- Joined
- Jan 7, 2014
- Location
- Munich, Germany
connecting a 10k resistor in parallel should help
-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
FEATURED Building PWM Controller for 4 wires PWM fan
- Thread starter Brutal-Force
- Start date
connecting a 10k resistor in parallel should help
- Joined
- Apr 4, 2011
oops...
sorry, I think I misunderstood your problem here
you can try to omit the 3.6k resistor between the pot and gnd.
this will reduce your minimum and maximum output voltage.
as for the soldering and de-soldering part, you might want to take a 2nd look to my edited previous post for you
and you can use the same way to mount your resistor as you mount the 1000uF on your strip board
- Joined
- Nov 29, 2010
Thanks for the input guys. I've tried many combinations and no combination will give me 5-11v like I had on the breadboard. I remember that I had the exact same problem years ago when I first built this circuit and I just forgot about the circuit... until yesterday when I thought "maybe I can fix it!" and now I regret remembering hahaha - it's driving me crazy! 
inVain, I tried both your suggestions and neither work I'm afriad. Connecting wiper to other pot terminal doesn't do anything. I also tried reducing resistor value to ground the pot, and with 2.4K the voltage range is about 4v to 11v, and the pot works from 0-75% but nothing higher. Another thing I noticed is the voltage dropout, even with VERY low powered fans (83mA Gentle Typhoon) the voltage drop is 1v across the regulator. Input from PSU is 12.04v, output from LM1084 is 11.06v. I don't remember dropout being so big! Output with no load is 11.4v, not a lot better. Is my LM1084 faulty, or should I just use a 5K pot and everything will work perfectly?
inVain, I tried both your suggestions and neither work I'm afriad. Connecting wiper to other pot terminal doesn't do anything. I also tried reducing resistor value to ground the pot, and with 2.4K the voltage range is about 4v to 11v, and the pot works from 0-75% but nothing higher. Another thing I noticed is the voltage dropout, even with VERY low powered fans (83mA Gentle Typhoon) the voltage drop is 1v across the regulator. Input from PSU is 12.04v, output from LM1084 is 11.06v. I don't remember dropout being so big! Output with no load is 11.4v, not a lot better. Is my LM1084 faulty, or should I just use a 5K pot and everything will work perfectly?
- Joined
- Apr 4, 2011
connecting the wiper to the other terminal won't do anything different than leaving the other terminal N/C.
but this had save me from headache from picking which terminal on the pot should I connect, at worst only the pot direction will be reversed
also; with time, the wiper contact might be worn out.
by doing this, at least we could expect some "backup" from the other terminal to be still intact and make the necesarry connection.
when the pot connected directly to ground, you should see the low voltage reaching around 1.3v (or maybe less quite a bit).
with R1 value @1k2, the highest voltage should be around 11v (with some light load).
so, this should fix the knob range issue.
did you change your IC from the one that you had on the breadboard version?
your results more like a 317 to me
the "dead spot" on the knob was resulted by setting the output voltage higher than the input voltage.
for a 317, with a resistor network of 1k2 + 10k + 2k4 will result maximum output around 14v, and minimum around 3.8v.
and you need input around 16v for the "dropout" headroom.
since we use 12v as input, the 14v output will never be achieved.
thus, dead spot occur on the higher pot position.
if it's an LM1084, you should get 11.3v at max with input around 13v (including the dropout)
if it's really a 317, I'd try to substitute R1 with 1k8, and R2 with 4k7 while maintaing the pot @10k to get 5-11v range
you can find the equation needed on the datasheet, or an online calculator to cheat
most datasheet for these kinds of IC use pot less than 10k.
I even use 1k for my PAPST for quite some time.
using lower resistor value might increase quiescent power, but I had no ill experience and my PAPST responding better than if I use 10k.
but this had save me from headache from picking which terminal on the pot should I connect, at worst only the pot direction will be reversed
also; with time, the wiper contact might be worn out.
by doing this, at least we could expect some "backup" from the other terminal to be still intact and make the necesarry connection.
when the pot connected directly to ground, you should see the low voltage reaching around 1.3v (or maybe less quite a bit).
with R1 value @1k2, the highest voltage should be around 11v (with some light load).
so, this should fix the knob range issue.
did you change your IC from the one that you had on the breadboard version?
your results more like a 317 to me
the "dead spot" on the knob was resulted by setting the output voltage higher than the input voltage.
for a 317, with a resistor network of 1k2 + 10k + 2k4 will result maximum output around 14v, and minimum around 3.8v.
and you need input around 16v for the "dropout" headroom.
since we use 12v as input, the 14v output will never be achieved.
thus, dead spot occur on the higher pot position.
if it's an LM1084, you should get 11.3v at max with input around 13v (including the dropout)
if it's really a 317, I'd try to substitute R1 with 1k8, and R2 with 4k7 while maintaing the pot @10k to get 5-11v range
you can find the equation needed on the datasheet, or an online calculator to cheat
most datasheet for these kinds of IC use pot less than 10k.
I even use 1k for my PAPST for quite some time.
using lower resistor value might increase quiescent power, but I had no ill experience and my PAPST responding better than if I use 10k.
Last edited:
- Joined
- Nov 29, 2010
Thanks again for your help inVain. Nice idea about the 317 however I have never owned one, the only vreg I have ever owned is this 1084!
I suspected there might be a problem with R1 because the voltage was reaching a threshold at about 5K pot resistance, which it shouldn't, so I decided to remove R1 and put it on the breadboard instead. I tried some different values:
R1 = 2.4K, max output 8.6v (with 130mA fan)
R1 = 1.7K, max output 11.06v, min 4.4v, pot was almost normal 0-100% here.
I also tried the "original" LM1084 circuit posted by bing, which uses much lower value components: R1 = 120, Pot = 1K, R2 = 360. NO LUCK. Turning the pot even slightly caused the voltage to go to max @ 11.4v.
So this confirms that all my connections are OK, but the LM1084 is not behaving according to spec.
I think my LM1084 has been damaged somehow, maybe from oxidisation or excess heat from repeated soldering. Rather than continue with troubleshooting this circuit and components, I will go back to breadboard with new components and see if I can reproduce this problem.
I suspected there might be a problem with R1 because the voltage was reaching a threshold at about 5K pot resistance, which it shouldn't, so I decided to remove R1 and put it on the breadboard instead. I tried some different values:
R1 = 2.4K, max output 8.6v (with 130mA fan)
R1 = 1.7K, max output 11.06v, min 4.4v, pot was almost normal 0-100% here.
I also tried the "original" LM1084 circuit posted by bing, which uses much lower value components: R1 = 120, Pot = 1K, R2 = 360. NO LUCK. Turning the pot even slightly caused the voltage to go to max @ 11.4v.
So this confirms that all my connections are OK, but the LM1084 is not behaving according to spec.
I think my LM1084 has been damaged somehow, maybe from oxidisation or excess heat from repeated soldering. Rather than continue with troubleshooting this circuit and components, I will go back to breadboard with new components and see if I can reproduce this problem.
- Joined
- Apr 4, 2011
my pleasure,
I dig back my datasheet and gained some more knowledge myself
I'm still scratching my head, your 1084's quite match the calculation for the 317
with 1.7k, 10k + 3.6k, you should get around 3.9v min, and 11.2v max
considering the components tolerance, variations on the ICs reference voltage, all the wiring mess and such; your results quite make sense to me.
anyway,
the 317s are pin to pin compatible with the 1084, in the TO-220 package they are even interchangeable.
they are super cheap (I can get around 14 pcs of 317s in comparison of a single 108x
)
it's not such a bad idea to experiments with the 317 to get the best possible layout for the future 108x.
and you can always use the fan socket for easy access to swap the ICs around
I dig back my datasheet and gained some more knowledge myself
I'm still scratching my head, your 1084's quite match the calculation for the 317
with 1.7k, 10k + 3.6k, you should get around 3.9v min, and 11.2v max
considering the components tolerance, variations on the ICs reference voltage, all the wiring mess and such; your results quite make sense to me.
anyway,
the 317s are pin to pin compatible with the 1084, in the TO-220 package they are even interchangeable.
they are super cheap (I can get around 14 pcs of 317s in comparison of a single 108x
)it's not such a bad idea to experiments with the 317 to get the best possible layout for the future 108x.
and you can always use the fan socket for easy access to swap the ICs around
- Joined
- Nov 29, 2010
I think my LM1084 became a 317 because I had it in a drawer for four years, a process of backwards evolution ahahah
I have ordered some more components which will be with me in a few days. I have ordered two LM1084 and some other components so I can make a final fan controller. Maybe it will have a 0-100% PWM channel as well.
I have ordered some more components which will be with me in a few days. I have ordered two LM1084 and some other components so I can make a final fan controller. Maybe it will have a 0-100% PWM channel as well.
- Joined
- Nov 29, 2010
@foz182, I didn't make the LCD - it's an Alphacool display which is powered by USB and has its own software. They come up on ebay from time to time but I sold mine a while ago because it wasn't getting used. In the youtube video I have the LCD configured to read out a value from Everest monitoring software, which is now Aida64.
Here's an update on the LM1084 palaver: brand new components (arrived today), set out on breadboard the same way as my circuit is set out on strip board, and I'm able to replicate the exact same behaviour: pot turns 50% and I get full power. The next step for me is to set up the circuit exactly as I set it up on the breadboard when I tested it previously, five years ago(!). Only then will I figure out why it is not working properly. Failing that, I'm gonna buy a 5K pot, or a Micrel adjustable regulator LOL!!
Here's an update on the LM1084 palaver: brand new components (arrived today), set out on breadboard the same way as my circuit is set out on strip board, and I'm able to replicate the exact same behaviour: pot turns 50% and I get full power. The next step for me is to set up the circuit exactly as I set it up on the breadboard when I tested it previously, five years ago(!). Only then will I figure out why it is not working properly. Failing that, I'm gonna buy a 5K pot, or a Micrel adjustable regulator LOL!!
@LennyRhys, thanks for letting me know, hope you dont mind im trying to replicate your project from back then. Will try to pick one up online. Just one question, how would did you get tach from each fan to read into aida64 software?
- Joined
- Nov 29, 2010
Whenever I make a fan controller I usually connect the fan's tach signal directly to the motherboard. The monitoring section in Aida64 automatically reports fan speed as soon as it detects a tach signal, and it will even tell you which header you've connected it to (they have strange labels, like PSU fan, chassis fan etc.).
______
Finally my LM1084 saga is over! Electronics is an exact science (well, exact enough) and I must conclude that bing's circuit must have incorrect values for the resistors or the potentiometer. I have done more testing with different resistor values using a 10K pot and I'm able to get the pot working almost perfectly with higher value resistors:
Adj to Out = 2K (+/- 5%)
Pot to GND = 5K
Voltage range is approx 5-11v, but I've yet to test it on my "good" PSU (which is 12.05v solid on the 12v rail). My test PSU is pretty weak, at only 11.6v, which explains why I'm getting a max voltage of 10.58v.
RESULT!
Edit: with 1.6k and 5k I'm getting a range of 5.2v - 10.97v with the 10K pot. Tried it with an 800mA Delta (equivalent to six of my San Ace fans) and the upper end drops to about 10.93v, and the 3A Delta generates some serious heat! In any case I'm glad to have sorted this out as I can now think about building the circuit. I'm going to need one FAT heatsink on that IC though...
______
Finally my LM1084 saga is over! Electronics is an exact science (well, exact enough) and I must conclude that bing's circuit must have incorrect values for the resistors or the potentiometer. I have done more testing with different resistor values using a 10K pot and I'm able to get the pot working almost perfectly with higher value resistors:
Adj to Out = 2K (+/- 5%)
Pot to GND = 5K
Voltage range is approx 5-11v, but I've yet to test it on my "good" PSU (which is 12.05v solid on the 12v rail). My test PSU is pretty weak, at only 11.6v, which explains why I'm getting a max voltage of 10.58v.
RESULT!
Edit: with 1.6k and 5k I'm getting a range of 5.2v - 10.97v with the 10K pot. Tried it with an 800mA Delta (equivalent to six of my San Ace fans) and the upper end drops to about 10.93v, and the 3A Delta generates some serious heat! In any case I'm glad to have sorted this out as I can now think about building the circuit. I'm going to need one FAT heatsink on that IC though...
Last edited:
- Joined
- Apr 4, 2011
0.6v drop is quite good
by the way, you can use your case as a big fat heatsink for the IC
just make sure to insulate the IC properly
and that reminds me to the "open hardware monitor" (LINK)
really similar to the well known h/w monitor, but with open hardware monitor we can relabel FAN #1 to CPU FAN and such.
I also found the reported CPU temps (the "on die" CPU temps) were slightly higher compared to the h/w monitor.
I was wondering if anybody tried this open h/w monitor with modern AMD CPUs,
does it give any make sense temps report?
by the way, you can use your case as a big fat heatsink for the IC
just make sure to insulate the IC properly
and that reminds me to the "open hardware monitor" (LINK)
really similar to the well known h/w monitor, but with open hardware monitor we can relabel FAN #1 to CPU FAN and such.
I also found the reported CPU temps (the "on die" CPU temps) were slightly higher compared to the h/w monitor.
I was wondering if anybody tried this open h/w monitor with modern AMD CPUs,
does it give any make sense temps report?
- Joined
- Sep 7, 2013
0.6v drop is quite good
by the way, you can use your case as a big fat heatsink for the IC
just make sure to insulate the IC properly
and that reminds me to the "open hardware monitor" (LINK)
really similar to the well known h/w monitor, but with open hardware monitor we can relabel FAN #1 to CPU FAN and such.
I also found the reported CPU temps (the "on die" CPU temps) were slightly higher compared to the h/w monitor.
I was wondering if anybody tried this open h/w monitor with modern AMD CPUs,
does it give any make sense temps report?
I see no difference between the two.
View attachment 178282
CPU-Z and HWMonitor are acting weird today though. HWMonitor doesn't show my voltages and CPU-Z reports a Vcore of 1.4 which is not true.
Temps are correct though. I know that for sure.
- Joined
- Apr 4, 2011
I see no difference between the two.
View attachment 178282
CPU-Z and HWMonitor are acting weird today though. HWMonitor doesn't show my voltages and CPU-Z reports a Vcore of 1.4 which is not true.
Temps are correct though. I know that for sure.
I see, both of the SS were missing the voltage data.
that's weird, especially when you're using a sabertooth
I'm fully confident with the AMD's temps report up to the Phenom II era,
but anything afterwards were quite bizarre.
sub ambient temps reports were quite common for bulldozer /vishera, while the APUs were seemingly too high.
So I got curious if the s/w was reporting the right value
your temps are surprisingly terrific for that 6 cores CPU @4Ghz+
so, no drift between those two
thx a bunch for taking those SS
- Joined
- Nov 29, 2010
inVain said:
He has custom water cooling with 360 rad
Most of these programs allow you to set temp offset so that they report temps more accurately (Aida64 allows this). IMHO the best software for CPU temps is RealTemp, and I use it every day to monitor CPU temps.
Small update on the subject of my next fan controller: I found some cheap regulators by Micrel (MIC29302WT) which have max 350mV dropout at 3A. I got 3 of them for a little more than a single LM1084, so now I wish I hadn't bought two more LM1084 hahaha. Live and learn, as they say.
- Joined
- Apr 4, 2011
the water master race
he's really successful with his mobo temps as well
I've been stuck far too long with the free version of CPUID h/w monitor.
never realize the calibration features were there, all the time LOL.
when the first time I met RealTemp (or was it CoreTemp), I had no prior experience with anything "inverted", it really gave me some headaches back then
You're really lucky to have such vast options out there
he's really successful with his mobo temps as well
I've been stuck far too long with the free version of CPUID h/w monitor.
never realize the calibration features were there, all the time LOL.
when the first time I met RealTemp (or was it CoreTemp), I had no prior experience with anything "inverted", it really gave me some headaches back then
You're really lucky to have such vast options out there
- Joined
- Sep 7, 2013
Thanks guys!
I just quickly booted up Prime95 for that SS. Normally hits 50C MAX core temps. Temperature #1 is the socket temp. It will climb to within 10C of the CPU. #2 is the motherboard temps, it is usually 32C.
I can answer the temperature question on the AM3+ FX line.
The temperature diodes in the FX are only accurate once they hit +40C. It's not actually a die temp either, rather it's some sort of complex formula to the tjMAX temp of the CPU, 62.2C. Something like that. It's very bizarre...
Nobody knows why they did it like that... AMD Overdrive is the most accurate on these platforms, but it only lists the distance to the max temps. "Thermal Headroom" they call it.
I just quickly booted up Prime95 for that SS. Normally hits 50C MAX core temps. Temperature #1 is the socket temp. It will climb to within 10C of the CPU. #2 is the motherboard temps, it is usually 32C.
I can answer the temperature question on the AM3+ FX line.
The temperature diodes in the FX are only accurate once they hit +40C. It's not actually a die temp either, rather it's some sort of complex formula to the tjMAX temp of the CPU, 62.2C. Something like that. It's very bizarre...
Nobody knows why they did it like that... AMD Overdrive is the most accurate on these platforms, but it only lists the distance to the max temps. "Thermal Headroom" they call it.
- Joined
- Apr 4, 2011
Thanks guys!
I just quickly booted up Prime95 for that SS. Normally hits 50C MAX core temps. Temperature #1 is the socket temp. It will climb to within 10C of the CPU. #2 is the motherboard temps, it is usually 32C.
I can answer the temperature question on the AM3+ FX line.
The temperature diodes in the FX are only accurate once they hit +40C. It's not actually a die temp either, rather it's some sort of complex formula to the tjMAX temp of the CPU, 62.2C. Something like that. It's very bizarre...
Nobody knows why they did it like that... AMD Overdrive is the most accurate on these platforms, but it only lists the distance to the max temps. "Thermal Headroom" they call it.
This is the best answer I ever had.
Thanks a ton!
how about the APUs?
does the same answer apply?
I need to upgrade my system,
but these temps issues and the "minimum specs for OC" always hinder me to pull the trigger
edit:
about the tjMAX, does this mean that the AM3+ FX line were intended to never exceed 62.2C (safe operating temps were under 62.2C)?
Last edited:
- Joined
- Sep 7, 2013
No problem!This is the best answer I ever had.
Thanks a ton!
how about the APUs?
does the same answer apply?
I need to upgrade my system,
but these temps issues and the "minimum specs for OC" always hinder me to pull the trigger
edit:
about the tjMAX, does this mean that the AM3+ FX line were intended to never exceed 62.2C (safe operating temps were under 62.2C)?
Give or take a few C yes. 70C is the max on 6 core CPU's like the FX 6300, but we all recommend 62C.
I don't know about the APU's for sure but I do think they use the same system to get the core temps. Not 100% on that though.
