Bridge rectifiers are relatively cheap, doesn't pay to go low value.
Max.

 

I will look into choosing the right capacitor. If you know any further reference about selecting a right capacitor for this application, I would appreciate it. I could use the same capacitor as shown in post #1 and will probably work. But that's a design I found on the web. I just don't want to put a random capacitor without knowing why I put it there.

 

Does zero crossing help minimize effects of inrush current?

Hello again,

Yes it does, quite a bit. The reason why it works is because of the difference in voltage levels at the time of turn on. With no zero cross detection and a 120vac line, it is possible to switch on when the line is at 170v (the peak) when the caps have 0v across them and that means it is turned on into a virtual short circuit, so the surge current will be very high. Relays can not handle this kind of activity very well and the contacts could weld closed.
With zero crossing detection, the line voltage is applied to the caps when the line is near 0v so there is no extreme surge. The cap voltage then follows the line voltage as the sine progresses, but this is more normal similar to driving a capacitor with a sine voltage which is acceptable. after the first half cycle the caps discharge a little, then start to charge again when the line comes up near the cap voltage again and that means only a more normal current. So there is never a time when there is an extreme voltage differential.

The modern way to deal with this is to use a special high current surge limiting thermistor in series with the bridge. Because the load is so extreme too though i am not sure if you can find a device that can limit the turn on surge and also deliver the current that the load needs. They also get very hot which i dont like either.

The second relay idea may not be bad either although you will have to experiment with the turn on resistor and the timing. That is if there is no load at the time.

Some circuit require a little more design work than others. Non regulated DC wall warts have a similar setup where there is a bridge an capacitance, but the transformer has built in leakage inductance so it limits the turn on surge. Other circuits are not as simple, that's just the way it goes.

I am sure you can find a zero crossing detector circuit on the web they are talked about all the time right here too.

Is the load really 4 HP?

 

Another question:

Regarding the image in my first post I see two capacitors in parallel of 1500uF/200V. What's the criteria to choose them? Can I use just one capacitor instead of two? How do I calculate the value that I need? I mean, I know the voltage rating should be at least 200V, but how do I choose the capacitance?

The capacitor value depends on the amount of ripple that is acceptable. DC motor bearings are known to wear differently with large changes of voltage, but since the line voltage probably will not be synced to the motor rotation it may not matter as much. You do have to think a little though about how much ripple your motor can put up with. Larger ripple voltage would cause more heating for example. I would think 10 percent ripple as a max but you could look into this more. The internal inductance may help a little here.

 

Hello again,

Yes it does, quite a bit. The reason why it works is because of the difference in voltage levels at the time of turn on. With no zero cross detection and a 120vac line, it is possible to switch on when the line is at 170v (the peak) when the caps have 0v across them and that means it is turned on into a virtual short circuit, so the surge current will be very high. Relays can not handle this kind of activity very well and the contacts could weld closed.
With zero crossing detection, the line voltage is applied to the caps when the line is near 0v so there is no extreme surge. The cap voltage then follows the line voltage as the sine progresses, but this is more normal similar to driving a capacitor with a sine voltage which is acceptable. after the first half cycle the caps discharge a little, then start to charge again when the line comes up near the cap voltage again and that means only a more normal current. So there is never a time when there is an extreme voltage differential.

The modern way to deal with this is to use a special high current surge limiting thermistor in series with the bridge. Because the load is so extreme too though i am not sure if you can find a device that can limit the turn on surge and also deliver the current that the load needs. They also get very hot which i dont like either.

The second relay idea may not be bad either although you will have to experiment with the turn on resistor and the timing. That is if there is no load at the time.

Some circuit require a little more design work than others. Non regulated DC wall warts have a similar setup where there is a bridge an capacitance, but the transformer has built in leakage inductance so it limits the turn on surge. Other circuits are not as simple, that's just the way it goes.

I am sure you can find a zero crossing detector circuit on the web they are talked about all the time right here too.

Is the load really 4 HP?

Yes, the PMDC motor that I will control is 4HP.

 

The capacitor value depends on the amount of ripple that is acceptable. DC motor bearings are known to wear differently with large changes of voltage, but since the line voltage probably will not be synced to the motor rotation it may not matter as much. You do have to think a little though about how much ripple your motor can put up with. Larger ripple voltage would cause more heating for example. I would think 10 percent ripple as a max but you could look into this more. The internal inductance may help a little here.

I was looking for a mathematical approach to get the capacitance values. I'm still not sure how I can choose them. Just by inspection?

 

Is everyone just being coy and pretending this isn't going to be powered directly from AC mains?

Is this to be powered from single-phase 120 VAC? Is such a load permissible where you live?

Keep in mind that the power factor is going to be quite bad, and the more input filter capacitance you add the worse it will be - i.e.the RMS to average ratio increases with increasing capacitance. The capacitors will be subjected to high ripple current at twice (assuming single phase) the AC line frequency and at the PWM frequency, and should be rated accordingly. Using several capacitors in parallel will very likely make it easier to find capacitors with suitable characteristics.

Given the power required, again assuming 120 VAC single phase, I would consider a 35 ampere bridge rectifier to be marginal, at best.

A solid state relay for inrush limiting will require a fairly substantial heatsink unless it is shunted by an electromechanical relay. An SSR made using back-to-back SCRs would be preferable to one based on a triac, but that is likely to be the case for something with adequate current rating. Since the rectifier itself will require a large heatsink, it might be worthwhile to consider a controlled bridge rectifier made with two diodes and two SCRs and controlling the SCRs for inrush limiting at first, then as rectifiers.

NTC thermistor inrush limiters are dubious for this power level. It is possible to use multiple thermistors in series to accomplish the required effective resistance. They will indeed run very hot, and shunting with a relay might be desirable.

An electromechanical relay to shunt some other current limiting device would of course have to have adequate current rating. Any circuit using a relay must be designed carefully to be sure that the relay drops out on brief power outage.

 

Is everyone just being coy and pretending this isn't going to be powered directly from AC mains?

Is this to be powered from single-phase 120 VAC? Is such a load permissible where you live?

Keep in mind that the power factor is going to be quite bad, and the more input filter capacitance you add the worse it will be - i.e.the RMS to average ratio increases with increasing capacitance. The capacitors will be subjected to high ripple current at twice (assuming single phase) the AC line frequency and at the PWM frequency, and should be rated accordingly. Using several capacitors in parallel will very likely make it easier to find capacitors with suitable characteristics.

Given the power required, again assuming 120 VAC single phase, I would consider a 35 ampere bridge rectifier to be marginal, at best.

A solid state relay for inrush limiting will require a fairly substantial heatsink unless it is shunted by an electromechanical relay. An SSR made using back-to-back SCRs would be preferable to one based on a triac, but that is likely to be the case for something with adequate current rating. Since the rectifier itself will require a large heatsink, it might be worthwhile to consider a controlled bridge rectifier made with two diodes and two SCRs and controlling the SCRs for inrush limiting at first, then as rectifiers.

NTC thermistor inrush limiters are dubious for this power level. It is possible to use multiple thermistors in series to accomplish the required effective resistance. They will indeed run very hot, and shunting with a relay might be desirable.

An electromechanical relay to shunt some other current limiting device would of course have to have adequate current rating. Any circuit using a relay must be designed carefully to be sure that the relay drops out on brief power outage.

Indeed, the board will be powered from single-phase 120 VAC. The load will effectively be 4HP. However, I have other loads from 1 to 3 HP which I might intend to use with the circuit I'm trying to design. All of them are PMDC motors.@ 120VDC.

 

Yes, the PMDC motor that I will control is 4HP.

Would this be a treadmill motor by any chance?
Max.

 

Would this be a treadmill motor by any chance?
Max.

Yes, it actually is.

 

Yes, it actually is.

You probably have to take the 4HP with a grain of salt, I have done empirical testing on some and they do not always come out to what they are claimed to be.
When you consider that if this T.M. was used on a residential 120v service, the rating for a typical outlet is 15a.
Another issue is if a flywheel is fitted and retained, in the T.M. controller application the acceleration is very controlled so they, have controlled acceleration and are also are prevented from having a high rpm setting at switch on.
The controllers are either SCR bridge type or PWM, both directly across the supply.
Max.

 

If you use some sort of controlled bridge or SSR, peak current control could be better managed by working back from the end of a half cycle rather than from start forward.

If you start with at 180 degrees and slowly decrease the firing delay to 90 degrees, you can ramp the current up very slowly - you could keep the peak charging current to less than the normal input current if you wanted to. This assumes that the load will not be on until the capacitors are up to voltage, and it does mean that with each decrease in delay angle there will be a small current pulse. Getting all the way to 180 degrees is tricky with a triac but easy with SCRs with steered firing since you don't have to worry about crossing over from the very end of one half cycle to the start of the next. Of course both half cycles must be managed.

All of this does take a bit of extra circuitry, but the problem of managing inrush current is not trivial at the power level contemplated. It might be worth considering a FET or IGBT between a conventional bridge rectifier and the capacitors as an alternative to an SSR or controlled bridge.

 

I was looking for a mathematical approach to get the capacitance values. I'm still not sure how I can choose them. Just by inspection?

Hi,

From your other post, wow, 4HP is almost 3000 watts. That's 25 amps from a 120vac source. Most 120vac circuits are designed for 20 amps tops but if you hook into the panel directly or set up a 30 amp circuit you can get more. You cant use regular hardware though like the normal two or three prong plugs and receptacles.

You can choose the cap based on the discharge over one half cycle. That is a typical approximation. The cap voltage gets replenished every half cycle. That means we can start with:
dv/dt=i/C

and solve for C and get:
C=i*dt/dv

and now just plug in the values.
The time interval dt=0.010 for 50Hz and 0.008333 for 60Hz.
The current 'i' is maybe 25 amps.
The change in voltage is 'dv' and you might want to keep that under 12 volts or something like that.
At 50Hz we get a value of about 20833 uf.
At 60Hz we get about 17361 uf.
The ripple voltage approximately halves for twice the capacitance, and doubles for half the capacitance, so at 50Hz if we use 10000uf we will get about 24v ripple, and if we double to 40000uf we get about 6v ripple approximately.
You could also do a simulation.

You will need adequate heatsinking for the diodes. 50 amp diodes would be a good idea.

 

Hi,

From your other post, wow, 4HP is almost 3000 watts. That's 25 amps from a 120vac source. Most 120vac circuits are designed for 20 amps tops but if you hook into the panel directly or set up a 30 amp circuit you can get more. You cant use regular hardware though like the normal two or three prong plugs and receptacles.

You can choose the cap based on the discharge over one half cycle. That is a typical approximation. The cap voltage gets replenished every half cycle. That means we can start with:
dv/dt=i/C

and solve for C and get:
C=i*dt/dv

and now just plug in the values.
The time interval dt=0.010 for 50Hz and 0.008333 for 60Hz.
The current 'i' is maybe 25 amps.
The change in voltage is 'dv' and you might want to keep that under 12 volts or something like that.
At 50Hz we get a value of about 20833 uf.
At 60Hz we get about 17361 uf.
The ripple voltage approximately halves for twice the capacitance, and doubles for half the capacitance, so at 50Hz if we use 10000uf we will get about 24v ripple, and if we double to 40000uf we get about 6v ripple approximately.
You could also do a simulation.

You will need adequate heatsinking for the diodes. 50 amp diodes would be a good idea.

Seems like most treadmills controllers (those for DC motor @90-120VDC) allow big ripple because the capacitors values normally range from 1000 to 3000 uF.

 

Seems like most treadmills controllers (those for DC motor @90-120VDC) allow big ripple because the capacitors values normally range from 1000 to 3000 uF.

Hi,

Are they 4HP also?

If so, then you may get away with lower capacitance. Motors are not as picky as power supplies for example.

What frequency will your PWM be, and what are you going to use your motor for?

 

Seems like most treadmills controllers (those for DC motor @90-120VDC) allow big ripple because the capacitors values normally range from 1000 to 3000 uF.

Like I said, I would be highly suspect of the HP claims, when you compare a N.A. made 4HP motor to a Chinese version.
Plus it is usually found that the maximum power may only be accessed on limited occasions in the application.
Are you retaining the flywheel?
Max.

 

Hi,

Are they 4HP also?

If so, then you may get away with lower capacitance. Motors are not as picky as power supplies for example.

What frequency will your PWM be, and what are you going to use your motor for?

Most treadmill motors HP range from 1-3HP, others are 4HP. Normally they are @90-120VDC. Mine is 4HP which doesn't have the controller. That's why I'm designing my own one. I have other treadmills motors from 1-3.5 HP. As Max says, rarely one reaches the maximum power. I'm trying to design my own controller which I can use with those HP variations. A software/hardware adjustment can be made to support that range of loads. They are for treadmill purposes. I will use PI/PID to control speed and compensate for the load, but that's another topic. I will later open a post in the project forum to update the progress of this project. I know nowadays there are lots of controllers over there, but I find that I can learn valuable concepts by making one by myself.

I haven't chosen the PWM frequency value. I'm still designing the power stage. I have finished the first part where I'm using a triac with zero crossing instead of the relay (I will upload the circuit later). Now I'm working in the selection of the bulk capacitor, then I will work with the selection of a proper MOSFET/IGBT with a proper gate driver. I will use STM32 as the main MCU that will generate PWM, and process external sensors , as current sensor (I will read motor's current), speed sensor, etc.

 

So you are using it for T.M. purposes, most as you mention only use 1.5hp - 3hp.
You can always use appropriate gearing through a belt reduction as the typical 4krpm & 4HP is not required.
For my similar controller I based it around Tahmid's design but my application was bi-directional, I am using a slot opto for feedback, just playing with the tuning right now.
I am using a minimum of 5k PWM frequency.
Max.

 

So you are using it for T.M. purposes, most as you mention only use 1.5hp - 3hp.
You can always use appropriate gearing through a belt reduction as the typical 4krpm & 4HP is not required.
For my similar controller I based it around Tahmid's design but my application was bi-directional, I am using a slot opto for feedback, just playing with the tuning right now.
I am using a minimum of 5k PWM frequency.
Max.

So returning to the bulk capacitor, I will then get away with some value from 1000 to 10000 uF @200V.

 

I used 4400μf for 2.5HP.
Max.