http://www.diyaudio.com/forums/lounge/200865-sound-quality-vs-measurements-124.html#post3058676
(for an MSc, teaching at university, your written English is appalling bad)
Liching, a Constant Voltage Transformer is a form of ferroresonant voltage regulator. Transformer & Inductor Design Handbook (1,2,3rd ed) by McLyman has a decent description & design procedure. its an old technology, and I dont think its been mentioned in a textbook for 30+ years. they really do maintain a constant output voltage for varying input voltage and load.
English is not my mothers tonque. How is your German? Of zullen wij het verder in het Nederlands doen?
Andrew, yeah, hence my earlier comment about a unity power factor boost regulator. Its quite possible to make one of them 97% efficient (the best I've seen to date is 99.3%) and at these low voltages it should be fairly easy* to get going (* yeah right)
for the truly brave I'd suggest an offline PFC flyback with dual outputs. not easy to design, and ever so slightly lethal to debug. there is also the minor issue of EMI. at 100W the input EMI filter needs on the order of 80dB attenuation at 150kHz. and then there's the regulatory hurdles....*sigh*.
for the truly brave I'd suggest an offline PFC flyback with dual outputs. not easy to design, and ever so slightly lethal to debug. there is also the minor issue of EMI. at 100W the input EMI filter needs on the order of 80dB attenuation at 150kHz. and then there's the regulatory hurdles....*sigh*.
I know this, it uses a leaking transformer resulting a serial inductance. Its the same as an inductance between rectifier and cap filtering out the ripple. However the voltage is lower and thus gives lower amp performance
Its up to you to take me serious or not. As a matter of fact most Swiss citizen doesnt take Dutch serious met hun grote mond.
sort of, but you left out the saturating portion (which is the important part). it can be modelled as an LC filter followed by a saturable transformer. and whilst off-topic, it follows from the constant-droop saturable inductor I proposed earlier.
alternatively the leakage inductance could be cancelled out by series-resonance.
none of these are really serious solutions (although I imagine they would look kinda neat, and could probably be used to extract cash from audiophools). obviously a better approach is to use a very low leakage transformer, size the cap bank appropriately and use a decent layout.
Near saturation not in saturation because you will blow the transformer then. That makes this concept very critical. Just use a normal transformer and big cap. There is no need for stable voltage at all!
I beg to differ, maybe you should read the whole thread. There is a little more at stake here. I hope if we do not agree with you we shall not be shot.
Would you concur that 100 Farad would be a reasonable value reservoir cap for a 1 watt class AB amp. We are trying to establish when is bigger big enough. Even the argument of the battery has been touched early in the thread, according to you should we consider the battery CCC or just capacity. Does impedance play a role....?
I am not convinced that your M.Sc is standing you in good stead regarding the challenges being discussed and solved in this thread.
I am not convinced that your M.Sc is standing you in good stead regarding the challenges being discussed and solved in this thread.
the Jan 1990 IEEE Trans. on Power Electronics has a paper "A simple scheme for unity power factor rectification for high frequency ac buses" by Ridley & Vorperian, which is a bridge rectifier/capacitor filter fed through a tuned series LC network. their AC bus is 20kHz, but:
at 50Hz for Vdc = 70V, Pdc = 100W, Rdc = 49R and Q=2 the series L-C is Lo = 307mH, Co = 33uF and the RMS winding voltage is 63Vrms. for Q = 2/pi thats 149mH and 68uF. It should be feasible to design the transformer to have this value of leakage inductance.
it has almost unity power factor.
at 50Hz for Vdc = 70V, Pdc = 100W, Rdc = 49R and Q=2 the series L-C is Lo = 307mH, Co = 33uF and the RMS winding voltage is 63Vrms. for Q = 2/pi thats 149mH and 68uF. It should be feasible to design the transformer to have this value of leakage inductance.
it has almost unity power factor.
Terry,
others have tried to explain this unity power factor correction stuff to me. I am still not getting it sufficiently to understand how we can correct a capacitor input filter fed by a transformer.
Your post refers to a paper. Is it possible to post that paper? and if necessary follow up with an explanation of what it is telling me?
Maybe just a schematic would be enough to get me going in the right direction.
others have tried to explain this unity power factor correction stuff to me. I am still not getting it sufficiently to understand how we can correct a capacitor input filter fed by a transformer.
Your post refers to a paper. Is it possible to post that paper? and if necessary follow up with an explanation of what it is telling me?
Maybe just a schematic would be enough to get me going in the right direction.
Nico,
My last few posts have really just been a distraction. as you rightly point out the objective is to formulate some rules re. cap size, and I've been playing silly buggers with the transformer leakage inductance, which is quite unhelpful and probably ought to be ignored.
a summary is that the transformer coupling is jolly important, and should be maximised (IOW the leakage inductance should be minimised).
Andrew (and Nico) #457 is talking about putting a series-resonant L-C circuit in between the transformer and the rectifier. If this is tuned to resonate at the AC line frequency it will draw sinusoidal current (its a bandpass filter) and turns the xfmr leakage inductance (in this case referred to the secondary) from an inconvenient parasitic (that can be minimised but never eliminated) into a useful component.
and the rough numbers I calculated dont look that unreasonable. its very hard work designing a transformer for a specific leakage, but it can be done - eg an EE core with a split bobbin will have a well controlled leakage inductance. Its not really a good idea though, I was just waiting for a sweep to finish.
Andrew if you want I can email you a copy of the paper.
A unity power factor boost rectifier is an entirely different beast. take a standard boost converter (series L, FET to 0V, diode to Vout) and place it between the rectifier and the filter cap. then stick a multiplier in the SMPS controller, between the error amplifier and the current comparator. feed the multiplier from the full-wave rectified (but not smoothed) input. that forces the boost FET current to have a |sin(wt)| envelope, so the xfmr winding currents are sinusoidal.
http://www.st.com/internet/com/TECH...AL_LITERATURE/APPLICATION_NOTE/CD00004002.pdf
is for one operating off-line, but there is no reason why one cant run one from the secondary of a transformer (its a lot safer). HTH
My last few posts have really just been a distraction. as you rightly point out the objective is to formulate some rules re. cap size, and I've been playing silly buggers with the transformer leakage inductance, which is quite unhelpful and probably ought to be ignored.
a summary is that the transformer coupling is jolly important, and should be maximised (IOW the leakage inductance should be minimised).
Andrew (and Nico) #457 is talking about putting a series-resonant L-C circuit in between the transformer and the rectifier. If this is tuned to resonate at the AC line frequency it will draw sinusoidal current (its a bandpass filter) and turns the xfmr leakage inductance (in this case referred to the secondary) from an inconvenient parasitic (that can be minimised but never eliminated) into a useful component.
and the rough numbers I calculated dont look that unreasonable. its very hard work designing a transformer for a specific leakage, but it can be done - eg an EE core with a split bobbin will have a well controlled leakage inductance. Its not really a good idea though, I was just waiting for a sweep to finish.
Andrew if you want I can email you a copy of the paper.
A unity power factor boost rectifier is an entirely different beast. take a standard boost converter (series L, FET to 0V, diode to Vout) and place it between the rectifier and the filter cap. then stick a multiplier in the SMPS controller, between the error amplifier and the current comparator. feed the multiplier from the full-wave rectified (but not smoothed) input. that forces the boost FET current to have a |sin(wt)| envelope, so the xfmr winding currents are sinusoidal.
http://www.st.com/internet/com/TECH...AL_LITERATURE/APPLICATION_NOTE/CD00004002.pdf
is for one operating off-line, but there is no reason why one cant run one from the secondary of a transformer (its a lot safer). HTH
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.