So I designed a new product, specifically a thermostatically regulated PWM fan controller for motorcycles that either already have or have been converted to an electric fan.

I own a Honda CX500 Custom, often referred to as a SilverWing and there are still a ton of them in use due to their longevity. (long story) Most of them came with a manual fan connected to the end of the camshaft but 30+ year old plastic fans have a habit of coming apart and taking the radiator with them. The only replacements are usually also already old so a lot of people are converting by taking off the plastic fan and mounting a small electric in there controlled by a simple thermostatic switch glued to the top of the radiator.

That works, but it wasn't accurate as there's always going to be hot water hitting the top of the radiator and if you're traveling at any speed it doesn't really need a fan to bring the coolant back down to temperature. The solution? An LM35 temp sensor IC in the TO-220 package at the bottom of the radiator right near the return hose feeding a PWM control circuit that has some adjustments as to when and how fast it runs depending on the temp that's sensed. I'll eventually post the circuit when I get the BOM and PC board layout pix & files completed.

The problem? The ideal IC I designed it around is old but I had to keep the design as through-hole so people could buy just the PC board or a kit of parts from me if they wanted to assemble it themselves. Just in the one little forum on my particular bike interest has already spiked like crazy and now that I'm starting to get the BOM together I'm finding that the IC is almost non-existant in supply. I got all but two that Mouser had, DigiKey said they had 128 last night, this morning they said they only had 100 but when I finally got a PO together it turned out they found the other 128 so I've now cornered the market and will have 278 of them in stock. Microchip Direct (with whom I've got an account) only has 24 left but are scheduling a production run in February if you get your requirements in early. Farnell has 50, but that's across the ocean from me. I haven't a clue how demand is going to go once word starts to get around the other motorcycle forums, it'll even work for cars or anything else water cooled for that matter and it's possible this may be as big of a hit as that guy who designed all those signs everyone helped with not long ago.

Anyhow, facing a potential demand that would be kinked by shortage of the IC I've about talked Mouser into putting in an order with Microchip so that at least someone will have them in stock. In the meantime I'll be glad to sell a few of what I've got if anyone wants to build one of these things once I release the rest of the files.

What the heck, here's a schematic of the prototype as it sits but I don't have any values listed on it yet. I will tell you that the IC is:
http://ww1.microchip.com/downloads/en/DeviceDoc/21756b.pdf
and the temp sensor is a simple LM35 in the TO-220 case.

Pretty much a raw schematic but you can get an idea by looking at it:
http://www.innoengr.com/images/fan_controller.jpg

When this is all done I think I'm going to submit it to the forum as a full project.

 

Interesting, but there sure are a lot of parts in this circuit.

Had you considered a small PIC with a built-in PWM, like a PIC12F683? It would take some programming to get the functions desired, but seems to me that it could be simplified quite a bit from what you have now.

Real estate on a motorcycle is generally pretty limited. I'd think you would want the smallest package that you can get away with. It would make assembly/troubleshooting a lot easier to have fewer parts, too.

 

.......................................................

 

That is a good number of parts, but it does have a lot of features.

The only suggestion I'd make is a small resistor before the filter caps into the regulator to help damp voltage spikes and other power supply noise.

What will the enclosure be? Motorcycles aren't usually a "Friendly Environment" unless you live in Florida.

 

Points noted, but I'm not familiar with PIC or any other controllers but despite the complexity the circuit appears to have it can be easily placed on a small PC board if properly laid out and etched as it's far simpler than it appears. Simple analog, very basic and very forgiving. You've got to remember that about 1/3 of the people interested just want an entire kit of parts they can assemble themselves. A good portion of these people may never have assembled anything in their lives thus documentation is going to prove to be a challenge.

A small resistor on the input side to the regulator is a good idea but the electrical system on bikes is by nature pretty lame to start with barely ever going above battery voltage unless it's in a heavy charging mode. The idea is valid though since this is designed to be universal and I don't know the nature of all the bikes some might use it on in time or if anyone wanted to use it on something else.

Anyone care to suggest a value?

The enclosure will of course be a die-cast aluminum box with a screw lid, allowing them access to the adjustments until they've got it set the way they want it then they can put silicone seal on the edge of the top when they close it up the last time yet still get in there later if need be. It's not designed to be mounted out in the open air, at the size I'm shooting for it will hide under the seat or in an unused area under one of the side covers.

 

.......................................................

 

I'm sharing the design with the public, all the way down to blank boards or the full-sized drawings and Gerber/NC drill files along with the documentation. It's too simple to duplicate in so many ways that protecting it isn't worth it and I'll make a fair enough profit from those that just buy the boards, complete kit of parts or a fully assembled and tested product.

I'll agree, some things are worth protecting but this isn't one of them, if demand ends up far exceeding my expectations then I might convert the whole thing over to a PIC format but there's far too much analog involved.

 

It would help a great deal if you'd fill in values for components, and what opamps or comparators you're using. As far as the input resistor, that'll be tough to figure out without having detailed component specifications.

Just so you know in advance, I'm going to make a stink if the parts you specify are not rated for the full automotive temperature range. You'd really rather hear just me complaining instead of hundreds of bikers that want your head impaled on a pole, right?

 

I haven't published the BOM yet but it isn't hard to figure out. The few electrolytics are all 105*C and the rest of the components are Mil-spec or rated to handle what they'll see during a lifetime - quality 1% resistors only under 1/100th of their rated load aren't going to fail due to it being warm inside the enclosure. I seriously doubt it will ever see any heat to speak of but I'm a stickler for spending the extra 5 or 10 cents to go with good caps simply due to the greatly increased life expectancy when operated at normal outside temperatures..

 

For instance, an average full current rating of less than 3A for the fan motor isn't going to make this MOSFET think it's really doing anything:
http://www.nxp.com/documents/data_sheet/PSMN2R0-30PL.pdf

 

Can you afford to change the temperature sensor?

I've always liked Microchip's analog line - good prices and a wide variety of components. If specs aren't critical (like low noise or accuracy), then they're pretty good.

Here's one temperature sensor in TO-92.

http://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en027103

It operates from -40°C to +150°C, well clear of automotive ratings.

If it's crucial that it's in TO-220, then the LM35 is available from Farnell (3k quantity available):

http://uk.farnell.com/national-semiconductor/lm35dt-nopb/temp-sensor/dp/1469235

Not sure if you can get the same deal in the USA.

For the linear regulator I'd recommend you choose one rugged enough to survive automotive conditions. MIC2940(-5.0) is a good choice: full auto temp range (-40°C to +125°C) as well as -20V/+60V surge survival, pretty much a requirement when operating from the battery.

http://www.micrel.com/_PDF/mic2940.pdf

 

The LM35 is cheap, available from most anywhere and the TO-220 mounting capability was of importance. Had I gone with a TO-92 they'd have to use thermally conductive epoxy to mount it plus some sort of insulation around the rest of the package lest it be too influenced by normal outside air constantly rushing over it. In the TO-220 package the die is right on the mounting tab so most all of the thermal energy transferred to it will go right to the die. Not to mention that fact that it introduces a fair amount of hysteresis into the setup. Contrary to the specs they're good for a lot higher temps than the data sheet shows but the idea is never to let your coolant get much above the thermostat rating which is 180*F on our bikes and probably 195*F on the newer ones.

Hope this makes sense.

[EDIT:] I can easily beat Farnell's price at a number of USA wholesalers and I don't have to deal with the extra shipping. I don't have a direct account with National but they're in good supply most everywhere.

The kicker was cornering the market on that stupid IC I designed it with. They've got similar ones that would require minor "tack on a component" board mods and they've also got them in surface mount but then I'd have to redesign the board I've already started on and I've already stated my reasons for trying to keep this at a level that anyone can build without having to learn SMD.

 

The LM35 is cheap, available from most anywhere and the TO-220 mounting capability was of importance. Had I gone with a TO-92 they'd have to use thermally conductive epoxy to mount it plus some sort of insulation around the rest of the package lest it be too influenced by normal outside air constantly rushing over it. In the TO-220 package the die is right on the mounting tab so most all of the thermal energy transferred to it will go right to the die. Not to mention that fact that it introduces a fair amount of hysteresis into the setup. Contrary to the specs they're good for a lot higher temps than the data sheet shows but the idea is never to let your coolant get much above the thermostat rating which is 180*F on our bikes and probably 195*F on the newer ones.

Hope this makes sense.

[EDIT:] I can easily beat Farnell's price at a number of USA wholesalers and I don't have to deal with the extra shipping. I don't have a direct account with National but they're in good supply most everywhere.

The kicker was cornering the market on that stupid IC I designed it with. They've got similar ones that would require minor "tack on a component" board mods and they've also got them in surface mount but then I'd have to redesign the board I've already started on and I've already stated my reasons for trying to keep this at a level that anyone can build without having to learn SMD.

The IC handles the PWM, right? Why is it important for the speed of the fan to slowly vary? Wouldn't it work just fine with an on-off control, like a thermostat?

Also, you could eliminate some components (at the expense of extra wires) if you used a logic output temperature sensor. These take a single resistor to set the temperature threshold, and output a logic high or low. You could use a rheostat in place of the resistor, or a potentiometer with the wiper wired to A or B, to allow for the set point to be adjusted.

Here is one in TO-220, although it has 5 pins. It even has a built in 2°C hysteresis - just add a driver transistor and you basically have what you need. Page 7 has an example application which should do what you want.

http://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en010742

With this, you could omit the 5V supply, but remember to surge protect the 12V line, as the 18V rating of the sensor may be exceeded. I would suggest a resistor inline with a zener with a breakdown voltage of 16V or so would work.

 

Well son, it's a matter of what we've found on most motorcycles. Most people that have already converted have used a thermostatic switch on the top of the radiator but you end up with the fan cycling on and off at an unreasonable rate, partially because it's sensing the output water from the engine which is usually enough to trigger it then the fan comes on full blast, cools it down in very short order then goes back off again for a short time until the water gets hot enough to trigger the switch.

Although I've allowed the option to sense the temp at that area it's far better to look at the temperature of the water returning from the radiator to the engine - when you're moving at a fairly decent rate the fan isn't necessary or may only need to run at a far slower speed than full blast. This changes of course if you're stuck at a stoplight in 101* weather where the fan really needs to run to some extent. It's far less strain on the engine and thermostat if you're feeding back a fairly regulated temperature from the radiator.

Yes, I could eliminate some components but I'd lose the functionality of making it fairly universal to be used on most any bike and the ability to tweak it to your climate. I've got friends riding up in Canada in the middle of the snow that would rarely need the fan even when they're stopped, then I've got people in Arizona or Australia that need at least a partial fan when they start climbing a long hilly run. Once set to the type of fan and bike you've got it's pretty much a close it up device.

A lot of people with the type of bike I've got have been adapting 5" fans from Kawasakis or certain Ducatis but I ended up using a 7" monster off of a Suzuki GSXR1100. We're limited as to what will fit without cutting the camshaft off after you remove the factory fan. Others that are already fan equipped go through the same on/off cycles just as a car normally does but I hate seeing temps rise way up then way back down once the fan kicks in, can't be the best for the engine.

For instance the Honda CX650I waits until the engine is way over temp before it kicks its fan in, reduces the water temp down to almost the point where the thermostat closes completely then goes through the cycle again. Not my idea of the best way to regulate an engine's temperature.

I've also provided Fault and Temp outputs which some may or may not use.

 

Well son, it's a matter of what we've found on most motorcycles. Most people that have already converted have used a thermostatic switch on the top of the radiator but you end up with the fan cycling on and off at an unreasonable rate, partially because it's sensing the output water from the engine which is usually enough to trigger it then the fan comes on full blast, cools it down in very short order then goes back off again for a short time until the water gets hot enough to trigger the switch.

If you find this is a problem, use a 555 timer as a monostable or Schmitt trigger with capacitor to add a delay to the fan, with values chosen well, you could get a delay of 1 minute or more. It might be worth considering a microcontroller.

Although I've allowed the option to sense the temp at that area it's far better to look at the temperature of the water returning from the radiator to the engine - when you're moving at a fairly decent rate the fan isn't necessary or may only need to run at a far slower speed than full blast. This changes of course if you're stuck at a stoplight in 101* weather where the fan really needs to run to some extent. It's far less strain on the engine and thermostat if you're feeding back a fairly regulated temperature from the radiator.

The temperature sensor can be anywhere, it just requires 4 wires (or 5) instead of 3. You can keep the resistor on the main PCB. I'm no biker, so I really don't know if having 4 wires vs. 3 is a problem.

Yes, I could eliminate some components but I'd lose the functionality of making it fairly universal to be used on most any bike and the ability to tweak it to your climate. I've got friends riding up in Canada in the middle of the snow that would rarely need the fan even when they're stopped, then I've got people in Arizona or Australia that need at least a partial fan when they start climbing a long hilly run. Once set to the type of fan and bike you've got it's pretty much a close it up device.

Again - it would probably work with any temperature you like. Look at the datasheet; the temperature can be adjusted from -55°C to +125°C by using a resistor anywhere from ~50k to ~200k. If you use a 100k resistor in series with a variable 100k resistor, you could adjust the temperature from 5°C to 125°C (41°F to 257°F.)

A lot of people with the type of bike I've got have been adapting 5" fans from Kawasakis or certain Ducatis but I ended up using a 7" monster off of a Suzuki GSXR1100. We're limited as to what will fit without cutting the camshaft off after you remove the factory fan. Others that are already fan equipped go through the same on/off cycles just as a car normally does but I hate seeing temps rise way up then way back down once the fan kicks in, can't be the best for the engine.

Possibly not - although, you're only talking about a ~2°C variation - you'd get more variation from environmental conditions.

For instance the Honda CX650I waits until the engine is way over temp before it kicks its fan in, reduces the water temp down to almost the point where the thermostat closes completely then goes through the cycle again. Not my idea of the best way to regulate an engine's temperature.

This probably isn't the best way but see above, this chip doesn't do that.

I've also provided Fault and Temp outputs which some may or may not use.

Yes - I suppose these could be useful, and this sensor does not provide them. It's a matter of whether or not you think they'll be needed. Would most people use these outputs? Would they prefer a product costing more, for something they might not use? The FAULT output could be generated by watching the fan current, I presume this is what your circuit currently does. Temperature isn't available directly from that IC, but you could probably use a normal temperature sensor and a Schmitt comparator to do the job if you really need TEMP output.

 

It is a comprehensive system. Is a low coolant alarm already on the bike? No circulation would give false temp, unless another temp sensor is on the heads/crankcase.

The other question is how much current can the fan generate when spinning freely at highway speeds? Diode rating is my question, I guess.

 

This doesn't bypass the normal temperature gauge or warning system for overheating thus that's not an issue. Most bikes have the sensor for that in the main thermostat housing which feeds directly to the stock gauge or the controlling electronics.

Yea, we went through all the tests on whether the fan might spin when riding around at 100 mph but once the wind has passed through the radiator it's minimal if any to get the fan blades to move so that MUR440 diode will handle anything generated.

The Temp output is coming directly from the LM35 if you look closely at the schematic, also adjustable as to *F or *C and has an offset adjust so you can tweak it to more closely represent the incoming temp if you like. Most people may not use it but it's there and will handily interface with one of the old dot/bar LM setups.

The Fault indicator merely shows if the fan is having trouble keeping the outgoing temp down enough. (or upper temp if you've got it hooked up that way) If you remember how a radiator works the water pump puts it in at the top and the theoretically cooler water comes out at the bottom. If the coolant level gets low the radiator isn't going to work worth a darn and your return temp is going to be hot enough to trigger it. It's rare to have a coolant leak on a bike without you knowing it, usually ends up as a spray that burns your knees or you can't help but see the smoke and loss of performance far in advance should you blow a head gasket.

Luckily the series of bikes I originally designed this for rarely have problems, they're built like aircraft engines. On occasion we might have to replace a few things around 60K when most have already been to the graveyard. 300K miles isn't unusual for one of these with absolutely minimal maintenance that, unlike the bikes of today, can be done by the even partially mechanical amateur with a minimal of tools. If necessary (and it rarely is) you can drop the entire engine in an hour.

500cc, 50 HP, 10K redline (although they'll go far higher)
OHV/pushrod, 4 valves per cylinder, twisted twin configuration
Water cooled, shaft drive, mag wheels and the first tubeless tires
0-60 is about 5 seconds, top speed around 108 and while it would go faster it starts rev limiting.
450- 550 lbs depending on what model and how it's dressed out, fairly top heavy due to design.

These handle like a dream and have all the power and low end torque than you could imagine. They're hard to come by but even barn finds that have been sitting for 20 or more years are easily restored. With any luck you can pick one up for free - $1,000 then with a little going through the normal stuff you'll have a classic back on the road. Most reside in other countries or are owned by (oddly enough) aircraft A&P techs over here.

Poor pictures, but here's a couple of links as to what a sideways twin actually looks like:
http://shoc.hu/data/files/news-228/honda_cx500c.jpg
http://www.motorcyclespecs.co.za/Gallery/Honda Cx500C 80.jpg
http://images.canadianlisted.com/nlarge/1983-honda-silverwing-gl650-interstate_4642109.jpg

 

Those are some nice bikes, especially for the price!

I didn't exactly realize which bike until I saw the pictures. I remember them being somewhat "ahead of their time" when introduced, but haven't owned one.

 

Tom, I like your sensor but I've already got 100 hours of R&D into the design and about the same to go considering all the documentation needs to be written. The IC does provide the Fault output signal but the temp output stuff I needed to design myself. Big deal when you're using an LM35, only thing was making it switchable between *F & *C so those not in the US could have something they're more used to reading and I had the extra two op amps to use.

As it stands it's not only functional but has some extras, it's also user friendly as far as building it.

When I get to feeling better I'll start sifting through the huge pile of papers devoted to this one project and finally make up a full BOM. I have plenty of parts in hand which I've found is almost a necessity when you go to laying out a PC board as component dimensions advertised aren't always exact. Some time back I designed a rather complicated board that was about 8" x 12" and I got burned on two things - manufacturer's specs on the spacing on a couple of items weren't exact and the rotary switch involved wasn't keyed according to their drawings.

 

Those are some nice bikes, especially for the price!

I didn't exactly realize which bike until I saw the pictures. I remember them being somewhat "ahead of their time" when introduced, but haven't owned one.

They have considered bringing the design back but they're far too expensive to reproduce. I will tell you one thing, when I was riding it on/off last year before I put it into the restoration process I would often come out of a store and find a crowd gathered around it. Even the Harley owners love them, they've either had a SilverWing or knew someone that did have one and it brought backa lot of stories and memories for them.

So extremely easy to restore at a minimum cost and most all of the parts are still available but be prepared to do your own work, a normal Honda dealer won't touch them for fear they'll run into a hard to find part they can't get. No matter, these were built back when everyone worked on their own things and they sure made these the easiest of any ever made.

I'd suggest having a garage and keeping your eyes on craigslist, normally all you have to do is replace a few of the old seals most of which can be done without even removing the engine. If they've sat for some times you'll need to go through the carbs but that's no problem either if you pay attention to detail.

You're welcome to peruse out local forum, there are three forums out in the world about these but I think ours is the most comprehensive. http://www.choppercharles.com