rounding up to the nearest whole number. I have two wndgs where the result is 2.1 layers & 0.4 layers, in this case would one round 2.1 up to 3, or down to 2? 0.4 up or down? I figured in this case 3 layers, fitting the 0.4 in the gap.
This wind is also a nightmare in that one heater wndg has to be insulated to 1350v, it's the heaters to a CRT who's cathode is at -1350v ish. I have Nomex 4thou or 0.12mm thick, looking at the specs it should be ok. I also have to wind several high current heater windings, EG 5A & 8A, the spec calls for two wndg's in parallel, this I've found is a big eater of window space due to the room the terminations take up.
My window calcs show I should be ok, but I know from experience calculations & reality often don't dovetail. Anyroad, there's not much else I can do apart from go for it, do the best job I can & hope it turns out ok.
Thanks for all the tips on work hardening and practical tfmr winding, regards, Andy.
This wind is also a nightmare in that one heater wndg has to be insulated to 1350v, it's the heaters to a CRT who's cathode is at -1350v ish. I have Nomex 4thou or 0.12mm thick, looking at the specs it should be ok. I also have to wind several high current heater windings, EG 5A & 8A, the spec calls for two wndg's in parallel, this I've found is a big eater of window space due to the room the terminations take up.
My window calcs show I should be ok, but I know from experience calculations & reality often don't dovetail. Anyroad, there's not much else I can do apart from go for it, do the best job I can & hope it turns out ok.
Thanks for all the tips on work hardening and practical tfmr winding, regards, Andy.
Perhaps these photos will interest someone. My project to convert the Grommes 240 into TubeCAD EL84 did not work out due to the small size of the chassis. So I disassembled the power transformer in the hope that in the future I will be able to figure out how to wind a new one.)
Some interesting details:
Some interesting details:
- metal strip on top (photo 1)
- plates press laminates on both sides (photo 2 - bottom right corner).
Attachments
Is the transformer defective? From your pics it does look like it's been over heating. The metal strip is curious, looks like aliminium, I don't think it's a magnetic flux shield though as it doesn't go all the way round, saying that it could be Mu metal and may be a magnetic shunt.
The transformers look good quality, they have plenty of laminations so should make a decent amp. As for re-winding the mains tfmr you might find the video's I've done on the Tektronix tfmr rewind informative. Have you any experience of tfmr winding?
Andy.
The transformers look good quality, they have plenty of laminations so should make a decent amp. As for re-winding the mains tfmr you might find the video's I've done on the Tektronix tfmr rewind informative. Have you any experience of tfmr winding?
Andy.
Started to wind the next layer, I want to put some Nomex paper between them, as a test to see if the turns lay better. Popervitch & Wolpert in their book/PDF recommend cutting the insulation paper oversize, because, as you know the paper gets pulled inwards, leaving gaps at the sides. Gaps = no insulation.
Looking at both 50AE's & LinuksGuru's pics, I see neither of you do this. What is more confounding I notice your top layer/turns aren't spilling, dropping down onto the layer beneath.
Another issue I've found is if I make the paper over size, then it won't fit well around the corners of the bobbin, to get it to do that I have to make nicks, small cuts into the sides of the paper so that it goes round the corners neatly.
Tips? Andy.
Looking at both 50AE's & LinuksGuru's pics, I see neither of you do this. What is more confounding I notice your top layer/turns aren't spilling, dropping down onto the layer beneath.
Another issue I've found is if I make the paper over size, then it won't fit well around the corners of the bobbin, to get it to do that I have to make nicks, small cuts into the sides of the paper so that it goes round the corners neatly.
Tips? Andy.
I'm taking advantage of the paper thickness and hardness to hold the end layers, and making sure pretty much all layer are of closest to equality length. It is a primary law for lowest leakage inductance anyway.
I also let the finer wire layers slightly longer than the thicker wire layers, because winding the thicker wire layer is better supported by a longer bottom finer wire layer. And thicker wire has a ; larger chance to fall.
I'm using 0.1mm paper for wires up to 0.5mm, and 0.2mm paper from 0.55 to 1.12mm.
The more the corners are sharp, as they are in the beginning of winding, the highest chance for the top winding to crush the dielectric and fall down, at the corner. The pressure is highest there and it is good practice to lower winding RPM to feed the layer length more gently.
I also let the finer wire layers slightly longer than the thicker wire layers, because winding the thicker wire layer is better supported by a longer bottom finer wire layer. And thicker wire has a ; larger chance to fall.
I'm using 0.1mm paper for wires up to 0.5mm, and 0.2mm paper from 0.55 to 1.12mm.
The more the corners are sharp, as they are in the beginning of winding, the highest chance for the top winding to crush the dielectric and fall down, at the corner. The pressure is highest there and it is good practice to lower winding RPM to feed the layer length more gently.
I use thick adhesive fiberglass tape around naked coil to get rid of sharp corners.
And yes, low RPM for the 1st layer.
Thanks for both your responses, your thoughts are invaluable. The big difference though is you can design your build before hand & adjust wire gauge, core size etc to a point, I'm stuck with what I have, winding a mains tfmr & wan't to use every bit of the window. Not grumbling though, practical winding methods are applicable to both mains and audio tfmrs.
I rounded off all the corners of the bobbin before winding. Looking back at both of your work, I notice neither of you wind right up to the sides, is that because it will put stress on the bobbins they will fail? Having done a few test winds this afternoon I sussed that at the start of a wind it's good to start with less tension.
I wound another layer of the primary today after lots of tests trying with/without Nomex etc. Good point re thicker wire needs heavier insulation. The turn count per layer was consistent and near calculations, as was the depth of the wind so far.
Thanks again, Andy.
I would have thought the opposite was true? I have seen a toroid mains tfmr with an open secondary of wire 1mm plus, but that was due to a construction fault. Most failed tfmr's I've dissected have had the thinner windings burnt & open circuit or under engineered.
I rounded off all the corners of the bobbin before winding. Looking back at both of your work, I notice neither of you wind right up to the sides, is that because it will put stress on the bobbins they will fail? Having done a few test winds this afternoon I sussed that at the start of a wind it's good to start with less tension.
I wound another layer of the primary today after lots of tests trying with/without Nomex etc. Good point re thicker wire needs heavier insulation. The turn count per layer was consistent and near calculations, as was the depth of the wind so far.
Thanks again, Andy.
Last edited:
Several years ago with support from the local DIY community and friends, I build Amity amplifier. Excellent design by Lynn Olson. A very good and skilled friend did the assembling and after many hours of work it was completed. Amity was soo good in combination with Wright's RTP3 that a friend ''stole'' it from me. Without Amity I start thinking what to do next. During Covid period (again with the help of another friend) I complete UGS UP amp and still enjoy it the same way I enjoy Amity. It's really crazy how UGS UP in feedback-less mode is good match with Tannoy Kensington's.
Still, I want to have another Amity amp. So now, when i know for sure Amity concept work, I decided to go completely crazy and try to finish Amity DHT or Karna. 90% it will be Amity DHT since I don't need extra gain of input tube and one transformer less can be only good thing.
In the process of collecting parts, one huge problem appear: transformers. Not really and issue for output, but interstage was a huge challenge. I made inquiry first with Monolith magnetics. From some reason it didn't work. Communication was very slow and I simply give up. One problem was interstage. They don't have suitable IT in their portfolio.
Thanks to this diyA, i found that there's other alternatives, one very highly rated was Muse Coils. So I contacted Alexander and everything goes super smoothly. He wound me nano-crystalline IT and Amorphous core output transformer. Needless to say that they are peace of art. OPT is 6kg a peace. Can't wait to start assembling this think. 8 Rod Colemans reg's, separate PS for 2a3 and 300B...4 box beast that will be. Still waiting for my brother to finish power transformers and chokes (all on double C cores, 8 extra irons). In fact I'm missing chokes for 2a3 so Alexander already offered help. Thanks in advance
With permission of Alexander I can send some measured data of transformers (or he can do it). But I'm very satisfied so far with experience of working with him. And hope soon I will comment sound of Minotaur series transformers build by Alexander from Muse coils.
Igor
Still, I want to have another Amity amp. So now, when i know for sure Amity concept work, I decided to go completely crazy and try to finish Amity DHT or Karna. 90% it will be Amity DHT since I don't need extra gain of input tube and one transformer less can be only good thing.
In the process of collecting parts, one huge problem appear: transformers. Not really and issue for output, but interstage was a huge challenge. I made inquiry first with Monolith magnetics. From some reason it didn't work. Communication was very slow and I simply give up. One problem was interstage. They don't have suitable IT in their portfolio.
Thanks to this diyA, i found that there's other alternatives, one very highly rated was Muse Coils. So I contacted Alexander and everything goes super smoothly. He wound me nano-crystalline IT and Amorphous core output transformer. Needless to say that they are peace of art. OPT is 6kg a peace. Can't wait to start assembling this think. 8 Rod Colemans reg's, separate PS for 2a3 and 300B...4 box beast that will be. Still waiting for my brother to finish power transformers and chokes (all on double C cores, 8 extra irons). In fact I'm missing chokes for 2a3 so Alexander already offered help. Thanks in advance
With permission of Alexander I can send some measured data of transformers (or he can do it). But I'm very satisfied so far with experience of working with him. And hope soon I will comment sound of Minotaur series transformers build by Alexander from Muse coils.
Igor
