DP832 - Fan Replacement 2018

[Gandalf_Sr]

Go back and read my first post.  There's a picture of the PCB with a large resistor that I drew a red rectangle around in one of the pictures.  That resistor carries the current drawn by the fan and the voltage across that resistor is what the DP832 uses to detect whether the fan is running or not.  You need about 1.2V across it when the fan is running, achieve that and the DP832 will not have issues with failed fan messages.  Its really simple, Ohms law.

[Gandalf_Sr]

I got a second DP832 a week ago and ordered a new fan, same one as before. I took the lid off today and the PCB is a new one and there is no big resistor!  I took pictures and will add soon but what did I do?

I measured the current draw of the old fan at 120 mA, the new one is only 60 mA so I added a 200 2W resistor (4 x 820 1/4 W resistors in parallel) and put that across the fan connection.  My calculations say that it only draws 0.7W so 2W seems reasnable.

It runs fine. I loaded up the supply and the fan cranked up to cope.

[mbedwani]

Looks like Rigol has changed the Fan to new model (Protechnic) in by new PSU. I used a Noctua NF-A8 FLX 3 pin FAN, Here is what I did.

https://bedwani.com/wordpress/2021/07/07/making-the-rigol-dp832-psu-quiet/

Mark.

[pushrax]

I have an older DP832, and finally got around to doing this mod. Thank you to everyone who posted useful information here. First I removed the grill with a dremel, and then replaced the fan.

My unit's board had a 7.5 ohm current sense resistor like the original poster. I measured around 1.2V across the resistor during boot up when fan spins up to max (~160mA draw). My replacement fan (some old DeepCool fan I had laying around) measured around 0.7V during that step (~90mA draw).

1.2V/0.7V ≈ 13Ω/7.5Ω. I desoldered the resistor and replaced it with 13.5Ω (two 27Ω 0.25W resistors in parallel). The unit is working perfectly and is super quiet now.

The total power across the current sense resistor peaks ~100mW, so minimal waste.

[sequoia]

I also ended up replacing the stock fan, as it was unbearably loud...

Used also Noctua NF-R8 redux-1800 (non-PWM version).

My unit has V03.02 top and bottom PCBs (dated 2018-06-12) and the fan current sense resistor is now done using three SMD resistors (although looks like through-hole resistor could still be used with this PCB, see attached photo...R210 vs R206-R208)
 
Didn't feel like taking everything apart to be able to swap out the current sense resistors, so made a short adapter cable from 2-pin JST to 4-pin PC fan connector (female), so no need to modify fan at all.... 

I ended up also using a parallel resistor in the adapter cable to avoid fan error... 330ohm 2W resistor brought up the current current consumption within 5mA or so to the original stock fan (my unit came with a Protechnic MGA8012HR-O25 fan).

[Kugelsicher]

Got a DP832 and, same as everyone here, noticed that it can be very loud. I found this thread and read about everyone's experiences and fixes, which made me want to do some testing to validate some of the theories. From the start, my concerns were the lack of headroom on the heat sink temperature at max load and the reduction in airflow potentially causing the smaller heat sinks that aren't monitored to overheat.

My DP832, purchased in August of 2022, has "Digital Version" 00.01.16 and came with a Protechnic MGA8012HR-O25 series b fan.

Here are the 80mm 12VDC fans included in my testing along with the specs the listed on their datasheets:
Listed Specs
NMB 3110SB-04W-B50-E00                 140mA   41.3CFM    0.150inH2O 30.0dBA
Protechnic MGA8012HR-O25 series b   240mA   37.2CFM   0.123inH2O 34.6dBA (Stock)
Noctua NF-R8 redux 1800 3-Pin           110mA   31.0CFM    0.141inH2O 17.1dBA
Noctua NF-A8 FLX 3-Pin                         70mA   29.7CFM   0.196inH2O 16.1dBA

First, I measured the 3 current sense resistors individually and in parallel:
Resistor: 16.2 || 16.2 || 100 = 7.49 Ohm (7.52 Ohm Measured)



Next, I measured the actual current and flow rate. The airflow coming out of a fan is like an inverted vortex which makes extrapolating the CFM from wind speed very difficult as the velocity is much higher at the edges. The air flow is also at an angle outward, away from the fan axis, introducing sine error. To solve this, I designed a simple flow collimator to straighten out the flow and make it relatively even across the exit surface. Then, I just multiplied the measured ft/min by the area of the exit on the collimator to calculate CFM. I used a Modern Device MD0555 Anemometer for the measurements. Note, the collimator restricts flow so the CFM measurements shouldn't be compared against the no-load value provided on their datasheets; but this test shows the relative performance of each fan and is a base-line for later tests. Also know that although I powered the fans from an external supply, was mindful of consistent placement of the collimator and sensor, and allowed the multimeter reading to stabilize before recording the reading, there were still enough uncontrolled variables to cause significant error on all of the following measurements. I'm showing 3 digits of precision but the final digit is definitely within the noise of my measurement setup.




On the Bench Measured Specs @ 12.4V
NMB 3110SB-04W-B50-E00                 140mA   34.8CFM
Protechnic MGA8012HR-O25 series b   125mA   30.7CFM
Noctua NF-R8 redux 1800 3-Pin           104mA   23.5CFM
Noctua NF-A8 FLX 3-Pin                          54mA  21.9CFM

I then installed each fan inside the case, placed the collimator at the exhaust, and measured the installed CFM for each fan.
Installed, Stock Case Measured CFM @ 12.4V
NMB 3110SB-04W-B50-E00                              12.9CFM
Protechnic MGA8012HR-O25 series b                11.3CFM
Noctua NF-R8 redux 1800 3-Pin                        10.8CFM
Noctua NF-A8 FLX 3-Pin                                     7.9CFM

Following this, I cut away the metal that makes up the stock fan guard, replaced it with a standard 80mm wire fan guard and measured the installed CFM again (still with the collimator at the exhaust).
Installed, Cut-Away Case with Wire Guard Measured CFM @ 12.4V
NMB 3110SB-04W-B50-E00                              28.4CFM
Protechnic MGA8012HR-O25 series b                25.8CFM
Noctua NF-R8 redux 1800 3-Pin                        14.2CFM
Noctua NF-A8 FLX 3-Pin                                    17.0CFM (I retested this to make sure I hadn't swapped the value with the other Noctua; it is correct)

As you can see, cutting away the stock fan guard and replacing it with an off the shelf wire guard has a huge positive affect on airflow. I did not measure noise during these tests.

Back when the stock fan was installed, I soldered some sense wires to the fan's current resistor and fan header and ran them out of the case to determine how the fan was controlled. I used these wires to measure the current and voltage of the fan and also to apply a load resistor in parallel with the fan to adjust the fan's apparent current draw. The values are below for you to judge for yourself; but my conclusion was that the fan is voltage controlled not current controlled. When the load was increased, the voltage decreased 3% while the current increased 41%. The small additional V-drop with added load is due to ~380mOhm resistance in traces or something between the voltage sensing & the fan connector
IDLE (No load on PSU) comparison:
Without vs With 220 Ohm Resistor in Parallel with Fan
Fan Voltage:          8.25 to 7.99 V
Resistor Voltage: 0.622 to 0.877 V
Current:                   83 to 117 mA (calculated)

FULL LOAD (enough load for the DP832 to max out the fan)
Without vs With 220 Ohm Resistor in Parallel with Fan
Fan Voltage:           12.6 to 12.2 V
Resistor Voltage:  0.969 to 1.355 V
Current:                  129 to 181 mA (calculated)

Next I reinstalled the Noctua NF-A8 in the case (accidentally blowing inward), plugged it into the DP832, and fully reassembled the power supply. I then ran a few timed load tests using a DC electronic load to compare these values to the same tests I had run earlier with the stock Protechnic fan. Note, a low voltage, high current load results in the most dissipation in the pass transistor because the 30V channel transformer taps are listed as 15V, 25V, 33V (probably significantly higher actual AC voltage for margin). So at 1V output the transistor is burning 15V transformer tap - 1V out = 14V times the current.


The stock results are first.

--STOCK HARDWARE--
Fan: Protechnic MGA8012HR-O25 series b
Fan Voltage Range:   8.22  - 12.60 V (Measured at empty diode pads beside connector)
Resistor:             0.646 - 0.969 V (Measured at Current Resistor test points)
Current:               86    - 129   mA (Calculated from V_resistor)
NOISE TEST
Conditions: Measured in final seconds of TEST 3 with meter sitting on top of the PSU at the very front, facing forward and on the flip-stand
Max Noise: 62 dbA
TEMP TEST 1
Conditions: 30V, 3A Load for 1 Hour
Final Temp: 72.7 C
Ambient Temp: 24 C
TEMP TEST 2
Conditions: 1.4V, 3A Load for 30min
Final Temp: 89.8 C
Ambient Temp: 24 C
TEMP TEST 3
Conditions: Ch-1 & Ch-2 in Parallel, Each Set to  1.4V, 2.45A; for 10min
Final Temp: 74.8 C
Ambient Temp: 25 C


--NEW FAN BLOWING INWARD, BUILT-IN GUARD CUT AWAY, WIRE GUARD ADDED--
Fan: Noctua NF-A8 FLX 3-Pin
Fan Voltage Range:   8.600 - 13.17 V (Measured at empty diode pads beside connector)
Resistor:              0.273 - 0.403 V (Measured at Current Resistor test points)
Current:                36    - 54    mA (Calculated from V_resistor)
NOISE TEST
Conditions: Measured in final seconds of TEST 3 with meter sitting on top of the PSU at the very front, facing forward and on the flip-stand
Max Noise: 54 dbA
TEMP TEST 1
Conditions: 30V, 3A Load for 1 Hour
Final Temp: 53.6 C
Ambient Temp: 25 C
TEMP TEST 2
Conditions: 1.4V, 3A Load for 30min
Final Temp: 61.5 C
Ambient Temp: 25 C
TEMP TEST 3
Conditions: Ch-1 & Ch-2 in Parallel, Each Set to  1.4V, 2.45A; for 10min
Final Temp: 55.2 C
Ambient Temp: 25 C

Here I noticed that the fan was blowing the wrong way and reversed it.

--NEW FAN BLOWING OUTWARD, BUILT-IN GUARD CUT AWAY, WIRE GUARD ADDED--
Fan: Noctua NF-A8 FLX 3-Pin
NOISE TEST
Conditions: Measured in final seconds of TEST 2 with meter sitting on top of the PSU at the very front, facing forward and on the flip-stand
Max Noise: 57 dbA
TEMP TEST 1
Conditions: 30V, 3A Load for 1 Hour
Final Temp: 76.2 C
Ambient Temp: 26 C
TEMP TEST 2
Conditions: 1.4V, 3A Load for 30min
Final Temp: 89.6 C
Ambient Temp: 25 C

At this point I was very disappointed, having reduced the noise a bit but not having improved the temperatures at all. The contrast of the inward blowing versus outward blowing fan tests made it clear to me that the airflow path was a key problem. So, I designed (and redesigned several times) a flow guide for the inside of the case and printed it in a flame retardant rated PC-ABS filament (the prototypes were all in PLA but I didn't feel safe leaving that inside something that can fail over 100C). The final tests were with the final version of the flow-guide installed.

--NEW FAN BLOWING OUTWARD, WITH INTERNAL AIR GUIDE, BUILT-IN GUARD CUT AWAY, WIRE GUARD ADDED--
Fan: Noctua NF-A8 FLX 3-Pin
NOISE TEST
Conditions: Measured in final seconds of TEST 2 with meter sitting on top of the PSU at the very front, facing forward and on the flip-stand
Max Noise: 56 dbA
TEMP TEST 1
Conditions: 30V, 3A Load for 1 Hour
Final Temp: 62.7 C
Ambient Temp: 26 C
TEMP TEST 2
Conditions: 1.4V, 3A Load for 30min
Final Temp: 75.2 C
Ambient Temp: 25 C



Finally, I tested the smaller heat sinks and found the temperatures to be with 10 C of room temp in most cases. With a large load on the 5V rail, the tall heatsink on the right side (viewed from the front) of the upper board reached 53 C after 15min. This was the highest value measured on the smaller heat sinks. I was sick of testing at this point and didn't write most of the values down; so I don't have a fully dataset for this.

I think this testing makes clear that most of the noise is coming from the airflow past the internals of the power supply and not the fan itself. This makes it more difficult to reduce than expected. I would also advise against simply installing a quieter & slower fan as the heat sinks get very hot with a large load even with the stock fan jet-engining air out of the case. I think a combination of a Noctua fan, cutting away the built-in fan guard area, and adding a wire guard offers a decent improvement without increasing loaded temperatures. Another option is to install a quieter fan blowing inward instead of out of the case. This results in excellent cooling of the primary heat sinks but will increase the temperature inside the case, which isn't good for the electrolytic caps, smaller heat sinks, or transformer (whose internal temperature may already be quite high). It would probably also cause the voltage reference to drift and affect accuracy.

If anyone has any other ideas, please let me know. I'd be up for more testing in a few weeks and the fan still isn't quiet at high loads.

[tv84]

  Nice testing!! 

[Hexley]

If anyone has any other ideas, please let me know.

I wonder if the exhaust air is not being drawn **through** the heat sinks, but rather around them. From the photos, it looks like the natural path would be through the channel between the outer edges of the heat sinks and the adjacent sides of the chassis. That looks like the path of least resistance. But it would not be the path of most effective cooling for the main heat sinks.

An interesting test would be to add a couple of small fans inside the case, blowing directly into the intake area of the heat sinks. These would encourage the air to flow through the inner parts of the heat sink. That hot air would then get expelled by the normal exhaust fan.

The intake fans could be quite small I think -- and therefore very quiet. They could poach power from the main fan since there is power to spare with the retrofitted fan. Perhaps the fans could be mounted with some small printed brackets.

Just thinking out loud, so to speak.

[Kugelsicher]

An interesting test would be to add a couple of small fans inside the case, blowing directly into the intake area of the heat sinks. These would encourage the air to flow through the inner parts of the heat sink. That hot air would then get expelled by the normal exhaust fan.

I like this idea and I think I have 4 refurb 40mm 12VDC Sunon fans that I can use to test this out. If so, I'll try it out in a couple of weeks.

[Veketti]

Old topic, but still valid. I got fed up to the fan noise and cut out the fan cover. That made the noise so much lower that no need to replace the fan. Did the same thing to my DS2000 scope but there no noticeable differences. I quess I need to change the fan on that scope then..

[Veketti]

Tried two different fans which were not yet tested here. Both Sanyo fans:
San Ace 9RA0812S4001. 38.6CFM / 64.8m3/h 0.210 in H2O / 52.3Pa 31dbA 80x25mm
San Ace 9RA0912H40011 52CFM / 87.6m3/h, 0.19 in H2O / 47.3Pa 31dbA  92x25mm

That 80mm fan would have been perfect, but it draw 40mA less current than the original and gives fan error. It was around 2dbA more quiet than the stock fan without the grill. Still having better performance. 28.7°C temperature with one channel on 10mA load.

As I had the thingverse 80->92mm adapter printed I installed the San Ace 92mm fan. Same noise than the original fan without the grill 40dbA, but that thing moves some serious amounts of air. So what I learned from this is that these San Ace fans are the highest quality fans that I've seen. Each of them seem to be balanced, which I haven't noticed on other brands. Also blade design is good that it is quiet for the amount of air it moves.