TOP272EN Output 500V / 300mA - Expert Advice Please
I'm consider to design power supply with TOP272EN. Since audio amplifier may consume over 400W, most likely it will be optimal to use 2xTOP272EN to power each channel separately.
Requirements:
1) Input, - universal AC 100-240VAC
2) Output (2 x TOP272EN) - 2 x 429V / 459V / 500V / 300mA (selectable by switch), 2 x 6.3V / 8A, 2 x 100V / 50mA
3) Lowest ripple and HF noise possible.
There are some tips and questions which come into my mind when I took a quick look at TOPSwitch datasheet.
1) Transformer construction mentions air gap. Is air gap necessary? Is there any DC current passing through transformer winding(s)?
2) I'm consider to use amorphous or nanocrystalline tape wound toroid core instead of EI ferrite in order to reduce size, leakage inductance and HF noise. Any thoughts?
BTW, is it necessary to ground metal toroid core?
3) Design Example Report 218. There as ia capacitor #2 - 2.2nF 250VAC, connecting transformer primary pin 1 and secondary pin 10/11. What is the purpose of this capacitor?
Additionally, there is a bug in DER-218 schematic - pin 4 of transforer connected to nowhere. This is probably PDF conversion glitch.
4) In order to get 500V output voltage - 2 stacked full bridge rectifiers, reducing voltage requrements to diodes. For example, 250V + selectable 179V / 209V / 250V. Are 600V diodes will be enough for this purpose?
I found this tip on the forum (thread was discussing 400V Output PS):
The diode has to withstand the output voltage plus the additional reverse voltage induced when the primary side MOSFET turns on. If the transformer turns ratio (primary: secondary) is 1:3 then for a 375 V bus voltage (265 x sqrt 2) the addition diode stress is 1125 V giving diode minimum diode rating of 1125+400=1525 V.
Is that correct or misleading statement?
5) What happens if one by mistake reverses polarity of primary or secondary? On prototyping stage, everything is possible. Is TOP272EN design crash proof? I meat reverse start and end of winding, not completely swap connection.
6) Does it makes sense to replace L1 DC choke with common mode choke in order to reduce HF noise?
7) What is the best material for EMI / FRI common mode chokes - ferrite or powdered Fe-based nanocrystalline?
8) 5 - 10 sec. soft startup (gradual increase of output voltage) - is it somehow possible?
Thanks in advance for any suggestion(s).
Comments
Hi, Pi-Traveler !
Thanks a LOT for so comprehensive and professional reply! Your suggestion to use HiperPFS + HiperTFS is greatly appreciated. PFS725EG + TFS764HG seems like the most optimal choice for 400W peak, 150 - 300W continous supply.
There are still some design issues which needs to be hammered out:
1) Standby circuit needs to be removed completely. RDR249 - which components should be taken out? Is there any design reference for HiperTFS w/o standby?
2) Does TFS have the same reflection of multiplied voltage from secondary to primary (on main, not standvy flyback transformer)? Is it still safe to use for example 2 stacked 125V secondaries with voltage doublers in order to get 500V?
3) I don't fully understand your statement - You won't want to use a toroid for a transformer design. You'll likely run into leakage inductance problems. Toroids have very low leakage inductance compared to EI, or in case of small sized tranformers with multiple windings this is not the case? Please correct me if I'm wrong here. I'm not stubborn on toroids, I'm keen to get lower HF noise possible. May be large ferrite pot core will be the best in this application?
4) May be it makes sense to build additional low-voltage secondary winding with dummy load, and connect TL431 there, thus eliminating multiple zenners on 500V output?
Thanks in advance.
LinuxGuru -
1) Standby circuit needs to be removed completely. RDR249 - which components should be taken out? Is there any design reference for HiperTFS w/o standby?
You will need some kind of standby power supply to provide the power neccessary to operate the HiperPFS and HiperTFS converters. You don't have to make this standby power supply available as an output of the power supply though.
2) Does TFS have the same reflection of multiplied voltage from secondary to primary (on main, not standvy flyback transformer)? Is it still safe to use for example 2 stacked 125V secondaries with voltage doublers in order to get 500V?
The reflected voltage issue with flyback power supplies isn't an issue in this case. With a forward converter, the MOSFET is delivering power to the output when it is ON and energy is being stored in the forward converter output inductor. I'm not sure why you'd want to use voltage doublers. That seems like it would just make the design more complicated. You could fairly easily do two 250V volt secondaries that share a common return and are symetric around a common reference though. This is fairly standard in analog designs where you need symetric voltage rails such as with op-amps and other amplifiers.
3) I don't fully understand your statement - You won't want to use a toroid for a transformer design. You'll likely run into leakage inductance problems. Toroids have very low leakage inductance compared to EI, or in case of small sized tranformers with multiple windings this is not the case? Please correct me if I'm wrong here. I'm not stubborn on toroids, I'm keen to get lower HF noise possible. May be large ferrite pot core will be the best in this application?
It really depends on the application. For inductors, toroids work great. For transformers there can be issues with manufacturability (they're much more expensive to wind), issues with leakage inductance (it can be harder to get tight coupling between the primary and secondary windings), etc. In audio situations, while you do want to keep high-frequency noise to a minimum, low-frequency noise is much more problmatic. This is because high-frequency noise is typically easy to mitigate with component choice, transformer design, PCB layout, shielding and simple low-pass filtering. Removing low-frequency noise (especially noise in the audio frequency range) is much harder to filter as it will typically require high-value components (tend to be large physically).
4) May be it makes sense to build additional low-voltage secondary winding with dummy load, and connect TL431 there, thus eliminating multiple zenners on 500V output?
Unless you can get GREAT coupling between the two windings, it'd be better to just modify the feedback circuit so it's sensing directly. Primary side feedback topologies (check out of LinkSwitch-II products for many examples and write-ups) can work really well but at higher powers can be difficult to work with.
-The Traveler
Hi, The Traveler,
Thanks for feedback!
I'm still missing one point. Yes, HiperPFS requires auxiliary 18-25V auxiliary power supply (RDR 236). How standby PS of HiperTFS (which is connected AFTER HiperPFS) can be used int his case (RDR 249)? HiperPFS won't start becuase it requires auxiliary supply from HiperTFS standby which is powered off. Please explain.
May be HiperPFS required auxiliary must be powered on from TinySwitch for example?
LinuxGuru -
A PFC boost stage has no isolation and doesn't require the IC to be switching for there to be an output voltage. The output voltage won't be 380 volts as it would with HiperPFS running but there will still be 120-370 volts present depending on the line input voltage (80-265 Vac rectified).
In RDK-249, the standby power supply runs as long as the input voltage is somewhere between 120-370 VDC. Once the standby supply is up and running, HiperPFS can then be turned on which will bring the output of the PFC stage to ~380 VDC. Once the PFC stage is running, the main output (forward converter) on the HiperTFS can be enabled.
Since it looks like our Hiper product family parts are going to be what you need for your project, I would suggest posting any further questions in the high-power forum. I know a fair amount about our Hiper product family ICs but the engineers answering questions in the high-power forum should be able to address specific questions about those products in much better detail than I can. As we haven't added the Hiper products into PI Expert yet (just PI XLS but we're actively working on it), the engineers answering questionss in the high-power forum will likley also be able to help with some of the specifics of using PI XLS to create your PFS/TFS power supply.
I hope this has been helpful. Thank you again for your interest in our products. If I can be of further assistance, please let me know.
-The Traveler
Hi, The Traveler,
Yes, thanks, your posts are very helpful since I can't get at a glance all details and nuances how your products are working.
Another thing to consider - LM431 / feedback circuit. HiperTFS should have 4 outputs - 500V (varable power consumption), and 2x6.3V + 1x100V (quite constant power consumption). Is it really necessary to connect LM431 / feedback to the 500V output? Dropping excess voltage from 500V will require a string of zenners or 50K / 7-.-10W resistor to get current down to 10mA, something I wish to eliminate.
May be its more logical to connect LM431 / Feedback to 6.3V output, or this not optimal becuase of constant power consumption (in opposite to 500V where power consumption varies at very large extent)?
LinuxGuru -
How you setup your feedback circuitry will depend on a few factors:
1 - which output has the most critical noise requirements.
2 - power levels involved.
3 - transient load conditions.
The outputs not directly providing feedback to the IC will tend to have a bit more low-frequency noise, their regulation won't be as tight and might have less than ideal transient load responses.
Typically, you will want your feedback to be provided by your highest power output. In your design, your highest power output is the output that also sees the most load transients (from what you're telling me). I think this would probably make it the ideal choice for providing your feedback.
An LM431 doesn't pull very much current to run the opto in the feedback circuitry. If you're using zeners to limit the LM431 voltage to 20V or so, you'd need 480V of zeners in series. At 10mA, the zeners would be dissipating ~5W (480V x 10mA).
-The Traveler
1) Transformer construction mentions air gap. Is air gap necessary? Is there any DC current passing through transformer winding(s)?
2) I'm consider to use amorphous or nanocrystalline tape wound toroid core instead of EI ferrite in order to reduce size, leakage inductance and HF noise. Any thoughts?
BTW, is it necessary to ground metal toroid core?
3) Design Example Report 218. There as ia capacitor #2 - 2.2nF 250VAC, connecting transformer primary pin 1 and secondary pin 10/11. What is the purpose of this capacitor?
Additionally, there is a bug in DER-218 schematic - pin 4 of transforer connected to nowhere. This is probably PDF conversion glitch.
4) In order to get 500V output voltage - 2 stacked full bridge rectifiers, reducing voltage requrements to diodes. For example, 250V + selectable 179V / 209V / 250V. Are 600V diodes will be enough for this purpose?
I found this tip on the forum (thread was discussing 400V Output PS):
The diode has to withstand the output voltage plus the additional reverse voltage induced when the primary side MOSFET turns on. If the transformer turns ratio (primary: secondary) is 1:3 then for a 375 V bus voltage (265 x sqrt 2) the addition diode stress is 1125 V giving diode minimum diode rating of 1125+400=1525 V.
Is that correct or misleading statement?
5) What happens if one by mistake reverses polarity of primary or secondary? On prototyping stage, everything is possible. Is TOP272EN design crash proof? I meat reverse start and end of winding, not completely swap connection.
6) Does it makes sense to replace L1 DC choke with common mode choke in order to reduce HF noise?
7) What is the best material for EMI / FRI common mode chokes - ferrite or powdered Fe-based nanocrystalline?
8) 5 - 10 sec. soft startup (gradual increase of output voltage) - is it somehow possible?
Some questions and observations about your design:
-The Traveler