Steve, the capacitor is indeed supplying 3.5A to the load, but it has to get recharged during the peaks of the line voltage. That recharge time is only a few degrees. During that time, all the coulombs delivered to the load have to be replaced, so the peak current is high. It's in the opposite direction of the load current, but it still causes power dissipation in any series resistance. If you make the capacitor larger, the average power dissipation increases.
Do you have the ability to run simulations? You might be able to calculate the dissipation, but a simulator will do it for you.

EDIT: I re-ran the sim with a smaller similation step size (10us), and the average power dissipation of the 60mΩ resistor was 5.65W (I previously reported 4.5W). This is an accuracy issue with Spice, not a change in the circuit.

 

Hi Ron,

I have been running simulations to come up with the inrush part of the design. I just simulated it in Multisim and am getting 1.43W of dissipation under load, which is far from your 5.65W value. Perhaps I didn't set up my schematic properly or my spice software is poor?

*** UPDATE ***

For some reason, the model for the full-wave bridge was wrong. I changed it, now I am getting 8W of dissipation under load. Maybe it is time for a relay or many mosfets in parallel?

Did you import the model for the transistor I am using? I was considering using a relay, but I really don't like relays that much. I know they make long-life relays, but I cannot see them exceeding a MOSFET's life.

Steve

 

Maybe I should just use an NTC device? There seems to be no free lunch with this inrush stuff. Even the relays are being difficult to select. I suppose that I need to ensure that there is no motor draw during inrush and startup, otherwise the relay will take a beating.

Steve

 

Hi Ron,

I have been running simulations to come up with the inrush part of the design. I just simulated it in Multisim and am getting 1.43W of dissipation under load, which is far from your 5.65W value. Perhaps I didn't set up my schematic properly or my spice software is poor?

*** UPDATE ***

For some reason, the model for the full-wave bridge was wrong. I changed it, now I am getting 8W of dissipation under load. Maybe it is time for a relay or many mosfets in parallel?

Did you import the model for the transistor I am using? I was considering using a relay, but I really don't like relays that much. I know they make long-life relays, but I cannot see them exceeding a MOSFET's life.

Steve

Steve, I was just using LTspice with their generic diode model, a 2200uF cap, and a 60mΩ resistor. I just tried it with the MOSFET. The transistor dissipation was 4.4W. Below is the latest schematic I simulated.
A thermistor (NTC) seems like it might be a good candidate.

 

Ron,

Thanks very much for doing this, I appreciate your time and effort

I used the same load as in your schematic, and am getting just over 4W, so we are finally in agreeance. I really don't like the idea of using an NTC, mainly because it is like having a tiny heater inside the enclosure. Since the final prototype will be used around a milling machine, I wanted to fully seal it. Sealing in a 260 degree multi-watt heater isn't necessarily a good thing to do.

I think I am going to scrap this idea, unfortunately, these mosfets are some of the best I could find for the voltage. I could parallel 3 of them, but then it gets expensive and I'd have to mount them all to the heatsink. I will either decide on using a relay or an NTC. Technically, I could use both of these, since it wouldn't mean catastrophic failure within a second for the NTC, whereas it would for the resistor.

Steve

Back to the drawing board!

 

Okay, I ended up running with the relay bypass idea, plus put one on the input. This isn't exactly a simple schematic, but I feel confident that it will work.

I used NTCs that can take the rating, even if the relay doesn't bypass them. I want to bypass them for many reasons: less wasted power, less heat in the enclosure, and I do not need to worry about a cool down period.

The process is as follows:
If the spindle is enabled, the microcontroller will check state : spindle_power. If true, then it will enable the PWM to the optogate. If not enabled, it will start the turn on procedure. It will power the input relay, then wait until the optoisolator detects the 16V supply startup. After a short delay (0.5s?), it will bypass the NTC inrush resistors. It will set the spindle_power state to high.

Do you guys think this is okay? It's a bit over complex, eventhough I tried to keep it really simple.

Steve

 

Just be aware that the repetitive peak current through your relay contacts will be on the order of 50 amps.

 

I see that in my simulations now, except their much higher than 50Amps, probably because most of the lines are ideal. I added in some realistic resistances, etc.

How do people do this with regulator drives? Am I trying to hard to smooth the DC? Should I just be full-wave rectifying, then adding a small capacitance? Then, PWM a quasi-fullwave rectified voltage? This seems like a simple task, yet it has been quite daunting.

Steve

 

For relays, isn't the average current what really matters? I suppose that if you close the contact, or open it, during a peak, then you may cause an arc which could weld, or somehow damage the contacts.

Steve

 

What problems do you think you might have if you don't limit inrush current?

 

Ron,

It isn't the limiting of inrush current that I am questioning, hence the NTC resistors in-line. I am switching these over when the capacitors are charged to reduce heat, power consumption, and to cool them down in case the circuit needs to be reset.

You mentioned that the relay bypassing the NTCs will see surges of 50A. I am wondering if these surges will destroy the relay.

Steve

 

Ron,

It isn't the limiting of inrush current that I am questioning, hence the NTC resistors in-line. I am switching these over when the capacitors are charged to reduce heat, power consumption, and to cool them down in case the circuit needs to be reset.

You mentioned that the relay bypassing the NTCs will see surges of 50A. I am wondering if these surges will destroy the relay.

Steve

I was wondering the same thing. I don't have much experience with relays.

 

Maybe I will build and find out the hard way

I'm doing the PCB layout right now, so give it a few weeks and I shall return with excitement or disappointment !

Steve

 

I finished up the schematic and PCB for my cnc controller with the DC drive. I will be sending the design out today and, once finished exams, I will solder and test it. I should know by next month if everything is okay If it works, I will create a project posting with gerbers, code, etc.

The servo portion is adapted from a design by Jeffery Kerr, except I took out a couple things that I felt were unneeded. But, as I said, I will report back if everything works well.

Steve

https://www.freedfm.com/freedfm/0007089702153894/results/plots.htm

Notes:

I didn't calculate resistor values on the PC interface page for the infrared spindle speed sensor, but I will update them once the design works.

I realize some of the annotations went over the cap values, these will be fixed.

The board is designed to be encased, then many connections soldered to panel mount connectors.

 

i need 8 channel relay board schematic

 

I might have designed the DC drive for no reason At least I can still run a D/A signal from the microcontroller to the drive, and I have the 15V supply to make a decent 0-10V converter for the VFD.

Finally, there are decent spindles available on ebay for a good price.

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&ssPageName=STRK:MEWAX:IT&item=250332735281

I have been searching ebay for something like this for years! I just need to find out if I get 220 single-phase into the VFD. Apparently the runout is 5 microns, but I find that really hard to believe. Hopefully it is between 5-10 microns though, since I want to do some PCB milling eventually and don't want to be breaking bits.

Steve