[ZytheEKS]

So a while back I bought a bunch of TFC Triebwerk 2600RPM 55mm thick fans from someone here on OCN. They're great fans, move a lot of air and come with a TON of accessories. The problem is they're loud, not obnoxiously loud but I have a hunch once I throw 12 of them on radiators they will be. XD The answer to that is simple, throw em on a fan controller. The problem is they draw 1.5 amps nominal current, so even if I bought one of the really nice 36watt fan controllers I'd be having to put two fans per channel and need to control them all separately, and even then I'd need two fan controllers to use them all. Well ever since then I've been working on a fan controller and after a few weeks of playing with different components, seeing what works better or worse, playing with different ways of controlling it all I finally have a working prototype hooked up on my breadboard.

The controller is really basic yet gives me loads more control than most of the analog fan controllers on the market right now. I've only tested it on a single fan on the circuit , and won't be able to test with more until I get it all mounted up as I'm not willing to push that much current through my breadboard, but I'm get full range of voltage control between 0-11.7volts, then I can control the fans further by limiting the current the fans get. I tried hooking a fans RPM wire into a mobo to get an idea of what kind of RPMs I got minimum but at minimum the fan was spinning too slow for the tach to register on the comp. I got out my trusty wrist watch and I got around 50RPM on my GT AP15s, so given the fact I will never need to go that slow I'm satisfied with that.

When this is all put together I should be able to get up to 20amps (240 watts) worth of fan run off a single channel controlled by two knobs, one for voltage and one for current.

The current control circuit (Now with less chips)


The voltage control circuit:


A quick video of a GT @ 12v being current controlled, from slowest to fastest then back to slowest. The potentiometer (knob you turn) is off screen.

The bearings on those GTs man, they just keep spinning and spinning with next to no resistance.

 

[ZytheEKS]

Reserved for end results

 

[Lucas Bezerra]

May I ask why making a current based controller AND a voltage based instead of building just one? What's the difference?

 

[ZytheEKS]

Quote:

Originally Posted by Lucas Bezerra

May I ask why making a current based controller AND a voltage based instead of building just one? What's the difference?

Some fans undervolt REALLY well, others can barely go below 10volts without having issues and I wanted to future proof the controller. As far as having voltage AND current goes, due to the nature of how I'm controlling current after I drop it to below 50% current some fans like to stutter, almost like little hiccups. They speed up and slow down at regular intervals, creating a weird noise which is reflected by the current jumping 100mA as it speeds up and slows down, it really can't be seen with the GT but with my TFC Triebwerk fans which is what I'm making this for it was pretty bad. I'd considered using a capacitor to keep the current steady but the issue is caps that could keep up with the potential max load of this circuit would cost as much as all the other hardware combined. I added voltage control via a voltage divider circuit with a static R2 and a voltage controller resistor as R1 and I get complete voltage control since I can adjust the resistance on R1 almost indefinitely. After testing the new circuit on my TFC Triebwerk the current fluctuations seemed to stop. I'm using the GT for all the circuit testing ATM because my multimeter can't measure voltage on the TFC, maximum current of 600mA on the voltage pin. I burned out the fuse trying to test it on my Triebwerk. I actually got around to replacing the fuse on my multimeter today, and did a short video on voltage control here.

Notice how it doesn't go nearly as slow as current control.

 

[ZytheEKS]

Next step is to modify the control circuit a bit to give me a longer range of control. You can't see it in any of the videos I have but the first 50% of the potentiometers (the knobs) do absolutely nothing, then the next 30% gives you the full range of control, then the next 20% continues to do nothing. This should be a pretty easy fix.

 

[Lucas Bezerra]

Quote:

Originally Posted by ZytheEKS

Some fans undervolt REALLY well, others can barely go below 10volts without having issues and I wanted to future proof the controller. As far as having voltage AND current goes, due to the nature of how I'm controlling current after I drop it to below 50% current some fans like to stutter, almost like little hiccups. They speed up and slow down at regular intervals, creating a weird noise which is reflected by the current jumping 100mA as it speeds up and slows down, it really can't be seen with the GT but with my TFC Triebwerk fans which is what I'm making this for it was pretty bad. I'd considered using a capacitor to keep the current steady but the issue is caps that could keep up with the potential max load of this circuit would cost as much as all the other hardware combined. I added voltage control via a voltage divider circuit with a static R2 and a voltage controller resistor as R1 and I get complete voltage control since I can adjust the resistance on R1 almost indefinitely. After testing the new circuit on my TFC Triebwerk the current fluctuations seemed to stop. I'm using the GT for all the circuit testing ATM because my multimeter can't measure voltage on the TFC, maximum current of 600mA on the voltage pin. I burned out the fuse trying to test it on my Triebwerk. I actually got around to replacing the fuse on my multimeter today, and did a short video on voltage control here.

Spoiler: Warning: Spoiler!

Notice how it doesn't go nearly as slow as current control.

Nice, didn't ever think about the idea of current controlling... I'll be here looking for the updates, good luck! (if it works would you mind posting here the PCB layout and the components needed?

. If you can't, no problem, it's a little painful thing).

 

[ZytheEKS]

Quote:

Originally Posted by Lucas Bezerra

Nice, didn't ever think about the idea of current controlling... I'll be here looking for the updates, good luck! (if it works would you mind posting here the PCB layout and the components needed?

. If you can't, no problem, it's a little painful thing).

Oh it works... All the components are rated for way more current/volts than I'll be using and I've tested the circuit in it's entirety and it all works. I just need to get it on a thick enough PCB to support the current. If I tried to put this on a standard 1oz/ft^2 PCB I'd need 6cm wide leads to allow that much current and even then I'm not sure it would work, then it would just burn out on the contacts. I'm going to use some 7oz/ft^2 PCB and beef up the leads with a bit of solder, or maybe gold plate it with a couple layers. 1cm PCB leads should be sufficient for that. One thing about this design is it's going to need a fan, and chipsinks on the fan cooler itself. If the fans combined draw 20amps nominal current and they're only getting 12amps then the other 8 are converted into thermal energy, and so on. The same goes for most fan controllers on the market with the acception of the Lamptron ones that use a self contained PWM circuits. The difference is they're usually using 3 amps absolute max, dissipating 36watts of heat isn't a big deal, a passive chipsink can handle that. Potentially dissipating upwards of over 100watts? Truth be told though I'm not that worried, I have voltage control divided between five static resistors, and three voltage controlled resistors, and current control divided between three voltage controlled resistors so I just need a really low RPM fan to make sure air is actually moving.

If it all works as well as I predict I might even order a few 10+oz/ft^2 PCBs from a professional and make a few controllers to throw on the OCN marketplace. Can't tell you how many threads I've seen of people struggling to put all their radiator fans on a single controller. XD

 

[Lucas Bezerra]

Quote:

Originally Posted by ZytheEKS

Oh it works... All the components are rated for way more current/volts than I'll be using and I've tested the circuit in it's entirety and it all works. I just need to get it on a thick enough PCB to support the current. If I tried to put this on a standard 1oz/ft^2 PCB I'd need 6cm wide leads to allow that much current and even then I'm not sure it would work, then it would just burn out on the contacts. I'm going to use some 7oz/ft^2 PCB and beef up the leads with a bit of solder, or maybe gold plate it with a couple layers. 1cm PCB leads should be sufficient for that. One thing about this design is it's going to need a fan, and chipsinks on the fan cooler itself. If the fans combined draw 20amps nominal current and they're only getting 12amps then the other 8 are converted into thermal energy, and so on. The same goes for most fan controllers on the market with the acception of the Lamptron ones that use a self contained PWM circuits. The difference is they're usually using 3 amps absolute max, dissipating 36watts of heat isn't a big deal, a passive chipsink can handle that. Potentially dissipating upwards of over 100watts? Truth be told though I'm not that worried, I have voltage control divided between five static resistors, and three voltage controlled resistors, and current control divided between three voltage controlled resistors so I just need a really low RPM fan to make sure air is actually moving.

If it all works as well as I predict I might even order a few 10+oz/ft^2 PCBs from a professional and make a few controllers to throw on the OCN marketplace. Can't tell you how many threads I've seen of people struggling to put all their radiator fans on a single controller. XD

I see... I'll PM you with some ideas and theoretical questions, if that is not a problem

 

[ZytheEKS]

Got a bunch of goodies in the mail today, most importantly my new VRMs. I switched the voltage divider circuit with a variable R1 to an array of Adjustable volt regs, much MUCH more stable voltages, and lower heat density on the components when I give it bigger loads. Still have to figure out how to supply a 7v source to my potentiometers, multiple voltage dividers obviously isn't going to work all right, and my mosfets don't seem to respond to having a static resistor hold the R1 @ 7k on my 10k pot. Even more odd doing the 7volt trap on my molex seems to work on the pot.

I'll play with it more tomorrow, but I'm all brained out for the night.

On a side note trying to find copper clad thicker than 6oz/ft^2 is proving to be very irritating. If anyone has any sources/ideas please let me know.

 

[ZytheEKS]

Just an update, ordered some 7oz/ft^2 copper clad, and will be etching the printed circuit board as soon as that arrives. Then I can finally try this all with a 10+ amperage load.

 

[ZytheEKS]

This will more or less be the final circuit design, the actual PCB will have much more VRMs in parallel to allow for the higher current, as well as more VCRs on the current control section to allow for lower heat density.



Red: Power and ground, will be supplied from a 6pin PCI-E power line.
Dark Red: Power to fans, will be supplied from 8gauge braided wire via screw down terminals.
Blue: Signal from potentiometers which will be housed on a separate PCB attached via 22gauge wires as bridges.
Dark Green: Volt Regs
Light Green: Voltage Controlled Resistors.

Will try to get another vid up either tonight or tomorrow of the effects of Current control vs voltage control on one of the Treibwerk fans. The fan is really awkward on current control @ low RPMs.

Thankfully I have both.

 

[IT Diva]

I'm always a sucker for a good electronics DIY project . . . .

Have you tried using a PWM control IC like an SG3525 with a mosfet to see if the Triebwork fans respond to having the supply PWM'd ?

I built one to see if the D5 Strong with a 24V supply would be able to be speed controlled, but found I'd have to pull the pcb out and do some reverse engineering so that the control circuitry would have a relatively constant voltage, while just the coil's supply would be PWM'd.

A cap at the control circuit, with a series Schottky diode on its Vcc supply could accomplish that easily.

I wasn't willing to risk trashing the pump trying to get the board out, but a relatively cheap fan might be a much different proposition.

If you could get PWM to work, the parts count and thermal dissipation would drop dramatically.

Darlene

 

[ZytheEKS]

Quote:

Originally Posted by IT Diva

I'm always a sucker for a good electronics DIY project . . . .

Have you tried using a PWM control IC like an SG3525 with a mosfet to see if the Triebwork fans respond to having the supply PWM'd ?

I built one to see if the D5 Strong with a 24V supply would be able to be speed controlled, but found I'd have to pull the pcb out and do some reverse engineering so that the control circuitry would have a relatively constant voltage, while just the coil's supply would be PWM'd.

A cap at the control circuit, with a series Schottky diode on its Vcc supply could accomplish that easily.

I wasn't willing to risk trashing the pump trying to get the board out, but a relatively cheap fan might be a much different proposition.

If you could get PWM to work, the parts count and thermal dissipation would drop dramatically.

Darlene

Yeah I thought about PWM as an option, but decided for the first go at it I wanted to keep it analog For now, I know this design works, and I want to see how it works with 12 1.5amp fans, so 18 amps. In theory it all should work, but this is electrical so I don't plan for it to be that set and forget type setup. If I throw on some chipsinks and put a low RPM fan I'd imagine thermal dissipation wouldn't be an issue, as with only 1 VRM, which seems to be the hottest component, I can control my TFC fan at nearly any voltage without it failing. That was in open air with no chipsink nor airflow, so once I throw it all on the PCB and drop in the stuff it should work fine. I might experiment with PWM at a later date, it would definitely be a fun alternative.

Just an update: Got the PCB in, gotta order the etching supplies and I can go at it. I've also decided to add a deadman switch, as the voltage control forces you to run @ 1.25 volts below the input voltage. I'm going to install the three pin deadman to turn the thing off, turn it on in only current control mode, and turn to on in voltage control and current control. Also probably add some caps on the volt reg circuit to make sure I'm getting stable voltage @ higher loads.

 

[ZytheEKS]

Well after WAY too long of trying different ways to etch the PCB ranging from photo resistant etch, to toner transfer, to hardmodding an Epson C84 inkjet printer to print ferric chloride resistant ink directly on the PCB I couldn't get any of the DIY PCB strategies to work for thicker PCB. So after a few months of messing around with all those trying to get em to work, I said screw it and started a proper CAD file for it. Problem is I had to custom make most of the throughole components so that combined with procrastination and other projects it took forever. Anywho, I finished up this design here:

Problem is it would cost to much to get a few one off PCBs that big, (5" x 7") so I'm going to cut down the VRMS into stacks of 5 instead of 10, reroute the electrical connections and get a new estimate. Cutting down just 2 inches would mean up to $50 less

Anyways, I just wanted to anybody interested to know the projects not dead, just been on the backburner after all the failed PCB attempts.

-Z

 

[IT Diva]

Have you thought about going at it the other way around . . . .

Instead of trying to mask off some areas to create the tracks, which with very thick copper requires too much etch time to prevent massive undercutting issues, look at using a mill or router like setup, to just machine thin spaces in the copper cladding to effectively break it into traces.

It will look like a nearly solid copper surface with some little bare lines running about.

For really simple boards, I've done it with a dremel. In such cases it can be faster than laying out a design and etching it.

Your board is physically larger than ones I've done that way, but, it's simple enough to manage a prototype via dremel once you layout the board with that plan in mind.

The easiest way to try to lay it out, if you're using Dip Trace layout software, is to place your smt parts, and then use "copper pours" for the traces so you can shape them as you need. For thru hole parts like the connection points, use very small diameter pad / hole sizes. The copper pour will automatically stay a tiny bit away from a pad, but overall, once you drill the hole, the copper will be right at the connector pin coming thru.

Once you get it how you want it, print it, toner transfer it to your pcb, and then instead of etching, mill / route the copper tracks off.

Darlene

 

[ZytheEKS]

Quote:

Originally Posted by IT Diva

Have you thought about going at it the other way around . . . .

Instead of trying to mask off some areas to create the tracks, which with very thick copper requires too much etch time to prevent massive undercutting issues, look at using a mill or router like setup, to just machine thin spaces in the copper cladding to effectively break it into traces.

It will look like a nearly solid copper surface with some little bare lines running about.

For really simple boards, I've done it with a dremel. In such cases it can be faster than laying out a design and etching it.

Your board is physically larger than ones I've done that way, but, it's simple enough to manage a prototype via dremel once you layout the board with that plan in mind.

The easiest way to try to lay it out, if you're using Dip Trace layout software, is to place your smt parts, and then use "copper pours" for the traces so you can shape them as you need. For thru hole parts like the connection points, use very small diameter pad / hole sizes. The copper pour will automatically stay a tiny bit away from a pad, but overall, once you drill the hole, the copper will be right at the connector pin coming thru.

Once you get it how you want it, print it, toner transfer it to your pcb, and then instead of etching, mill / route the copper tracks off.

Darlene

That's..... A really good idea. Wish I'd thought about that before spending all those hours in Eagle CAD making the parts. XD I think I still have a few 7oz/ft^2 clad out in the garage. I might pull out my dremel and try that tomorrow, or sometime this week. At this point I'm already invested in the CAD design, but it might be nice to see what kind of thermals I'll get on the PCB and VRMS pushing 20 amps through it. Thanks for the idea.

-Z

 

[IT Diva]

Quote:

Originally Posted by ZytheEKS

Quote:

Originally Posted by IT Diva

Have you thought about going at it the other way around . . . .

Instead of trying to mask off some areas to create the tracks, which with very thick copper requires too much etch time to prevent massive undercutting issues, look at using a mill or router like setup, to just machine thin spaces in the copper cladding to effectively break it into traces.

It will look like a nearly solid copper surface with some little bare lines running about.

For really simple boards, I've done it with a dremel. In such cases it can be faster than laying out a design and etching it.

Your board is physically larger than ones I've done that way, but, it's simple enough to manage a prototype via dremel once you layout the board with that plan in mind.

The easiest way to try to lay it out, if you're using Dip Trace layout software, is to place your smt parts, and then use "copper pours" for the traces so you can shape them as you need. For thru hole parts like the connection points, use very small diameter pad / hole sizes. The copper pour will automatically stay a tiny bit away from a pad, but overall, once you drill the hole, the copper will be right at the connector pin coming thru.

Once you get it how you want it, print it, toner transfer it to your pcb, and then instead of etching, mill / route the copper tracks off.

Darlene

That's..... A really good idea. Wish I'd thought about that before spending all those hours in Eagle CAD making the parts. XD I think I still have a few 7oz/ft^2 clad out in the garage. I might pull out my dremel and try that tomorrow, or sometime this week. At this point I'm already invested in the CAD design, but it might be nice to see what kind of thermals I'll get on the PCB and VRMS pushing 20 amps through it. Thanks for the idea.

-Z

Try the free version of Dip Trace for laying out your PCB:

http://diptrace.com/download-diptrace/

For a simple board, you should have a grip on how to work the program in a night or so.

What were you using as a transfer process when you tried the toner transfer method?

It's what I always use for all but the simplest boards.

It does require a small laminator and the blue Pulsar brand "toner transfer paper".

Once you transfer the toner, you have to use the green film to "set" it.

I can probably help a lot with making DIY PCBs . . . . some kind of custom electronics is always a unique part of my builds.

Darlene

 

[ZytheEKS]

Quote:

Originally Posted by IT Diva

Try the free version of Dip Trace for laying out your PCB:

http://diptrace.com/download-diptrace/

For a simple board, you should have a grip on how to work the program in a night or so.

What were you using as a transfer process when you tried the toner transfer method?

It's what I always use for all but the simplest boards.

It does require a small laminator and the blue Pulsar brand "toner transfer paper".

Once you transfer the toner, you have to use the green film to "set" it.

I can probably help a lot with making DIY PCBs . . . . some kind of custom electronics is always a unique part of my builds.

Darlene

I was using high gloss photo paper pressed through a cheap laminator for 20-30 passes. I'll look into that Pulsar brand transfer paper. If I take another crack at it I'll also probably order some proper hydrogen peroxide, that medical grade crap diluted the mixture down WAY too much. Thanks for the info.

-Z

 

[IT Diva]

Quote:

Originally Posted by ZytheEKS

Quote:

Originally Posted by IT Diva

Try the free version of Dip Trace for laying out your PCB:

http://diptrace.com/download-diptrace/

For a simple board, you should have a grip on how to work the program in a night or so.

What were you using as a transfer process when you tried the toner transfer method?

It's what I always use for all but the simplest boards.

It does require a small laminator and the blue Pulsar brand "toner transfer paper".

Once you transfer the toner, you have to use the green film to "set" it.

I can probably help a lot with making DIY PCBs . . . . some kind of custom electronics is always a unique part of my builds.

Darlene

I was using high gloss photo paper pressed through a cheap laminator for 20-30 passes. I'll look into that Pulsar brand transfer paper. If I take another crack at it I'll also probably order some proper hydrogen peroxide, that medical grade crap diluted the mixture down WAY too much. Thanks for the info.

-Z

If you're going to take another run at toner transfer, here's a good link . . . http://www.pcbfx.com/main_site/pages/products/overview.html

These guys have the toner transfer paper and the green sealing foil. You can also order both from Digikey or Mouser.

It's also important to have the right laminator that has high enough heat, and puts the heat into the rollers, instead of before them, so you have heat and pressure at exactly the same time.

It looks like they had to go to a new laminator, the Apache AL13P, as the original brand, (GBC) changed their design since I got mine years ago . . . . but $80 from Amazon (and you do a little mod to it to get more roller pressure) isn't too bad. . . . The GBC ones cost that much 10 years or more ago.

I found if I wet sand the PCB material with 600 and 1500 paper and wipe it with acetone or 91% alcohol, I get a good transfer. I usually use 2 passes, but never more, it just mushes it into a mess.

Good luck,

Darlene

 

[ZytheEKS]

A couple of insomnia fueled sessions later I finished the CAD files for the power board. I just need to finish the control board with the POTs on it now, and I should be ready to order it. Going to use a 3 point switch to either go to the VRM circuit, override the VRM circuit and go straight to the MOSFET array (aka the current control section), or turn the board off. Diodes are handy little things.

-Z