Dear Ju,

Hope you are well and all other colleagues,

There are some solutions:

The first solution is to use a transformer to step up the voltage and then rectify this voltage then filter out the output from the rectifier. You did not tell how the amount of the ripples contained in the output DC voltage. If it is difficult to raise the voltage in one step to the required voltage you have then you have to cascade the transformers.

The other solution is to convert the AC 50 Hz in DC and using this DC to feed a class D high frequency oscillator then following the high frequency oscillator by a ferrite cored transformer followed by a rectifier and filter.

The second solution will lead to a much more compact design.

Best wishes

This must be done through an incremental transformer and AC/DC converter.

dear Mahdi Saadatmand thanks for reply. but it was so general :)

Dear Ju Cheng, this material (via the link below) may be useful for you. Go through and see how you can amend the design to suit yours using any of the available Circuit Simulation software.

https://www.google.com/url?sa=t&source=web&rct=j&url=https://physlab.lums.edu.pk/images/2/24/Hv_power_supply.pdf&ved=2ahUKEwiG9cHI3t7oAhUKyxoKHd51BEUQFjAJegQIBxAB&usg=AOvVaw1zRPMO483LIwFa6EvaBMii

dear Abdulrahaman Otuoze thank you for your help, but that is not a "High Power" one.

Change the transformer

Dear Ju Cheng ,

instead of an answer, I have some questions first, sorry:

Does the picture show a commercial device which fits your requirements?

Do you just need the power supply for your work, or is the development of the device the core of your work?

If you just want to use it: For the sake of time as well as money, buy a device. The price will include part of the development costs but if you do the development yourself then this will take the whole development costs (at least for material, and you will need some special parts which are pretty expensive).

If the development is an inherent part of your work: Well, an interesting time stretches before you. ;-) General ideas:

I guess your first approach will have an efficiency of 0.7 .. 0.8. So, the max input power will be in the range of 1900 W .. 2200 W. The first and second stage (rectifier and push-pull stage or H-bridge (made from IGBTs?)) have to cope with this power.

First main decision: Switching frequency. Low frequency (20 kHz) results in low power loss in the switches, low loss in a large transformer, inexpensive rectifiers, and large (expensive) capacitors. Higher frequency (50 kHz .. 200 kHz or above) means higher power losses or more expensive switches, transformer core, rectifiers, but smaller transformer and smaller capacitors.

Regarding the capacitors behind the rectifier on the secondary side: How much AC voltage is acceptable at the output?

The problem of controlling the output voltage remains: If all works fine with 20 kV and nearly 80 mA, I would first try driving the switches by a PWM signal. One prerequisite for this approach is that the capacitance behind the secondary rectifier is so large that it takes a considerable number of halfwaves to charge it to the target voltage. Two problems with a large capacitance before the output: How do you discharge it if the user wants to decrease the voltage without load? And in the case of a short circuit or a spark at the output, the power delivered by a large capacitor might be dangerous.

Two ideas in order to have small or no capacitors at all at the output:

1. Use two H-bridges (or similar switch configurations) and two transformers, and drive the bridges with a phase difference of 90°. => During the switching process of one bridge the other bridge delivers power.

2. Instead of driving the bridges by a PWM signal, vary the supply voltage of the bridges in the range 75 V .. 300 V, perhaps by charging a large but cheap (400 V or so) capacitor by a PWM driven switch or (probably simpler) by PFC.

Do you have access to a three-phase power outlet? Using a three-phase rectifier would help by reducing the necessary magnitude of the subsequent capacitor.

dear Joerg Fricke

Thanks a lot for your very detailed answer.

yes that is a commercial version of what I want

and I need to make it by myself based on some reasons :)

could you also please introduce me some good references about this subject?

Hi Ju Cheng ,

I'm sorry that I don't know of any recommandable books or articles on the subject. The physical laws remain the same all the time, of course, but the practical limits are shifted by the technological improvements, especially in the semiconductor industry. So, publications older than, say, 10 years, would explain the problems correctly but wouldn't give the optimal solutions.

Provided you have a solid knowledge of electromagnetics and semiconductors, you can start by relying on the information published by the manufacturers of the parts you potentially need. Many of them as well as distributors (RS Components, Mouser, Digi-Key, Farnell, Arrow, to name but a few) provide parametric search tables on their web sites. With the help of those you can get an impression whether parts with certain properties exist, and if so, you can download the datasheets.

By reading the datasheets, you will find application hints and in some cases references to application notes of the respective manufacturer.

I would start with a supply of fixed 20 kV without feedback control: Rectifier (no problem), capacitor (no problem), bridge of MOSFETs or IGBTs (parametric search), heat sinks for those transistors (passive/active ?), drivers for those transistors (low side, high side, param. search), diodes providing a path for the current through the primary winding during/after switching (param. search), the core of the transformer (param. search, appl. notes), diodes for the secondary side (param. search), and high voltage capacitor (param. search). The bridge can be driven by any microcontroller (pay attention to the necessary delay between switching off one transistor and switching on another). You'll need a high voltage probe for a scope, a wideband (some MHz) inductive current probe would be helpful, too, and a suitable load (1,5 kW!), of course.

If this circuit works satisfactory you can go on to variable output voltage, and finally to feedback control.

Dear Joerg Fricke

Awesome explanation!

Thank you so much for your help and your time.

Dear Ju Cheng,

Using power electronics converters can solve your problem.

I suggest to you: to cascade:

a non-controlled rectifier (2kW) ---> a continuous bus ----> An elevating chopper ~ 300V - 5..20kV + its control.

Good luck.

Regards.

El Mahfoud Boulaoutaq

thank you El Mahfoud Boulaoutaq

could you also please provide me some references?

Dear Ju,

One way to make such a power supply (if you do not mind about power factor correction) is to implement the following stages:

1. AC to DC: with a full bridge rectifier this will give you a DC link voltage of around 320 V.

2. DC to AC (or pulsating DC): either unidirectional or bidirectional DC pulses, depending on converter topology.

3. High frequency power transformer: selection based on selected switching frequency and its secondary insulation must be rated for 20 kV continuous rating.

4. AC (or pulsating DC) to DC: half bridge or full bridge rectifier (high voltage diodes) based on selected converter topology.

The converter topology could be a push-pull, a flyback ,etc. The flyback transformer can be designed with two output windings, a high voltage (e.g. 5-20 kV) and a low voltage (e.g. 5-20 V). In this way you can monitor the low voltage winding to regulate the output, the high voltage output will be cross-regulated.

Another way is (no power factor correction):

1. AC to DC: full bridge rectifier.

2. DC to AC (pulsating DC): through high frequency transformer.

3. Diode-capacitor voltage multiplier

This will be less efficient that the previous way, however, the diodes-capacitors on the high voltage side do not need to be rated for the full 20 kV. Their rating depends on the input voltage of the voltage multiplier.

Note: whatever you choose to design, make sure you follow the guidelines for working with high voltage equipment. During the testing, always wear safety glasses and cover everything with a box (can be transparent if you want to have a view of the circuit) because components can explode.

Remember that 20 kV at 80 mA is lethal.

Dear Ju,

Hope you are well and all other colleagues,

There are some solutions:

The first solution is to use a transformer to step up the voltage and then rectify this voltage then filter out the output from the rectifier. You did not tell how the amount of the ripples contained in the output DC voltage. If it is difficult to raise the voltage in one step to the required voltage you have then you have to cascade the transformers.

The other solution is to convert the AC 50 Hz in DC and using this DC to feed a class D high frequency oscillator then following the high frequency oscillator by a ferrite cored transformer followed by a rectifier and filter.

The second solution will lead to a much more compact design.

Best wishes

Dear Ju,

The simplest solution was provided by Mr Abdelhalim Zekry, use of an AC step-up transformer.

To achieve regulation of the output voltage, you can design an AC to AC converter in order to vary its output voltage from 50 to 200 V AC for example. Then use a (or as many required) 50/60 Hz step-up transformer to step the voltage from 200 V AC to 15 kV AC. Then use a high voltage bridge rectifier to convert the AC to DC.

The output DC voltage will be measured and transmitted to the power converter via a device which provides galvanic isolation (e.g. isolation amplifier).

Your response "I need to make it by myself based on some reasons :)" does not explain much. This grand adventure will be very educational and will probably take you more than 2 versions before you meet your relatively unspecified specifications. Remember, you will need to address fault conditions, too. Do not work on this project alone (for safety reasons).

Good luck!

Carl

Hi all,

I would advice against a solution with a 50/60 Hz transformer (or a cascade thereof). While the first two stages (transformer(s), rectifier) look pretty simple, and the weight of the transformers of more than 20 pounds each doesn't matter as long as the power supply is not meant to be used in a mobile application, Pandora's box opens in stage three:

Suppose a ripple voltage of 50 V = 1% of 5 kV is tolerable. With a bridge rectifier and 50 Hz, the intervals between charging of the output capacitor are (roughly) 10 ms. For 80 mA, this results in a capacitance of 16 uF.

1. Look up the price for 16 uF @ >= 20 kV.

2. @ 20 kV, the energy stored in this capacitance is 3200 Ws.

An electrostatic precipitation requires high voltage (to produce ionisation - corona) but low current. A simple circuit that has been used for decades is the Cockroft-Walton circuit, also known a DC voltage multiplier, cascade rectifier, "DC ladder, see https://www.google.com/search?q=cockroft-walton+voltage+multiplier+circuit&rlz=1C1NHXL_enUS729US729&oq=cockroft-walton&aqs=chrome.2.0j69i57j0l5.15777j0j8&sourceid=chrome&ie=UTF-8 .

There are many older and new book that describe the circuit and its operation. No need for transformers, just diodes and capacitors.

Please let me know if you need more information. I used such circuits up to 1.5 MV DC for more than 30 years. I can be reached at [email protected] and ggela@berkshireelectric. com.

I would suggest using a step-up transformer based on the space you can accommodate in your unit. Then build the AC to DC converter and increase your voltage using Cockcroft-Walton voltage multipliers. The first decision would be to choose the right transformer in terms of size, which defines the AC voltage the rest of the circuit has to deal with.

Can you share any of the details about your project that may help to focus our responses to your goals? A quick Google search using "dust precipitator HV current draw" showed me more than I'd ever considered on the topic, leading me to believe that our comments (although all correct) may not be optimal for your specific needs. There is a significant range of secondary considerations that might guide your design choices.

The easiest thing to do, is purchase a HV supply in the form of a brick, which uses low voltage I/O for reading the V / I of the output, and driving I-Limits and V-Level using whatever regulation circuit you have or construct as per your application. They are compact, well insolated (isolated are available as well) and very compact.

If you need to create this yourself, I designed a 2.5kV @10mA PMU, and it was straightforward, using a custom-wound toroid measuring approximately 2-inches overall diameter, 1/2" thick. Druiven with pulsed 12V from an oscillator, and the only tricky part, can be current sensing -- but at high voltages, you can get away with a sense resistor. However, in my case, I used an opamp output pin to maintain a virtual ground, configured as a current mirror, with the feedback input and the value of the feedback resistor(s) would scale and provide a voltage from which the HV current could be derived. The oscillator was gated by a control signal, and the V regulation just used an HV divider, spaced out appropriately on standard FR4. The VME type connectors allow for specialized pins, which include a very well isolated blind mating HV connector and HV wire. So the whole thing can be cleanly implenmented.