Fender Rumble 500v3 Fan Noise - My Troubleshooting and Resolution Journey

[dORSY]

With all of the physical restrictions in the cooling airflow path of the R500 amp

and this is the part that's not clear. what's the required pressure here?
any excess is basically a "waste".
if the fan is driven well (by temp sensors, targeted temps) it should run at a higher curve spot == moar pressure and flow when needed.
seems like the stock fan is optimized for around 10cfm (horizontal part in the middle), which the noctua can easily do.

also:
MTTF > 150,000 h and a 6-year warranty
vs.
the 1-year(?) that sunon has according their PDF You quoted

which says something about the durability and lifespan.

more on the topic (by noctua): https://www.noctua.at/en/expertise/tech/nf-a12x25-performance-comparison-to-nf-f12-and-nf-s12a

 

[agedhorse]

There is significant pressure drop due to the slot grilles that Fender uses on these models. There are 3 different schemes deepening on if it's a combo or head and also which head version (one uses a 90 degree duct, similar to the straight ducts used on some Genz Benz models), and each scheme has different losses across the grille and through the duct. Historically, Fender's slotted grill is about 50% open, which is why static pressure curves are important.

 

[Never2Late]

if the fan is driven well (by temp sensors, targeted temps) it should run at a higher curve spot == moar pressure and flow when needed.

Agreed, if fan speed was controlled by temperature sensors, it might self correct up to the airflow limits of the fan.

BUT the circuit block diagram doesn’t show any temperature sensors feeding the fan speed. It only shows power amp output and power amp clip detection as control variables. If that’s accurate, then inadequate airflow would result in hotter components all the time.

That was part of what I tried to share in post 2 of this thread, apparently not clearly.

Voltage to the fan from the amp remained at 6V during all of bench testing, even after heat sinks of amp components became hot to the touch.

MTTF > 150,000 h and a 6-year warranty
vs.
the 1-year(?) that sunon has according their PDF You quoted

which says something about the durability and lifespan.

Also agreed that Noctua fans are well built, maybe better than the SUNONs.

BUT the Fender Rumble Club thread has almost 2000 members and over 4 million views, and documented failures of the stock Rumble v3 fans (in combos and standalone amps) have been extremely rare. Their durability doesn’t appear to be an issue.

I do like Noctua fans (in my computers). But even if they are built better, that still doesn’t make them suitable for this application.

I’m not defending SUNON. And I’m not happy that my stock fan was whining (even if it’s not an issue for 99% of other owners).

I am pointing out that the Fender Rumble amps have proven themselves to be reliable as designed and built, and that includes component cooling.

Is it overkill? Maybe, maybe not. But if @agedhorse, with his lifetime of designing and repairing amplifiers, is telling us to to stay with the stock fan, we really should be listening.

And if there’s a suitable silent factory replacement available for the few of us that want it (the MB fan), I don’t understand the pages and pages of debate.

 

[dORSY]

There is significant pressure drop due to the slot grilles that Fender uses on these models. There are 3 different schemes deepening on if it's a combo or head and also which head version (one uses a 90 degree duct, similar to the straight ducts used on some Genz Benz models), and each scheme has different losses across the grille and through the duct. Historically, Fender's slotted grill is about 50% open, which is why static pressure curves are important.

that noctua fan can overcome way more than 50%. numbers on required pressure? still can't see maths beyond repetitive "curves" :/

 

[agedhorse]

The Rumble 500 V3 documentation that I have does include proportional control of the fan. At normal temperatures the fan's quiescent voltage is biased to about 6V and the fan turns slowly, this is important for the rest of the circuit to track. As the amp heats up (it takes a pretty good signal and load to do so), the fan voltage increases ultimately to about 12V.

It's the tracking of the fan from the 6V offset to the variable set-point speed that the circuit depends on the voltage to CFM curve of the fan, and this curve is affected by the static pressure since that spec varies with speed. To achieve good thermal tracking requires designing around the curve slopes and the available CFM at each point on the curve. There are several non-linear variables involved here. Fortunately good first order approximations can work well if properly chosen.

Note that there may be different versions out there, including the combos that may use a different implementation since the chassis are quite different. I do not have docs on all versions.

 

[agedhorse]

that noctua fan can overcome way more than 50%. numbers on required pressure? still can't see maths beyond repetitive "curves" :/

I don't know, I haven't tested it.

 

[dORSY]

documented failures of the stock Rumble v3 fans

that: we don't know.
what I know that I've seen many whining posts (pun intended) about noisy fans there. and well, this one also

if fan speed was controlled by temperature sensors

we got an employee here that _could_ answer most of our questions, but refuses to or cannot.
edit: it does have temp control! we just got the info above!

components became hot to the touch.

does mean nothing, most components, like the class-D amp ICs can operate at high temps. it's in the specs.
Operating junction temperature range –40 150 °C
Overtemperature error 145 155 165 °C
and so on. (i know that for example the rumble100's chip shuts off at 154 C)

i myslef would definately not be on the opposite side of the Fender lawyers seeing what's up lately. so I understand.

 

[agedhorse]

does mean nothing, most components, like the class-D amp ICs can operate at high temps. it's in the specs.
Operating junction temperature range –40 150 °C
Overtemperature error 145 155 165 °C
and so on. (i know that for example the rumble100's chip shuts off at 154 C)

You are just stating a single specification on a single component. Junction temperature by itself doesn't matter because the heat must be able to exit the part, and de-rating factors are used to account for thermal resistance of the package/mounting system and the ambient temperature. The module designer doesn't care about junction temperature other than to insure that there is more than adequate thermal capacity after the de-rating factors are calculated.

There are many components that have temperature coefficients, and the design must account for all of these variables so that the amp remains stable throughout the operating temperature range. Since the timing control is so critical, thermal drift is a VERY important part of a reliable design.

NONE of the temperatures you stated above apply to the design of these modules because the international safety standards specify the maximum allowable temperature under normal, abnormal and fault conditions For example, while the transformers used in these modules have a class B insulating system (130 deg C), they are limited by the regulations to 105 deg C and on most modules there is temperature monitoring of these parts that feeds into the thermal shutdown management circuit. Depending on the module, there may be over a dozen different points that are monitored. This is exactly what I deal with when designing (reliable) products using these modules, and I have 20 years of commercial product experience around this family of products, I am not a hobbiest.

Additionally, most components simply can't operate anywhere near these temperatures, not if the amp is expected to last more than a few minutes.

 

[dORSY]

You are just stating a single specification on a single component. Junction temperature by itself doesn't matter because the heat must be able to exit the part, and de-rating factors are used to account for thermal resistance of the package/mounting system and the ambient temperature. The module designer doesn't care about junction temperature other than to insure that there is more than adequate thermal capacity after the de-rating factors are calculated.

There are many components that have temperature coefficients, and the design must account for all of these variables so that the amp remains stable throughout the operating temperature range. Since the timing control is so critical, thermal drift is a VERY important part of a reliable design.

NONE of the temperatures you stated above apply to the design of these modules because the international safety standards specify the maximum allowable temperature under normal, abnormal and fault conditions For example, while the transformers used in these modules have a class B insulating system (130 deg C), they are limited by the regulations to 105 deg C and on most modules there is temperature monitoring of these parts that feeds into the thermal shutdown management circuit. Depending on the module, there may be over a dozen different points that are monitored. This is exactly what I deal with when designing (reliable) products using these modules, and I have 20 years of commercial product experience around this family of products, I am not a hobbiest.

Additionally, most components simply can't operate anywhere near these temperatures, not if the amp is expected to last more than a few minutes.

and if You were following, my statement was a reply on "hot to the touch". which _does_ _not_ matter. You can burn Yourself on those parts while they are well operating in spec range. yes, even the filter caps that are mostly rated to 95C and above.

lets make an educated guess from an idiot like me!

i'm not saying that Fender uses some customized ICEpower amp+power boards, but let's just assume. /wink
let's assume they know what they're saying as a manufacturer.

Thermal Design
Thermal design is generally a great challenge in power amplifier systems. Linear amplifier designs operat-
ing in class A or AB are normally very inefficient and therefore equipped with extensive heat sinking to
keep the transistor junction temperature low. The ICEpower125ASX2 is based on highly efficient ICEpower
switching technology providing high overall efficiency characteristics at all levels of operation.
Part of the “component” philosophy of the ICEpower125ASX2 module is to provide a self-cooled compo-
nent thus eliminating the need for special attention to thermal design.
The ICEpower125ASX2 module is designed for music reproduction, which means that the output power of
the amplifier will never be continuous. Research has shown that the long-term RMS level of any music
signal does not normally exceed 1/8th of the maximum non clipped RMS value, rated output power and
the power supply is therefore designed for large short-term power handling and lower continuous power
handling. If the average output power of the ICEpower125ASX2 exceeds 65W @ 4 (SE-mode with both
channels driven) or 70W @ 4 (BTL-mode) for a long time at 25°C ambient temperature, the module will
reach its maximum allowable temperature and the temperature protection will be activated

their board designs can, and sometimes do operate passively cooled in closed boxes.
let's assume their spec is valid of the 80+ overall efficiency of their combined power supply+amp
let's assume other components are low power, like preamps/IO board/EQ and such (they are). this means everything else dissipates almost no/marginal heat if at all.
let's assume they tell the truth about avg. max nonclipped RMS. (1/8) and let's double it just for fun/safety/overkill!
let's do math: 500W/4 = 125 x 0.2 > we're still around a 25W heat dissipation range that needs to be pushed out (in my experience it's waaaay less, even the consumption from wall is mostly below that...)
ICE has off the shelf boards, they got temp protection and temp monitor outs, which goes 0V when temp is high.
let's make the educated guess that this correlates to Your statement above about the fan control: if it goes hot, fan goes wroom. when it's cool, fan goes "idle" at half max voltage ~6V. (also driven by power consumption, but... You get the point /wink on the next line
let's assume Your statement was correct above, and the fan rarely ramps up under normal (not test) conditions.

what does this say to me? that the stock fan can (and usually do) cool the unit near it's lowest setting, at 6Vs.
and this is what Never2late was experiencing: hot to the touch, fan is still at low RPMs.

what this all say to _me_: the fan is for blowing out excess heat from the box, so it does not stuck and overheat. it is _not_ against a heatsink/radiator or anything like that. (static pressure...)

yes, I know, we (and Fender) design products for idiot customers, like us, musucians making loud noises. so we overbuild (and cut as much cost as could be, cause corporate "overlords"), noone likes returned/failed products.

again, i'm just an idiot making educated guesses and assumptions based on similar products and public info here

 

[toughluck]

dORSY: This is very much not so. You're looking at the charts, but you're not reading them right.

Let's start with a realistic version of the chart:

Note 1: I'm using the maximum values for airflow and static pressure of Noctua NF-A6x25, but the values in between were NOT empirically measured, they are derived synthetically (I edited this sentence for clarity)
Note 2: I'm not going into discussions of acoustic noise and power consumption. This is way beyond the scope of what's discussed here, and although it's important, the Noctua fan would likely be quieter and draw less current.
The thin blue line is if the change was linear. In an axial fan, it's physically impossible to get to the right/above this curve.
Indeed, if it wasn't for physical limitations, the line would actually be a hyperbole, with infinite pressure possible at zero airflow (and infinite airflow at zero pressure).

The red line is what's realistically possible.
You need to follow it from the right to the left. Airflow falls more or less linearly, until there's suddenly a knee. That knee is the stall region of the fan. It's where the airflow separates from the fan blades due to pressure buildup on the exhaust side of the fan causing turbulence.

Indeed, if the fan had no stall, then maximum static pressure would be higher than what's indicated on the chart, the line would be steeper (like the yellow line). But then, it's likely that the fan would also have even higher airflow and higher static pressure, and would be even further to the right.

There are ways to mitigate stall and to make a more predictable fan, but they reduce maximum airflow without any other benefit (dark green line on the chart), so there's no point making a fan like that.

Still with me so far? Okay.
So here's another chart:

This one is for airflow vs. static pressure at a set of RPMs, going down from 3000 to 1000 and following the typical scaling rule (linear for airflow, quadratic for static pressure).
So far, so good... Now let's add another level of complexity to the chart:

This is the previous chart with the lines dashed and thinned and with application curves added.

This is where the confusion stems from: @dORSY, you were referring to the dashed lines, while @agedhorse was referring to the thick lines
When you engineer a cooling solution, you will look at the fan's behavior in a given application.
You will never have a perfectly sealed, nor a perfectly free air system.
Of course, airflow will change linearly. So triple RPM will give you triple airflow. But it might not be enough to cool the hot components inside.

And system impedance is the built-in function of the device. You have to work with it, and some fans might look worse on paper, but will actually achieve significantly higher airflow in a high impedance system thanks to higher static pressure.

Take those two Fractal Design fans I mentioned — the high flow fan can push a lot of air into a computer case, but if you put it on a radiator, it will struggle and it won't push nearly enough air to cool it.

 

[dORSY]

dORSY: This is very much not so. You're looking at the charts, but you're not reading them right.

Let's start with a realistic version of the chart:
View attachment 7499681
Note 1: I'm using the maximum values for airflow and static pressure of Noctua NF-A6x25, but these values are NOT empirically measured, they are only possible estimations.
Note 2: I'm not going into discussions of acoustic noise and power consumption. This is way beyond the scope of what's discussed here, and although it's important, the Noctua fan would likely be quieter and draw less current.
The thin blue line is if the change was linear. In an axial fan, it's physically impossible to get to the right/above this curve.
Indeed, if it wasn't for physical limitations, the line would actually be a hyperbole, with infinite pressure possible at zero airflow (and infinite airflow at zero pressure).

The red line is what's realistically possible.
You need to follow it from the right to the left. Airflow falls more or less linearly, until there's suddenly a knee. That knee is the stall region of the fan. It's where the airflow separates from the fan blades due to pressure buildup on the exhaust side of the fan causing turbulence.

Indeed, if the fan had no stall, then maximum static pressure would be higher than what's indicated on the chart, the line would be steeper (like the yellow line). But then, it's likely that the fan would also have even higher airflow and higher static pressure, and would be even further to the right.

There are ways to mitigate stall and to make a more predictable fan, but they reduce maximum airflow without any other benefit (dark green line on the chart), so there's no point making a fan like that.

Still with me so far? Okay.
So here's another chart:
View attachment 7499683
This one is for airflow vs. static pressure at a set of RPMs, going down from 3000 to 1000 and following the typical scaling rule (linear for airflow, quadratic for static pressure).
So far, so good... Now let's add another level of complexity to the chart:
View attachment 7499684
This is the previous chart with the lines dashed and thinned and with application curves added.

This is where the confusion stems from: @dORSY, you were referring to the dashed lines, while @agedhorse was referring to the thick lines
When you engineer a cooling solution, you will look at the fan's behavior in a given application.
You will never have a perfectly sealed, nor a perfectly free air system.
Of course, airflow will change linearly. So triple RPM will give you triple airflow. But it might not be enough to cool the hot components inside.

And system impedance is the built-in function of the device. You have to work with it, and some fans might look worse on paper, but will actually achieve significantly higher airflow in a high impedance system thanks to higher static pressure.

Take those two Fractal Design fans I mentioned — the high flow fan can push a lot of air into a computer case, but if you put it on a radiator, it will struggle and it won't push nearly enough air to cool it.

and we are pushing air into a -computer, oh wait!- amp case here. that's what i'm saying from the start...
we still dunno the stastic pressure requirement. and your graphs were made up to begin with. and now You're iterating on it! (which of the 6 lines applicable in an r500?) we could draw a 100 more... for fun.
what airflow do we need to push out the <30W of heat? (not much TBH as the stock fan also operates at it's lowest, 6V point most of the time - according to above info in this thread)
also, when You say "NOT empirically measured" seems to me you're suggesting that Noctua lies in their specs... okay!

 

[toughluck]

and we are pushing air into a -computer, oh wait!- amp case here.

It doesn't really matter for two reasons: One, the fan sucks air out of the amp. See those curves I drew above? Now consider that there are actually four curves: Fan mounted externally and blowing into a case, fan mounted internally and blowing into a case, fan mounted externally and drawing air out of a case, fan mounter internally and drawing air out of a case.
And there are four more curves: Fan mounted backwards and running backwards in each of these positions.

Two, it's always a closed system and will always have some impedance. Overall, if a fan is able to provide a specified amount of airflow through a system with its impedance at a specified temperature, it will be adequate for the application. If not, ten it won't.

we still dunno the stastic pressure requirement. and your graphs were made up to begin with. and now You're iterating on it!

The graphs were made up, but I expect them to be close to the real deal. And I just wanted to point out that we don't know how efficient the Noctua fan is. But if you took the chart you made up, and drew a straight line between the start and end point, it would still end up below and to the left of the Sunon fan curve.

also, when You say "NOT empirically measured" suggest me you're suggesting that Noctua lies in their specs... okay!

That wasn't my intention at all!!! I edited that sentence in my post. What I meant is that we only know the start and end points, the values in between are synthetic.

 

[dORSY]

It doesn't really matter for two reasons: One, the fan sucks air out of the amp. See those curves I drew above? Now consider that there are actually four curves: Fan mounted externally and blowing into a case, fan mounted internally and blowing into a case, fan mounted externally and drawing air out of a case, fan mounter internally and drawing air out of a case.
And there are four more curves: Fan mounted backwards and running backwards in each of these positions.

Two, it's always a closed system and will always have some impedance. Overall, if a fan is able to provide a specified amount of airflow through a system with its impedance at a specified temperature, it will be adequate for the application. If not, ten it won't.

The graphs were made up, but I expect them to be close to the real deal. And I just wanted to point out that we don't know how efficient the Noctua fan is. But if you took the chart you made up, and drew a straight line between the start and end point, it would still end up below and to the left of the Sunon fan curve.

That wasn't my intention at all!!! I edited that sentence in my post. What I meant is that we only know the start and end points, the values in between are synthetic.

again, we are talking about a box that needs to be ventillated. pushing out <30W of heat. can the noctua fan do that? ofc it can.

 

[toughluck]

@agedhorse, after re-reading your comments, I realized that we were talking about the same thing, but from two different perspectives. I was talking about the specific curve of each fan, before applying it to the system in question, while you were talking about the application curve for a given impedance.

@dORSY:
I don't know the system impedance of that Fender combo, but if the specific curve of a replacement candidate fan is below and to the left of the stock fan, it's likely not going to work, unless one or more of these apply:
1. The stock fan is drastically inefficient, and the curve drops very sharply. This isn't likely. Axial fans aren't exactly a new thing, they're a commodity and while the silent PC market led to the design of remarkably quiet high airflow/high pressure fans, they're still limited by physics and the fan would have to be broken to not roughly follow the typical fan curve. But there's really no reason for a commodity fan like that to exist (it would be a specialized fan used in compressors and the like).
2. Stock fan's static pressure scales with a higher order power of RPM than square. That would mean that the fan performs poorly at low speeds and needs to ramp up to speed to perform well. This isn't impossible and doesn't make for a bad fan.
3. Replacement fan is very efficient at lower RPM. Higher efficiency would bring it to par with the stock fan. This is synergistic with 2.

Designers would follow the requirements for each fan and would tune the cooling controller around the characteristics of each fan. It might be possible that they overengineered the solution and the stock fan is way overspecified for the purpose, and any replacement PC fan with roughly the same specs would have worked.

But here is where it gets hairy: We don't know that they did. If they did, the replacement would work. But if they didn't, the amp would overheat. Or it wouldn't, perhaps it would run fine in completely standard conditions.
And if they did overspecify, they assumed that at the very worst, there would be an option available to run the fan on higher speed, up to maximum, and it would achieve the desired cooling capacity. Like in hot weather, or at higher elevations, or when the amp is really pushed to its limit, or when the fan has aged.
The replacement fan might still be enough even in those extreme situations, but it's a gamble where you risk breaking the amp for small to no gain*.

Considering that the Rumble 500 doesn't have a temperature sensor and only relies on power output of the amp for variable speed fan control suggests that it never gets hot enough in normal operation to worry about it, but there's no way to be sure of that. But the lack of temperature control also suggests that it's dangerous to rely just on assumptions — I'd even say that it's way more critical to use the proper fan here than it would be on higher power amps that do have overtemperature protection.

*) PC fans are a peeve of mine. The whole craze for quieter fans stemmed from the fact that when active cooling became necessary in home PCs about 30 years ago, manufacturers just slapped on the cheapest available fans that they used in servers. The fans ran loud and people were annoyed.
That didn't mean that quiet fans were not available, audio industry was specifying and using them for years, but many were bespoke parts, others ran on atypical voltages (for PCs), and you couldn't exactly look it up and find the companies online or order components from them.

And this resulted in premium PC fans being way too expensive, unjustifiably so.

A lot of such markets exist. When I needed a bag to carry a digital mixer, I repurposed a photography bag that I no longer used, but I realized I need a slightly larger bag that I just didn't have lying about, and looked for one. I found out that photography bags cost around twice to as much as five times more than a DJ/lighting/mixer bag from the same brand. Two different categories in the store, roughly the same bag, at half the price or less.

 

[dORSY]

the lack of temperature control

read back. our tech said the opposite (and general, available, similar ICE boards indeed got temp sensor outputs, linked to both power supply side and amp side, which goes from 5V to 0V when the unit gets hot)

 

[toughluck]

Take a look at the diagrams from the wiki:
Rumble 800 Combo:

Rumble 800HD Head:

Rumble 200/500 (head/combo):

Note that the "TEMP/OVERCURRENT PROTECTION" block is very conspicuously missing from the last diagram.

I'm not saying there is no overtemperature protection built in. But it's not present in the diagram. At the same time, the temperature/overcurrent protection block in the Rumble 800 diagrams appears to only take the (electric) signal from the input side of the amplifier, which is absolutely viable, but there's no accounting for actual output current — which will differ significantly depending on cab impedance.

And on the yet another hand, connecting headphones would cut out the power amplifier. That means that regardless of the model, it would immediately stop the fan. If it doesn't stop immediately, then it's proof that there's more to it than just the direct control. And if anyone was so inclined, then putting the amp in a hot room, connecting headphones and letting it heat up regardless would prove that it takes temperature into account if the fan does start up.

 

[dORSY]

which will differ significantly depending on cab impedance.

ICE says efficiency goes UP at higher power (makes sense):

Total power efficiency %
Po = 100W 4 ohm 75
Po = 400W 4 ohm 80,4
Po = 500W 4 ohm 81,4
Po = 250W 8 ohm 86,4

and also up by a couple of %s 4ohm vs 8.
we are still at <30W typical heat dissipation. even doubling their documented average power consumption of 1/8 of rated max.

also:

RUMBLE 500 = ICEPower 125ASX2
I can see the thermal pin there. Hope it's not just for looks.

 

[agedhorse]

There is no thermal monitoring output but there is an over-temp shutdown function that provides an overtemp output signal when the amp enters over temp protection shutdown.

That said, these circuits are not perfect (though they are getting better) and are present to improve survivability in the event of a fan failure, or a passively cooled application being driven beyond its save duty cycle.

 

[agedhorse]

ICE says efficiency goes UP at higher power (makes sense):

Total power efficiency %
Po = 100W 4 ohm 75
Po = 400W 4 ohm 80,4
Po = 500W 4 ohm 81,4
Po = 250W 8 ohm 86,4

and also up by a couple of %s 4ohm vs 8.
we are still at <30W typical heat dissipation. even doubling their documented average power consumption of 1/8 of rated max.

also:
View attachment 7499726
RUMBLE 500 = ICEPower 125ASX2
I can see the thermal pin there. Hope it's not just for looks.

I hope you don’t think 1/8-power for bass guitar applications is acceptable… it’s not even close.

 

[dORSY]

I hope you don’t think 1/8-power for bass guitar applications is acceptable… it’s not even close.

have You measured average consumption from wall when playing (not testing with the ususal sinewave signals, but playing)? I did with an r100 cranked up to uncomfortably high levels some years ago. will not spoiler and let You do the same! curious how an r500 does in that regard!
(also note that I was generous and doubled it. for a reason.)