Hello everyone,
I'm planning to make PP output transformer for UL KT88 amp. Nothing new, +470-480V B+, UL, around 4.5k Ra-a.
What I have at hands now - M6 EI 120-53 lams, split bobbin.
Windings in one bobbin's half - 1/2 P-S-P-S-P-S-1/2P (4 primary, 3 secondary)
where 1/2P is 3 layers (74 turn each) and P - 6 layers (74 turns)
Secondary(S) is 4 layers: 2 layers(114 turns)+2 layers(114 turns) in parallel = 114 turns in one section
All secondary sections also in parallel (both halves of bobbin)
Overall - 2664 turns primary, 114 sec. , so Ra-a is ~4.4k (without considering DC resistance, just turns ratio). Other data - ~160 Ohm primary(80+80), ~0.32 Ohm secondary.
Main question - will it be OK as is, or cross-connection of primary windings is necessary? Any advice would be appreciated.
I'm planning to make PP output transformer for UL KT88 amp. Nothing new, +470-480V B+, UL, around 4.5k Ra-a.
What I have at hands now - M6 EI 120-53 lams, split bobbin.
Windings in one bobbin's half - 1/2 P-S-P-S-P-S-1/2P (4 primary, 3 secondary)
where 1/2P is 3 layers (74 turn each) and P - 6 layers (74 turns)
Secondary(S) is 4 layers: 2 layers(114 turns)+2 layers(114 turns) in parallel = 114 turns in one section
All secondary sections also in parallel (both halves of bobbin)
Overall - 2664 turns primary, 114 sec. , so Ra-a is ~4.4k (without considering DC resistance, just turns ratio). Other data - ~160 Ohm primary(80+80), ~0.32 Ohm secondary.
Main question - will it be OK as is, or cross-connection of primary windings is necessary? Any advice would be appreciated.
Attachments
Last edited:
Cross connecting the primary parts surely will minimize leakage inductance without increasing inter and intra winding capacitances. So go for it, but be aware of correct phasing.
Best regards!
Best regards!
Great, thank you.
If we have 3 primaries, cross connecting seems pretty straightforward (just cross-connect middle sections with right phasing), but I'm planning 4 primaries - so, which option is better - 1 or 2?
Of course, I can test it in practice, but always good to hear feedback from more experienced colleagues. I've made many SE amps and few class A PP's in past 20 years, but UL transformer with this windings layout and class AB is something new for me.
If we have 3 primaries, cross connecting seems pretty straightforward (just cross-connect middle sections with right phasing), but I'm planning 4 primaries - so, which option is better - 1 or 2?
Of course, I can test it in practice, but always good to hear feedback from more experienced colleagues. I've made many SE amps and few class A PP's in past 20 years, but UL transformer with this windings layout and class AB is something new for me.
Attachments
Cross connecting the primary parts surely will minimize leakage inductance without increasing inter and intra winding capacitances. So go for it, but be aware of correct phasing.
Best regards!
Not necessarily it increases capacitance. In this case, voltage gradients from primary to secondary do not change.
However it does decrease leakage that happens of the transition from class A to class B operation, when only one primary effective half remains active.
Cross-coupling is recommended for class A-B operation.
FEV, any particular reason your anode layers are on the top? Do you want to minimize anode to core to anode capacitance?
Here's how to cross-couple:
Assuming the secondary is at zero ground potential, all capacitances remain the same.
Cross-coupling becomes symmetrical with an odd number of primary layers.
Your sketch of number two will give higher leakage inductance IMHO, because one half has a lower primary to secondary turns than the other. 2 layers of 1/2 P on 3S vs. 2 layers of 1P vs 3 S
Last edited:
Yes, from my previous expirience with SE OTs it minimize capacitance and show better frequency response (but it were SE and non-split bobbins...)FEV, any particular reason your anode layers are on the top? Do you want to minimize anode to core to anode capacitance?
And hence my questions here... I "feel" that my windings layout may be non-optimal or plain wrong for UL AB amp.Cross-coupling becomes symmetrical with an odd number of primary layers.
Maybe I shoud go for 4 secondaries with 3 primaries in beetween?
Thanks alot for your help!
https://www.diyaudio.com/community/...without-global-nfb.384100/page-5#post-6991884
"If the bobbin has dual or more chambers, connect parallel one coil from each chamber.
Each individual secondary coil must be surrounded (in a sandwich fashion) by two primary windings in the same chamber."
"If the bobbin has dual or more chambers, connect parallel one coil from each chamber.
Each individual secondary coil must be surrounded (in a sandwich fashion) by two primary windings in the same chamber."
I totally agree with you. Did I write something else?Not necessarily it increases capacitance. In this case, voltage gradients from primary to secondary do not change.
Why? With even numbers of primary sections they can be distributed in the two bobbin halves with equal numbers anyway.Cross-coupling becomes symmetrical with an odd number of primary layers.
Best regards!
My bad, Kay. Didn't have my cup of coffee.
As for the symmetry, I meant for the vertical axis only geometry. You have a point, that with an odd primary turn number, the equal turn rule to secondary will not be satisfied. With three internal primary layers for example, that would translate to one primary active layers on the one side and two active layers on the other.
In this PP case, I find it can be worth starting the anodes at the top, because capacitance between the two A-A regions and the core will be high, 4 times compared to one A region vs ground capacitance. Especially if using a tight, thin coil former.
However you will still have some A-A capacitance, depending of your filling factor, the distance between the outside layers and the core legs.
A-A capacitance is located into the primary region only and not much evil compared to a primary to secondary capacitance. It will effect the peaking (series RLC) resonance, that is easily dampened by load.
Primary to secondary capacitance can bring much more chaos, especially if you series connect the wrong secondary layers. But it seems you are using all secondaries in parallel and that acts as self-screening shunt.
The most capacitance is located however between the primary and secondary layers, because most voltage gradient difference and physical closeness are there.
In a SE transformer still, I see no reason to start the anode connection at the top. It's slightly beneficial to begin at the bottom, because of the lower Mean Turn length, hence less surface area and less capacitance to the following secondary layers.
You're good to go for cross-coupling with the first schematic.
As for the symmetry, I meant for the vertical axis only geometry. You have a point, that with an odd primary turn number, the equal turn rule to secondary will not be satisfied. With three internal primary layers for example, that would translate to one primary active layers on the one side and two active layers on the other.
Yes, from my previous expirience with SE OTs it minimize capacitance and show better frequency response (but it were SE and non-split bobbins...)
And hence my questions here... I "feel" that my windings layout may be non-optimal or plain wrong for UL AB amp.
Maybe I shoud go for 4 secondaries with 3 primaries in beetween?
Thanks alot for your help!
In this PP case, I find it can be worth starting the anodes at the top, because capacitance between the two A-A regions and the core will be high, 4 times compared to one A region vs ground capacitance. Especially if using a tight, thin coil former.
However you will still have some A-A capacitance, depending of your filling factor, the distance between the outside layers and the core legs.
A-A capacitance is located into the primary region only and not much evil compared to a primary to secondary capacitance. It will effect the peaking (series RLC) resonance, that is easily dampened by load.
Primary to secondary capacitance can bring much more chaos, especially if you series connect the wrong secondary layers. But it seems you are using all secondaries in parallel and that acts as self-screening shunt.
The most capacitance is located however between the primary and secondary layers, because most voltage gradient difference and physical closeness are there.
In a SE transformer still, I see no reason to start the anode connection at the top. It's slightly beneficial to begin at the bottom, because of the lower Mean Turn length, hence less surface area and less capacitance to the following secondary layers.
You're good to go for cross-coupling with the first schematic.
Last edited:
In my experience, the coupling factor k between the primary and secondary should be as high as possible (ideally k=1). This will ensure the least stray inductance. Stray inductance can be measured on the primary side when the secondary is short circuited. Stray inductance forms a resonant circuit with stray capacitance at high frequency, ideally well beyond the audio range. 100 kHz is OK. Hence my observation, that each secondary should be between two primary windings of the same tube half, on the same section of the transformer. Stray inductance is our biggest enemy. Stray capacitance is decided at the design phase by deciding on how many primary windings will be used, e.g. 3 primaries in each section, 2 secondaries in between them.
You're right with your statement about noxious stray inductance. Anyway, in an output transformer that is dedicated for class AB or B operation, stray inductance between both primary halves needs to be minimized as well. This can be done as shown by 50AE in his coloured scetch (#4). Otherwise we'll increase the xover switching transients. For a class A PP OPT each primary half could be wound into one bobbin chamber.
Best regards!
Best regards!
@lcsaszar , I find that leakage inductance or shunt capacitance priorities depend on the transformer's application. Leakage inductance on its own only causes HF smooth first order roll-off. However, when combined with the (wrong) distribution of capacitance, it will resonate, sometimes into bad frequency regions.
When it comes to resonances, there comes a lot of confusion, especially when trying to find an explanation how the observed resonance is formed.
I've observed many cases where a very low resonance, audio band dipping resonance can be observed at specific interleaving geometries, despite the main calculated leakage inductance being lower. I believe additional leakage inductances of higher values potentially exist within a transformer. For example, physically distant primary layers can form a resonant bridge via capacitive difference between secondaries connected in series. I believe this capacitance bridge activates a new potential leakage inductance between both primary halves, which can be much bigger than the main transformer Ls. Empirically, I have invented a few principles to minimize, what I call, tertiary leakage inductance.
-Keep secondary layers capacitively bridging far distant primary layers paralleled. This acts as a capacitive shunt, I call it self-screening.
-You can series connect secondaries if capacitance difference between the layers is equal.
-Guarantee equal voltage potential of all secondary layers related to the primary. This is achieved via parallelling all layers, or using screens between layers. The screens do not need to be necessarily grounded, but paralleled.
-Reduce the voltage gradient between primary to secondary layers. This is the hardest to do at anode potential layers. It is easiest to achieve with no interleaving, where the primary ot secondary package is reverse wound. But gets harder to impossible with more interleaves. I am using a capacitance dumping technique, which involves splitting the biggest anode potential primary chunk into smaller chunks, then starting the anode layers from inside the chunk to the outside. That dumps capacitance into the primary region, where it is much less offensive.
A year ago,I built two line-out education purpose transformers to demonstrate this, that are deliberately high-leakage transformers (70mH), involving S-P-S interleaving. Three layers of secondary, 22 layers of primary, then another three secondaries.
The first one involves a normal winding pattern, where the anode starts at the bottom, close to the S layer. The second transformer has, I call it Snail winding, where the primary chunk anode ending begins within the middle of the chunk, then progressively advances to the B+ zero potential end towards the secondary layers. Back and forth from the middle to the ends. It is not an entirely free lunch, as the capacitance is dumped into the primary region, however you can observe it is much less offensive there, judging by the frequency response comparison.
When it comes to resonances, there comes a lot of confusion, especially when trying to find an explanation how the observed resonance is formed.
I've observed many cases where a very low resonance, audio band dipping resonance can be observed at specific interleaving geometries, despite the main calculated leakage inductance being lower. I believe additional leakage inductances of higher values potentially exist within a transformer. For example, physically distant primary layers can form a resonant bridge via capacitive difference between secondaries connected in series. I believe this capacitance bridge activates a new potential leakage inductance between both primary halves, which can be much bigger than the main transformer Ls. Empirically, I have invented a few principles to minimize, what I call, tertiary leakage inductance.
-Keep secondary layers capacitively bridging far distant primary layers paralleled. This acts as a capacitive shunt, I call it self-screening.
-You can series connect secondaries if capacitance difference between the layers is equal.
-Guarantee equal voltage potential of all secondary layers related to the primary. This is achieved via parallelling all layers, or using screens between layers. The screens do not need to be necessarily grounded, but paralleled.
-Reduce the voltage gradient between primary to secondary layers. This is the hardest to do at anode potential layers. It is easiest to achieve with no interleaving, where the primary ot secondary package is reverse wound. But gets harder to impossible with more interleaves. I am using a capacitance dumping technique, which involves splitting the biggest anode potential primary chunk into smaller chunks, then starting the anode layers from inside the chunk to the outside. That dumps capacitance into the primary region, where it is much less offensive.
A year ago,I built two line-out education purpose transformers to demonstrate this, that are deliberately high-leakage transformers (70mH), involving S-P-S interleaving. Three layers of secondary, 22 layers of primary, then another three secondaries.
The first one involves a normal winding pattern, where the anode starts at the bottom, close to the S layer. The second transformer has, I call it Snail winding, where the primary chunk anode ending begins within the middle of the chunk, then progressively advances to the B+ zero potential end towards the secondary layers. Back and forth from the middle to the ends. It is not an entirely free lunch, as the capacitance is dumped into the primary region, however you can observe it is much less offensive there, judging by the frequency response comparison.
- Home
- Amplifiers
- Tubes / Valves
- PP output transformer sanity check