They are very flat ( wider of course ). In a 1U rack one can fit a 100VA O-core.
I only see advantages over classic toroidals. In theory, O-core is optimum, efficiency, iron and copper losses, flux leakage.
Do you see drawbacks ?
Do you recommend a manufacturer.
I am looking for low powers: Less than 40VA
I only see advantages over classic toroidals. In theory, O-core is optimum, efficiency, iron and copper losses, flux leakage.
Do you see drawbacks ?
Do you recommend a manufacturer.
I am looking for low powers: Less than 40VA
Last edited:
Cost perhaps? Larger diameter mounting washer? For the same cross-sectional area of core the O-core is thicker than square cross-section core?
I do not understand.
The core section is round, a perfect circle, isn’it ?
So, it is flatter than the rectangular ones. May be, I misunderstood ?
My main concern here is to fit power supplies in a 1U rack ( 1 U is 44 mm hight, about 40 mm available inside ).
Project: 32V 400mA and 50V 120mA.
Linear PSUs I have seen inside 1U racks were with R-core transformers.
R-core are flatter than E I, but O-core is the flattest.
I am afraid of this.
A circle with radius 1cm has a diameter of 2cm and an area of 3.1416 cm^2. A square of side 2cm has an area of 4 cm^2. So a square of the same area as the circle would be narrower than 2cm - simple geometry.
"The VA rating is directly proportional to the cross-sectional area of the core, no matter what shape it is. Normally those with round core are made up of ferro magnetic wire in stead of flat tape wound into a circular shaped and there is no sharp edge and need no end caps. They are slightly more efficient as the windings follow the curvature of the core closely. They should not really cost more, it would virtually use the same weight in material. Maybe because they look more esoteric or audiophile like.
You are right.
A disq of diameter 2 cm has an area of 3.1416 cm2. So, given the same VA a supposed square core toroidal is flatter than a O-core toroidal.
However, classic toroidals are way higher because they do not have square cores but rectangular cores with a quite hight H/W ratio.
Looking for flat transformers.....to fit inside a 1U rack.
Are they square core toroidals or rectangular core toroidals with core W > H ?
Or elliptic core toroidals with core W / H > 1.2732 ?
Last edited:
A rack unit is 1+3⁄4 inches (44.45 mm), but how much space do you have inside?
However it shouldn't be so darn difficult to find a toroid transformer that fits your case, just as an example this 40VA toroid is 28mm high:
https://sklep.toroidy.pl/en_US/p/TTS0040-Transformer-TS40VA230-voltage-to-50-V/292
However it shouldn't be so darn difficult to find a toroid transformer that fits your case, just as an example this 40VA toroid is 28mm high:
https://sklep.toroidy.pl/en_US/p/TTS0040-Transformer-TS40VA230-voltage-to-50-V/292
Thanks for the link.
Inside, I saw 39 mm for a aluminium rack. I think there is a little more with steel, may be 41mm.
Toroids from jamestransformer.com are much higher than the ones at sklep.toroidy.......😕
Could they have very different core shape, what is wrong ?
Inside, I saw 39 mm for a aluminium rack. I think there is a little more with steel, may be 41mm.
Toroids from jamestransformer.com are much higher than the ones at sklep.toroidy.......😕
Could they have very different core shape, what is wrong ?
Last edited:
Not wrong , just different.
Maybe it is wise not to use odd transformers when a 1U high casing is a fact. If one chooses normal PCB mount molded toroids there is no issue with the height of 1U till the 15VA types. With 25 to 50VA one has a too small casing when choosing 1U as the transformers themselves are already 37.5 mm. Add about 2 mm for the PCB and then at least 5 mm clearance from upper and lower side to chassis/cover for both heat and safety. That is a minimum of 50 mm. Then it is either having a few smaller toroids or choosing another casing.
Your 1U project has 12.8W and 6W requirements. So you could use a 15VA and a 10VA PCB mount type depending if the mentioned currents are nominal or maximum values. If the 15VA is too tight (with maximum of 12W) you already have to use 2 x 10VA in parallel (yup, 3 x 10VA in total) or switch to a higher casing. Or use SMPS......
The question is what a too small casing will bring when the transformer almost touches the casing. In my experience not positive both mechanically, magnetically and thermally. Choosing too small casings is a habit leading to time loss (cooling challenges, ventilation slots required etc.) but on the other hand more practical sized units. Choices.
Maybe it is wise not to use odd transformers when a 1U high casing is a fact. If one chooses normal PCB mount molded toroids there is no issue with the height of 1U till the 15VA types. With 25 to 50VA one has a too small casing when choosing 1U as the transformers themselves are already 37.5 mm. Add about 2 mm for the PCB and then at least 5 mm clearance from upper and lower side to chassis/cover for both heat and safety. That is a minimum of 50 mm. Then it is either having a few smaller toroids or choosing another casing.
Your 1U project has 12.8W and 6W requirements. So you could use a 15VA and a 10VA PCB mount type depending if the mentioned currents are nominal or maximum values. If the 15VA is too tight (with maximum of 12W) you already have to use 2 x 10VA in parallel (yup, 3 x 10VA in total) or switch to a higher casing. Or use SMPS......
The question is what a too small casing will bring when the transformer almost touches the casing. In my experience not positive both mechanically, magnetically and thermally. Choosing too small casings is a habit leading to time loss (cooling challenges, ventilation slots required etc.) but on the other hand more practical sized units. Choices.
Last edited:
From Rod Eliot ESP site I understand there is a 3,3 factor from transformer VA and PSU DC Watts.
And it seems to be un-regulated DC Watts where I think about 40V un -regulated for 32V out and 60V un-regulated for 48V out.
And it seems to be un-regulated DC Watts where I think about 40V un -regulated for 32V out and 60V un-regulated for 48V out.
OK manufacturers numbers is a factor 0.8 (unregulated, only rectified/filtered). I think Rod knows his theory better than I do but IMHO using a 15VA transformer for just 4VA load is not standard.
All superfluous voltage will be dissipated as useless heat. If you would choose uLDO regulators you can have smaller VA transformers.
For 32V output regulated I would choose a 2 x 15V in series. With Schottky rectifiers and enough bulk cap capacitance and an uLDO regulator a 2 x 12V one might be doable. Challenge!
I misread 50V for 5V so for 50V (and now it suddenly is 48V 🙂) regulated you could use a 2 x 22V in series. Possibly a 2 x 18V is a possibility depending on the brand/type. The challenge here is to prove this is possible 🙂
It helps to calculate in reverse. You need for instance 48V output. Add ripple voltage, 2 x Uf rectifiers and the regulators dropout voltage to know the minimum. The deviation from standard with "sharp calculation" is to use uLDO, at least double the amount of capacitance and use Schottky diodes or LT4320 ideal rectifiers and check all possible scenarios plus measuring to see if things work out OK.
All superfluous voltage will be dissipated as useless heat. If you would choose uLDO regulators you can have smaller VA transformers.
For 32V output regulated I would choose a 2 x 15V in series. With Schottky rectifiers and enough bulk cap capacitance and an uLDO regulator a 2 x 12V one might be doable. Challenge!
I misread 50V for 5V so for 50V (and now it suddenly is 48V 🙂) regulated you could use a 2 x 22V in series. Possibly a 2 x 18V is a possibility depending on the brand/type. The challenge here is to prove this is possible 🙂
It helps to calculate in reverse. You need for instance 48V output. Add ripple voltage, 2 x Uf rectifiers and the regulators dropout voltage to know the minimum. The deviation from standard with "sharp calculation" is to use uLDO, at least double the amount of capacitance and use Schottky diodes or LT4320 ideal rectifiers and check all possible scenarios plus measuring to see if things work out OK.
Last edited:
The most anoying factor is the +/- 10% Mains variability. With this in mind, LDO doesn't help much.
I will use my favorite discrete positive regulator, based on TL431 ( not LDO, but not bad at this at all ).
My favorite is about 12 parts. A CCS ( 4 parts ), a TL431 ( two for 48V ) a BJT, resistors and caps.
I plan about 2 V peak to peak ripple using 2000 uF per Amp.
I do not want the transformer(s) on a PCB, no space but just a pad about vibrations.
I will use my favorite discrete positive regulator, based on TL431 ( not LDO, but not bad at this at all ).
My favorite is about 12 parts. A CCS ( 4 parts ), a TL431 ( two for 48V ) a BJT, resistors and caps.
I plan about 2 V peak to peak ripple using 2000 uF per Amp.
I do not want the transformer(s) on a PCB, no space but just a pad about vibrations.
Last edited:
Well, I do this all the time and even have that 10% deviation not giving issues at all. In most cases the voltages under load are quite higher than one assumes and one ends up with still too much headroom with -10% deviation of mains voltage! Of course this is one off work so there is freedom to try out. In Germany one never sees mains voltage at 230V anymore, quite the contrary 🙂
If you have predefined requirements as a favorite regulator with predefined dropout voltage then the whole situation is different. Then all calculations need to be adapted to 1. the casing limits, 2. the favorite regulators specific dropout voltage. That are 2 extra challenges in an already existing challenge to do this with linear PSUs 😀 If you include another specific requirement you'll have a nice Gordian knot. If you go with the flow you could use a fine SMPS and be done with it. A designers easy way out so to speak.
* uLDO, not LDO. The one I was testing a few weeks ago had under 0.15V dropout voltage even at 5A.
If you have predefined requirements as a favorite regulator with predefined dropout voltage then the whole situation is different. Then all calculations need to be adapted to 1. the casing limits, 2. the favorite regulators specific dropout voltage. That are 2 extra challenges in an already existing challenge to do this with linear PSUs 😀 If you include another specific requirement you'll have a nice Gordian knot. If you go with the flow you could use a fine SMPS and be done with it. A designers easy way out so to speak.
* uLDO, not LDO. The one I was testing a few weeks ago had under 0.15V dropout voltage even at 5A.
Last edited:
There you have the third specific requirement 🙂 The fourth specific "no PCB" requirement and the fifth specific "no space but a pad" (cooling?, safety?) make it quite difficult to solve altogether. Also availability of low VA chassis mount toroids is quite frankly bad at the moment. I used those by preference but had to switch to molded PCB mount toroids.
Curious how you will solve this but please mention specific requirements beforehand to avoid time loss of those responding.
Last edited:
Oups....2000uF reservoir caps is to target 5 volt pp ripple, according to Eliot ESP.
For my chosen 2Vpp ripple I should have writen 5000uF per Amp capacitor
For my chosen 2Vpp ripple I should have writen 5000uF per Amp capacitor
Mounting toroids, I think simply a couple of Zip locks, no fancy chassis mounting.
My thinking started after repairing few 1U rack équipements ( delays, reverbs, preamps ). They were using C-core transformers simple mounted on the chassis with screws, close to it a PCB with diodes, caps, LM317 and LM337. Most had trouble at the PSU, some at the display backlight.
My project is an octo microphone preamp.
PSU reqirements:
Fit inside1U rack.
30V ( or 32 ) 400mA for 8 preamps with overload LED
48V ( or 50 ) 120mA for Phantom power.
1.25V drop ( V un-regulated to Vout regulated ). But......
I fired the simulation at the 48V regulator. It shows i can lower the un-regulated voltage down to 49.25V still getting regulation 113dB PSRR.
I think the actual possible drop is better, because the simulator bumps on "not finding the DC operating point" when I try to go lower than 1.25V drop.
The CCS with two bjts, I am using is able to do better, I suspect the simulation doesn’t know about bjt near saturation behavior.
It seems, the simulation doesn’t want VCE < VBE. A BJT at low current is able to do VCE = 0.1V. So I think, the real circuit drop is 0.75V.
I did not fire the simulation at the 30V regulator, the drop is likely the same.
My thinking started after repairing few 1U rack équipements ( delays, reverbs, preamps ). They were using C-core transformers simple mounted on the chassis with screws, close to it a PCB with diodes, caps, LM317 and LM337. Most had trouble at the PSU, some at the display backlight.
My project is an octo microphone preamp.
PSU reqirements:
Fit inside1U rack.
30V ( or 32 ) 400mA for 8 preamps with overload LED
48V ( or 50 ) 120mA for Phantom power.
1.25V drop ( V un-regulated to Vout regulated ). But......
I fired the simulation at the 48V regulator. It shows i can lower the un-regulated voltage down to 49.25V still getting regulation 113dB PSRR.
I think the actual possible drop is better, because the simulator bumps on "not finding the DC operating point" when I try to go lower than 1.25V drop.
The CCS with two bjts, I am using is able to do better, I suspect the simulation doesn’t know about bjt near saturation behavior.
It seems, the simulation doesn’t want VCE < VBE. A BJT at low current is able to do VCE = 0.1V. So I think, the real circuit drop is 0.75V.
I did not fire the simulation at the 30V regulator, the drop is likely the same.
@Nico Ras
If you ask me and SMPS would not be my choice (it seems about everyone elses choice though) but the obvious so higher power efficiency, lower weight, less heat, smaller size.
However another new requirement that popped up is that it is for a microphone amplifier which (IMHO) directly excludes SMPS.
Too bad even the best SMPS needs a linear regulator or LDO to be good enough for small signal (analog) audio 🙂
Mchambin, mounting toroids with ziplocks is cursing in the church. There is a certain élégance in making stuff sturdy/safe/power efficient.
If you ask me and SMPS would not be my choice (it seems about everyone elses choice though) but the obvious so higher power efficiency, lower weight, less heat, smaller size.
However another new requirement that popped up is that it is for a microphone amplifier which (IMHO) directly excludes SMPS.
Too bad even the best SMPS needs a linear regulator or LDO to be good enough for small signal (analog) audio 🙂
Mchambin, mounting toroids with ziplocks is cursing in the church. There is a certain élégance in making stuff sturdy/safe/power efficient.
Last edited: