Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

AC-DC power supply design

Status
Not open for further replies.

blazini36

Newbie level 5
Joined
Dec 3, 2018
Messages
10
Helped
0
Reputation
0
Reaction score
0
Trophy points
1
Activity points
114
I'm trying to design a self contained DC bus power supply. Specifically it's for a 400v 2200w AC servo drive. Trying to get this all down on 1 PCB for simplicity. I've been using this drive for some time with just 240v mains rectified to 340vDC and some filter caps. The inrush current on charging the caps from flat is rather large, Which for a while I was getting by just upsizing the fuses. I could use the NTC ICL with a contactor bypass controlled externally but like I said I'm trying to get this all down to a PCB with a heatsink and I don't know of any suitable magnetic contactors.

I whipped up a schematic just to put the design down to something I can look at. I'm not very familiar with IGBTs and triacs which is where I need the most help on this. I have 2 triacs in the circuit to cut off the split single phase mains input controlled with gate drivers. There is an NTC on the DC bus side to handle the inrush. Trying to come up with a means of bypassing the ntc after the caps have charged but this is where I'm having the most trouble. I'm not really sure how to derive a suitable voltage or current reference to operate the gate of the IGBT or even if that is the best solution.

There's notes on the drawing for 120/240v input but that'll require a bit more thought so at the moment I'm only concerned with 240v input.

View attachment output.pdf
 

you run the risk of frying the igbt - try a relay ...

- - - Updated - - -

p.s. you only need one triac - the other is redundant - also the thermistor could be 2 x 10W 68E resistors in series, with no triacs on the mains - just a switch - then close the relay after Vcap = 250VDC say - or just after 2 seconds ....
 

you run the risk of frying the igbt - try a relay ...

Prefer not to use a relay if possible. If I were to use a relay the NTC has to go on the line side since I have found no PCB mount relays that can handle the DC voltage/current. Not using any off-board contactors/relays.

p.s. you only need one triac - the other is redundant -

Yes but one line remains live to the rectifier, I may remove the second triac but it's not my main concern atm.

also the thermistor could be 2 x 10W 68E resistors in series, with no triacs on the mains - just a switch - then close the relay after Vcap = 250VDC say - or just after 2 seconds ....
Not sure I see a benifit of the resistors over the NTC. Was planning on using a something like a SL22 5R012 that costs under $2. The triacs are there so that an enable signal can turn on the power supply through the linear regulator. A "switch" would wind up being another relay. I prefer solid state components over magnetics. Controlling the IGBT or relay or whatever it winds up being that bypasses the current limiting device is the main question. I need a voltage or current reference from the board to switch this bypass, I do not want to control it externally. So the switch has to operate from a voltage reference on the DC bus side when it gets above 250v or so as you say or when the inrush current levels off, but how do I get this reference? I can find no suitable shunts and haven't figured it out otherwise yet.
 

Hi,

I don't think that a 2200W power supply with just rectifying the input to a capacitor bank is according nowadays rules.
I'd use a PF stage with soft start to smoothly charge the capacitors.

Klaus
 

Suppose you pre-charge the capacitors directly from an auxiliary power supply? Install a plain diode and resistor, calculated so caps take a second or two to reach 90 percent running voltage. Power ratings can be small due to short operating time .

Turn it on first briefly, then apply full power. The diode prevents accidental backward current flow. Put several diodes in series to create a threshold voltage so it won't conduct during brief voltage drops in your circuit.
 

Suppose you pre-charge the capacitors directly from an auxiliary power supply? Install a plain diode and resistor, calculated so caps take a second or two to reach 90 percent running voltage. Power ratings can be small due to short operating time .

Turn it on first briefly, then apply full power. The diode prevents accidental backward current flow. Put several diodes in series to create a threshold voltage so it won't conduct during brief voltage drops in your circuit.

Not really sure what I would use as an external power supply, again trying to keep everything self contained.

If you omit the IGBT from the drawing there, the only actual issue with the DC bus side is the after the caps have charged the NTC should not really remain in the circuit as it will continue to have enough resistance to heat up and drop it's own resistance. Controlling the bypass device is all I really have to figure out with that.

I'm not aware of every way that exists to do this but a couple of ideas that I've had so far. A current shunt feeding a current monitor IC that will in turn operate the IGBT, relay whatever. Problem there is that no suitable shunts seem to exist. At that current level they are huge and not suitable for a PCB.

if it were a lower voltage bus a resistor divider might make sense to derrive a < 20v control voltage from the caps to operate the IGBT gate. I see huge power consumption from resistors when trying to calculate a suitable divider. Something similar can likely be done with zener diodes or similar, any insight on this?
 

Hi,

I don't think that a 2200W power supply with just rectifying the input to a capacitor bank is according nowadays rules.
I'd use a PF stage with soft start to smoothly charge the capacitors.

Klaus

I was looking at power factor correction ICs briefly. Do you have any suggestions that are more specific?
 

To illustrate here's a simple simulation of 2000uF charging through resistor and diode from 240 VAC source (peaks=340V). Timeframe 1 second.

2000 uF cap chrg 1 sec thru 80 ohm resis diode supply 340 VAC.png

Looking at your schematic it could be done by adding a switched resistor in parallel with each of your triacs. These can take the place of your NTC thermistor since all do a similar job for a while after power-up. It's a question which works better, a thermistor in the path all the time, or resistors in the path briefly.

What watt rating should the resistor be? The net dissipation is about 120W for one second. However it's hard to be sure of the amount of stress resulting from high power applied for a second or two, then idle for a long time. You may find 5 or 10W is sufficient.
 
To illustrate here's a simple simulation of 2000uF charging through resistor and diode from 240 VAC source (peaks=340V). Timeframe 1 second.

View attachment 157417

Looking at your schematic it could be done by adding a switched resistor in parallel with each of your triacs. These can take the place of your NTC thermistor since all do a similar job for a while after power-up. It's a question which works better, a thermistor in the path all the time, or resistors in the path briefly.

What watt rating should the resistor be? The net dissipation is about 120W for one second. However it's hard to be sure of the amount of stress resulting from high power applied for a second or two, then idle for a long time. You may find 5 or 10W is sufficient.

Hmm, how about putting the NTC in parallel with the triac rather than resistors,The NTC has the advantage of reducing it's own resistance if something goes wrong where the resistor remains a heater until it burns up.

It was never the intention to "leave the thermister in the path the whole time", bypassing it after inrush was always the intention.
 

It was never the intention to "leave the thermister in the path the whole time", bypassing it after inrush was always the intention.

Another possibility is a choke filter. This is useful for smoothing current spikes which are typical with capacitor filters. A choke might go on the AC side or it might go after the diode bridge (or both places). Perhaps in the area of 10-100 mH. If you choose too great a Henry value, it causes voltage drop to your load. It needs to handle the Ampere amount in your system. 2200W at 240VAC calculates to 9A.

There are also the sensible suggestions in posts #2 & 4. I didn't mean to distract from them.

- - - Updated - - -

Also consider that power factor error comes into play when capacitors and inductors are involved.
 

Another possibility is a choke filter. This is useful for smoothing current spikes which are typical with capacitor filters. A choke might go on the AC side or it might go after the diode bridge (or both places). Perhaps in the area of 10-100 mH. If you choose too great a Henry value, it causes voltage drop to your load. It needs to handle the Ampere amount in your system. 2200W at 240VAC calculates to 9A.

There are also the sensible suggestions in posts #2 & 4. I didn't mean to distract from them.

- - - Updated - - -

Also consider that power factor error comes into play when capacitors and inductors are involved.

Something like this after the rectifier?
https://www.digikey.com/product-detail/en/tdk-electronics-inc/B82622S0173L030/495-B82622S0173L030-ND/10413388
 

I realize that the idea of the resistor or NTC in parallel with the triac defeats the purpose of the triac. The triac is there as the "switch" that turns on the power supply, running the current limiter in parallel with the triac would be using the triac as the bypass and require an external on "switch".

I came across an application note which shows a circuit that Is mostly along the lines of what I was thinking. It shows a Mosfet as the bypass device which I substituted for the IGBT as I don't know of any mosfets that meet the voltage/current reqs. I'm not sure how to calculate the value of the passives required to make this work. I drew the schematic with no values.

https://www.vptpower.com/wp-content/uploads/downloads/2017/06/Inrush-current-DC-DC-converters-AN008-1.pdf
 

Attachments

  • output1.pdf
    36 KB · Views: 47

Something like this after the rectifier?

Yes, although its specs should be chosen to suit your system. Ampere saturation rating needs to be much higher than 9A. This simulation is a simple power supply, comparing the effects with a choke filter and without a choke.

power supply diode bridge choke compare to no choke current spikes.png

Notice the severe current spikes without the choke. The choke reduces stress on components.

Unfortunately choke filtering gets expensive. You may prefer to consider a greater Henry value which allows use of a lower saturation rating, and possibly less bulk and less expense.
 

Hi,

2200W at 240VAC calculates to 9A.
9A RMS .... is the optimum value and only true for non phase shifted pure sinusoidal signals.

For a rectifier - capacitor solution the RMS current (which is responsible for heating of the wires) is much higher.
An additional L will reduce the RMS current, but it still will be much higher than 9A.

In case the RMS current is higher by a factor of 1.5 ... then the generated heat in the wires is higher by a factor if 1.5^2 = 2.25.
The temperature rise of the wires will be higher almost by the same factor.
Don't overload your house wiring .... don't risk fire.

In my eyes the PFC stage is the only solution
* according mains quality regulations (all the simple rectifier solutions may be illegal, check the regulations of your country)
* to keep RMS current close to 9A
* with low power loss soft start feature

Klaus
 

Hi,


9A RMS .... is the optimum value and only true for non phase shifted pure sinusoidal signals.

For a rectifier - capacitor solution the RMS current (which is responsible for heating of the wires) is much higher.
An additional L will reduce the RMS current, but it still will be much higher than 9A.

In case the RMS current is higher by a factor of 1.5 ... then the generated heat in the wires is higher by a factor if 1.5^2 = 2.25.
The temperature rise of the wires will be higher almost by the same factor.
Don't overload your house wiring .... don't risk fire.

In my eyes the PFC stage is the only solution
* according mains quality regulations (all the simple rectifier solutions may be illegal, check the regulations of your country)
* to keep RMS current close to 9A
* with low power loss soft start feature

Klaus

Whether PFC is absolutely necessary or not, I'm not sure. I have used the setup of some caps with a rectifier and some contactors for over a year. Not that it's a bad idea, I'm just not sure I have the experience to implement it. I did spend a bit of time looking at a TI PFC chip and calculating the requirements to do the circuit. Looking at TI specifically because the datasheets are far more helpful than most other manufacturers. I'll see how far I get with this.
 

Whether PFC is absolutely necessary or not, I'm not sure.
According to IEC 61000-3-2 Limits for harmonic current emissions (equipment input current ≤16 A per phase), professional equipment (defined as: equipment for use in trades, professions, or industries and which is not intended for sale to the general public) with a total rated power greater than 1 kW doesn't need to comply with harmonic current standards. It may be however necessary to achieve a permission by the supply utility to connect it. Presently most utility companies are ignoring harmonic current load of professional customers.

Regarding your design, I don't understand the purpose of input triac switch and particularly not the intended 120/240V configuration. If you have 120 VAC single phase input, you need a voltage doubling rectifier, as usually implemented in 120/240V PC power supplies.
 

According to IEC 61000-3-2 Limits for harmonic current emissions (equipment input current ≤16 A per phase), professional equipment (defined as: equipment for use in trades, professions, or industries and which is not intended for sale to the general public) with a total rated power greater than 1 kW doesn't need to comply with harmonic current standards. It may be however necessary to achieve a permission by the supply utility to connect it. Presently most utility companies are ignoring harmonic current load of professional customers.

Regarding your design, I don't understand the purpose of input triac switch and particularly not the intended 120/240V configuration. If you have 120 VAC single phase input, you need a voltage doubling rectifier, as usually implemented in 120/240V PC power supplies.

Regarding the PFC concerns, this particular power supply is meant to work with a Servo drive that does not have it's own DC bus. It's a fairly common drive and I can certainly guarantee 90% of the users of the drive do not gu through trying to implement a PFC circuit. Caps, a recifier, an inrush limiter, controlled with magnetic contactors or SSRs is about the gist of what typically gets done. I'm just at a stage where I like to challenge myself to PCB designs and the PFC is another solution to the original problem so there's no reason not to if I can get it going.

The triac is for turning on the power supply from an external enable. The Power supply is not specifically for 300+ vdc, it is dependent on the input voltage. Again it's typically just a rectifier and some caps, at 120v input you would expect ~170vdc and at 240v input you would expect ~340vdc. This is intentional as it's somewhat universal but all of the components are specified to handle higher voltage or current. The exact output voltage isn't critical, if it's a 200v servo it would require a 120v input. if it's a 400v servo it could be either 120v or 240v input and possibly be speed limited by lower voltage.

The whole PFC thing might change the way that goes though.
 

Hi,

I'd use a PFC stage not only becasue of power factor correction or mains quality.

I'd use it because it solves the problem with charging high value capacitors.
There are many ready to buy ICs with relatively simple circuits.
Timing, inrush current ... can be easily adjusted. Low power loss, low heating.
And many really good application notes. I assume there are development boards available, too.

Klaus
 

Hi,

I'd use a PFC stage not only becasue of power factor correction or mains quality.

I'd use it because it solves the problem with charging high value capacitors.
There are many ready to buy ICs with relatively simple circuits.
Timing, inrush current ... can be easily adjusted. Low power loss, low heating.
And many really good application notes. I assume there are development boards available, too.

Klaus

Any suggestions as to a particular relatively simple IC? Right now I'm looking at UCC28180
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top