In a output transformer we want to have the highest inductance/leakeage inductance ratio.
I would have thougth that a shell type transformer with quadratic centerleg would give the highest ratio because of highést Afe/wirelength, but now i stumbled over this artikle from Crowhurst.
Can some one smarter than me figure out what those optimum ratios are, and why they are referenced to window heigth?
A/B = window length /window heigth
C/B = centerleg width/window heigth
D/B = stack thickness/window heigth
Thank you
I would have thougth that a shell type transformer with quadratic centerleg would give the highest ratio because of highést Afe/wirelength, but now i stumbled over this artikle from Crowhurst.
Can some one smarter than me figure out what those optimum ratios are, and why they are referenced to window heigth?
A/B = window length /window heigth
C/B = centerleg width/window heigth
D/B = stack thickness/window heigth
Thank you
Last edited:
It doesn't make much sense, at least for me, to think in terms or primary inductance to leakage inductance ratio.
For a PP transformer, the primary inductance limiting parameter is mostly your core permeability and turn count. If zero magnetizing DC current exists, in theory inductance can safely go to infinity. In practice there is always a bit of DC imbalance between the PP tubes, so there is a bit of headroom allowed. Some designers gap their PP transformer because of this reason.
For a SE transformer, you need to take off permeability anyway in order to allow for a bigger amount DC magnetizing current. This is done by introducing a bigger airgap or using a distributed gap low permeability core.
When designing an OPT, you have to usually start with
1. PP or SE application.
2. Power output for the lowest frequency
3. Primary inductance vs Idc current (SE)
4. Maximum allowed leakage inductance and shunt capacitance for your application.
5. Maximum amount of losses.
It's a huge back-and forth process, especially for beginners. With some practice, you'll get a feeling which interleaving type suits your application. You'll have to do a few calculation trials with some variety of calculation of different leakage inductance and capacitance vs the chosen number of turns and core area. The bigger the core, the better, if you can handle the size, cost and weight. As a bigger core allows for less turns for the same power required.
Some interleaving configurations do not tolerate a high amount of primary turns due to potential dip resonances getting close to the audio band. Although I've designed output transformers with 9k primary turns without issues, I wouldn't recommend more than 2-3k primary turns for beginners, until you get a feeling of how distributing the capacitance in different interleaving configurations work.
SE transformers are harder, not only due to the magnetizing current eating your flux density headroom, but because of the uneven capacitance distribution among primary to secondary layers, which can bring troublesome resonances if not studied well. Unfortunately it is not simple to explain within one page of words.
SE, PP class A vs PP class B OPT can differ very much in the interleaving strategy for optimal work. I usually design my PP transformers for worst case duty, class B, where one primary package shuts off during the operation cycle.
For a PP transformer, the primary inductance limiting parameter is mostly your core permeability and turn count. If zero magnetizing DC current exists, in theory inductance can safely go to infinity. In practice there is always a bit of DC imbalance between the PP tubes, so there is a bit of headroom allowed. Some designers gap their PP transformer because of this reason.
For a SE transformer, you need to take off permeability anyway in order to allow for a bigger amount DC magnetizing current. This is done by introducing a bigger airgap or using a distributed gap low permeability core.
When designing an OPT, you have to usually start with
1. PP or SE application.
2. Power output for the lowest frequency
3. Primary inductance vs Idc current (SE)
4. Maximum allowed leakage inductance and shunt capacitance for your application.
5. Maximum amount of losses.
It's a huge back-and forth process, especially for beginners. With some practice, you'll get a feeling which interleaving type suits your application. You'll have to do a few calculation trials with some variety of calculation of different leakage inductance and capacitance vs the chosen number of turns and core area. The bigger the core, the better, if you can handle the size, cost and weight. As a bigger core allows for less turns for the same power required.
Some interleaving configurations do not tolerate a high amount of primary turns due to potential dip resonances getting close to the audio band. Although I've designed output transformers with 9k primary turns without issues, I wouldn't recommend more than 2-3k primary turns for beginners, until you get a feeling of how distributing the capacitance in different interleaving configurations work.
SE transformers are harder, not only due to the magnetizing current eating your flux density headroom, but because of the uneven capacitance distribution among primary to secondary layers, which can bring troublesome resonances if not studied well. Unfortunately it is not simple to explain within one page of words.
SE, PP class A vs PP class B OPT can differ very much in the interleaving strategy for optimal work. I usually design my PP transformers for worst case duty, class B, where one primary package shuts off during the operation cycle.
Last edited:
Hi 50AE, .
After stumbling over that Crowhurst paper I find ,that the core dimensionratio that can give the highest ratio of primary inductance to leakage inductance is of prime importabce, at least to me.
Why should i blindly sidestep the obvious advanteges that translate to wider bandwidth, lower distortion a.s.o.?
Gapped or ungapped core, in both cases i want to maximize low end inductance (thats why a dc premagn core gets gapped in the first place)
At the same time, i want to get away with the minimum leakage inductance and if a certain core dimension ratio offers an advantage i want a core
that gets as close as possible to the ideal.
Anyway, unable to figure em out myself, i only asked and want to know the optimum dimension ratios (Crowhurst attachment #1).
and to understand the reasoning behind those ratios.
Thank you for your reply
B.t.w, did you receive my email?
After stumbling over that Crowhurst paper I find ,that the core dimensionratio that can give the highest ratio of primary inductance to leakage inductance is of prime importabce, at least to me.
Why should i blindly sidestep the obvious advanteges that translate to wider bandwidth, lower distortion a.s.o.?
Gapped or ungapped core, in both cases i want to maximize low end inductance (thats why a dc premagn core gets gapped in the first place)
At the same time, i want to get away with the minimum leakage inductance and if a certain core dimension ratio offers an advantage i want a core
that gets as close as possible to the ideal.
Anyway, unable to figure em out myself, i only asked and want to know the optimum dimension ratios (Crowhurst attachment #1).
and to understand the reasoning behind those ratios.
Thank you for your reply
B.t.w, did you receive my email?
Last edited:
The easy and most hermetic answer to that is: always use the smallest core that can do the job..
One reason is that leakage inductance can always be reduced for a given primary inductance by increasing the interleaving. The same can be done for capacitance with vertical sectioning as shown in the article. At the end of the day this is not an endless process because the more interleaving/sectioning the more the additional insulation will rob space. So the best ratio between primary and leakage inductance may exist on its own but does not necessarily turns into the best solution because efficiency (low DC resistance) is a primary parameter too. A transformer is ALWAYS a compromise among many factors with opposite requirements.
In my experience interleaving with 4P and 3S to 6P and 5S is the optimal range for 99% of OTPs. Less is not really good (good for guitar OPTs , in general), more will rob too much space. Horizonal sectioning, none to double cave/double bobbin. By combining these in optimal way on the smallest core that can do the job one can reach up to 150 KHz bandwidth without many troubles. The output device play a fundamental role as well and this is another reason why there is no universal single answer. More than 150KHz ( at -3 dB) is just technical exercise adding nothing to performance in my opinion.
For small signal, input and interstage transformers is a different story because power delivery is not a primary requirement anymore and having a larger bandwidth than the OPT in use is desirable as it will make sure that they don't become the bottleneck for frequency response.
One reason is that leakage inductance can always be reduced for a given primary inductance by increasing the interleaving. The same can be done for capacitance with vertical sectioning as shown in the article. At the end of the day this is not an endless process because the more interleaving/sectioning the more the additional insulation will rob space. So the best ratio between primary and leakage inductance may exist on its own but does not necessarily turns into the best solution because efficiency (low DC resistance) is a primary parameter too. A transformer is ALWAYS a compromise among many factors with opposite requirements.
In my experience interleaving with 4P and 3S to 6P and 5S is the optimal range for 99% of OTPs. Less is not really good (good for guitar OPTs , in general), more will rob too much space. Horizonal sectioning, none to double cave/double bobbin. By combining these in optimal way on the smallest core that can do the job one can reach up to 150 KHz bandwidth without many troubles. The output device play a fundamental role as well and this is another reason why there is no universal single answer. More than 150KHz ( at -3 dB) is just technical exercise adding nothing to performance in my opinion.
For small signal, input and interstage transformers is a different story because power delivery is not a primary requirement anymore and having a larger bandwidth than the OPT in use is desirable as it will make sure that they don't become the bottleneck for frequency response.
Last edited:
Thanks for the replies guys, but i find it regrettable that Crowhusts findings get so ligthly brushed aside.
IMO, maximal BW is a very desirable goal in OPT design and any means for improvement are worthwile. Sadly, sidestepping my question and instead bringing forward stuff that may be of interest to a beginner but does nothinģ to improve my undeŕstanding of Crowhursts reasoning does not help me.
45
As to vertical sectioning, to my understanďing, it can only increase leakage ìnductance in a transformer of usual 3-4/1 width/heigth window ratios. Offcourse, in a pp shell typé of transformer it is done for symmetry despite the increasing léakage
As to the smallest core, in respect to BW, you certainly have a point.
As you said, it is all about compromises.
I still hope someone would chime in able to shed some ligth on that Crowhurst thingy...
IMO, maximal BW is a very desirable goal in OPT design and any means for improvement are worthwile. Sadly, sidestepping my question and instead bringing forward stuff that may be of interest to a beginner but does nothinģ to improve my undeŕstanding of Crowhursts reasoning does not help me.
45
As to vertical sectioning, to my understanďing, it can only increase leakage ìnductance in a transformer of usual 3-4/1 width/heigth window ratios. Offcourse, in a pp shell typé of transformer it is done for symmetry despite the increasing léakage
As to the smallest core, in respect to BW, you certainly have a point.
As you said, it is all about compromises.
I still hope someone would chime in able to shed some ligth on that Crowhurst thingy...
Last edited:
I only added something more (efficiency) to the big pot hoping to make it more clear. Probably the best way to read that article is starting from the last paragraph where it is clear that the overall solution is application dependent: mode of operation, source impedance, load impedance....even opinions! I do not mean opinions about the sound by opinions about techincal specs. For example, if you tell me that the power rating must be rated at 20Hz I would disagree because it's a waste 99.9% of the time. Simply too many variables. The tables and the tricks in the article are meant to be consulted and used to save time without repeating calculations every time.
The only things that are always valid, independetly from the application, are that leakage inductance is minimum for square cross-section (EI and C) cores however the difference between square and moderately non-square is small and not worth loosing some sleep. It's not a big deal. More important instead is how the space is utilized as best efficiency is always achieved using half winding space for the primary and half for the secondary. Probably this is the best indicator about "optimal core dimensions" because core dimensions fix the winding space available and if half of that is not enough to meet target specs (so again no universal answer) then one has to decide if to go on and accept some compromise or move to another size.
The only things that are always valid, independetly from the application, are that leakage inductance is minimum for square cross-section (EI and C) cores however the difference between square and moderately non-square is small and not worth loosing some sleep. It's not a big deal. More important instead is how the space is utilized as best efficiency is always achieved using half winding space for the primary and half for the secondary. Probably this is the best indicator about "optimal core dimensions" because core dimensions fix the winding space available and if half of that is not enough to meet target specs (so again no universal answer) then one has to decide if to go on and accept some compromise or move to another size.
P.S.
The article is not even exhaustive about possible winding schemes. For example, Z-type windings, bifilar, trifilar are not considered. These are not just useful for interstage or small signal transformers but also for OPTs. A full bifilar primary removes the necessity of a double cave or double bobbin even for an ideal PP OPT, for example, and solves the problem of mode of operation (Class A or B) all at once. I am just in the process of completing a 11K PP OPT, rated 100W at 28Hz for 1T induction where the core alone is 7Kg. There is no double cave or double bobbin. Its "little" brother (5K, 75W) done in the same way can do 110KHz -1dB....
The venerable Radiotron Designers book also considered the case of random winding with optimal winding space filling, showing it can work really well. This just say that calculations and evaluations are just the first step and actual techinque, winding machine, winder skills will normally tell what can be done and what not.
The relevance of all this is that saving insulation space using one specific technique, for example, is also a way to determine optimal core dimensions. Sometimes it's "economical", sometimes it's not. Economical can be interpreted from different perspectives, too.
At the end of the day, there is no single answer mostly because parasitic capacitances are not easily quantified before hand as they also depend on how it's actually done. The numbers in those tables are based on appriximations and/or assumptons. They are just guidelines.
The article is not even exhaustive about possible winding schemes. For example, Z-type windings, bifilar, trifilar are not considered. These are not just useful for interstage or small signal transformers but also for OPTs. A full bifilar primary removes the necessity of a double cave or double bobbin even for an ideal PP OPT, for example, and solves the problem of mode of operation (Class A or B) all at once. I am just in the process of completing a 11K PP OPT, rated 100W at 28Hz for 1T induction where the core alone is 7Kg. There is no double cave or double bobbin. Its "little" brother (5K, 75W) done in the same way can do 110KHz -1dB....
The venerable Radiotron Designers book also considered the case of random winding with optimal winding space filling, showing it can work really well. This just say that calculations and evaluations are just the first step and actual techinque, winding machine, winder skills will normally tell what can be done and what not.
The relevance of all this is that saving insulation space using one specific technique, for example, is also a way to determine optimal core dimensions. Sometimes it's "economical", sometimes it's not. Economical can be interpreted from different perspectives, too.
At the end of the day, there is no single answer mostly because parasitic capacitances are not easily quantified before hand as they also depend on how it's actually done. The numbers in those tables are based on appriximations and/or assumptons. They are just guidelines.