| zerohead_ak47 |
 |
|
|
| rejithcv |
| Is it really possible to obtain such a high power with a flyback topology? |
|
|
| N-Channel |
Sure, if you've got a BIG enough core, IGBT, and are using #0 battery cable for your secondaries. ;)
Shouldn't the 68KW/220mF cap (R2C2) filter go to the HV DC Bus, and not one leg of the AC Mains?
Steve |
|
|
| zerohead_ak47 |
| quote: | Originally posted by rejithcv
Is it really possible to obtain such a high power with a flyback topology? |
labgruppen high power amplifiers using flyback smps |
|
|
| Eva |
| Some component values are probably wrong. For example, mains filter capacitor can't be 100uF, it should be 1000uF or bigger. |
|
|
| Workhorse |
| Evita, once you mentioned that Flyback is not good for KW power levels, but now i think your mind has changed after viewing the Lab Gruppen 10KW peak Flyback;) |
|
|
| Eva |
It may be scaled up, or several converters may be paralleled, but EMI performance and efficiency is still not good at such high power levels. Voltage and current stress on transistors and diodes is higher, and primary side and secondasry side silicon is not used at the same time, so silicon uitilization is 50% lower than push-pull. Also, the required transformer is considerably bigger and prone to leaking tons of flux...
Also, that 10KW amplifier does not actually use a 10KW flyback but more likely a 1-2KW one with current limiting. The rails should start sagging above a certain continuous power output. Note that tracking amplifiers are not only much more efficient than class AB with music signals but also achieve a considerable reduction in power consumption. Power efficiency of class AB with music seems to be around 10-20% (depending on how heavy the bass content is and how much compression was employed). |
|
|
| TOINO |
| quote: | Originally posted by Eva
Also, that 10KW amplifier does not actually use a 10KW flyback but more likely a 1-2KW one with current limiting. . |
That is the R14 function of published schema (as you know much better than me).
Exactly the same principle is used on the refereed “10Kw” amplifier PSU, but with “R14” value adjusted for cycle by cycle current limiting.
| quote: | Originally posted by Eva
(depending on how heavy the bass content is and how much compression was employed). |
Hhmm.. Eva I appreciate when you are so reasonable and objective in your comments…:clown::yes: |
|
|
| darkfenriz |
| Passive mains filtering/PFC ? No, thanks... |
|
|
| Eva |
Well, I forgot to mention that class D and rail tracking tend to exhibit 80-90% power efficiency regardless of the crest factor of the music signal, that's why those amplifiers can play high power music with a small PSU. Most of the power drwan from mains supply is just delivered to the loudspeaker. Plain class AB playing music tends to waste into heat 80-90% of the power delivered by the PSU.
PS: Get some reasonably sized class D module, like UCD180/400/800, and use it to play your favourite music into a full range speaker at clipping threshold. Then measure average current consumption.You will probably have a hard time believing how much class D music you can get out of, say, 50W PSU power. |
|
|
| Adnancoskun |
I think so same as zerohead, the flayback PSU using at some power amplifier and usually welder apparatus, but you must cooling to IGBT very well,,
eline salk veysel:) |
|
|
| sixtek |
Hi,
I would build a 28V/40A SMPS.
Would this PSU a good start?
Thx, |
|
|
| zerohead_ak47 |
| quote: | Originally posted by sixtek
Hi,
I would build a 28V/40A SMPS.
Would this PSU a good start?
Thx, |
good lucks |
|
|
| Workhorse |
Hi Zero,
Have you tried it in your setup... |
|
|
| zerohead_ak47 |
| quote: | Originally posted by Workhorse
Hi Zero,
Have you tried it in your setup... |
transformer not ready...waiting.... |
|
|
| ChocoHolic |
hm... everything is possible, but not everything makes sense...
Does anybody have the link to the full appliction note?
My search at ONSEMI did not show this beast.
I am really wondering about the trafo design. They use 3mm wires at 40kHz? Strange! Even if we consider a single layer winding (no need to consider proximity effect) and ignore that flybacks ussually have airgaps with catastrophic fields (leading to eddy current issues in the winding...) .... even then, just from the skin effect wires above 1mm are a waste of copper. At 40kHz already at 0.32mm below the surface the current density down to 30%. The inner area of the copper doesn't carry much current.
:scratch:
Which core do they use? |
|
|
| star882 |
| quote: | Originally posted by Eva
Well, I forgot to mention that class D and rail tracking tend to exhibit 80-90% power efficiency regardless of the crest factor of the music signal, that's why those amplifiers can play high power music with a small PSU. Most of the power drwan from mains supply is just delivered to the loudspeaker. Plain class AB playing music tends to waste into heat 80-90% of the power delivered by the PSU.
PS: Get some reasonably sized class D module, like UCD180/400/800, and use it to play your favourite music into a full range speaker at clipping threshold. Then measure average current consumption.You will probably have a hard time believing how much class D music you can get out of, say, 50W PSU power. |
Class D sure is efficient! I have an old 30w Philips amp (USB input) that can play really loud without distortion, while only getting warm in the process.
My current amp is 260w, also by Philips. It has S/PDIF input and is very amazing for efficiency. TI PurePath technology is simply amazing. I'm surprised why it's not used more and advertised more as it's environmentally friendly (uses less than half the power of a traditional analog amplifier) and cheaper at high power. With digital sources, the quality is better as the D/A conversion is done at high power levels. |
|
|
| zerohead_ak47 |
Final schema
i have got all the electronic components but gapped etd59 transformer |
|
|
| Khron |
| Any chance of seeing a PCB design? |
|
|
| Workhorse |
Zero
Your Flyback cannot work with a 600V igbt with mains voltage at 240VAC, you need at least 1200V Igbt to do the job... |
|
|
| zerohead_ak47 |
| quote: | Originally posted by Workhorse
Zero
Your Flyback cannot work with a 600V igbt with mains voltage at 240VAC, you need at least 1200V Igbt to do the job... |
test clamping network for 600v igbt |
|
|
| Eva |
| The voltage reflected to the primary side is 60V + 1V from the diodes * 64T / 20T = 195V. Thus good primary-side clamping is required for operation at 230V input and above. |
|
|
| TOINO |
:worship: Princess of Asturias, Visigod Queen:worship:
Primary rectified voltage 310V + 195V= 505V
Maybe a stretched safety margin? |
|
|
| Eva |
It's not a matter of margin. Unless the tranformer is really well made to exhibit extremely low leakage inductance, the collector voltage will rise up to avalanche upn turn off. The schematic shows no clamp, only an RCD snubber that is probably not going to provide good enough clamping effect.
Capacitors across the diodes help in reducing the turn-off spike but make turn-on behaviour worse. |
|
|
| ChocoHolic |
Let's make a rough guess:
According the newer schematic the refelected voltage would be 284V. Means already without leakage inductance the IGBT would see values close to 600V.
Say 310+284V=594V.
In the snubber cap 594V will mean a energy of 388uJ.
Then let's consider 5uH leakage. This will have stored 250uJ if we consider an switching event at 10A.
These 250uJ will be added to the 388uJ. So you get 638uJ in the snubber cap, means it will go to 762V.
Not only your IGBT is underrated also the 400V snubber cap.
And I it might be worse, because I have doubts that you will be able to realize 5uH.... The transformer itself is not so easy to design and you will additionally see an additional upwards transformed portion from all parasitic inductances of your secondary circuit loop(s).
Last but not least take care in this game.
In most offline applications people are considering 400V as relevant voltage for the safety isolation, which would required a layout AND transformer desgin that ensures 8mm creepage (EN60065, Table 11).
In your case with the high reflected voltage and leakage peak... well, in such a design the requirements might be even higher... |
|
|
| ChocoHolic |
hmmm, the second snubber with 47nF should be helpful to keep the peak less than my first calculation....
Nevertheless, especially with the new output voltages and turns ratio of your new schematic you will need more 600V for the switch and also for the snubber caps. |
|
|
| zerohead_ak47 |
for 1kw gapped E70/33/32 transformer
 |
|
|
| Eva |
I have experience with long air gaps. Yours is very large and requires litz wire to prevent the windings from overheating due to Eddy currents, even at a few A only.
Are you isolating primaries and secondaries properly? |
|
|
| zerohead_ak47 |
| now not to be |
|
|
| Eva |
Keeping the windings away from the gap helps, this is achieved by using the outer portion of the winding space and leaving the inner portion unused. Using a gap in the three legs instead of a center-leg gap also helps a lot, but at the expense of stray magnetic fields that require some way of shielding.
You may consider several smaller transformers or the kind of iron powder toroid cores used for class D (low-perm low-loss RF iron powder and Arnold's MPP & Hi-Flux materials) as alternatives.
|
|
|
| ChocoHolic |
...yes, litz would be a big step forward even in ungapped designs. See posting 16. In flybacks with large gap litz becomes a key factor.
The eddy currents are not just a small loss factor, in your design you will really get a brute hot spot in the windings which are close to the gap...
Of course for first looking if it works at all, the thick wires are better than nothing, but be carefull at higher power. You will find quite massive transformer heating burn the isolation in the inner area close to the airgap.
...I know HF litz is hard to get and pain to work with...
Regarding the other material like iron powder, MPP, cool u etc...
I am not sure that you really can make use of their higher saturation levels. These materials have remarkable higher core losses, so you still might have to use a lot of turns. Remaining advantage could be that you get them in lower permeabilities and might get rid of the air gap. |
|
|
| zerohead_ak47 |
| pcb is ready right now..but i am not sure that if there is a problem about routing . |
|
|
| Eva |
I'm sorry to say this but your PCB design is terrible. You may consider several improvements for much lower inductance in power loops. However, the most important improvement that you should consider is to add proper clearance (at least 5mm) between the secondary side and the primary side, and this includes the tracs going to the optocoupler.
BTW: Series inductance just after the secondary side diodes will only make the primary switch turn-off spike worse. |
|
|
| ChocoHolic |
Besides the giant loop inductances and missing safety, there seems to be a real circuit error:
Two of the four output caps should be connected before the 10uH chokes. In the PCB all are behind.
Looks like you are using software, without synchronisation between schematic and PCB?
Anyway, the most serious issue are the safety creepages & clearances.
I do not know to which standard Eva is referring, from my point of view you would need at least 8mm, may be even more. Referring to EN 60065 table 11.
If you tell me the worst rms values at 10% overvoltage of AC input for following voltages:
-DC-Rail: (probably close to 400V)
-Across primary of transformer at full power: xxxV?
-Voltage between collector of the IGBT vs protective earth: xxxV?
Then I can search for you the correct creepages and clearances in the EN60065. |
|
|
| TOINO |
ChocoHolic, are you sure about the 8mm safety distance?
The distance between optocoupler led pins and the other side is 7.6mm (.3) :rolleyes: |
|
|
| ChocoHolic |
I am referring to the EN 60065 (Audio, Video and similar). May be there are other standards, which define diffrent values..... You never know....
For 400V I am sure 8mm would be right if we assume dirt level II and and materials with low CTI value.
Dirt level II should be OK for in house use equipment.
If you are going to use materials with better surface creepage properties then the creepages could be reduced.
If the CTI value is ranging between 400...600: 5.6mm
If the CTI value is 600 or higher: 4mm
But nobody in DIY is able to control the CTI value of his materials. We have to use what we get and mostly without detailed specification.... So we should assume a low CTI value and design for 8mm.
Besides creepages also clearances are defined and there not only the rms value is relevant but also peak and DC values.
In our relevant voltage range a clearance of 5mm should be fine. |
|
|
| ChocoHolic |
| A proper safety opto coupler would be i.e. the CNY66. |
|
|
| zerohead_ak47 |
hi ChocoHolic
-Across primary of transformer at full power: xxxV?
-Voltage between collector of the IGBT vs protective earth: xxxV?
590 volt max peak |
|
|
| ChocoHolic |
Glad to read that your peak values are lower than expected. In fact this value is looking like you have just a very minor peak....
Please note from the peak values I can tell you not the required creepages, but clearances only.
Clearances with 5mm is OK (referring to table 10).
For the creepage the standard is counting the rms value only.
From your schematic I would expect something like:
-300Vrms across the primary winding.
-430Vrms across the IGBT.
-380Vrms form collector of IGBT to protective earth.
If we consider 10% overvoltage then the highest voltage would be 473V.
EN60065 is giving creepage values for 400V (8mm) and
for 600V (12.6mm). Fortunately EN60065 is explicitely allowing to interpolate for voltages inbetween, -the found value must rounded up to the next 0.1mm grid.
For 473Vrms you will end up in 9.7mm creepage requirement.
BTW:
I guess I am not the only one who is curious about pictures and screen shots! :) |
|
|
| Eva |
| 5mm is the distance routinely found on most consumer equipment. You know, rules are not the same for everyone. |
|
|
| ChocoHolic |
| quote: | Originally posted by Eva
5mm is the distance routinely found on most consumer equipment. You know, rules are not the same for everyone. |
I would expect that most of the manufacturers are able to ensure the proper CTI value of their PCBs.
More difficult are the CTI properties of transformer isolation materials. (Designing safe transformers is master mind fun anyway. 3D creepage thinking is always good for some pitfalls...)
...well, and then there are of course some cruel guys, who do not care about anything and simply sell unsafe products... |
|
|
| sawreyrw |
| This core is not intended for use as a transformer. Its intended to be used as a high current inductor. |
|
|
| zerohead_ak47 |
| but kool mu meterial for flyback smps |
|
|
| Eva |
There are essentially two problems with toroids in off-line flybacks:
- The required 3KV mains isolation is quite hard to achieve
- At 1000W, the resulting leakage inductance may easily require a 100W active clamp in the primary side. |
|
|
| ChocoHolic |
Cool Mu for flybacks is only suitful if you want to operate a low frequencies between 20kHz-30kHz and when you have not continuous power, but short term high power with low average power.
In fact the last condition is reality in SMPS for audio.
But Eva is right, torroids make it very difficult to ensure safety isolation - except you are going to use special tripple isolated wires....
If you want to stick with Cool Mu, I would propose that you search for a reasonable E-shape Cool Mu.
E-core-bobbins allow proper isolated winding constructions with margin tapes at the sides for correct creepages. Furtheron E-core-bobbins make it easy to go for interleaved windings in order to reduce the leakage inductance. |
|
|
|
