Ive been playing around with OPT_da to come up with the specs for winding an OT for a pair of 807 tubes running in UL in class AB1.
I have been reading the guide after translation, but feel I have some variables wrong.
Firstly, i am after the primary plate resistance (Rp) the datasheet for 807 only shows load resistance, so I just went with 40K as used on a pair of 6L6.
Im not sure how much this differs with an 807, but it must be close.
I am also not too sure on what core size to select for a given wattage, and i may have this wrong what ive selected, but i went with EI78 as i feel that looks the right size for other transformers ive seen rated at similar power.
The other thing I dont see in this software is how many layers thick of laminations that I need stacked.
Anyway, I have the majority of lamination sizes available along with copper wire.
Personally, i get the feeling that the number of turns on the primary is overly high, but i could be wrong.
See chart below.
TIA 🙂
I have been reading the guide after translation, but feel I have some variables wrong.
Firstly, i am after the primary plate resistance (Rp) the datasheet for 807 only shows load resistance, so I just went with 40K as used on a pair of 6L6.
Im not sure how much this differs with an 807, but it must be close.
I am also not too sure on what core size to select for a given wattage, and i may have this wrong what ive selected, but i went with EI78 as i feel that looks the right size for other transformers ive seen rated at similar power.
The other thing I dont see in this software is how many layers thick of laminations that I need stacked.
Anyway, I have the majority of lamination sizes available along with copper wire.
Personally, i get the feeling that the number of turns on the primary is overly high, but i could be wrong.
See chart below.
TIA 🙂
Here's some information that might be useful when designing UL output transformers:
http://www.r-type.org/articles/art-140.htm
http://www.r-type.org/articles/art-140.htm
Yes that looks about right for a UL transformer, the direction is different than I would have done it however, looks like the centre tap on this is split and each side of it goes to the start and end of the winding.
It should make little difference. The load impedance dominates. Try 20k and 80k. If they come up <10% different in Turns or in Hz, then you don't need a real good guess. Most of the classic audio power pentodes aimed at Rp=10*Rl because higher cost more to make and lower hurt power/gain, but it is a very broad optimum.
Ok, that's helpful thanks.
That just leaves me with what core to pick from really.
Have I selected too small a core?
I dont see anywhere in the list that mentions core stack thickness. It appears to be calculating core weight which I guess could be measured on scales, but not sure if this is how it's done. What if it's too much(or less) to fit the bobbin? I don't see anywhere to specify bobbin size, but it appears to be calculating bobbin depth.
I assume this refers to the total thickness of all the windings on the bobbin?
What exactly is the total hcu measurement in mm?
Does anyone know how accurate this program is? Would be interested to hear what results people have had with it.
That just leaves me with what core to pick from really.
Have I selected too small a core?
I dont see anywhere in the list that mentions core stack thickness. It appears to be calculating core weight which I guess could be measured on scales, but not sure if this is how it's done. What if it's too much(or less) to fit the bobbin? I don't see anywhere to specify bobbin size, but it appears to be calculating bobbin depth.
I assume this refers to the total thickness of all the windings on the bobbin?
What exactly is the total hcu measurement in mm?
Does anyone know how accurate this program is? Would be interested to hear what results people have had with it.
The center leg of EI78 lams is 2.6 cm wide. To achieve an iron cross section of 6.4 cm², you need to stack it 2.5 cm high.
Anyway, this core appears to be too small wrt to 30W @ 20 Hz...
Best regards!
Anyway, this core appears to be too small wrt to 30W @ 20 Hz...
Best regards!
Ok this is what I was unsure of.
I think an EI78 core is 78mm x 60mm or so.
Your probably right, that does look a bit small.
What size should I be going for ideally?
I think an EI78 core is 78mm x 60mm or so.
Your probably right, that does look a bit small.
What size should I be going for ideally?
A set of EI78 laminations measures 78 x 65 mm. The window is 39 x 13 mm. Stacked to a height of 25 mm, resulting in an iron cross section of 6.4 cm² as seen in your screenshot, they might do 30 VA in a mains transformer @ 50 Hz or 20 % more @ 60 Hz, but surely not 30 W as an output tranny @ 20 Hz. Even not if they were M6X or some better GOSS, what never was the case afaik.
For 30 W @ 20 Hz you'll need an iron cross section of 15 cm² or a bit less with better material. Or you'd go for C or double C cores.
Best regards!
For 30 W @ 20 Hz you'll need an iron cross section of 15 cm² or a bit less with better material. Or you'd go for C or double C cores.
Best regards!
As Kay Pirinha says, that core is way too small for 30W @ 20Hz. I doubt that with more than 8000 primary turns it will make already 10 KHz decently, in fact I read more than 400 mH leakage inductance! That's enormous. You should aim at something like 10 mH or less (interleaving helps). Anyway, the transformer in that sheet is quite impossible to make because if try to put 8000+ turns onto that small EI78 with 0.28 mm (0.334 mm with double insulation) copper wire you will have no space left for the secondary before you finish the primary! Best efficiency is AWALYS achieved when half of the winding space is used for the primary and half for the secondary, hence you have to pick the right core size to meet the power rating.
In general, you can already assess that such enormous number of primary turns will make the transformer way inefficient for its task. You should always aim at 88-89% efficiency at least. Better if more. If you ignore this and accept such low efficiency, having a massive DC resistance in series will have a negative effect on the sound. Guaranteed.
Because winding a transformer properly takes quite some time, good quality materials have become more expensive, I say that is not worth doing if there is such big compromise.
For good performance you need at least a core weight around 2.5 Kg. EI96 (metric) laminations with cross-section of 32x50 mm is the minimum you should use, IMHO.
For 30W @ 20Hz and Bmax=1T (to stay safely away from saturation and leave some headroom for little DC unbalance) you will "only" need to wind 3400 primary turns. That's quite a big difference. Hope that helps.
In general, you can already assess that such enormous number of primary turns will make the transformer way inefficient for its task. You should always aim at 88-89% efficiency at least. Better if more. If you ignore this and accept such low efficiency, having a massive DC resistance in series will have a negative effect on the sound. Guaranteed.
Because winding a transformer properly takes quite some time, good quality materials have become more expensive, I say that is not worth doing if there is such big compromise.
For good performance you need at least a core weight around 2.5 Kg. EI96 (metric) laminations with cross-section of 32x50 mm is the minimum you should use, IMHO.
For 30W @ 20Hz and Bmax=1T (to stay safely away from saturation and leave some headroom for little DC unbalance) you will "only" need to wind 3400 primary turns. That's quite a big difference. Hope that helps.
Last edited:
I suspect some of those red numbers mean something is wrong already. You can see that you are way off the max height (Hcu) allowed by that small bobbin.
P.S.
0.31 mm total wire diameter (0.315 mm to be more precise) means that it's using single insulation layer. meh!😉
Last edited:
Yes, I didn't have EI96 in mind, but with an iron cross section of 16 cm² it would do.
@nzoomed: Note that interleaving consumes much of the winding window for increased insulation. Hence, there's a demand for the next bigger core not only for lower Bmax reasons.
Due to the bias plate current, I'd go for a bit more than 50 % of the winding space for the primary.
Best regards!
@nzoomed: Note that interleaving consumes much of the winding window for increased insulation. Hence, there's a demand for the next bigger core not only for lower Bmax reasons.
Due to the bias plate current, I'd go for a bit more than 50 % of the winding space for the primary.
Best regards!
It didn't occur to me that some cores could be imperial, makes total sense though, im totally forgetting that most of this stuff was made prior to metric.
Could complicate things if I need to duplicate an existing transformer I've got that is potentially using imperial laminations. If I go to the nearest size in metric would it typically be too detrimental? I guess it would be sager to go slightly larger in core.
This explains why I couldn't find bobbins the right size when I was rewinding a transformer on an old philips radio.
It must have had an imperial core. I ended up 3d printing a bobbin.
@Kay Pirinha
I'm pretty sure I have plenty of ei96 available.
I need to take a good look at my inventory here, but I've got enough materials to make me a lifetime supply of transformers.
I believe there is no harm in having a larger core, it's probably safer to have more iron than not enough and risk over saturating it.
Your right about interleaving taking up more space, since it's push pull I won't be needing air gaps, so that helps.
I've pulled apart a commercially made transformer and it only has 3 sections of primary with the secondary split into two.
This seems to be the most common layout.
The thing that gets me is the layers of paper between each layer of winding.
Is this necessary for an OPT? I think its only there to serve as extra insulation between the wire as I believe the enamel on older wire was not as good as the more modern stuff.
I dont know if this is done with more modern transformers, but its probably a good thing if there is less space taken up on the bobbin with paper and replaced with wire instead?
It's a long time since then, but when I wound my OT's, I wound it meticuously turn by turn per each layer and insulated with thin foil/paper between each layer. Insulation between primary and secondary winding sections needs two or better three layers of that foil/paper.
You're right with that P-S-P-S-P interleaving arrangement. Divide the primary into four and make the middle part two times the turns count of the inner and outer parts (i. e. half the total count). This allows you to get the same ohmic resistance for both primary halves, if that counts: Connect the inner and outer P part in series. The beginning needs to be connected with the end of the middle part and becomes CT.
Best regards!
You're right with that P-S-P-S-P interleaving arrangement. Divide the primary into four and make the middle part two times the turns count of the inner and outer parts (i. e. half the total count). This allows you to get the same ohmic resistance for both primary halves, if that counts: Connect the inner and outer P part in series. The beginning needs to be connected with the end of the middle part and becomes CT.
Best regards!
Ok that seems easy enough, iirc, I think pat turner says on his guide that he used .05mm paper between windings and .5mm between primary and secondary.
I've just been examining a whopping big transformer rated at 30w with similar specs to what i need and the laminations measure 115 high, 95 wide and stacked about 75mm deep! This thing is bigger than the output transformer I had wound for my 100w guitar amp!
I've also got a 12w transformer that measures 102x89mm can't remember how deep off the top of my head though.
I am looking at doing non destructive tests to work out the windings. I've managed to make measurements by winding 10 turns of fine wire around the core, but I believe you can also do it with an LCR meter and measure the inductance on each winding.
I've just been examining a whopping big transformer rated at 30w with similar specs to what i need and the laminations measure 115 high, 95 wide and stacked about 75mm deep! This thing is bigger than the output transformer I had wound for my 100w guitar amp!
I've also got a 12w transformer that measures 102x89mm can't remember how deep off the top of my head though.
I am looking at doing non destructive tests to work out the windings. I've managed to make measurements by winding 10 turns of fine wire around the core, but I believe you can also do it with an LCR meter and measure the inductance on each winding.
Do you really need full power down to 20Hz? If you required 30W at 28Hz, you would get power bandwidth (i.e. -3 dB at the same max induction) at 20Hz. If you don't listen to pipe organs all the time then it's something to take into serious consideration as you will get advantages in all other areas, starting from the number of turns you need and wire size on the very same core. For the 32x50 core you would only need 2430 turns instead of 3400. Quite a difference....
Besides, for better overall behaviour I would discard P-S-P-S-P configuration. It's not enough for a 7K well behaved transformer. 1/2P-S-P-S-P-S-1/2P would be significantly better.
If you use premium insulators like Nomex 410 instead of generic paper (for transformer use), you won't need 0.5 mm. About half of that will be fine.
https://www.gbstape.com/dupont-nome...MI_LWMsau9_QIVtGDmCh0imwkOEAAYAiAAEgIUT_D_BwE
Besides, for better overall behaviour I would discard P-S-P-S-P configuration. It's not enough for a 7K well behaved transformer. 1/2P-S-P-S-P-S-1/2P would be significantly better.
If you use premium insulators like Nomex 410 instead of generic paper (for transformer use), you won't need 0.5 mm. About half of that will be fine.
https://www.gbstape.com/dupont-nome...MI_LWMsau9_QIVtGDmCh0imwkOEAAYAiAAEgIUT_D_BwE
Yes, better interleaving will reduce stray inductance furthermore. But I think it's hard to get equal DC resistances for both halves with this arrangement, if that matters at all.
What about dividing the primary into four equal sections? P-S-P-S-P-S-P?
German musical equipment manufacturers Dynacord and Klemt, maybe even more of them, used to divide their primaries into six equal sections, with five secondaries interleaved.
Best regards!
If you make the external primaries equal to the internal ones, will get more leakage inductance. The reason is that the inner primaries are sandwiched between two secondaries while the external primaries only "see" one secondary. So if the external primaries have 1/2 turns the magnetomotive force is heavenly distributed and get lower losses, generally speaking 10-15% more leakage if make primaries all the same.
I just thought 20hz was the common range, i see a few that are marked 20-20,000hz
I will look into that, either way, the amount of wire used is not too much concern to me as I have so much of the stuff.
Does a 1/2P-S-P-S-P-S-1/2P configuration mean that the two primary sections in the middle are the second half split into 2 giving 2 quarters?
Something else I was going to ask is regarding the ultralinear taps, is the ~40% the percentage of the number of turns, or the impedance (or inductance perhaps)?
Ive made measurements when testing a transformer, and found that the voltage was only 6.2V between the CT and the UL tap, and 18.6V between the UL tap and the end of the winding. Was reading 24.9V across the CT and the end, so 40% as far as voltage goes should have been close to 10V, now measuring with a multimeter, it was reading similar ratios as far as resistance went too.
Regarding the secondary, lets say i keep 8 ohms, is the 4 ohm tap at 50% the number of turns?
I have a transformer that has 2x 4 ohm secondaries and you either wire them in series to get 8 ohms or parallel to get 4.
Im assuming both windings are the same number of turns, but was told it doesnt work that way.
OK i might look into this, gives me a bit to think about!