Will this psu work as intended?

[ajrenzetti]

I'm currently designing a circuit which will require around 100VDC 400mA (exact voltage and amperage can vary a little). I've designed the circuit for the power supply, which will be separate from the main circuit for my project.

I've built the schematic to reflect what the suggested circuit should look like for my project on the IC (TL783) datasheet. The resistors following the first TL783 IC are 30W each on a TO-220 package (which will be on a heat sink just in case), so 40 watts of power shouldn't be a problem for them. The two pads at the top labeled SW in and SW out are for the wires which will connect to an external flip switch rated for 115AC mains voltage. The board needs to provide the proper output, + or - a few volts and mA, from 115V AC mains, and provide short circuit protection. If you look at the TL783 datasheet, you will see a few example circuits, two of which are being used in this design. When i test this board out, I'll be attaching the output to a 20ohm 100W resistor to simulate the load, as both the power supply and H-Bridge circuit will be sharing the same "on/off" switch, but through a relay to isolate the voltages. When I go to test this board for the first time, it will be connected to an AC isolation transformer. The board has ratsnests all connected to neutral to help isolate some of the emf that would probably result from operation of this circuit.

Here's what I want to know before actually ordering the parts and building the board:
1) Obviously, will this circuit work?
2) If not, does it short out?
3) Where is the fault located and what can I do to fix it?

While I have experience in high voltage high power (such as induction heater and tesla coil) application design, I'm not really used to using adjustable voltage regulators like the TL783 for high voltage applications (usually I use low voltage regulators).

My apologies if the schematic and board are hard to read. I a bit new to eagle; used to using proteus, but I haven't installed it to my new pc. Plus I hear creating G-code for cnc routing is easier on eagle, especially when using Mach3 to control the CNC.

Hopefully this site doesn't have a rule stating against questions regarding transformer-less power supplies. I didn't see one in the terms of agreement, but I may have missed it when skimming through the terms.

 

[ronsimpson]

Let me take one piece at a time.
1) Here is probably where you started. That 6.25 resistor is for 200mA so 3.1 ohm is for 400mA. From OUT to ADJ there is 1.25 volts. Your two (1.5 ohm 30 watt resistors are over kill) The 3 ohms is right. But the wattage will be low.
2) You need "connecting dots" where line connect.

Then you used this to keep the voltage across the TL783 below 120 volts.

That works! BUT.
3) What happens when there is too much current and the top TL783 opens up and takes all the voltage? Does it also need "too much voltage protection"? I do not have time to think about it. BIG question.

4) There is a circuit with a diode from OUT to IN. Please do that if this is a "bench power supply".

 

[chemelec]

When you Rectify 115 VAC and Filter it, you get 162 VDC.

 

[kubeek]

A PSU connected directly to mains is a really bad idea. Do you realize that both the high and low side of that circuit will be a direct electrocution hazard as both the positive and negative rail will be at line potential?

 

[ajrenzetti]

Thank for the responses. The reason for this being a transformer-less psu is due to spacial constraints. Believe me, I do know the dangers of connecting directly to mains, and had i the option to do so i would have added a transformer. This supply, however, needs to be small, and must carry approximately 100v at 40W, and it is very tough to find such a psu that doesn't cost too much.

Thanks for the info, ronsimson and chemelec, I forgot about the higher voltage output of rectified AC voltage, and what would happen to the first TL783 since it can only output up to 125V. I was wondering if flipping the switches would solve the problem. The TL783 has a max output of 700mA, and the chip is more than capable of taking in higher voltages. So if I lowered the voltage first to an appropriate 100v output, the current limiting chip would be able to lower the amperage down to 400mA.

As for the comment about the wattage being low, I am not quite sure i understand you. It needs to be 40w +-10%, which is also why i have 2 30w resistors in series in the output: The entire output current would flow through them, therefore anything less than 40w may cause some heat issues.

so, do think switching the ic circuits would help? do you think i should replace the current sourcing limiter with a current sinking limiter circuit?

 

[Tony Stewart]

THe Cap is way too small.. 220uF * 20 Ω = 4.4ms or only 1/2 of the 8.3ms cycle.
try 1mF

I might consider a triac dimmer or 20A SCR

 

[ronsimpson]

Here is a thought. I know it is not quite right. Max input voltage is 140 volts.
Input voltage should work from 102DC to 140DC and the output will be 100V at 400mA.
The thing I don't like about your linear regulator is heat. It will run hot. Under a short circuit it will get very hot.
This switcher will not be hot and should survive a short with out heating. 90% efficient.
The circuit below is small. Remove the second IC for 400mA. Change the feedback resistors to get 100V out.
If you are interested in switching power I could work up something better.

 

[ajrenzetti]

ronsimpson,

Thank you for this info! Its exactly what i need. I just have one small question:

The info in chemelec's post is rather concerning; rectifying 115ac gets you 162vdc. While I'm not 100% sure that 162vdc is correct, i do know rectifying an ac current yields a higher voltage. so in order to prevent this i was considering two things: either use a resistor voltage divider using resistors of the correct wattage, or using the second part of my circuit to lower the rectified ac voltage to a compatible amount (125v like in the datasheet example). I was also considering using the circuit Tony Stewart suggested to lower the voltage.

Do you believe i should use one of those circuits to lower the voltage, or simply use the one you suggested without them? The major components will be attached to heat sinks cooled by a small fan, but this circuit will need to run for over an hour without failing.

 

[ronsimpson]

110 vac x 1.414 = 155V
115 x 1.414 = 162V
120 x 1.414 = 170V
If I were to design these for production, where we make millions, I would plan on it surviving 200 volts DC input.
Tony Stewart's ides is interesting.
I think a higher voltage switcher is the best answer.
A fan is a sign you have too much power loss in the box. (and not efficient) Fans are one of the first things to fail.
A good switcher will not need a heatsink.

How good must the 100 volts be? What is the load? Will the load all wise be 400mA or will it vary from 0 to 400mA?
What I am trying to learn is who good a supply you need.
Example: Can you live with 110V at 0mA and 95V at 400mA?

 

[ajrenzetti]

the load needs to be approximately 100vdc at 40 watts, but the voltage and amperage can vary by +- 10% as long as it doesn't exceed 40 watts, as that is what the device this supply is rated at (although 45w may prove ok to use with proper cooling). This supply will be powering an H-bridge circuit, so the dc load must be as smooth as possible, as well as provide some over-current and short circuit protection. as for the fan, i tend to use them for many of my circuits just as a precaution, although most of the circuits i build are much more powerful, so i guess its more of a force of habit than a necessity.

 

[ronsimpson]

Do you have LT Spice? If so you could see this run.
100khz switcher. Probably will work up to 140VAC in.
I have not testes what happens at low line. Probably needs some work.
Regulation is not right yet. Need some work!
For the first shot it almost work.
There are some noise problems yet.
R10 250 ohms is the load.
I have to remove the 17 volt supply and probably remove the U2 before done.
I don't remember what the current limit is set to but I think 600mA.
The IC is 8 pin small. M1 is to-220 type.

 

[Mikebits]

I would also change the voltage rating in Cap C2 shown in the original drawing. It should be at least a 300v rated cap.

 

[ajrenzetti]

The only spice cad software I've ever used was orcad capture (pspice) and the spice simulation from proteus, but the program always seemed to glitch up and never had the IC's I wanted to use. I'll see if lt spice has a demo version so i can try it is see how i like it.

as for R9, what wattage do you suggest i use for it? I've never been comfortable using resistors directly on the circuit that has the most power flowing through it, although i guess L5 is there to help.

 

[ajrenzetti]

I'm going to attempt creating a simulation profile on proteus and orcad of the circuit you showed me. It may take a while, as i only have orcad lite, and it seems i cannot import pspice library files into that version. As for proteus, i'm having trouble importing the library file i downloaded from linear technologies; the program doesnt seem to recognize it. Do you know of any way i can simply download the spice model script for that part? with that i can just rebuild the part and place the spice text in the script editor. With proteus i should be able to do an interactive simulation, if the circuit isn't too complex for it to do so.

 

[Tony Stewart]

THe Cap is way too small.. 220uF * 20 Ω = 4.4ms or only 1/2 of the 8.3ms cycle.
try 1mF

I might consider a triac dimmer or 20A SCR

 

[Tony Stewart]

Forget a linear supply . You will need at least 1mF @>120V rating which cost $35 (1), more if driving LEDs without flicker. What is your actual load device?

Define your cost and volume targets and expected ripple V max.

For a series power resistor , if you were dead set using a linear supply, you can use some halogen bulbs in parallel with a 10A bridge and 1mF cap. Halogen R with 3~ 5 Ohms cold resistance, putting in series or parallel depending on voltage.

then you can use an adjustable 3 terminal regulator to drop from 110~130 to 100V

To prevent no load overvoltage, you need something to preload the supply when load is off otherwise it can go up to 170V.

 

[ronsimpson]

as for R9, what wattage do you suggest

R9=1/4 watt
It really does not have current flowing through it because it is shorted out by the coil.

LT Spice is free.

 

[ajrenzetti]

Ok, so i gave up on getting orcad and proteus to simulate my circuits correctly, so i decided to try ltspice and was able to get it working. I was even able to import a TL783 spice model into it and got it to work properly. There is one thing i am confused about. The rectified 115V AC input does not go anywhere near the projected 170VDC. Is this because i am only using a single phase AC line voltage, or is it just the way the simulator works?

 

[ronsimpson]

The rectified 115V AC input does not go anywhere near the projected 170VDC.

That is because in spice a wave form is defined by its peak voltage not its RMS voltage.. So the line voltage is 160 volts.

Try this file. I changed to a "constant off time" PWM that is stable. I need to remove the 17V supply. I did that in another version and it works fine. (had to add some components)

You probably know. Push the "run" button. Then with the "probe" select nodes to watch. In the case of the output voltage that is not referenced to ground you want (node 2) - (node 10). The easy was is to either select node 2 and then in the "scope" change the formula by adding -node10. OR Put the prove on node 2, hold down the mouse button, move to node 10, release the button.

 

[ajrenzetti]

I ran the simulation and see what you mean. This should definitely help. I've also created this small power supply in LTspice using the LTC7138 you mentioned earlier, however instead of a rectifier bridge, I used a simple diode, since only one side will be ac when i plug it in (I will of course ensure that the plug is designed to be oriented a specific way to avoid reversing the polarity). Would this sort of supply work? or will the voltage still exceed 170VDC?