first box didn't explode

There's nothing un-professional about admitting that a panel is held on by screws. And making them easy to use.

I hate gadget with cleverly hidden fasteners.

Presumably this is a very reliable piece of kit? So I would't expect to be removing the front panel every 5 minutes. Therefore for the 99.99% of it's lifetime it is not being repaired I think I would prefer a nicer screwheadless front panel.

Can't be sure yet, but our stuff usually is.

Probably not every 5 minutes.

I don't think any of our customers care, or even notice, what screws we use. They care about what the box does.

The enclosure is a standard design from Protocase, laser machined per our Solidworks file. Done is better than perfect.

If the screws were on the inside, access to them would be horrible. I like this as-is.

Show us some equipment that you've designed.

How loud can a small chinese fan be? It doesn't look like a 6" 110 watt Nidec Servo fan. It's also possible the fan sucks too. NMB makes some of the quietest square fans if that's what you need. They really are different from the rest, but they don't seem to mention that in the spec sheets for some reason. You can also get "reverse" flow fans where the output side is not the side with the label and struts. Keeping the fan blades away from your grill may also cut down on racket. Nidec and NMB both offer these.

I think you mentioned each amp had either a heatsink or fan too. Small fans run faster and tend to be louder too. Maybe your noise problem isn't with the big fan.

I work in a different field, but I don't know of any places cool air is found on a flat surface, inside a rack, just to the right of the mounting ears.

Sencore, the former maker of bench test equipment is still in business making computer based video something devices. They had rackmount computer type device at trade show a few years ago with the cover off. It had a surprisingly weird air flow pattern like a fan in the power supply as exhaust and another in the case doing the same thing but no air intake, or the cool air intake was on the back- whatever it was, I had to go take a look at it for a while and chat with the guys at the booth about it.

It's a small 60 CFM fan. I guess it's spinning pretty hard.

They really are

That was our observation: the laser-cut grille makes the fan whine.

Both, on the bottom of the board. They run cool at 120 watts out each.

Small

It is. We unplug the big fan when we're working on the software and FPGA. The little ones are quiet.

You should get out more.

this guy has to be messing me me, not you.

cover the entire front panel with those silkscreened adhesive labels like they use on membrane swithes. Got to make sure fixing it requires peeling it apart or poking holes to fix it.

So, record the noise and FFT it. Compare to the fan blade spin rate and the grille periodicity.

For white-ish noise, you want the grille to be spiral or aperiodic. A whine means there's a harmonic being excited. Probably.

Some noise is good (the presence of turbulence next to the hot bits gives a cooling benefit).

And make sure the replacement adhesive label costs several times more that the switch underneath it you are replacing (Broan range hood).

Not really. In the expensive server world, cooling has been solved within limitiations of the density of air and how much you can even phyiscally pull though a piece of machinery. dissipating well over 500 watts in a rack mount case 1U high isn't sorcery anymore. I can tell you it doesn't involve three little fans off in the corner doing their own thing and one big fan mounted on the side of the chassis. Throw some plexiglass on the top of your case and light a cigar by your air intake to see what's going on. They even sell smoke in a spray can for testing optical smoke detectors, never used the stuff myself though.

even older business class desktops from IBM, HP and even Dell really works of art when it comes to simple, quiet cooling. The lucent guy going on about tuned baffles or whatever and air impedance isn't joking.

There's a cheapest at home depot type broan stove hood here. Do they actually hide stuff under that label? I see the light and fan speed switch mount through it. The fan is weak as hell, can barely run on the low speed, so I suspect it's time for cleaning and fresh oil.

Tobacco is disgusting. We use incense sticks.

My customers fill a bay of racks with expensive, high-power stuff. They usually have a modular floor that's pressurized with cold air. The botttom of a rack is open for cable access and cold air. It's cool inside the rack, along the sides, where we intake our air. The sides are actually the best place to get cool air. And a lot quieter than front-panel intake.

We do have thermistors on the heat sinks of the three class-D amps. The TI chips have thermal shutdown, so protect themselves, so the real reason to have the thermistors is to know that the small fans are working properly during factory test. Chips like the TI amps really should report their die temperature. Most FPGAs do nowadays.

It's a beautiful piece of gear. There are 7 boards inside, three different designs. The controller board has a uP, an FPGA, DACs, ADCs, relay drivers, Ethernet, USB, lotsa switching regs from +48, all sorts of stuff. I was hoping it would work first time, exactly as designed, without any ECOs, but people are complaining that the POWER green LED is too bright, so I may have to change two resistors. The light pipes are really good. Bummer.

It looks like everything else works first try. That's what good design reviews can accomplish. My guys and girls are eager to catch me making dumb mistakes. I've trained them to do that.

I rented a competitor's instrument to see what it looked like inside. It arrived from the rental outfit with maybe a dozen "do not remove or calibration void" stickers inside and out. Some extensive research on Amazon found the identical sticker, so I bought a reel.

The insides were pretty awful.

On the version I have the switch indicators show through the label. The screws that mount the circuit board are under the label. To get the board out to replace the switches you have to take the label off. The first time I destroyed the label. On the replacement I left about half the adhesive backing on. It holds OK and hopefully I can get the label off without destroying it. I may know soon - after several years the switches are starting to fail again.

I haven't had any trouble with the fan on mine.

I don't smoke but prefer tobacco over incense.

That's 15+ years behind the times in datacenter cooling, but you're also not really dealing with high power densities either.

PWM the indicators, sell a software unlock or license code to enabled this extra feature.

I came across a weird kit in a pile of junk that consisted of a #2 philips driver with the equipment supplier's name on it. Totaly unnecessary cost right there. The bag also had a sheet of new round warranty void stickers. Never did figure our the story behind that. Maybe a service tech left it by accident, not really sure.

The used regular BNC connectors, soldered to pads on the board, with no vias. Naturally, a little shock broke the pads. Hence the wires.

The melf resistors are strange too.

Cute. They're just "aged" resistors and therefor very stable. What's behind the 6-32 looking screws that nearly clamps on the edges of the chip resistors?

My favorite are dodgy 0.250" male faston terminals soldered onto boards that somehow weren't soldered right at first and then cracked from age or being manhandled. I'm just guessing 50% of furnace failures are caused by this.

seeing as those are ground nodes (4x 200R in parallel makes 50R) probably a metal stand-off attached to the case,

I didn't even consider the fan and light switches might mount on a PCB for these. I just assumed the switches have those wings with little stepped ridges and a mess of wiring behind them.

Is there a good way to get those sticker things off?