Switching DC step-down from 320V to 160V

[soldar]

I want to build a switching step-down circuit from 320 Vdc (rectified 230Vac) to half that.  It would handle no more than about 250mA output. Output voltage does not need to be much stabilized with output current variations. As long as it stays within, say 120 and 180 V it should be OK

I would like to get some help completing the attached circuit. R1 represents the max load but the load can be anywhere from zero to max. I believe the value of L1 is not critical and I have many taken from PSUs

I believe as long as there is a minimum load a 50% switching duty cycle would maintain roughly a 50% output voltage and the circuit could work without feedback which would simplify things a lot.

- The main question is what control circuit circuit to use, with or without feedback. What frequency to use. The higher the frequency the higher the losses so I am guessing somewhere between 15 Khz and 40 Khz.

- What switching element to use. It needs to be fast switching and withstand 400V

- What fast diode


Edited to add: probably a crowbar at 190V at the output in case the switch fails closed.

[Andy Chee]

What is the nature of your proposed load? This will dictate some of the answers to your questions. Without feedback, the output voltage will vary with line & load.

Incidentally, a crowbar across mains is a very bad idea!

Would prefer to see an isolated design, though perhaps you have a reason for not doing so?

Is this supply for a vacuum tube or something?

[Berni]

The easiest is to just use one of those chips you find in switching wallwarts. That way you get all the waveform generation and driving from one chip. Like for example a TEA1731

Using the classical buck topology is a bit tricky because your switch is way up there at the high voltage supply rail. With your output current that would make it 40W so a flyback topology is also not so good because the size of the inductor would start to become rather large. But you could use a forward converter topology to make the transformer more compact while having a low side switch.

Tho you do have chips like the NCP1060 that can do buck topology directly down from rectified mains, but this is a bit high in power for a single chip solution like that.

[soldar]

What is the nature of your proposed load? This will dictate some of the answers to your questions.

Is this supply for a vacuum tube or something?

I do not know all the details and I would rather not get sidetracked with this. Not inductive. Just assume resistive from 0 to 250 mA at the extreme. Probably about half that most of the time.

Without feedback, the output voltage will vary with line & load.

I already said that is acceptable as long as it doesn't go above, say 180V.

Incidentally, a crowbar across mains is a very bad idea!

I cannot see the problem with fuse and crowbar for last resort protection.

Would prefer to see an isolated design, though perhaps you have a reason [for not doing so?

Simplicity. Isolation is not needed although, of course, it is always good.

[NiHaoMike]

What you're really looking for is a "buck PFC", it can be done with many standard switching supply controllers if you can have the switch on the low side although the feedback would likely need an optoisolator.
If you're always running it from sine wave AC, you could cut the wave using the method the SR087 does. The power factor would be poor but not an issue for such a small load.

[Andy Chee]

If you want zero feedback, 50/50 duty cycle, you can’t go past a simple 555 astable mulvibrator.

Switching device can practically be any mains rated MOSFET, dictated by switching frequency and package size requirements.

Crowbar across mains is generally bad due to the lack of any supplementary method of current limiting (fuse not withstanding). Crowbars are better suited to isolated designs, where natural current limiting of the secondary transformer windings offers some protection.

[soldar]

The easiest is to just use one of those chips you find in switching wallwarts. That way you get all the waveform generation and driving from one chip. Like for example a TEA1731

Using the classical buck topology is a bit tricky because your switch is way up there at the high voltage supply rail. With your output current that would make it 40W so a flyback topology is also not so good because the size of the inductor would start to become rather large. But you could use a forward converter topology to make the transformer more compact while having a low side switch.

Tho you do have chips like the NCP1060 that can do buck topology directly down from rectified mains, but this is a bit high in power for a single chip solution like that.

Well, I would need a concrete and specific circuit diagram. With generic ideas it would take too much work for me because I am not familiar with these circuits.

Regarding "high side switching" it just depends on how you look at it. It can be low side.

[soldar]

What you're really looking for is a "buck PFC", it can be done with many standard switching supply controllers if you can have the switch on the low side although the feedback would likely need an optoisolator.
If you're always running it from sine wave AC, you could cut the wave using the method the SR087 does. The power factor would be poor but not an issue for such a small load.

This sounds very interesting. Could you give me more specific information?

[langwadt]

ir2153 ?

[soldar]

If you want zero feedback, 50/50 duty cycle, you can’t go past a simple 555 astable mulvibrator.

Switching device can practically be any mains rated MOSFET, dictated by switching frequency and package size requirements.

Yes I have thought of starting to experiment with a 555. I would need to supply it at 12 V which is not a problem as I have small transformers. The first thing I need would be a switching element of adequate voltage and frequency.

[soldar]

  ir2153 ?

looks interesting. I will study it. Thanks

[Andy Chee]

I would need to supply it at 12 V which is not a problem as I have small transformers.

You may not even need a transformer. A conventional zener diode shunt regulator hanging off the 320V rail should be able to power the 555.

[langwadt]

I would need to supply it at 12 V which is not a problem as I have small transformers.

You may not even need a transformer. A conventional zener diode shunt regulator hanging off the 320V rail should be able to power the 555.

that is where the ir2153  comes in, it is basically a 555, a zener, and a half bridge gatedriver all in one so8

[srb1954]

Why not just use a standard off-the-shelf switching PSU and modify it, if necessary, to give the desired O/P voltage or O/P current?

A commercial PSUs are so cheap these days and you would be hard pressed to do your own custom design for a lower price unless you are planning in making them in high volumes.

A commercial device would give additional benefits like:

• a nice case
• isolation
• EMI filtering
• voltage regulation
• current limit

and maybe even some safety certifications.

[soldar]

that is where the ir2153  comes in, it is basically a 555, a zener, and a half bridge gatedriver all in one so8

I see the ir2153 has been replaced by the IRS2153D. I will have to study it and that will take me some time. Thanks.

[langwadt]

that is where the ir2153  comes in, it is basically a 555, a zener, and a half bridge gatedriver all in one so8

I see the ir2153 has been replaced by the IRS2153D. I will have to study it and that will take me some time. Thanks.

is the same thing, it just doesn't need an external diode for the high-side driver boot strap

[jbb]

First, a general recommendation: unless you’re making high volumes and cost is a major concern, I thoroughly recommend you use a silicon carbide (SiC) diode in the switching path. It’ll generate way less irritating electrical noise (so it’s easier to make your circuit behave) and have slightly lower loss. May also decrease losses in the power MOSFET (specifically turn on switching losses).

Never used one myself, but how about a LinkSwitch-TN2 device? It’s an integrated buck converter controller and MOSFET.

https://www.power.com/products/linkswitch/linkswitch-tn2

[DavidAlfa]

Just assume resistive from 0 to 250 mA at the extreme. Probably about half that most of the time.

But does it needs voltage adjustment, or will it be powered at 160V all the time?
640 Ohms will dissipate 40W at 160V, if it's just a heating element, you could run it on PWM 25% duty cycle from 320V.

Orhm has several non-isolated high-voltage dc/dc converters:
https://www.rohm.com/products/power-management/ac-dc-converters-ics/ac-dc-converters-ics-pwm-qr?PS_ProductSupplyStatusText=Recommended#parametricSearch

https://www.rohm.com/products/power-management/ac-dc-converters-ics/ac-dc-converters-ics-pwm-qr/bm2p141x-z-product

[Zero999]

that is where the ir2153  comes in, it is basically a 555, a zener, and a half bridge gatedriver all in one so8

I see the ir2153 has been replaced by the IRS2153D. I will have to study it and that will take me some time. Thanks.

Use the IRS2153D, with a couple of MOSFETs to drive a transformer via a suitable AC coupling capacitor. If you want isolation use a 2:1 transformer, otherwise go for an autotransformer, which will require half the space. Put a bridge rectifier makes with high speed diodes and a smoothing capacitor on the secondary. If it's an isolation transformer, then a Y capacitor will need to be connected from the secondary to either the primary or earth, for RF suppression.

[soldar]

First, a general recommendation: unless you’re making high volumes and cost is a major concern, I thoroughly recommend you use a silicon carbide (SiC) diode in the switching path. It’ll generate way less irritating electrical noise (so it’s easier to make your circuit behave) and have slightly lower loss. May also decrease losses in the power MOSFET (specifically turn on switching losses).

Never used one myself, but how about a LinkSwitch-TN2 device? It’s an integrated buck converter controller and MOSFET.

https://www.power.com/products/linkswitch/linkswitch-tn2

Very interesting. Thanks. I have looked at the datasheet and it took me a while to figure it out but I think I have the basics. It provides an example with a 12V output where the output voltage is set by a divider formed by R1 and R3 so I know how to set the output voltage to whatever I want. R1 = 200K and R3 = 2K4 should give 169 V output. Other components also need to be adjusted for voltage and current but that sets the output voltage.

But all the examples are with low voltage outputs and I do not know if it would work with something like 170 V output. I cannot see anything that says it wouldn't but I do not want to assume anything.

I do not understand why D4 serves any purpose or is necessary.

If I could get my hands on a few samples I would definitely play and experiment. Mouser price is about $1 but then shipping is $20. I will need to look for some other source.

Looks promising. Thanks.

[soldar]

Just assume resistive from 0 to 250 mA at the extreme. Probably about half that most of the time.

But does it needs voltage adjustment, or will it be powered at 160V all the time?
640 Ohms will dissipate 40W at 160V, if it's just a heating element, you could run it on PWM 25% duty cycle from 320V.

Orhm has several non-isolated high-voltage dc/dc converters:
https://www.rohm.com/products/power-management/ac-dc-converters-ics/ac-dc-converters-ics-pwm-qr?PS_ProductSupplyStatusText=Recommended#parametricSearch

https://www.rohm.com/products/power-management/ac-dc-converters-ics/ac-dc-converters-ics-pwm-qr/bm2p141x-z-product

Thanks. That seems the same basic idea as the LinkSwitch except that with Rohm I can only see fixed output voltages.

[DavidAlfa]

The regulator uses its own VDD as the feedback, but higher outputs can be set using a zener in series.
Thus if it meant for 12V, adding a 150 V zener will do it, this method is also shown in the datasheet.

[soldar]

The regulator uses its own VDD as the feedback, but higher outputs can be set using a zener in series.
Thus if it meant for 12V, adding a 150 V zener will do it, this method is also shown in the datasheet.

Thanks. That would work although it is quite a bit more awkward than the two resistor voltage divider.

I will search to see what I can obtain because Mouser and other similar suppliers charge too much for postage.

[soldar]

that is where the ir2153  comes in, it is basically a 555, a zener, and a half bridge gatedriver all in one so8

I see the ir2153 has been replaced by the IRS2153D. I will have to study it and that will take me some time. Thanks.

Use the IRS2153D, with a couple of MOSFETs to drive a transformer via a suitable AC coupling capacitor. If you want isolation use a 2:1 transformer, otherwise go for an autotransformer, which will require half the space. Put a bridge rectifier makes with high speed diodes and a smoothing capacitor on the secondary. If it's an isolation transformer, then a Y capacitor will need to be connected from the secondary to either the primary or earth, for RF suppression.

Thanks but for now the other direct switching options look more attractive to me. But I will keep it in mind.

[Terry Bites]

Without isolation from mains power you're builidng a death trap.

[NiHaoMike]

This sounds very interesting. Could you give me more specific information?

https://www.microchip.com/en-us/product/SR087
It's limited to 50V but might be possible to use for higher by using a zener diode to lift the low side. Or just replicate the design using a common comparator chip. (I have done something similar although at low voltage, deriving about 6V from a 12V transformer which then gets regulated to 5V with a LDO, the 17V or so from the main rectified supply is used for some of the analog circuits on the same board.)

Without isolation from mains power you're builidng a death trap.

Isolation would only make things a little safer at 160V.

[DavidAlfa]

You microwave oven uses nearly 4KV, and guess, it's not isolated.
Calling it a death trap, but you don't know where this is being applied.

[Wolfram]

Without active feedback, your buck converter will enter discontinous conduction mode (DCM) under light load conditions, and the output voltage will no longer be proportional to the duty cycle. So practically you need either active feedback, or to replace the diode with a controlled switch in order to operate in forced CCM, both are valid solutions. Active current limiting is also strongly recommended. This project comes to mind as a potential source of inspiration: https://ludens.cl/Electron/latsup/latsup.html

[soldar]

Without active feedback, your buck converter will enter discontinous conduction mode (DCM) under light load conditions, and the output voltage will no longer be proportional to the duty cycle. So practically you need either active feedback, or to replace the diode with a controlled switch in order to operate in forced CCM, both are valid solutions. Active current limiting is also strongly recommended. This project comes to mind as a potential source of inspiration: https://ludens.cl/Electron/latsup/latsup.html

I believe that is the purpose of R4, to provide a minimum load but, you are correct that it needs to be taken into account.

I had no visited ludens.cl in quite some time. Good to see it is still around. That circuit would work for me although it is way more powerful than I need.

A plus for that circuit is that it uses UC3842 of which I must have many in my junk pile. Against it is that I would need to wind the inductance with a secondary winding to supply the IC. I will consider it even though the power I need is 40W or less.  Thanks.