![WP_20170227_006[1].jpg](/community/data/attachments/556/556173-de963b5d51cf6b782f840b321bef44a7.jpg)
![WP_20170227_007[1].jpg](/community/data/attachments/556/556188-62e3aedf8e9656424fad561c9a10ca92.jpg)
If I want 300v DC out of the reg,should the trafo be 250v AC or 300v AC then?
250Vac is 353.6Vpk +- mains tolerance.
Assume -6% for mains low and you get 332.3Vpk.
subtract 1.4Vf for a solid state rectifier and you have over 330Vdc on the smoothing capacitor.
That leaves ~ 30V to be dropped through the regulator to give your target of 300Vdc.
Why would you think that 300Vac would be required?
Assume -6% for mains low and you get 332.3Vpk.
subtract 1.4Vf for a solid state rectifier and you have over 330Vdc on the smoothing capacitor.
That leaves ~ 30V to be dropped through the regulator to give your target of 300Vdc.
Why would you think that 300Vac would be required?
![WP_20170228_001[1].jpg](/community/data/attachments/557/557596-ecb5fb0aff87e42ff023336a5d4456e0.jpg)
Almost unreal situation. The more voltage you ask - the less conducting angle for rectifier - the more current should be supplied to replenish the energy in the filtering capacitor - the more capacitance you should use to support the current. More current - less efficiency due to active and reactive loss - more heat - possible failure.
You counted amplitude voltage - it means the diodes would conduct just a tiny fraction of the period and even in the situation the output voltage = amplitude voltage is unachievable.
There is also the consideration of ripple voltage at the output filter of the raw part of the PS which should be taken into account. Big ripple voltage can spoil the SSHV job really.
What does it mean? That you should design your raw voltage more carefully than just guess "what voltage the transformer may be". Long time PSUD2 is here - simple and elegant solution to estimate everything.
Quote:
Originally Posted by Ionmw
Hi Salas,
great, not shabby at all! Finally something in the ball park.What detail of component or module you think is most critical for increasing this parameter?
What would be the max. expected value (PS still nice behaving) for current slewrate capability assuming 2Adc at 160V?
Thanks,
Those questions involve a full scaled up power components redesign to be answered
Please, would you care for a direction (sketch or name or link or even a hands-on comment on the most critical details for such redesign)?
To remind the problem, the output current requirement is up to 200mA in 25Ohm+700uH load (class A amp at 2A/150V bias) with mandatory request to provide I_load harmonics too (-3dB at >500KHz), if this is at all feasible.
Thanks,
First you need to locate and choose high amperage and high voltage depletion mode Mosfets. Then an analogous enhancement mode output Mosfet. Then you got to see what dissipation is going to be involved. That last one could be enormous and impractical. Depends a lot in the amp's idle to peak ratio. A shunt has to burn at peak+ at all times. Assuming that's doable you then analyze the open loop. Which most likely will prove slow because devices at ease with 2A at 150V will be unavoidable big and slow. A typical capacitance multiplier with a big BJT series pass element is rather a much better bet.
Thanks! Any example for each kind of FET you mentioned? Each one goes where in your circuit?
On your later suggestion: what really bothers me is a simple question ... if using such a typical circuit I can obtain my specs, then why I find it so hard to see real world examples.Why, in contrast, I see only MOSFET based high power high voltage regulators that are only a bit faster (25KHz I would call very fast - but still not cutting the 1MHz open specs)?
I say, if such ultrafast B+ source would exist with typical capacitance multiplier BJT series, would not then make this thread uninteresting?
Then, for both reasons: 1. lack of schematics of BJT series with such specs and 2. here reality-check I see 5000+ replies, logical conclusion is that some reasons must exist for these facts. What are in your opinion those reasons?
Regards,
My IMHO for Ionmw:
I can't see the initial picture in the post where you justified the need for 1MHz bandwidth. Taking into account that "fundamental frequency" is marked as 100kHz and "relevant harmonics" as 1MHz: you were speaking about square wave. While I understand the nice wishes about wide bandwidth it seems in the audio world it is better to stick with reality. And the reality suggests that 100kHz square wave is far from audio band.
The simulation of Salas' shunt reg suggests that it saves its effectiveness at least to 1MHz, but it is greatly depends on parts used and the PCB-layout. I doubt the simulation has had any prove in real-life measurements. The second problem - high current needed. Measures for the conversion could take the "simplistic" to the "severe". This was already pointed to by Salas.
I can't see the initial picture in the post where you justified the need for 1MHz bandwidth. Taking into account that "fundamental frequency" is marked as 100kHz and "relevant harmonics" as 1MHz: you were speaking about square wave. While I understand the nice wishes about wide bandwidth it seems in the audio world it is better to stick with reality. And the reality suggests that 100kHz square wave is far from audio band.
The simulation of Salas' shunt reg suggests that it saves its effectiveness at least to 1MHz, but it is greatly depends on parts used and the PCB-layout. I doubt the simulation has had any prove in real-life measurements. The second problem - high current needed. Measures for the conversion could take the "simplistic" to the "severe". This was already pointed to by Salas.
Sorry, can't catch your point here. The Salas' schematic does not relate in any way to "capacitance multiplier". The main problem of such multipliers arises from the emitter-(source-, cathode-) follower layout with stability problem due to phase shifting. The Salas' schematic mainly overcome the problem by design and this is the main reason why it is good. And it could be "translated" to BJT, you just have to do it yourselfes. I do not have any motive to do it considering the schematic works!
Hello Poty,
Thanks for reply!
Doctor's orders (aka. Salas) to look into typical series BJT capacitor multiplier. I am a simple person in these matters and honestly until this project I did not appreciated the science (and need) for good power supplies. Lesson learned, big thanks to Salas!
The current (I_load) waveform requested by amplifier from PS is sawtooth, not square. PS is constant voltage but floating, as amp is a current out Class A floating bridge. Bias is 2.5A and signal is 200mA (same in PS as in amp too, 1:1). Also requested is a THD+N<1mA @ 100KHz sawtooth. High voltage requirement comes from the load, current and frequency: it is an inductor(16Ohms+0.7mH).
My project is to adapt an audio amplifier for a laboratory experiment where I need a precisely linear 10T/sec induction coil (spectroscopy stuff). Yes, for audio this is definitely beyond over-engineering, actually I would say it is hyper-engineering (=nonsense). Yet, the know-how plus hands-on plus fundamental components selection principles applies. Also some frequency range is common to audio: because in this experiment the inductor must also perform in the 5KHz to30KHz range with 1A amplitude @5KHz (that's why I put bias so high). But for PS the problem is "simple": same bandwidth is requested due to the essential THD+N specs.
So here I am asking for help from my audio engineering friends, freaks and colleagues. If the amp or its PS it will prove not good enough for my lab experiment, then I will have perhaps at least an hyper-engineered audio amplifier for home (as my consolation prize), or a very expensive +1KW heater for cold winters.Win-win !
If the experiment works good I will mention your all kind support in the resulting article!PS.: If it will fail... I still don't have the proper "current driven" PA loudspeakers though ... so this could be a next project if money is enough
Back to my problem, I have a new question: do you have this simulation project in LTSpice or Tina and could it be shared with me?