How would I determine what toroid I need to use? They don't look overly complicated to make, so I could do that if I could just find the core I needed.

Also, if I'm using a capacitor to provide the initial 36V to the solenoid, could I use a lower amperage DC/DC to charge the capacitor? That way I wouldn't need to build a 10A unit, I could just use one of the ones you linked on ebay.

The truth is initially when the capacitor of a dc/dc is empty, the dc/dc needs to supply a large current.

A 10000uF cap. does not take long to charge up using a dc/dc in this power class. Considerably less than 1 sec.

Yes you could use a dc/dc with let say 2 Amps. capability.

Given the fact that the duty cycle for discharge from 36v to 12v is very low.

10000uF only as reference for a starting point, even if I think this value might work.

Okay, it shouldn't be too hard to make sure the DC/DC goes off when the solenoid comes on.

I don't really want to overload the DC/DC, I don't want to have to worry about stressing the components and having potential premature failure issues.

I like the capacitor idea, I think it will work nicely, and if I can use a lower amperage DC/DC, then it will allow me to keep the heat down.

So just to run through everything, to make sure I have understood everything up to this point...

Start with a 12V source, send that to the DC/DC and get 36V. Send 36V to the capacitor to charge it, then run the capacitor to a MOSFET that is activated by my IC output. Run 12V source in parallel with the capacitor with a diode to prevent reverse current. When it is time for the solenoid to be fired, it receives 36V from the capacitor, until the capacitor is discharged, at which point it receives 12V from the source to remain closed until the IC ends its firing sequence.

Does that sounds right?

yes you fully understand it.

Don't worry about overloading a dc/dc for short time. They need to have such capabilities anyway.

If youhave for instance 5 Amps. steady current, normally you would use a 10 Amps MOSFET, or better 15 Amps.

If you work a MOSFET very close to it's margin, chances for failure are much greater. BJT (for instance power NPN) are a bit more robust.

I use for instance a TIP36C, with capability of 32 Amps, and surge capability of about 100 Amps. That never will be reached because the wires are too thin... Max. current is only 12 Amps.

I made experience with MOSFETs working them more close to the margin, and they burned out instantly when overloaded (having high temp. already).

How does a boost converter compare to a push-pull converter in terms of cost? What about overall temperature during operation?

Depends if you can obtain suitable transformer cores. But the effort is much more, especially if you have not build circuits like this one before.

Be aware there are many different magnetic materials used for toroid cores.

Yes they are not difficult to make, but sometimes all the space is filled up with magnet wire.

If the magnetic properties are unknown or not fully known all you can do is to add magnet wire, and try the coil in a circuit, and see about the efficiency.

Some coils will result in rather low efficiency because their parameters don't match the circuit. It is possible to improve it trying different diameter magnet wire and so different turns.

When I built my TL494 circuits, all cores I had here produced rather low efficiency. In the end, I used a ready-made 700uH toroid inductor, which works quite well.

So if you produce a PCB, prepare it for different coils, maybe a screw terminal would be appreciate.

And over-dimensionate the MOSFET or BJT.

I don't think max. efficiency is so much important here, but something close to 70% is desireable for any SMPS or dc/dc circuit.

The choice of power level must be made by you, if you want to try 2 Amps capability only, to charge a capacitor, even eventually use a small power resistor to limit charge current, or to build 5 Amps, 8 Amps etc., and overload it for a short moment.

 

The truth is initially when the capacitor of a dc/dc is empty, the dc/dc needs to supply a large current.

A 10000uF cap. does not take long to charge up using a dc/dc in this power class. Considerably less than 1 sec.

Yes you could use a dc/dc with let say 2 Amps. capability.

Given the fact that the duty cycle for discharge from 36v to 12v is very low.

10000uF only as reference for a starting point, even if I think this value might work.

Okay...so with a 10000uF cap, I could use something like this -

http://www.ebay.com/itm/120W-DC-DC-...069?pt=LH_DefaultDomain_0&hash=item3a688955f5

- to charge it up to 36V, right? The max output amperage rating on that thing is 5A.

I had to look up capacitance calculations for this, it's been a while since I've seen it.

With a 10000uF capacitor, 35V supply, and I want 210W from the cap to power the solenoid...

P = V^2/R
210W = 35V^2/R
R = 5.833 ohms

time constant = capacitance * resistance

time constant = 0.1F * 5.833 ohms = 0.5833 seconds

Which, apparently, is the amount of time it takes for the voltage to drop to 37% of its original value?

So, by that math, after 0.5833 seconds the capacitor would still be producing 25.9V.

I only need 36V for 0.08 seconds, so it would appear that you are correct, and that a 10000uF capacitor would work perfectly.

If that DC/DC I linked on ebay can charge a 10000uF cap, then I think I've got this part of my system figured out.

yes you fully understand it.

Don't worry about overloading a dc/dc for short time. They need to have such capabilities anyway.

If youhave for instance 5 Amps. steady current, normally you would use a 10 Amps MOSFET, or better 15 Amps.

If you work a MOSFET very close to it's margin, chances for failure are much greater. BJT (for instance power NPN) are a bit more robust.

I use for instance a TIP36C, with capability of 32 Amps, and surge capability of about 100 Amps. That never will be reached because the wires are too thin... Max. current is only 12 Amps.

I made experience with MOSFETs working them more close to the margin, and they burned out instantly when overloaded (having high temp. already).

I've definitely planned to use over-dimensioned MOSFETs. I was planning to use these to activate the solenoids. The amperage is way overkill (80A), but they are rated at 255W, which I felt would be sufficient to power the solenoids at 210W.

http://www.newark.com/fairchild-semiconductor/fdp3651u/n-ch-mosfet-100v-80a-to-220ab/dp/20M1178

Depends if you can obtain suitable transformer cores. But the effort is much more, especially if you have not build circuits like this one before.

Be aware there are many different magnetic materials used for toroid cores.

Yes they are not difficult to make, but sometimes all the space is filled up with magnet wire.

If the magnetic properties are unknown or not fully known all you can do is to add magnet wire, and try the coil in a circuit, and see about the efficiency.

Some coils will result in rather low efficiency because their parameters don't match the circuit. It is possible to improve it trying different diameter magnet wire and so different turns.

When I built my TL494 circuits, all cores I had here produced rather low efficiency. In the end, I used a ready-made 700uH toroid inductor, which works quite well.

So if you produce a PCB, prepare it for different coils, maybe a screw terminal would be appreciate.

And over-dimensionate the MOSFET or BJT.

I don't think max. efficiency is so much important here, but something close to 70% is desireable for any SMPS or dc/dc circuit.

The choice of power level must be made by you, if you want to try 2 Amps capability only, to charge a capacitor, even eventually use a small power resistor to limit charge current, or to build 5 Amps, 8 Amps etc., and overload it for a short moment.

Eventually we may contract out the construction of our boards, so I wouldn't have to worry about making the toroids. If that DC/DC I posted up above will work though, I may just use those, at least for prototyping.

I definitely think I have a much better grasp on all of this now.

Do you have a schematic for one of these ebay DC/DCs?

Also, can you explain what the two capacitors on the DC/DC boosters are actually for? Is it just for filtering, or is it for storage?

 

Okay...so with a 10000uF cap, I could use something like this -

http://www.ebay.com/itm/120W-DC-DC-...069?pt=LH_DefaultDomain_0&hash=item3a688955f5

- to charge it up to 36V, right? The max output amperage rating on that thing is 5A.

I had to look up capacitance calculations for this, it's been a while since I've seen it.

With a 10000uF capacitor, 35V supply, and I want 210W from the cap to power the solenoid...

P = V^2/R
210W = 35V^2/R
R = 5.833 ohms

time constant = capacitance * resistance

time constant = 0.1F * 5.833 ohms = 0.5833 seconds

Which, apparently, is the amount of time it takes for the voltage to drop to 37% of its original value?

So, by that math, after 0.5833 seconds the capacitor would still be producing 25.9V.

I only need 36V for 0.08 seconds, so it would appear that you are correct, and that a 10000uF capacitor would work perfectly.

If that DC/DC I linked on ebay can charge a 10000uF cap, then I think I've got this part of my system figured out.

The module certainly can charge 10000uF.
It might even be possible to use smaller cap. value than 10000uF.

I don't have the schematic, but I have modified the PCB, and that is the feedback input. Supply it with a fake voltage, and the module will stop putting energy through the coil. Which is the same as to turn it off.

I exchanged the MOSFET/Flyback diode as well...was not easy because it is glass fibre PCB.

I've definitely planned to use over-dimensioned MOSFETs. I was planning to use these to activate the solenoids. The amperage is way overkill (80A), but they are rated at 255W, which I felt would be sufficient to power the solenoids at 210W.

http://www.newark.com/fairchild-semiconductor/fdp3651u/n-ch-mosfet-100v-80a-to-220ab/dp/20M1178

yes this is a good idea. 80A well you see these thin leads, they are not capable to carry 80A. TO220 has a thermal carry capability of about 50W with good cooling.

I think these are appreciate to use for prototype.

What I have used for my circuits are large TO2 case devices, with higher thermal capability. Means if they are overloaded, they don't heat up so fast.

Eventually we may contract out the construction of our boards, so I wouldn't have to worry about making the toroids. If that DC/DC I posted up above will work though, I may just use those, at least for prototyping.

I definitely think I have a much better grasp on all of this now.

Do you have a schematic for one of these ebay DC/DCs?

Also, can you explain what the two capacitors on the DC/DC boosters are actually for? Is it just for filtering, or is it for storage?

No I don't have a schematic, I searched for UC3843, but found very few on the web, and none in the original datasheet.

The TL494 has a user manual in addition to the datasheet.

Yes I explain the purpose of the capacitors.

One is used to buffer the input voltage. The inductor will produce sharp spikes, and having 1 or two metres cable, dc/dc converters actually can't work anymore (without input buffer capacitor).

Having a generator with 80A capability, the value that you need for input buffer cap. depends on the wire gauge for the cabling, and the length.

Batteries, small 50 Hz transformers etc. all have internal resistance, so the buffering is used, the inductor current is not continuous/constant.

And also if you use long cables, you need buffering for the wire inductance/resistance.

some 100uF to 1000uF or 2000uF might be sufficient.

The output capacitor is kind of a storage capacitor!

It is required at least a small value. It depends on the ripple current that is permitted. For a solenoid this is not a problem.

I have used at some point of time only 100uF output storage cap., + large reactor choke as filter, with good results.

For the reason high voltage caps are expensive, while a choke does not care much about voltage.

dc/dc converters usually have 220uF, 470uF or 1000uF output storage capacitor.

It might be possible to increase it to 10000uF, but not neccessarily. Large capacitors may need some kind of current limiting.

You have calculated for 10000uF, so I think smaller value might work maybe 3300 uF.

 

How does a boost converter compare to a push-pull converter in terms of cost?

Depends on many factors.

What about overall temperature during operation?

Again, many factors but it would likely be more efficient than a boost so less total power to be dissipated.

Looking at the schematic you posted, it looks like it would have four coils?

No, it has one transformer which will be physically smaller than the inductor used in the boost for the same wattage.

I guess I've determined that I only need 210W for 8ms at a time (10% duty cycle on solenoid), after that I only need 21W (100% duty cycle on solenoid).

If a capacitor can provide the wattage/current that I need for the first 8ms, then does that mean I could use a lower wattage converter (of either type)?

I guess you could use a big storage capacitor and a boost to pump energy into it. The cap would have to be big enough to hold the voltage up during the 8 ms (210W) interval current drain.

At 10A drain and a 1V drop over an 8 ms duration, the cap would have to be at least 80,000 uF. An 80,000uF cap rated for 40V would be HUGE.

 

Okay...so I was finally able to find a schematic for a boost converter that I could understand and that didn't have a bunch of extra stuff on it. From that, I was able to apply it to a TL494 and add in my solenoid. The schematic should be attached.

I know I am missing a few things, it seems like the TL494 has a bunch of extra pins and I'm not sure what they are for.

I also would like to stop pulsing Q1 when Q2 is activated. What this should do is stop producing 36V from the DC/DC out, but since the 12V in is still connected through L1 and D1, I should get 12V when the boost converter is "off". This should work perfectly in giving me 36V from C3 and providing 12V when C3 runs out of energy.

I just need to figure out how to stop Q1 and turn on Q2 at the same time.

From here, I guess all I need to do is select components.

Let me know what you guys think!

 

I went through and tried to select components for everything in the schematic I posted. Ignore the designations on the schematic, those were just components I picked in eagle for the symbol.

Okay...

For the capacitors, I went ahead and picked 50V capacitors. I didn't want to limit myself to 35V, and since you can't buy 36V capacitors, I went ahead and jumped up to 50V.

Edit: Will raising the voltage limit of the capacitor affect that amount of usable capacitance? For example, if I only charge a 1000μF, 50V cap to 36V, would my usable capacitance be, say, 720μF?

C1 - I'll actually run two capacitors in parallel here, two of these.
I wanted 2000μF since my input will be 80A, but 2000μF caps apparently aren't all that common, so two 1000μF in parallel should work just as well.

C2 - I'm not sure if C2 is actually necessary. Takao, you mentioned using a 10000μF or 3300μF cap here, and now that I have a schematic drawn up, I can see why. C2 and C3 are essentially the same thing. Nevertheless, if I do need C2, I will just use the same cap as C1.

C3 - Haven't decided 3300μF or 10000μF yet. I'll probably try both. Here's the 3300μF I picked out. Here's the 10000μF.

D1 & D2 - I'm not really sure if I picked out a good diode here. I know I need a Schottky, and 10A is about 75% higher than what I am expecting (5.8A). Here it is.

R1 - Apparently R1 may not be necessary if the TL494 has a 'sense resistor' built in. Anyone know if this is the case? I have no idea how to pick resistors, so I haven't picked any.

R2 - No idea what to pick out here. Apparently this governs the actual output voltage of the booster. What do I do?

R3 - Same as R2, not sure what to pick.

Q1 & Q2 - 80A, 100V, 255W should be plenty.

L1 - This toroid looks to be about the same size as the one on the ebay boosters, so I figured I'd experiment with it first. As far as the windings go, is it just copper wire? Or is it insulated inductor specific wire? The 150W ebay boosters had 18 windings with 3 separate wires. That shouldn't be too hard to build, right?

IC1 - There seem to be a bunch of different versions of the TL494...I just picked the first one on the list. Will this one work?

It looks like I'm going to have 4 components in TO220 casings (D1,D2,Q1,Q2). What do I do about adding a heat sink? The ebay boosters mentioned that they can sustain lower amperages without any fans, and their heat sinks weren't all that big. Where can I find heat sinks similar to those that I could stick 2/4 TO220 cases on?

Thanks!

 

I went through and tried to select components for everything in the schematic I posted. Ignore the designations on the schematic, those were just components I picked in eagle for the symbol.

Okay...

For the capacitors, I went ahead and picked 50V capacitors. I didn't want to limit myself to 35V, and since you can't buy 36V capacitors, I went ahead and jumped up to 50V.

Edit: Will raising the voltage limit of the capacitor affect that amount of usable capacitance? For example, if I only charge a 1000μF, 50V cap to 36V, would my usable capacitance be, say, 720μF?

No capacity is the same, only voltage is lower.

However I can see a few problems ahead for the schematic. I would not recommend to produce a printed PCB from the current schematic.

For instance you need to connect the feedback voltage to one of the comperators, and on the other comperator input, you need a reference voltage. I use 7805 sometimes, divided into 1/2, so 2.5v reference.

There are two comperators, the other is for current sensing.

The output drive is often done using the built-in transistors.

C1 - I'll actually run two capacitors in parallel here, two of these.
I wanted 2000μF since my input will be 80A, but 2000μF caps apparently aren't all that common, so two 1000μF in parallel should work just as well.

C2 - I'm not sure if C2 is actually necessary. Takao, you mentioned using a 10000μF or 3300μF cap here, and now that I have a schematic drawn up, I can see why. C2 and C3 are essentially the same thing. Nevertheless, if I do need C2, I will just use the same cap as C1.

C3 - Haven't decided 3300μF or 10000μF yet. I'll probably try both. Here's the 3300μF I picked out. Here's the 10000μF.

yes looks good, capacitor values are not that critical.

D1 & D2 - I'm not really sure if I picked out a good diode here. I know I need a Schottky, and 10A is about 75% higher than what I am expecting (5.8A). Here it is.

this would be a starting point. They need a small cooling grid. If you don't use cooling for the time being, then you must monitor temperature very carefully, and in case of high temp., switch off immediately.

R1 - Apparently R1 may not be necessary if the TL494 has a 'sense resistor' built in. Anyone know if this is the case? I have no idea how to pick resistors, so I haven't picked any.

no it is not neccessary, but you connected it to the voltage reference?
Sometimes 0.1R etc. is added in the output path, but I don't think in the switcher transistor path. And one of the comperators is then used to sense current. But it can be left out.

R2 - No idea what to pick out here. Apparently this governs the actual output voltage of the booster. What do I do?

R3 - Same as R2, not sure what to pick.

Essentially the comperator will compare this to the reference voltage. So it is kind of a voltage divider.

For instance 20K pot. with 1K or 470 Ohm limiting inside the feedback path will work. Also 10K or 50K it is not critical.

Q1 & Q2 - 80A, 100V, 255W should be plenty.

I would not trust these ratings. TO220 can dissipate 50W with good cooling.
This is not the output power, it is the loss inside the transistor. So about 20 to 30% of the output power.

But in general, this device can work as a starting point.

L1 - This toroid looks to be about the same size as the one on the ebay boosters, so I figured I'd experiment with it first. As far as the windings go, is it just copper wire? Or is it insulated inductor specific wire? The 150W ebay boosters had 18 windings with 3 separate wires. That shouldn't be too hard to build, right?

I must say this one is very likely too small. Magnetic properties are unknown. You need a larger Toroid with 25 to 30mm diameter. Or maybe a ready-made coil. I made experience with various Toroids + wire, none of them produced good efficiency. I don't think this small core can carry more than 1A or 2A, if it will work at all.

IC1 - There seem to be a bunch of different versions of the TL494...I just picked the first one on the list. Will this one work?

they are pretty much all the same.

It looks like I'm going to have 4 components in TO220 casings (D1,D2,Q1,Q2). What do I do about adding a heat sink? The ebay boosters mentioned that they can sustain lower amperages without any fans, and their heat sinks weren't all that big. Where can I find heat sinks similar to those that I could stick 2/4 TO220 cases on?
Thanks!

These small heat sinks can carry 3W to 6W typically.
So they won't be enough very likely.

Maybe try VGA coolers (from eBay).
They are quite efficient, I have measured cooling capability of 50W for the smaller one's. They are very easy to use actually.

The metal pad is exposed upside down on the PCB, and then the cooler is pressed onto it.

But, you need to make some major changes to the schematic.

Have you seen this Thai circuit for TL494 that I uploaded here sometimes?

You only need to change it for boost configuration.
I have also made various changes, removed a lot of components, and added DTC.

Do you have the user manual for the TL494 available? (not the datasheet)

 

No capacity is the same, only voltage is lower.

However I can see a few problems ahead for the schematic. I would not recommend to produce a printed PCB from the current schematic.

For instance you need to connect the feedback voltage to one of the comperators, and on the other comperator input, you need a reference voltage. I use 7805 sometimes, divided into 1/2, so 2.5v reference.

There are two comperators, the other is for current sensing.

The output drive is often done using the built-in transistors.

Oh, this is just the first draft. Like I said, I'm very new to all of this. I just figured it would help if I had a starting point that I could build on.

What are the comparator inputs? I honestly can't tell from the data sheet. What do the comparators do?

For the output drive, you're saying that Q1 is unnecessary because the TL494 can do it? Can the TL494 handle the amount of current coming off of L1? Or am I misunderstanding something?

yes looks good, capacitor values are not that critical.

Okay. Do you think that C2 is necessary? Or can I eliminate it?

no it is not neccessary, but you connected it to the voltage reference?
Sometimes 0.1R etc. is added in the output path, but I don't think in the switcher transistor path. And one of the comperators is then used to sense current. But it can be left out.

I've attached the schematic that I was referencing when I designed the one above. I wasn't sure if I was connecting it to the correct pin or not. Where is that supposed to go?

Essentially the comperator will compare this to the reference voltage. So it is kind of a voltage divider.

For instance 20K pot. with 1K or 470 Ohm limiting inside the feedback path will work. Also 10K or 50K it is not critical.

What is a 20K pot? Potentiometer?

I'm not really following what you mean here. I don't fully understand the TL494 or what all of the various pins are for.

I must say this one is very likely too small. Magnetic properties are unknown. You need a larger Toroid with 25 to 30mm diameter. Or maybe a ready-made coil. I made experience with various Toroids + wire, none of them produced good efficiency. I don't think this small core can carry more than 1A or 2A, if it will work at all.

Okay...I just looked again on Newark, apparently I was looking in the wrong section for pre-made inductors. I found some that look like they might work. I don't know how much inductance I need, but there are plenty that are rated for 10A.

What about this one? 33UH, 11.1A.

These small heat sinks can carry 3W to 6W typically.
So they won't be enough very likely.

Maybe try VGA coolers (from eBay).
They are quite efficient, I have measured cooling capability of 50W for the smaller one's. They are very easy to use actually.

The metal pad is exposed upside down on the PCB, and then the cooler is pressed onto it.

Cool, I'll look into those. I don't want to use a fan if I don't have to, I don't want the additional noise from that. Not to mention, if it's getting that hot, it will make locating the control box that much more difficult.

But, you need to make some major changes to the schematic.

Have you seen this Thai circuit for TL494 that I uploaded here sometimes?

You only need to change it for boost configuration.
I have also made various changes, removed a lot of components, and added DTC.

Do you have the user manual for the TL494 available? (not the datasheet)

The schematic that you uploaded on the first page of this thread? I saw it, but to be honest, I had some difficulty reading it. I understand it a little bit better now, but when I first looked at it I was a bit overwhelmed.

I don't have the user manual for the TL494, I tried to find it but I didn't know exactly what I was looking for, or where to look.

 

I found something offered by TI that helps calculate the amount of inductance you need for a bunch of different kinds of switched power supplies.

Here's what it gave me for a boost converter conforming to my specs.

Granted, I only requested a 2A output here. I don't think I'll need more than that, as long as C3 will be charged quickly enough to fire the solenoid.

 

TO220 can dissipate 50W with good cooling.

No, it certainly can not. Even with an infinite heatsink, that would put the junction temperature over 200C.

 

Here is said document attached.

33uH seems to be a very low value.
Many TL494 schematics I have seen use a value of some 100 uH.
Or they use HF transformers.

In the manual there is explanation for all the pins.
Yes a pot means potentiometer. Switching regulators always have a feedback path. And often a so-called error amplifier.

The TL494 has two: One for voltage, one for current. The current amplifier does not need to be used. It also has an internal voltage reference. Which is then connected to one of the inputs of the voltage error amplifier. It is normally divided down (to be able to adjust to very low voltage).

In your case you do not need to divide it down, however. Since it is a fixed voltage boost converter.

C2 is neccessary.

The internal transistor can only drive very small currents, some 200mA.
It is used to drive the MOSFET gate!

You have to read the manual carefully. There are two modes, one is for push and pull, the other for single phase (means for inductor based converter).

I have also found on my hard drive a Motorola datasheet which actually contains some schematics.

If you can get away without cooling is something I can not fully estimate. You need to work out the deactivation somehow, and then switch on the solenoid + provide holding current.

Why not design for just 2A, and see if it is enough? This won't require expensive components.

For inductor get maybe 300uH and see how it works. I am however not sure about the value for boost converter, and 24 volts delta voltage. 30 uH seems to be too low to me.

I would also add adjustable resistor for the frequency. Then observe the input current, and adjust for the lowest input current.

There are many small things you need to consider for a SMPS to work well. For instance you need a basic load, some 1K Ohms or the like, which always draws some current from the output. Not a problem for 36V I think...

 

No, it certainly can not. Even with an infinite heatsink, that would put the junction temperature over 200C.

I have seen this printed in a datasheet.

I have also observed a compareable amount of dissipation, including heat up to about 70C with the cooling fan spinning. However 2X IRF MOSFETs were used for each cooler. So the amount of distribution is only 25 Watts to 30 Watts each.

The cooling system I used was quite effective, direct contact to the metal tab, using thermal grease, and having a large aluminium structure with fins.
However large is subjective. The cooling capability for these fans is about 50 Watts.

I was just saying, the information from the vendor with 250W or 200 Watts is incorrect. Dealing with power semiconductors, the ratings are not always to be taken literally. Having some cable on the PCB, as well supply cable, unless it is very heavy gauge, large currents of 80 or 100 Ampere can not really exist. Using rather thin wire even limits such currents to the 1A range.

 

Here is said document attached.

33uH seems to be a very low value.
Many TL494 schematics I have seen use a value of some 100 uH.
Or they use HF transformers.

In the manual there is explanation for all the pins.
Yes a pot means potentiometer. Switching regulators always have a feedback path. And often a so-called error amplifier.

The TL494 has two: One for voltage, one for current. The current amplifier does not need to be used. It also has an internal voltage reference. Which is then connected to one of the inputs of the voltage error amplifier. It is normally divided down (to be able to adjust to very low voltage).

In your case you do not need to divide it down, however. Since it is a fixed voltage boost converter.

C2 is neccessary.

The internal transistor can only drive very small currents, some 200mA.
It is used to drive the MOSFET gate!

You have to read the manual carefully. There are two modes, one is for push and pull, the other for single phase (means for inductor based converter).

I have also found on my hard drive a Motorola datasheet which actually contains some schematics.

If you can get away without cooling is something I can not fully estimate. You need to work out the deactivation somehow, and then switch on the solenoid + provide holding current.

Why not design for just 2A, and see if it is enough? This won't require expensive components.

For inductor get maybe 300uH and see how it works. I am however not sure about the value for boost converter, and 24 volts delta voltage. 30 uH seems to be too low to me.

I would also add adjustable resistor for the frequency. Then observe the input current, and adjust for the lowest input current.

There are many small things you need to consider for a SMPS to work well. For instance you need a basic load, some 1K Ohms or the like, which always draws some current from the output. Not a problem for 36V I think...

According to the power stage designer tool from TI, 33uH should work, however going higher doesn't hurt at all. So, it's probably simply safer to use a larger inductor, which I will do. Maybe this one?

I think I finally understand the error amps. I used the Motorola sheet you posted to come up with what I believe to be the proper way to regulate the output voltage. I just need two resistors, R3 needs to be 3 times the resistance of R2. According to the Motorola sheet, in this configuration, Vout = Vin (R3/R2). Since I want 36V out and I have 12V in, that's a ratio of 3:1. Does this seem to be correct?

I've grounded OUTC (Pin 13) so that the TL494 acts in single phase, like you mentioned.

I think I figured out the deactivation. I put transistor Q3 in as a PNP transistor. So, when the solenoid is activated by the signal from my microcontroller (not the TL494), the path from the TL494 to Q1 is cut off, which stops the pulsing that creates the 36V output from L1. Because the 12V input is still connected to the solenoid via L1, D1, and Q2, the solenoid will receive 12V once C3 is depleted.

I don't know if I have the frequency components set up correctly (Pins 4, 5, & 6). I found another schematic online for the TL494, so I just copied their setup for the frequency controls. You mentioned using an adjustable resistor for the frequency, would that mean that I need to replace R1 with an adjustable resistor?

Also, I added R5 as a 1k ohm resistor to put some load on the 36V output at all times.

Let me know what you think. If this schematic looks a little better, I'll go through and determine what actual components to use again.

 

Good improvement on the schematic.

Have you noticed DTC? That can be used to shut off the TL494 output.

Why use extra transistor?

However, larger MOSFETs need quite low-Ohms gate control, for which purpose also often a small power transistor is used. Otherwise the gate can not charge/discharge fast enough.

It also needs to be discharged! Not just disconnect like a bipolar transistor.

See the Thai circuit how the power PNP drive is done. Interestingly it also works for p-CH MOSFET without changes.

n-CH simply needs different polarity! But the given resistor values are realistic.

For the inductor well if you want to go with 33uH, I believe it's too low. But if the frequency is adjustable, maybe 33uH can be used.

What is the Q2 MOSFET, n-CH? Then you have current flowing through the MOSFET body diode all the time. You need to put that in the ground path, reverse it, or use p-CH MOSFET.

p-CH also needs different voltage level to drive, so n-CH in the ground path is easier. It is not exchangable into the Vcc path, unless the gate voltage is higher than Vcc. And I think it is in reverse in respect to what is required.

I would recommend eventually to run the circuit through LTSpice.

Instead of the TL494, you simply provide a square wave! Then at least you can get some idea if the polarities/voltage levels are OK.

This is the way I have saved a lot of time when doing MOSFET circuits, to see the required voltage levels in the simulation software.

 

Regards the error amplifier, I think one input receives potential from the voltage Reference, the other from the feedback.

It is very usual simply to have an adjustable resistor from Vcc output to GND, and then the wiper towards the feedback.

You can even divide the output voltage, and then route trough adjuster pot.

What you have drawn is somehow mixing the potential from the voltage Reference with the obtained feedback voltage.

 

Good improvement on the schematic.

Thanks! It's getting there. I've attached a revised version. I added in the second boost converter circuit that I'll need for the second solenoid, as well as the IC that will control the activation of the solenoids.

The left and right boost circuits are identical (or, at least they should be), so it's not nearly as complicated as it looks. Unfortunately, the numbering of the components did get a little messed up, though.

C1 is the input capacitor, for the boost converter, I just let it apply to both converter circuits. I figured simply getting a bigger one would work fine for both systems. That's the only component that I shared between the two systems, though.

Also, all of the grounds are common, I just felt that it looked a lot cleaner without a bunch of ground traces all over the place.

Have you noticed DTC? That can be used to shut off the TL494 output.

Why use extra transistor?

However, larger MOSFETs need quite low-Ohms gate control, for which purpose also often a small power transistor is used. Otherwise the gate can not charge/discharge fast enough.

It also needs to be discharged! Not just disconnect like a bipolar transistor.

See the Thai circuit how the power PNP drive is done. Interestingly it also works for p-CH MOSFET without changes.

n-CH simply needs different polarity! But the given resistor values are realistic.

You're right, I should have used the DTC. I had to read up on the DTC to figure out exactly what it does and how to make it work, but I think I figured it out. I don't know if I built that part of the circuit in the most efficient way possible, but I think it will work.

It's a PNP transistor with the DTC at the collector, and ground at the emitter. So, when the Tiny85 (Tiny22 in the schematic) sends a 5V signal to activate a solenoid, that signal also goes to the DTC and the PNP transistor. It's cut down to 3.3V by a resistor, as 3.3V on DTC is 100% dead time (completely off). When the transistor sees 3.3V, it cuts the ground to the DTC, so the DTC only sees the 3.3V signal.

For the inductor well if you want to go with 33uH, I believe it's too low. But if the frequency is adjustable, maybe 33uH can be used.

I will use a 300uH. It can't hurt, and they're about the same price, so why not?

What is the Q2 MOSFET, n-CH? Then you have current flowing through the MOSFET body diode all the time. You need to put that in the ground path, reverse it, or use p-CH MOSFET.

p-CH also needs different voltage level to drive, so n-CH in the ground path is easier. It is not exchangable into the Vcc path, unless the gate voltage is higher than Vcc. And I think it is in reverse in respect to what is required.

Q2 (now Q2 and Q6 on the new schematic) is an N-Ch MOSFET. You are correct, I had the drain and the source backwards. Haha. I fixed it in the new schematic.

I would recommend eventually to run the circuit through LTSpice.

Instead of the TL494, you simply provide a square wave! Then at least you can get some idea if the polarities/voltage levels are OK.

This is the way I have saved a lot of time when doing MOSFET circuits, to see the required voltage levels in the simulation software.

I will look that program up and do that as soon as I re-select components for this new schematic.

Regards the error amplifier, I think one input receives potential from the voltage Reference, the other from the feedback.

It is very usual simply to have an adjustable resistor from Vcc output to GND, and then the wiper towards the feedback.

You can even divide the output voltage, and then route trough adjuster pot.

What you have drawn is somehow mixing the potential from the voltage Reference with the obtained feedback voltage.

I'm really not sure what to do with the feedback in this thing. All of the schematics for boost converters that I have seen have had a capacitor on the feedback, and a ground on the other side. Maybe if I add a resistor after the junction between Input 1 and the Comp input?

The User manual is pretty vague regarding the Feedback, and they don't have any good descriptions for their examples.

I did switch Pins 1 and 2 around, I noticed that the Motorola PDF had the VREF going to Pin 2, not Pin 1.

Is Pin 3 (feedback) just a 5V output from the internal reference regulator?

If it is, then would it make sense to have a capacitor on Pin 3 to switch the sign of the electric potential? If Pin 2 has 5V input from VREF, then Pin 1 would need to be the opposite sign, so -5V. Or, at least ground. Which, right now, I think it is both.

I don't know, I'm a little lost on the comparator system.

 

I forgot to mention that I added an NPN transistor to run the inductor MOSFETs (Q1 and Q5). I think that should work fine, to make sure that the MOSFETs can switch on and off quickly enough.

I used a schematic I found on google to get the design from, but I think I should have looked at the Thai schematic like you mentioned. I'll have to revise that, most likely, it looks like the Thai schematic is more correct than the one I used.

 

Looks promising, more close to an actually working circuit.

Yes the feedback is covered only very vague. Agree.

You ground one comperator input, and then mix reference voltage and feedback together.

I have worked through this actually took me 1/2 hour or so, when I built my circuit.

Only logically- one comperator input receives 2.5V from the reference (or 5V also can be used for your voltage levels), and the other input receives voltage from the feedback adjustable resistor. Most switcher ICs have error amplifiers which are very similar- look up for instance LM2576 and MC34063.

It is also possible to use external voltage reference, for instance a simple 7805.

I don't actually use the pot on the full output voltage, like I do for MC34063 or LM2576, it is divided down by a resistor divider. But this is mainly due to high voltages, upto 60V, where a 10K pot eventually would burn out.

All adjustable resistor outputs have 470 Ohms or 1K protection.

Frequency and DTC are all adjustable in my circuit.

I would recommend to test the output stage in LTSpice, using only a square wave as driving voltage, to see if the levels are right. It is easy to do this actually, no need to use exactly the same components. Only voltage levels are important here.

I must say I have built some 100s of circuits, and rarely they behaved as expected. There have often been blatant errors, and I am not inexperienced.

But equipped with LEDs + multimeters, normally I figure out quickly what's wrong.

If you produce a PCB and there is some error, you can patch it up, but this can be a lot of effort.

Also screw terminals are highly recommended instead of soldering wires, for any kind of circuits that carry high currents.

 

The IRF MOSFET used to switch ON/OFF the solenoid would receive only 5V at the gate, and I am not sure if this will be enough.

Eventually add a driver stage as well, supplying it with 12V from Vcc via resistor, and then normally short it to ground with a NPN.

But this can be tested easily in LTSpice.

 

I forgot to mention that I added an NPN transistor to run the inductor MOSFETs (Q1 and Q5). I think that should work fine, to make sure that the MOSFETs can switch on and off quickly enough.

I used a schematic I found on google to get the design from, but I think I should have looked at the Thai schematic like you mentioned. I'll have to revise that, most likely, it looks like the Thai schematic is more correct than the one I used.

Yes it is fully working except there are many superfluous components.
I had no difficulty to base my circuit on that + made modifications.

The top part of the circuit you could add if you like to have a blinking LED for voltage regulation. But not needed at all.