mag, congratulations, if you have 95% of efficiency at first attempt, you are damn lucky EE, and you don't need improvement of your SMPS anymore! 
Actually, I've about 92-93% with full Ls*I^2 recuperations at 40A 11.5VDC regulated SMPS with ring core trafo, bifilar primary winding Ls=50nH(primary2primary), all of high current PCB lines are copper pour polygons with measured resistance (40mm max distance for GND, and 30mm for VDD), and this is certainly limit for using non shcottky (MUR1620 10A) rectifiers.
PS: I hope you mean 95% for SMPS, not for SMPS+AMP? BTW, 13mm^2 2M it's about 2.7mOhm, are you sure that your 6.9mOhm is correct?
Actually, I've about 92-93% with full Ls*I^2 recuperations at 40A 11.5VDC regulated SMPS with ring core trafo, bifilar primary winding Ls=50nH(primary2primary), all of high current PCB lines are copper pour polygons with measured resistance (40mm max distance for GND, and 30mm for VDD), and this is certainly limit for using non shcottky (MUR1620 10A) rectifiers.PS: I hope you mean 95% for SMPS, not for SMPS+AMP? BTW, 13mm^2 2M it's about 2.7mOhm, are you sure that your 6.9mOhm is correct?
I don't think 95 % is real. There are so much possible source of measuring errors.
mag!
2*2 turns of primary is OK, however there could be lower idle loss with 2*3 turns (but a little higher full-power loss). I don't want to make exact calculations. Secondary? It's up to you! If trafo is OK, drop will be less then 10 % (only on trafo).
mag!
2*2 turns of primary is OK, however there could be lower idle loss with 2*3 turns (but a little higher full-power loss). I don't want to make exact calculations. Secondary? It's up to you! If trafo is OK, drop will be less then 10 % (only on trafo).
In fact 95% efficiency sounds really good... Probalbly there are some measurement errors.
The calculated 6.9mOhms are not only the cable. I have calculated it measuring the voltage drop at idle (1.4A) from the battery + to the first filter coil. It includes the wire, the interconnections and the fuse.
BTW also my PCB uses polygons where high current is involved, to reduce inductance and resistance as much as possible. Check the attached .zip file
Now I am winding the new transformer with 10+10 turns secondary then stacked 2+2 turns primary and then other 10+10 turn secondary in parallel. Everything will be wound in bifilar
@ IVX
what is a typical figure for Ls un such a transformer?
thank you
The calculated 6.9mOhms are not only the cable. I have calculated it measuring the voltage drop at idle (1.4A) from the battery + to the first filter coil. It includes the wire, the interconnections and the fuse.
BTW also my PCB uses polygons where high current is involved, to reduce inductance and resistance as much as possible. Check the attached .zip file
Now I am winding the new transformer with 10+10 turns secondary then stacked 2+2 turns primary and then other 10+10 turn secondary in parallel. Everything will be wound in bifilar
@ IVX
what is a typical figure for Ls un such a transformer?
thank you
Typical Ls figure doesn't exist, trust me Winding approach is major factor, e.g. those bifilar primarys are 50nH, but they would be 200nH if primary A on the left side of ring and primary B on the right side. Also try to calculate your layout inductance, you'll be surprised BTW, I guess, layout isn't good enough for 95% of efficiency: Q16 and Q18 almost doesn't work, if L9=0Ohm you will get about 2mOhm from FUSE1 to TR1 even on 3oz PCB, what is the reason to go VDD (and GND too!) so long way if you can place TR1+MOSFETs instead of U6+etc?
Winding approach is major factor, e.g. those bifilar primarys are 50nH, but they would be 200nH if primary A on the left side of ring and primary B on the right side. Also try to calculate your layout inductance, you'll be surprised BTW, I guess, layout isn't good enough for 95% of efficiency: Q16 and Q18 almost doesn't work, if L9=0Ohm you will get about 2mOhm from FUSE1 to TR1 even on 3oz PCB, what is the reason to go VDD (and GND too!) so long way if you can place TR1+MOSFETs instead of U6+etc?Why you say Q16 and Q18 almost doesn't work?
It is because the connection between their drain and the trafo has higher resistance that the path from the other two mos?
The layout is made in this way for mechanical reasons. The mos are attached to an heatsink that is shorter than the PCB itself.
The PCB has 2oz copper weight and all the planes carrying high current has apertures on the top solder mask. On those apertures I have soldered plenty of flat copper wire to reduce the resistance as much as possible.
BTW: I am not aiming for 95+ % of efficiency. This is my first attempt with class-D and with power electronics in general.
In the life I am an RF engineer working with milli-amperes and micro-watts.... I am quite satisfied with the results I have got so far; I think that for sure there is margin for improvement and the trafo will be an huge achievement for me.
It is because the connection between their drain and the trafo has higher resistance that the path from the other two mos?
The layout is made in this way for mechanical reasons. The mos are attached to an heatsink that is shorter than the PCB itself.
The PCB has 2oz copper weight and all the planes carrying high current has apertures on the top solder mask. On those apertures I have soldered plenty of flat copper wire to reduce the resistance as much as possible.
BTW: I am not aiming for 95+ % of efficiency. This is my first attempt with class-D and with power electronics in general.
In the life I am an RF engineer working with milli-amperes and micro-watts.... I am quite satisfied with the results I have got so far; I think that for sure there is margin for improvement and the trafo will be an huge achievement for me.
mag said:
yeah, if you talk about 1.85mOhms Rds_on, you should think about PCB resistance too, because 3mOhm of VDD way will dissipate 30000mW at 100000mA.
Opened mask aperture it's a good idea, but soldering alloy is 9 times worse vs copper conductor, so for 50% resistance reducing you'll have to put .63mm of soldering alloy, it's very possible that PCB will bend.PS: I hope my critical points isn't rough at all, because not intended for it, rather for answer to your technical questions. Your project is nice job, no doubt, but I'm would make it differently, and I've explained why.
No problem IVX,
it is always good to know the opinion of other guys that for sure know better than me stuff related to power electronics. There is always margin of improvement.
I have opened the solder mask not to fill it with solder alloy (I know it will be quite useless due to the resistivity of the solder alloy compared to copper). I have opened it to solder thick copper wires (I use a flat 14 AWG tinned copper wire) to help the PCB vdd plane to carry high currents.
In any case this is an amp, designed to amplify music, hopefully not a full power sinewave at 1000W power....
Your 100000mA and 30000mW
will never be reached continously
it is always good to know the opinion of other guys that for sure know better than me stuff related to power electronics. There is always margin of improvement.
I have opened the solder mask not to fill it with solder alloy (I know it will be quite useless due to the resistivity of the solder alloy compared to copper). I have opened it to solder thick copper wires (I use a flat 14 AWG tinned copper wire) to help the PCB vdd plane to carry high currents.
In any case this is an amp, designed to amplify music, hopefully not a full power sinewave at 1000W power....
Your 100000mA and 30000mW
will never be reached continouslyHi IVX,
I have wound the new transformer and I will test it on the SMPS probably this week (I need to disassemble the complete amp, remove the old trafo and put the new one).
You are speaking about Ls energy recuperation, can you please explain me how does it works?
It it a metod of resonate out the stray inductance at switching frequency or other?
Thank you
I have wound the new transformer and I will test it on the SMPS probably this week (I need to disassemble the complete amp, remove the old trafo and put the new one).
You are speaking about Ls energy recuperation, can you please explain me how does it works?
It it a metod of resonate out the stray inductance at switching frequency or other?
Thank you
Hi all,
after having tried all the possibility to stack windings on the RM14 core and having wasted about 1km wire I have decided to go with a bigger core.
Now I am using a PM50/39 core in N87 material
my trafo is wound in the following way
-S1: 15 turns of 2*0.8mm secondary (bifilar)
-P1: 3 turns of 10*0.8mm primary (bifilar)
-S2: 15 turns of 2*0.8mm secondary (bifilar)
-P2: 3 turns of 10*0.8mm primary (bifilar)
-A1: 10 turns of 1*0.8mm auxiliary secondary
-S3: 15 turns of 2*0.8mm secondary (bifilar)
Then P1 and P2 are paralleled (the result is 3+3 turns of 10*0.8mm) and S1, S2 and S3 are paralleled (the result is 15+15 turns of 3*0.8mm). The auxiliary secondary is used to generate a 12V supply referred to the negative rail (only around 50mA average current).
On the PM50 there is much more room than in the RM14 making the whole winding process much more easy.
I need now to modify my PCB to accept the PM50 core but it is quite an easy task.
What do you think about such transformer. I hope that this week I can do some measurement on the trafo including primary and secondary inductance and stray inductance
after having tried all the possibility to stack windings on the RM14 core and having wasted about 1km wire I have decided to go with a bigger core.
Now I am using a PM50/39 core in N87 material
my trafo is wound in the following way
-S1: 15 turns of 2*0.8mm secondary (bifilar)
-P1: 3 turns of 10*0.8mm primary (bifilar)
-S2: 15 turns of 2*0.8mm secondary (bifilar)
-P2: 3 turns of 10*0.8mm primary (bifilar)
-A1: 10 turns of 1*0.8mm auxiliary secondary
-S3: 15 turns of 2*0.8mm secondary (bifilar)
Then P1 and P2 are paralleled (the result is 3+3 turns of 10*0.8mm) and S1, S2 and S3 are paralleled (the result is 15+15 turns of 3*0.8mm). The auxiliary secondary is used to generate a 12V supply referred to the negative rail (only around 50mA average current).
On the PM50 there is much more room than in the RM14 making the whole winding process much more easy.
I need now to modify my PCB to accept the PM50 core but it is quite an easy task.
What do you think about such transformer. I hope that this week I can do some measurement on the trafo including primary and secondary inductance and stray inductance
update:
I have wound my tranformer on a PM50/39 core and I have measured it.
Results
Total primary inductance=45.2uH (each half 11.3uH) (3+3t)
Total secondary inductance=1.18mH (each half 296uH) (15+15t)
Auxiliary secondary inductance=126uH (10t)
Leakage inductance pri-sec=205nH (secondary shorted)
Leakage inductance sec-pri=5.7uH (primary shorted)
Voltage ratio sec/pri = 5:1 (injected a 1Vpp sinewave at 100kHz on the primary and I got 5Vpp on the secondary).
These results seems to be acceptable. The next step will be mounting this transformer on my SMPS and see how it works
I have wound my tranformer on a PM50/39 core and I have measured it.
Results
Total primary inductance=45.2uH (each half 11.3uH) (3+3t)
Total secondary inductance=1.18mH (each half 296uH) (15+15t)
Auxiliary secondary inductance=126uH (10t)
Leakage inductance pri-sec=205nH (secondary shorted)
Leakage inductance sec-pri=5.7uH (primary shorted)
Voltage ratio sec/pri = 5:1 (injected a 1Vpp sinewave at 100kHz on the primary and I got 5Vpp on the secondary).
These results seems to be acceptable. The next step will be mounting this transformer on my SMPS and see how it works
For Push-Pull converter Ls primary2primary is a critical point. This inductance generate ugly voltage spikes from the Els=Ls*I^2/2, to recuperate Els you can use additional U source equal to VDD and in serial to VDD, next clamp spikes to this source through shcottky. It's for instance, other approaches does exist too.
Hi All,
I have measured the leakage inductance shorting primary A and measuring primary B.
BTW if I short the secondary too I see the inductance further decrease. Now with the instruments availables is quite impossible for me to measure inductances lower than 50nH.
I have mounted the new trafo on my amp, I have to mount it upside down because I don't have enough place on the PCB.....
Everthing is up and running, the rails are regulated to 50V and the amp is working. The only difference I see is a slightly higher power consumption at idle; it was 1.27A with RM14 and now I am around 1.8A with the PM50.
I have more losses at idle becouse of the bigger core size or other?
I have not yet done any power test with the new transformer, I am running it without heatsinks and I don't want to put power on this and blowing something.....
As soon as I mount back my amp in its enclosure I will do new power tests an I will let you know
thanks
-marco
I have measured the leakage inductance shorting primary A and measuring primary B.
BTW if I short the secondary too I see the inductance further decrease. Now with the instruments availables is quite impossible for me to measure inductances lower than 50nH.
I have mounted the new trafo on my amp, I have to mount it upside down because I don't have enough place on the PCB.....
Everthing is up and running, the rails are regulated to 50V and the amp is working. The only difference I see is a slightly higher power consumption at idle; it was 1.27A with RM14 and now I am around 1.8A with the PM50.
I have more losses at idle becouse of the bigger core size or other?
I have not yet done any power test with the new transformer, I am running it without heatsinks and I don't want to put power on this and blowing something.....
As soon as I mount back my amp in its enclosure I will do new power tests an I will let you know
thanks
-marco
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.