Anyone else's Neumann KH80 randomly reset themselves?

[Hugh Robjohns]

[Drew Stephenson]

Interestingly* one of mine went into standby mid-play last week. Hasn't happened again but nervously keeping an eye on it now.

* Probably only for this audience.

[Hugh Robjohns]

Hmmm.... this auto power feature does seem to be a weak point, doesn't it? I can excuse, to a degree, false wake-ups caused by electrical noise on the line. But false shut-downs when in use is much more troubling...

[Drew Stephenson]

Only happened once in 18 months ish (I can't actually remember where I got these) so not overly concerned at the moment. I can always turn the auto-standby off - they're not exactly a heavy power drain.

[Kayvon]

I've basically written off my KH80s now. It's really difficult to pin down intermittent faults like these. To buy the speakers again from Neumann at a reduced price is not something I'm particularly keen on doing having coughed up already.

It's funny, at the time I was keen on investing in a set of speakers that would last for a good long period of time. The Neumanns didn't make it much past 2 years.

I do appreciate that things can go wrong over time and typically I'm not one to buy new technology when it's first been released but since it was Neumann I forewent that prudence. Ah well.

[Kayvon]

Perhaps my previous post is a bit toys out the pram. They still work as monitor speakers without the auto-standby mode but the other annoying thing for me was that they never implemented the negligible latency mode.

Since my cumulative latency (Ableton Live I'm looking at you) was causing me recording issues (still does tbh) I was intent on cutting out any latency wherever possible. The extra monitoring latency on top of everything I could've done without.

Sometimes I think music technology is less of a hobby and more of a battle

[Hugh Robjohns]

Any speaker relying on dsp correction will introduce latency, and in general any 'reduced latency mode will have to compromise the filtering precision, especially at the low end... so its inevitably introducing a different compromise and potentially degrading performance. Perhaps that's why it hasnt been implemented.

If latency is a problem analogue speakers are the solution (ideally with a high sample rate source)

[Kayvon]

Any speaker relying on dsp correction will introduce latency, and in general any 'reduced latency mode will have to compromise the filtering precision, especially at the low end... so its inevitably introducing a different compromise and potentially degrading performance. Perhaps that's why it hasnt been implemented.

I read they managed to get the latency low enough that it became in their take not worth implementing it.

I see on the SOS news page there's some new HEDD monitors that have a switch on the back to toggle on/off their linear phase mode. Pretty neat though maybe I'll wait for the mk3 versions with all that on a remote control

[adambier]

Here's an update for those still having trouble with their KH 80 DSPs when set to go into standby through the Neumann.Control iPad app:

(1) I spoke with the speakers product portfolio manager at Neumann Berlin. False positives (waking in the absence of a signal) are always going to be a bit of an issue on the KH 80 DSPs, since they don't have a third ground prong, which means they're more subject to line noise. Early U.S. units can be retrofitted with a mod (just a single resistor swap; more recent units already have the new part) to make them more resistant to this issue, but the problem will likely still persist. Increasing the standby threshold to something high (max is 30 dB SPL @ 1m) will help, and/or connecting them to your source via a KH 750 sub, which has a third ground pin, may significantly reduce false positives. Apparently, the standby functionality is only used by a relatively small portion of the user-base; most people either leave the monitors on all the time or turn them off and on manually with their racks of other gear. Also note that the iPad app will allow you to set very low thresholds that, as a matter of physics, will be impossible for the monitors to distinguish from line noise. So be sure to pick a reasonably high threshold.

(2) A bunch of problems seem to be caused by bugs in the Neumann.Control app and associated network traffic. If you're experiencing weird issues like random power-cycling or false negatives (i.e., the monitors going into standby notwithstanding the presence of a strong signal), you should factory-reset your monitors and then reconfigure them.

(3) To remove other network traffic as a possible cause of problems, I recommend buying a $50 wifi router with at least 2 ethernet jacks on the back (or use an old one you have lying around). Do NOT connect the router to the internet or the rest of your network and just connect your two monitors to the router via ethernet and connect your iPad to this isolated network via wifi.

(4) Next, delete the applicable System configuration from the Neumann.Control app. This will also delete all alignments associated with that System. It is critical that you then force-quit the app (by double-clicking the home button on your iPad and then swiping the app up and away). Through experimentation, I've discovered that if you don't quit the app before resetting the monitors, even if you've deleted a System configuration, the app sometimes keeps the configuration data cached in memory and they may get pushed back out to the monitors upon their reset, which will mean you'll have to go back to square one.

(5) After you have deleted the System configuration and force-quit the iPad app, you can then reset your monitors. Note that the directions in the manual for doing a factory reset are confusing: they say to toggle the SETTINGS switch up and down until the logo "flutters" red, but it's more like a slow flash. And, most importantly, before that flashing occurs, the logo will turn white while you are in the middle of this process, which might make you think the reset has failed. In actuality, however, you just need to continue toggling the switch through the period of the logo being white, and eventually the logo will flash red to indicate a successful reset.

(6) You can now use the Neumann.Control iPad app on this isolated network, connected only to the monitors, to set up a new System and apply any Alignment you wish. When that's done, disconnect the ethernet cables from the monitors, and they should work as intended without any weird behavior other than occasionally waking from sleep due to line noise.

Thus far, they have worked perfectly for over a week. I've ordered the new MA-1 alignment microphone, and plan to use that to re-align my monitors. Because the desktop app for the MA-1 lacks a number of features found in the Neumann.Control iPad app (including standby settings), I will take a screenshot of the resulting EQ settings and then recreate them in a manual alignment on my iPad.

[adambier]

Also I'm still debating whether to get the KH 750 DSP, which apparently people are raving about. A big concern is that, while the manual suggests that you should use the sub's digital (BNC) outputs when connecting to a DAW interface, there's no good way to control the monitor volume in real-time (and digital attenuation results in a loss of bit-depth and possible dithering artifacts), but running your signal through the analog inputs means a D>A>D>A>D>A signal path when used with the KH 80s...and that is a LOT of conversion. Any thoughts on this would be welcome.

[zenguitar]

Thanks for a very helpful post

Andy

[Drew Stephenson]

Also I'm still debating whether to get the KH 750 DSP, which apparently people are raving about. A big concern is that, while the manual suggests that you should use the sub's digital (BNC) outputs when connecting to a DAW interface, there's no good way to control the monitor volume in real-time (and digital attenuation results in a loss of bit-depth and possible dithering artifacts), but running your signal through the analog inputs means a D>A>D>A>D>A signal path when used with the KH 80s...and that is a LOT of conversion. Any thoughts on this would be welcome.

Given the accuracy of conversion nowadays I wouldn't be worrying about half a dozen conversions.

[Hugh Robjohns]

... digital attenuation results in a loss of bit-depth and possible dithering artifacts, but running your signal through the analog inputs means a D>A>D>A>D>A signal path when used with the KH 80s...and that is a LOT of conversion. Any thoughts on this would be welcome.

Back in the late 90s I ran a signal through 50 D>A>D passes using Sony pro DAT machines (44.1/16 bit) and then asked countless BBC soundies to identify when they were aware of degradation. A few felt they could start to hear artefacts after ten generations, but most struggled to hear any difference until closer to 20 passes. And that was over twenty years ago... Converters are even better these days.

So I wouldn't be too worried about quality going through multiple converters.. The latency might be more of an issue, depending on how you're using the monitors.

But actually, digital attention isn't a problem. If you have dithering artefacts then the attenuator coding is broken. And a 'loss of bit-depth' is a total red herring. The wordlength defines the signal-to-noise ratio. Nothing more. Attenuating the signal level reduces the signal-to-noise ratio... Exactly as it does in an analogue system. And with 24-bit converters the digital system noise floor will be no worse than n blogger system's noise floor.

So again, it's not something I'd lose any sleep over.

Thanks for the suggestions about resetting the KH80s. I'm sure many owners will find that useful.

[adambier]

But actually, digital attention isn't a problem. If you have dithering artefacts then the attenuator coding is broken. And a 'loss of bit-depth' is a total red herring. The wordlength defines the signal-to-noise ratio. Nothing more. Attenuating the signal level reduces the signal-to-noise ratio... Exactly as it does in an analogue system. And with 24-bit converters the digital system noise floor will be no worse than n blogger system's noise floor.

Thanks, that's very helpful, Hugh. I wondered whether that was the case (since you're effectively only discarding unused bits), but I've seen so many different folks on various forums with all kinds of theories about this I wasn't sure.

The bigger issue with connecting monitors via a digital connection is that most interfaces just pass the bitstream through, without applying any attenuation via the control knob, but in a multi-purpose setup, the convenience (and safety, as a panic control, if there's an unexpected loud signal from the computer) is hard to give up. I'm curious if others have come up with workarounds.

[Mike Shand]

1) I spoke with the speakers product portfolio manager at Neumann Berlin. False positives (waking in the absence of a signal) are always going to be a bit of an issue on the KH 80 DSPs, since they don't have a third ground prong, which means they're more subject to line noise

Thank you for a very informative and useful post. I struggled with these problems and tried everything before finally giving up and switching them on and off manually.

[air2k]

Increasing the standby threshold to something high (max is 30 dB SPL @ 1m) will help,

Is this really true? My observation/tests showed no influcence of the standby level and the frequency of sporadic wakeups.
Additionaly, I could not find an hints in the manual/faq that the standby level is indeed connected to the wakeup level.
My guess is that the standby trigger is done in software (therefore configurable), while the wakeup-trigger is fixed in analogue circuits (therefore not configurable), as the manual mentions that during standby, almost all subsystems are unpowered.

To remove other network traffic as a possible cause of problems, I recommend buying a $50 wifi router with at least 2 ethernet jacks on the back (or use an old one you have lying around). Do NOT connect the router to the internet or the rest of your network and just connect your two monitors to the router via ethernet and connect your iPad to this isolated network via wifi.

Software engineer here. This statement is just wrong. Don't invest any time/money to do 'network isolation'.
I reverse engineered the network protocol (OSC based, BTW) between the neumann.control app and the KH 80 and random interference from other devices on the same network is impossible as long as they don't expecitly target for the IP addresses of the KH 80 and know how to speak with them.
BTW, I managed to control every parameter from my desktop machine without using the app as a proof of concept. If my free time would allow it, I would try to publish a free tool which can replace the ipad app.

[Hugh Robjohns]

You would certainly win a lot of friends if you came up with a usable mac/win desktop app to control kH80s!

[adambier]

Grrr, I'm sorry to report that after 1-2 weeks of problem-free operation (having gone through the protracted process described in my earlier post), a couple days ago I had two apparently random power cycles (which look to be standby while a strong signal is present, followed by a pause, and then the monitors waking up again--OTOH, it could be an unrelated reboot, but no way to be sure).

The first such incident involved only one of the monitors, but the second involved both, about 8 seconds apart.

I'm 95% sure this only happens when auto-standby is enabled (and do not have enough data to say whether this only affects things when both auto-standby and network-control are enabled, or whether auto-standby plus local-control also results in this issue, though anecdotally I suspect it's the former).

Without some kind of logging functionality, I'm at a loss for how to diagnose things further (apart from giving up and manually powering them off and on, or just leaving them on all the time).

The only other thing the folks from Sennheiser mentioned was potential mains power issues--I'm not totally clear one what exactly those would be, but I don't really want to buy a bulky online UPS just to ensure perfect sine waves and exactly 120V as an experiment.

Increasing the standby threshold to something high (max is 30 dB SPL @ 1m) will help,

My guess is that the standby trigger is done in software (therefore configurable), while the wakeup-trigger is fixed in analogue circuits (therefore not configurable), as the manual mentions that during standby, almost all subsystems are unpowered.

That's certainly possible, though a weird design decision given that the DSP and other DC circuitry probably use a tiny amount of power compared to the amp(s).

To remove other network traffic as a possible cause of problems, I recommend buying a $50 wifi router with at least 2 ethernet jacks on the back (or use an old one you have lying around). Do NOT connect the router to the internet or the rest of your network and just connect your two monitors to the router via ethernet and connect your iPad to this isolated network via wifi.

Software engineer here. This statement is just wrong. Don't invest any time/money to do 'network isolation'.

I reverse engineered the network protocol (OSC based, BTW) between the neumann.control app and the KH 80 and random interference from other devices on the same network is impossible as long as they don't expecitly target for the IP addresses of the KH 80 and know how to speak with them.

Definitely not a software engineer here, but, dunno, multicast issues?

Apart from comments from the Sennheiser rep that one should eliminate network issues as part of troubleshooting, my main concern was eliminating errant commands from the Neumann.Control app. Isolating the network doesn't technically affect that, but it does allow me to switch the iPad to the normal house WLAN immediately after pushing out the configuration, limiting the exposure during the period between that and physically disconnecting the monitors from the isolated network.

BTW, I managed to control every parameter from my desktop machine without using the app as a proof of concept. If my free time would allow it, I would try to publish a free tool which can replace the ipad app.

That would certainly be great. It's disappointing that the desktop software for use with the new MA-1 doesn't support System-level configuration, including standby settings or gain control. Neumann is obviously not primarily a software company, but this seems half-baked.

I really, really want to be happy with these monitors. I think they sound great. This is arguably a silly issue, but it's frustrating.

Any other thoughts (or potential solutions) are certainly most welcome.

[air2k]

The only other thing the folks from Sennheiser mentioned was potential mains power issues--I'm not totally clear one what exactly those would be, but I don't really want to buy a bulky online UPS just to ensure perfect sine waves and exactly 120V as an experiment.

I invested hours of testing after setting standby time to 1 minute:

At first, I measured my balanced outputs to the KH 80 for signal spikes (by using an audio interface connected to my high quality USB audio interface outs) but I was not able to measure them (looking at peak SNR) therefore I would rather rule this out as a cause.

1) KH 80 not connected to any signal source (leaving the XLR cables unconnected).
Ethernet cables are connected.
-> no sporadic wakeups (tested for 30 minutes)

2) KH 80 connected to an USB-powered Audio interface (without any signal, setting output level knob set to zero) with ground lift enabled on the interface.

a) Ethernet cables are connected.
-> sporadic wakeups on rather both KH 80 (after ~2 minutes), but they do not go into sleep reliably

b) Ethernet cables are disconnected
-> significant humming audible

3) KH 80 connected to an USB-powered Audio interface (without any signal, setting output level knob set to zero) with ground lift disabled on the interface

a) Ethernet cables are connected
-> sporadic, more random wakeups (first wakeup after ~5 minutes),
happens much more likely on one of the two KH 80 (even if I switch the audio channels), but usually sleeps again after 1 minute.

b) Ethernet cables are disconnected.
-> sporadic, less likely (compared to 2a) random wakeups (first wakeup after ~20 minutes),
happens much more likely on one of the two KH 80 (even if I switch the audio channels), but usually sleeps again after 1 minute.

4) KH 80 connected to via an active, battery-powered DI box (ground-lift disabled). The Di box is connected to an USB-powered Audio interface (without any signal, output level knob set to zero)
-> sporadic wakeups on rather both KH 80 (after ~3 minutes)
This also happens if I set a 20db or 40db attenuation at the DI box.
This is another hint that the problem has nothing to do with the signal lines (e.g. spikes, SNR etc.)

5) KH 80 connected to via an active, battery-powered DI box (ground-lift enabled). The Di box is connected to an USB-powered Audio interface (without any signal, output level knob set to zero)
-> extremely loud ground-loop-like-humming-noise (not present when ethernet cable IS connected). Hopefully the oberload protection at KH 80 worked. I'm sure my neighbors just woke up
Not sure if this originates from my rather cheap DI box or from an issue within the KH 80.

Taking the results of above tests and obvervations into account, especially the influence of a connected ethernet cable (connecting an ethernet cable will also likely cause the KH 80 to wake up!) I suspect grounding (earthing) issues with the KH 80's wakeup/standby functionality as the most likely cause (eventually because they only have a 2-pins main connection, so the grounding reference is missing).

This would also explain the randomness as well as the strong dependency on the actual setup/electrical environment.

To everyone affected: Remove your ethernet cables (if present) and see if this reduces unwanted wakeup/standup probabillity.

Definitely not a software engineer here, but, dunno, multicast issues?

Multicast is only used for device discovery (via mDNS). Actual communication to the KH 80 happens via regular (unicast) TCP connections.

[air2k]

I did further testing. Standby time still set to 1 minutes.

This time, I connected both KH 80 to an analog mixing console with the master fader set to the lowest output level - which means that it doesn't output anything at all)

Beside the mixer's 3-pin main power connection and the balanced connection to the KH 80, no other signal has been plugged in. Ethernet cables are not connected to the KH 80.

The KH 80 wake up in this set up as well. A little bit less often (~15 minutes) compared to my USB audio interface. And still conspicuous and similar to my previous tests: One of the KH 80 woke up much more often than the other (also when I swapped inputs).

This brings me to the conclusion:
* the KH 80 most likely have a serious engineering flaw regarding standby/wakeup functionality.
* there is a significant manufactoring variance which has an influence of the occurence probabilty (my 2nd KH 80 showed the issue 10 times less).

[Drew Stephenson]

Appropos of nothing one of mine just switched itself off mid-playback this arvo. And didn't come back on automatically, had to go round the back and switch off and on.

[Mike Shand]

Is this really true? My observation/tests showed no influcence of the standby level and the frequency of sporadic wakeups.

I certainly found the standby level had no apparent influence on frequency of wake ups.

[air2k]

Appropos of nothing one of mine just switched itself off mid-playback this arvo. And didn't come back on automatically, had to go round the back and switch off and on.

Although Auto-Standby is enabled? That's strange.

However, I turn Auto-Standby off, then I have the same behavior - but then the KH80 are always-on which I'd like to avoid (but didn't reboot/poweroff so far).

[Drew Stephenson]

Appropos of nothing one of mine just switched itself off mid-playback this arvo. And didn't come back on automatically, had to go round the back and switch off and on.

Although Auto-Standby is enabled? That's strange.

Yes, first time it's failed to re-awaken, only the second time it's gone to sleep mid play.
If it becomes more regular i'll just switch off the auto-standby.