Keysight 34461A Fan Noise

[BVH]

I had the 34461A before upgrading.  It used the same 5V fan with the identical disturbing "howl".  I bought mine from Amazon.  I reasoned "Why do all the wire cutting and resistor inserting when the higher quality Noctua fan with it's higher quality bearing system is a drop-in replacement."

[iMo]

I had the 34461A before upgrading.  It used the same 5V fan with the identical disturbing "howl".  I bought mine from Amazon.  I reasoned "Why do all the wire cutting and resistor inserting when the higher quality Noctua fan with it's higher quality bearing system is a drop-in replacement."

A quick 5cents fix?..

[EE-digger]

There's also a 0 cent fix.  I held a 1/2" thick pad of loose medical cotton on my 34465A fan exit and it silenced the noise.  A thin piece of open cell foam sheet will do the same thing.  Just be sure it's open cell and has no adhesive on it.  The material used to replace window ac filters is about right.

You're basically building a muffler.  Most fan noise is caused by the shearing of the air between the blades and housing, etc. 

[Messtechniker]

Just replaced the stock fan of my 34465A with a Noctua NF-A4x10 5V fan.
The howl is a bit less now, but still not good enough for me

The air flow provided by the Noctua is noticeably higher by the way.

1 k 0.05 % resistor tolerance measurement after a 1 hour
warm-up immediately after having made the change.

Before: -0.0021 % to -0.0020%
After: -0.0013 % to -0.0012%


Just for the record: list of parts and tools you will need for the replacement:

Parts:
1.) The Noctua NF-A4x10 5V fan, of course
2.) 2 M4x20 mm metal bolts (doubtful if plastic bolts will be better)
3.) 2 M4 metal nuts
4.) 2 M4 spring washers

Tools for disassembly and reassembly of your 3446x:
1.) T15 bit
2.) T20 bit
3.) Screwdriver for the 2 M4 bolts you will be using.

What I might be trying next (should Xmas be too boring):
Inserting a piece of cotton fabric between casing and fan.
Reducing fan speed a bit.
As a last resort: cut away the metal at the fan exit. Something I will definitely not try.

Edited: Minus sign for tolerance values

[Zucca]

The air flow provided by the Noctua is noticeably higher by the way.

Then the 100ohm 1 watt resistor could be a nice solution.

I would also measured the delta T (inside - ambient) before and after swapping the fan or installing the resistor.
I have a similar problem on my Agilent 66309D, which is lauder than a 1969 tractor compared to all others devices on my bench.

I swapped the old fan with a new high quality papst one:
https://www.tme.eu/de/details/612nn/dc-lufter-12v/ebm-papst/612-nn/

I took the old one out and I did an "outside device  test" comparing new and old.
I was so happy to hear almost nothing with the new papst and a stupid rattling noise in the old one.

Unfortunately, after installing the new one in the unit I was not even close to the noise of the outside test, I improved the situation but there was still too much noise.
I am now tempted to try to add some rubber padding where the fan is mounted, to me the fan rotation resonate with that big heat sink.

TLTR: changing the fan with a new high quality one is always good, but sometimes it does not solve the problem.

[Zucca]

I coudn't resist and I open up again my 66309D.

After a quick check I discover that the bottom part of fan was rattling against the PCB, so I put some absorbing material:



the noise is now decent.

Furthermore since the fan is mounted on a very long hollow heat sink, the acoustic is bad by design.
In other words that heat sink seems to work somehow like this:

[Messtechniker]

See graphs of the tolerance measurements on a 1 k resistor
using the stock fan and the Noctua replacement fan.

Findings:
1.) Due to the higher air flow of the Noctua, the 34465A stabilizes faster.
2.) A recalibration will be required.

Measurement conditions:
1.) 34465A off over night.
2.) 34465A switched on early in the morning.
3.) After starting up switched immediately to Ohms.
4.) Start logging immediately after having switched to Ohms.

Instead of these measurements I would rather have liked
to log the internal temperature of my 34465A. But in the manual
there is apparently no SCPI command for logging the internal temperature. 
Or is there a secret SCPI command for this?

If not, maybe this would be - if at all possible - an idea for the next firmware update.

Found the command SYST:TEMP?

Will repeat the measurements in the next few days.

[HighVoltage]

What software are you using for logging?

[Gandalf_Sr]

I get the argument that a fan change would justify the need for a recalibration but, having just had my 34461A recalibrated, I'm now not sure I want to change the fan.  If the 34461A has a temperature-controlled voltage reference (does it?) then, as long as the new fan is made to match the same air flow rate (perhaps using a resistor) then is a recalibration really needed?

Surely the calibration can't be that sensitive to the fan flow rate, it's more likely to be sensitive to the ambient temperature and, even then, Keysight would have to make a $1,200 meter have some sort of allowance for a 'normal' ambient temperature range e.g. 65 - 75 F?

[Messtechniker]

What software are you using for logging?

Home-grown logging based on "Profilab" from here:
https://www.electronic-software-shop.com/index.php?language=en

Been using it for years because I am not interested in programming as such,
but rather want to get measurement related things done easily and quickly. Since my (and everyone's at my sort of age) learning ability is degrading by and by over the years  , a program like Profilab
and others from Abacom for that matter, helps to ease the pain.

[Messtechniker]

If the 34461A has a temperature-controlled voltage reference (does it?) then, as long as the new fan is made to match the same air flow rate (perhaps using a resistor) then is a recalibration really needed?

Yes, I am thinking about a speed dropping resistor. Planning to make a quick and dirty air flow indicator based on  a suspended piece of thick paper to determine the resistor value which will cause the Noctua to produce the same air flow as the stock fan. But first I will be doing some internal temperature logging for both fans.

It would be nice to know how the temperature is measured, i.e. sensor type and location on the pcb. and within the circuitry. Pity no schematic is available as in the old days. Sigh. What I do know is that the temperature is measured in increments of 0.1 deg C. So a bit of averaging will be need to smooth things out a bit.

[iMo]

I made measurements on my 34401A (with the temperature sensor mounted inside).
The voltage reference is the same as yours (LM399). The TC of the meter at 10V range (Tdmm=38..43C) is aprox 5uV/C. Yours will be similar, imho, unless 34461A+ does a TC correction. I do TC correction externally (stm32).
The internal temperature Tdmm is always aprox Tdelta=17.5C higher than ambient Tamb (no fan).
With the fan the delta will be lower, but still Tdmm=Tamb+Tdelta.

[nctnico]

See graphs of the tolerance measurements on a 1 k resistor
using the stock fan and the Noctua replacement fan.

What kind of resistor? If this is a standard one then you are likely seeing the temperature dependance of the resistor. It makes no sense for the DMM to be sensitive to the amount of airflow because this is likely to vary anyway by the environment of the DMM, variations between fans, built up of dust and aging of the fan.

BTW I just ordered the Noctua fan with a low-noise-kit (adapter cables to reduce the speed of the fan).

[Messtechniker]

What kind of resistor? If this is a standard one then you are likely seeing the temperature dependance of the resistor. It makes no sense for the DMM to be sensitive to the amount of airflow because this is likely to vary anyway by the environment of the DMM, variations between fans, built up of dust and aging of the fan.

BTW I just ordered the Noctua fan with a low-noise-kit (adapter cables to reduce the speed of the fan).

Got me! It was an ordinary 1k 0.05% 35 year old metal film resistor. Probably with a TC of 50 ppm/K. It was simply a stopgap measure, because at that time I did not know how to measure the internal temperature of the DMM. Even so, the logging graphs above indicate that the higher airflow of the Noctua fan seems to stabilize the DMM readings more rapidly.

[Performa01]

It was an ordinary 1k 0.05% 35 year old metal film resistor. Probably with a TC of 50 ppm/K.

If a resistor is specified as 0.05% (500ppm) tolerance, it certainly has to have a better TC than 50ppm.

I just had a look at my old stock of precision components and most 0.05% resistors have a temperature coefficient of 4-5ppm; the worst one I could find was 10ppm.

Nevertheless the original objection is valid. You should not use Resistance measurements to verify the calibration of a DMM in the first place, even if you have a good resistor reference with TC<1ppm available. A DMM has its best accuracy and stability when measuring DCV. Furthermore you should use the base DCV range with direct connection (without internal attenuator), which is usually the 10/20VDC range. Quite obviously you need a stable (preferably 10V) reference voltage for this.

[nctnico]

A quick off-the-cuff calculation makes me think that the differences measured with the different fans are well within the 34461A's specification. The 34461A's spec also shows a few PPM of drift per degree Celsius of temperature change. So using a fan with more or less airflow is going to affect the 34461A's readings a little bit as well because the internal temperature will be different.

[Messtechniker]

See graphs of the 34465A internal temperature measurements
using the stock fan and the Noctua replacement fan.
Please note the different logging periods.

Measurement conditions:
1.) 34465A off over night.
2.) 34465A switched on early in the morning.
3.) After starting up, started logging immediately and left the lab.
4.) Did not enter the lab while logging.

Conclusion: not much to be seen here as to the different fans influencing the internal temperature. I'll leave the stock fan installed since there is not much to be gained by replacing it with the Noctua fan.
Cutting noise levels down to signifiant levels will here require a redesign of the air flow outlet, preferably with the assistance of an air flow expert (from the car or aircraft industry). Anyway some insights were gained.

[Messtechniker]

Here now the logs

[radhaz]

You could always get one of these, it has no fan.
https://www.eevblog.com/forum/testgear/iwatsu-voac-7602/

[rernexy]

Yesterday my 34461a started producing an extra sound, like a pulsing at high frequency. That sound can be heard in the first attached mp3 (isolated it using a high pass filter on the original audio).

Has anyone ever heard that in their 34461a, or in a related Agilent/Keysight bench DMM?

The sound it usually made is just the high pitched whine from the fan. That can be heard in the 2nd mp3 (isolated it using a low pass filter on the original audio).

The 3rd mp3 is the original unfiltered audio of the noise my 34461a now makes.
The jpg is a spectrogram of the audio, starting from powerup.

[rernexy]

Welp, it's definitely the fan: https://drive.google.com/file/d/1-zcULG_mnfcH52MxQbIPaJsYwtBfZU2V
Is the new sound just a sign that the fan is broken in now? 

[rernexy]

Dismantled the fan. Surprised it uses 2 bearings.

[rernexy]

Replaced the fan with a Multicomp MC002684, which is actually a Sunon MF40100V2-1000U-A99.
Spectrogram of the new sound is attached.

[NoisyBoy]

Good to see you got it repaired, it sounds like my 34461A with the new fan.

I know there has been some complaint about the fan noise on the Truevolt DMMs, but both of mine have been very quiet.  They are perhaps my quietest fanned equipment in my lab.

[tangram]

My 34465A was making one hell of a racket.  It's just over three years old now and never missed a beat, but it was howling like a husky in heat. 

Bought a Noctua NF-A4x10 5V, Premium Quiet Fan, 5V Version (40x10mm, Brown) from amazon and installed it.  It didn't come with the molex two-pin connector that the board uses, so just chopped the wires off the old fan and spliced them together. 

The one I bought from Amazon was the 3 pin version.  I'm not even sure that there is a two pin version of this fan, but just chopped the third (yellow) wire short and capped it with a little heat-shrink before soldering and heat shrinking the +ve and -ve.

Works a treat, blows a good amount of air and is barely audible.  Meter needed a good hoover inside anyway. it was getting a bit dusty in there.

Happiness is restored