Nidec TA350DC 4-Wire Fan Parallel Wiring: Why Do Fans Pulse and Not Run Together?

i have got a bank of these 4 wire fans "click(4 wire)":http://www.nidecamerica.com/fanpdfs/ta350dc.pdf

when connected in parallel they fail to work and just pulse slowly.
i am just using the 12v supply wires and they work individually connected.
i am guessing that each fan produces a signal which causes any others connected to fall over. i have tried isolating each with a diose and also tried putting an additional cap close to each unit all with out success. any ideas as to what i might have overlooked ?
thanks in advance

can the 12 volt line support the load of all 4 fans at once? use a DMM and measure the voltage as you add 1 then 2 then 3 then 4 fans. If the voltage drops then the power source you are using cannot supply enough power to power all the fans. Then you have to find an alternate 12 volt supply which can support the full load.
measure the voltage and then let me know what you find out - if the voltage does not drop its something else like a fan fail signal or another issue.

yes it can each run at aprox 1A and i have a 40amp smps . what i do see happening is the voltage shown on the display of the supply go erratic, so i have rigged up a simple transformer, bridge and capacitor set up, and the failure of the fans is the same

Hi

You Absolutely correct Bong !! 100/100 Did you like that ill do it again "Bong"


I’ve just checked the datasheet but there is no mention of how the pwm is controlled / or schematics for the internal of the fan assuming four fans in parallel and all are 12 volt operation all the same model number and also all the same current rating ( Which no fan ever is )

So suggestion here is use Fet control to switch on each fan or all four fans distributing the load evenly through each fan Very similar to why 4 led’s in parallel wont work I would imagine without current limiting resistors ( Thats better )


Other than this if the fans are three wire controls Not so simple you have to analyze the data being fed back and convert to pwm with individual switching on each fan

See below

http://www.bearblain.com/fan_speed_control.htm

Here is a diagram for you I've just put together very quickly on how too do this for you hope it helps ps let me know if it works ok for you and if you need a very quick pcb to do this

This way you can have all four fans running at the same time Im sure this will work for you

Whats the other advantage of doing it this way ( The correct way ) Yes Correct one fan goes down the others still run !! and ?? You can pwm each one of them individually There you go see below

thank you very much, i will give it a go and let you know.

No worries welcome ( make sure you heatsink the IRF's with insulation behind them I should have mentioned that ) In the meantime in the next half hour / hour Ill put a PCB design together for you Wont take me too long

Gives me something to do

Done see below for pcb , gerber plot files , layout files in pdf format and proteus design files Any questions shout

Only one problem with all of this Whilst doing all of this in for good will of course it doe rather present a problem Its costing people their lively hoods , jobs and yes would you just believe this one too Also leaves people unable to complete manufacturing and yes correct doesn’t pay the bills but not to worry what the hell This wind up in another huge row its bound too bit like 160 decibells blowing or perferating eardrums not that we dont have to tolerate this as well


Give up hmm

Hi Mark. I noted that the fans are 4 wire. (each?) The data sheet shows onl 2 wires. Where I used to work the 2 wire fans paralleled up quite well. I must be missing something, and it is obvious you know what that is. I realize that the motors have DC to AC converters, but if the PSU impedance is low enough, the motors should not see eachother. If you are able to throw alittle more light on this I would be most greatful. This is a paticularly interesting problem. Cheerssnd thanks

aaah I cant see that on the datasheet Sarah has supplied shows only two wired red and black

Thats what I have here see below direct from Sarahs Datasheet (only two wires shown ) Confusing !!

Also Sarah says works on its own ( Strange ) So Assuming that there is manufacturing problem( Possible ) and one of them or the net parralled load is different on all three / four of them the best way to control , solve that problem is control the current through each of them so that we can see which one or two or three , four is the giving rise to this fault thats assuming only two wire not four ( Which is what the diagram says otherwise its send me the fans to look at ) so that they can be tested with others of the same type

In other words switch all four on sequentially ( hopefully isolating the problem ) There is a method to what way I approach this



Its does says options ( I guess thats down to when you order specifics)
Alarm/Tachometer/Thermal Speed
Control/PWM


Speed Control Option which could be other two leads then but no mention of wiring diagram at all None shown except what you see in the gif image ( Not very good is it Very Poor Datasheet ! ) Still see what Sarah comes back with She could measure the current through each one on thier own and then all in parrallel Im sure its going to be what I think though

Sarah could also decouple the fan with a capacitor across each fan so that any strange feedback not mentioned is eradicated

If that is the case one will be for alarm the other will be FG feedback pulse which Ive refered to in above comment with reference to 3 wire fans considerably more difficult Not to worry See what she comes back with either way Sarah will still need fet control just far better to do this way prevents funny problems as well later on and provides descreet overall controll

( Would need to know how the alarm works is this no presence of pulse or Is this logic high / low switching ) In which case are they using a motor control device inside the fan with alarm out and pwm pulse out or is this in to the fan If its using a control specific ic rather than external components

Dont exactly tell you alot do they ? Im sure they expect you to magic this up somehow with limited information

What you need is the internals or block diagram might be an idea to supply those in the datasheet ( expecting alot i guess these days ) Of course flip side of this is when buying or importing these fans Whjy isnt the buyer asking these questions hmmm Aaah also interesting hmm ? Before supplying them

To be honest with you I’ve never seen a single design where they parallel all the fans They don’t

They controll those in parallel via transistors , fets independently controlling the current through each


When inductors are connected in parallel, the total inductance is less than any one of the parallel inductors' inductances

Last point here is ( What model of fan uhm no idea not mentioned )
M34709 62 12 10.2-13.8 500 current in ma

Lets see uh 4 times 500ma = Oh dear 2 amps but Sarah only has a 1 amp supply we are not quite sure as to how good the supply is either are we hmmm ? Nevermind my suppliers telling me they dont work and uh we havent even got a correct power supply to deal with the job on hand

Sounds about right that does Normal

Hmmm, a detailed reply
In my first post i mentioned the fans are 4 wire
they pull aprox 900ma when running
i have a 40a bench power supply (digimess )
connecing red and black wires to 12v the fans all run individually
and connecting the blue wire to a square wave i can control the speed (crudely with a ne555 circuit i breadboarded up )

but when i connect 2 in parallel the issues arise. what i find strange is that what ever signal is given out by the fans totally screws up with my power supply. the voltage output and current readings go haywire .
i tried it on a different supply, a maplin 5a bench supply and the results were the same. so

so what ever these things give back to the power line is really quite disruptive.

what i did do last night was to put diodes across each fan (as a back emf snubber ) and another diode in each positive line. after a hesitant start , both fans eventually spun up but it seemed only to happen once their speeds were in sync.

so i guess i need to try and kill any signals pumped back into the power line.

Grrrrrr,

I see that this mentioned I must admit i didnt notice thatI confused four fans with four wires but i did notice datasheet says (two wires ) hmmm not to worry !! I also noticed work individually


Maplins power supply is this switch mode as well ?


Did you try what I suggested ? Other words Did you build that setup I put together for you ?

What were the results then ?

Have you scoped the supply line disconnected ? , Connected ?

How did you pwm control the fan I know you used 555 but what driver did you use ? Did you use the schematic that I drew for you with PWM setup using a 555 Was in the input to the fets < 5volts ? hmmm

Can we have a schematic of your setup Complete ?

Did you do this with two fans connected with same circuitry ?

Have you decoupled the fans with a capacitor ? On each ?

What does the Alarm signal do ? hmmm wonder if that has something to do with this ? When is the alarm signal raised Under what conditions does this occur?

What type of alarm signal is this pulse , steady dc , if pulsed what is the duration ,type of waveform ?

What must the fan see in order to have no alarm signal ? to raise an alarm signal ? Interesting

****** Where is this detailed in the datasheet ? ****

When using just a single fan with pwm input did you scope the alarm signal ,you scope pwm lines , scope supply lines ?

One more thought is the alarm signalling line an Input or output in other words does the fan have to see some type of signal on the alarm signal wire in or is this an outbound signal ?

In other words does perhaps this fan have some type of input gating to stop the fan from running if it dosent, for arguments sake ,assuming that the alarm wire is in fact an input , perhaps the fan needs to see this under this assumption before the fan can reach max rpm . rotate correctly

Well with limited data I have to make assumptions

Can you upload a photo of your breadboard setup ? also can you upload you scopeed wave forms showing at which points you did these checks on your breadboard setup please ?

Joking of course " You cant have your pudding if you dont eat your dinner " , " How can you have any pudding if you dont eat your dinner "

Right goto your bedroom please Now !! Instantly !!
And you can have your dinner for breakfast tomorrow

Or if it were Tova It would be , " Why not ? < " Why ? " < why havent you done this , why not ? "

Looks like it could be brick in the wall again

If they are 4-wire, could we establish what the wires are, eg
Pin 1 - Ground - Black
Pin 2 - 12v - Yellow
Pin 3 - Sense - Green
Pin 4 - Control - Blue
Do you have the manufacturer name and model number for the fans you have?
If they came out of a PC, do you have the motherboard manufacturer and model number? You can then look up the manual details on the 4-wire connection. Intel m'brds used to be fond of 4-wire and some Asus.
The information provided under 'click' above is a 2-wire dc fan, with various voltage options according to exact model number. It does not seem to tally with your '4 wire fan' statement.

Its great to see all the helpful comments - could it be that some EMI from the fan is disrupting the control loop of the power supplies? one way to test this out is use a battery maybe a 12 v gel cell or such to see if the problem is still present. You could try a filter consisting of an inductor and capacitor (values depending on the frequency of interference) to decouple the fan from the power supply. The diodes are not a bad idea to keep any back emf out of the power supply. These fans are likely electronically commutated though so you would not think that they would generate a lot of noise. It would be good to see a description and schematic of that exact fan model with the 4 wire detailed specifications on the care and feeding of these things. the other wires could be - tach, alarm signals its just hard to tell. Do you have to use this model fan?

maybe contact nidec and let them know the model number and describe what is happening and see if they have an apps engineer you can talk with or any app notes on that exact model?
I know the Nidec rep in the SW US and if you need contact name info I can get it for you.

Sarah, have you got the power supply current limit set high enough, over 10 amps for example?

Sarah, you say a fan draws 900 mA, yet the datasheet shows the largest fan drawing only 500 mA. This is odd.

I believe these are brushless DC fans employing an H-bridge to supply AC to the motor. As such the input current consists of current spikes of perhaps 3X the average current. The power supply might not be able to respond quick enough to the current spikes but a suitable capacitor across the power supply output will enable it to power the fans. A value to use could be *4700 uF and having an ESR of less than 0.1 ohm*. This will account for the power supply taking 1 ms to respond to a current spike and for a 3 volt discharge of the cap (including ESR) at 10 amps.

I’ve had that problem, particularly with dc to dc converters Wont start up looks as though its tripping the supply and its not On start up it draws nearly thirty, forty amps for a few micro seconds then idles quite happily at 50 ma and runs A ok

I still think there is something else here hence why I said start with that circuit I drew You will isolate it that way

All you have to do is just simply provide the information I’ve asked asked Sarah to do Cant be that difficult may not be simple buts its not difficult if you carry out those tests

Its simple fault finding

Datasheet is a massive problem though Its not exactly Sarah fault this I blame the suppliers for this and the manufacturers typical of the non co operation you get same for service manuals I don’t know what you do about this either except don’t buy , sell the parts or equipment unless they supply the information you need

If everyone adopted that attitude they would get the message and they would have to supply the information

I know exactly what they’ve done here as well I have the same row with suppliers , same row with advertising I simply adopt one attitude either you provide the answers to my questions and you back them up with signed agreements or I don’t do business with you that simple

I do exactly the same with customers as well as soon as the arguments start


Off you go please I don’t need your business its simply not worth the aggravation , same for suppliers , same for manufacturers as well No matter how big or how small or what ever it is they want or you want

All sides I simply now adopt this policy that way they cant point a finger at you which is what they do all the time

You will get like this as well make no mistake!! You will get this way as well guaranteed and it’s the same when they ask you to do a job now I look at it think then work out what is that’s going on in the background 9/10 they looking for an excuse to blame someone for something that they’ve done or made a mistake with

I some case I take photos now I take notes i record everything they say because I know what they are So im advising you to do very much the same sorry come across this all the time these days all too often as well

They cost you everything at the end of the day and they will do it as well

I have had several thoughts,now that I know that the motors are 4 wire. Firstly, to avoid RFI problems (or at least to thwart them temporarily) power the motors from an accumulator. If voltage (particularly) and possibly current were to fluctuate, then the RFI is getting into your multimeter.
Secondly, it also may be important to have an individual control for each fan (if using the tach signal to compare to that from the 55 timer. Each motoer will need its oen control loop. The reference frequency being common to all four fans. You will need to be sure that the reference frequency has sufficient drive for the four fans as well. A CRO or DSO will be good to view the reference signal amplitude as fans are added.

thank you everyone for your input.
i have managed to sort this.
special thanks to mark for all of his efforts and off the wall remarks you really do need to get out more often lol

a simple solution with some components i had lying around which was to drive each fan through a simple lm317 voltage regulator. when coupled up to a simple transformer / bridge rectifier and cap works fine.

thank you again every one

Well done Sarah !! Which means , aaaah ha the layout I gave you was correct I knew it would be something along the lines of this Individual control of each fan ensuring that the current is distributed evenly through each fan

But I am really pleased you found this I hope too that from now on before you buy and components or equipment that you will always insist that the manufacturer , supplier etc always gives you all the data not just half the data and further that before you ever buy bulk quantities that you do so slowly that you also ask for sample to rigorously test and see if this complies with the datasheet and you will never go wrong , just through doing this you will also never let your fellow colleges down or hopefully your own company one day, unlike many other companies who have done what ??

Ignored this advice and consequently gone bust

On the subject of going out I whish I could, in fcat I would if i could find somewhere decent to go where people are friendly So if you ever in london look me up and you can take me out hows that for you


Good luck once again Well done