Posts by Jandalf

Quote from heldal

When the KPA is based on Ethernet/IP and there are PoE-enabled chip-sets readily available I struggle to even understand why they would deviate from the standard. What were they thinking?

I haven't seen any integrated one-chip solution for a PoE powered Ethernet interface yet. You still need a PHY (physical layer chip) and a power acquiring circuit and a (preferably isolated) step-down converter to bring the PoE voltage to a lower level as required by your device. There are integrated chips containing the power circuit and the converter and the Remote uses such a chip. Basically, everything required for an implementation supporting both modes on the Remote would have been already there.
However the Remote was released way later than the Profiler, where they had already made that bad decision to go for a propietary and spec violating solution. So they probably took the easy way, which was to just continue with that bad decision when designing the Remote instead of using the opportunity to get this right. As to what they were thinking, i have no idea and i'm not sure if we'll get an official comment on that.

Quote from Ruefus

To my knowledge, there is no requirement to support mode a and b.

Then i suggest that you check with ISO/IEC/IEEE 8802-3:2021(E):

(PD = Powered Device, which applies to the Remote)
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

33.3.1 PD PI

The PD shall be capable of accepting power on either of two sets of PI conductors. The two conductor sets are named Mode A and Mode B
[...]
The PD shall be implemented to be insensitive to the polarity of the power supply and shall be able to operate per the PD Mode A column and the PD Mode B column in Table 33–13.

NOTE—PDs that implement only Mode A or Mode B are specifically not allowed by this standard. PDs that simultaneously require power from both Mode A and Mode B are specifically not allowed by this standard
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

I would say that "specifically not allowed" leaves little room for interpretation (my teenage son might disagree though)

It's unfortunately slightly more complicated since the PoE part of Profiler and Remote do not fully comply with the corresponding IEEE specs.

The Profiler manual states that PoE switches compliant with 802.3af-2003 and 802.3at-2009 and using Mode A are supported.

That means, any PoE or PoE+ switch should work, as long as it uses "Mode A" for powering the connected devices, since the Remote only supports "Mode A" (that is one point where the Remote violates the spec, which requires powered devices to support both, "Mode A" and "Mode B").

Hooking up the Remote to a switch that uses "Mode B" will not work and even includes a slight risk of damage (here is the second point where the specs are violated. Though the Remote supports "Mode A" only, it still (ab)uses the signal pins specified for powering according to "Mode B" for powering the Remote when its connected directly to a Profiler. They could have just done that by implementing a spec compliant "Mode B" powering in Profiler and Remote or a spec compliant "Mode A" in the Profiler (then there would be no need for forum threads like this) but unfortunately they didn't. Instead, the Profiler supplies a voltage of 4V, expected by the Remote in a certain polarity to power a directly connected Remote, which also severely limits the possible cable length between Profiler and Remote).

So the simplified rules are: Use any PoE or PoE+ switch powering in "Mode A" only and don't connect the Remote or Profiler to a switch or injector using "Mode B". PoE++ switches (can) use all signal pairs for powering and should hence not be connected either.

BTW that is, why following specs is (or would be) such a great idea: You could just interconnect any Ethernet equipment without having to care about such details. There would not only be no risk of damage, it would simply work, literally, plug and play.......

As you said: That was back in the early 90s, which was at least 30 years ago now! Times have changed a lot since then and so have technical demands, common design practice, etc.
Of course, a poor four layer design can be worse than an optimized two layer design, but when comparing technologies you sensibly have to assume that both technologies are applied in an equally proper way. Otherwise you might as well compare a Formula One car to a road vehicle by putting a professional race driver on the road vehicle and an absolute driving beginner on the Formular One car and let them compete at a race track and when the race driver expectedly wins, you could say "look, a road vehicle is just as good as a race car for winning a car race".

You can easily mess it up with even a 10 layer design and more layers of course don't guarantee a good result, but they provide more possibilities, and as soon as you have to use fine pitch / high pin count BGA components in your design, you simply can't use a two layer board anymore, as you just can't fan out of your component with only two layers (especially when you don't even have a power and ground layer you can directly connect to).

When it comes to a clean power supply, a well designed four layer board with a dedicated power system will definitely have a better performance than whatever you can get out of a board that has to share it's copper between supply and signals (allowing only discontinuous power islands that necessarily have a quite big distance (~1.5mm) between both layers and hence an inherantly higher impedance). Even the most sophisticated layout can't eliminate physical laws.

If a certain application really needs that better performance, is of course a different question, but it definitely exists and the probability that the design will require that extra performance has constantly raised since the 90s. Since then, we've seen several shrinks in semiconductor process technology (we were at 800nm or 600nm process in the early 90s, now we are at 5nm or even 3nm), resulting in faster signal rise times and raising demands on signal and power integrity.

Anyway, obviously none of us will be able to convince the other here, so i suggest we close the discussion about PCB basics (which is most likely not very interesting for the rest of the community) and just agree to disagree.

I had just expected more than a two layer design in the Remote and from what i have seen of the layout, i'm sure the developer didn't make as much effort as you probably had to, when doing your layout in the 90s in order to get a good performance out of your design, so they had probably been better off with the four layer board. For a PCB with the size of the Remote mainboard I would roughly estimate the difference in cost between a two and a four layer PCB at less than 10EUR.

Quote from lbieber

I'll say it again. A well designed 2 layer board can be every bit as good as a 4 layer board. FULL STOP.

O.K., i have to give up. You got the killer argument, called full stop (or period) and that can't even be beaten by knowledge .

Quote from lbieber

IA well designed 2 layer board can be every bit as good as a 4 layer board. FULL STOP.

'm sorry but i have to completely disagree with that and i would like to provide some technical arguments as this makes a discussion a bit easier compared to only using punctuation marks like a full stop .

Of course there are still applications where a double sided PCB can do the job, but when it comes to digital designs, which nowadays have usually to be considered high speed designs either, two layers can never be as good as four (which is an abolute minimum and only sufficient for more simple designs).
One of the reasons for that is the fact, that the supply current drawn by clocked components in digital designs can no longer be considered mainly DC currents. While there is still a certain DC amount, a substantial part of the power consumption is in fact AC, as the components are quite dynamic and are controlled by (constantly rising) system clocks, resulting in corresponding current pulses. Depending on the frequency and rise time of these pulses, even small inductivities can result in substantial voltage drops along the way from the power source to the silicon die inside the chip package and these voltage drops can impact the stability of the circuit and result in severe electromagnetic emissions (once we leave the DC world, we are no longer dealing with only ohmic resistances but with impedances).

The only way to properly address such power demands is to provide a proper PCB power system with a sufficiently low impedance, that consists of a dedicated power plane and an adjacent ground plane, stabilized by appropriate capacitor groups on stratetic positions. The distance between the two planes should be as small as possible (50um is good, 100um or 200um still sufficient in many cases). Since you still need at least one further layer to place and connect your components and route your interconnects, you end up with a minimum of four layers here.

But we're not done yet. Besides the power demands you also have digital high speed signals in your system that travel along their signal traces on the board. Unfortunately the signals do not only need a copper trace to travel but they also need a proximate path for their return currents, usually provided by a ground layer below (or above) the layer the signal trace runs on. Now with two layers only you are not only unable to provide a proper power system, you also can't provide a continuous plane for the return currents and at each gap, whole or whatever discontinuity you have on the return path your return currents will cause emissions and you will face an impact on the signal quality.

As i said before, you may still get a basically functional system with only a doubled sided PCB (like the Remote), but that strongly depends on the application and when it comes to EMC (electromagnetiic compatibility = radiated emissions and susceptibility to Burst/Surge/ESD), the two layer design will never have the same performance as a four layer design.

Of course the CPU of the Remote does not exactly have a hard job. Driving a small monochrome display and a few LEDs, checking a few switches and sending and receiving a few Ethernet frames once in a while doesn't require the computing power of a Gaming PC and so the Cortex-M3 CPU picked for the Remote has a comparably moderate power consumption. However it still runs on a system clock of up to 100MHz so i wouldn't consider the Remote as a proper candidate for a two layer design (don't forget that we are not looking at 19,99.- consumer electronics, being produced in real large quantities, where even a few cents less in production cost make a huge difference. There is no point in risking EMI and signal integrity problems on a 420.- professional device just to save a few bucks).

Quote from Ruefus

I can't imagine why only one person has replied in just over a week.

Well, i suppose that most of the Remote owners here did never take apart their Remote to look what's inside (neither would i have, if it didn't fail or if i was offered a repair at a reasonable price) and an even smaller group of those who actually did, may have the background to comment on the technical aspects of this thread, so i actually did not expect a lot of comments from the community referring to these aspects.

As to why there is no "official" repsonse from the Kemper side yet, we can only guess. Maybe the guys who are responsible for design decisions are not on this forum, maybe they just don't care or maybe they don't have anything to say....

However besides complaining about the design of the Remote and Kempers repair policy, my intention was to share information on the "repair" cost of the Remote that i would have loved to get from the Support BEFORE sending my equipment to the service. I would then have saved at least an hour of carefully getting my Profiler and Remote ready for safe shipping and dropping it off at DHL, i would have saved the week i couldn't use my Profiler and i would have saved the ~18.- shipping cost.

Quote

I'm getting the feeling you blew your remote by plugging into Mode B?

Nope, i didn't use a PoE switch at all before the connection problem showed up.
However, connecting the Remote to a PSE (power supplying equipment) that supports Mode B only still shouldn't fry the Remote as they at least implemented some (very basic) protection to the Remote by connecting a 5V Z-Diode between the abused signal pairs, which will work as a crowbar when power is applied at reversed polarity (which is allowed by the spec) and limit the voltage to arround 5.1V in standard polarity. Further, a properly designed PoE Mode B PSE would not apply power to the Remote since the detection and handshake phase would fail.
I'm not sure if that also applies to any few-dollars-PoE-injectors you can find on the Chinese market (at least i wouldn't bet), so to be safe, you shouldn't connect the Remote to Mode B PSEs (it won't work anyway).
BUT: This is something, an end user shoudn't have to care about. The worst thing he should have to expect is that it doesn't work. The (even remote) possibility of damage is imho unacceptable.

To be clear: I'm not some Know-it-all that has too much time on is hands and hence inspects musical equipment to then engage in bitching about the design.
If the Remote didn't stop working or if Kemper would offer repairs at a reasonable price i wouldn't know about all the design flaws of the Remote.
But when i decided to repair the device by myself, i had to do some reverse engineering and it seems, the deeper i dig, the worse it gets.

Besides that they violated the IEEE specs by not supporting Mode B but abusing the Mode B signal pairs for some proprietary (and bad) way of powering, there are many more signs of bad design, like an unsuitable input capacitor (capacity and voltage rating too low) on the PoE Mode A power supply circuit, missing ESD protection on the chip side Ethernet signal lines and not using impedance controlled differential signal lines for the Ethernet signal pairs on the PCB design. They didn't even decide to use proper PCB technology by at least picking a 4-Layer board, which is mandatory for providing a clean and stable power system on the board and which is standard for decades already but they actually put a cheap 2 Layer PCB in a 420.- digital device .
All in all, the Remote looks to me more like the result of an internship or a bachelor thesis than a professional design made by the Kemper developers.

I think the question, if that suits a company like Kemper or not is worth a discussion. Any other opinions on this topic here?

And as for economic feasibility of board level repair:
I was able to repair my Remote within a few minutes. After resoldering the LAN-transformer (which solder joints looked like the component would come off the board any time soon) i got first connects to the Profiler again (unstable though) and after replacing the LAN chip (LAN8720A) everything works fine again (i also replaced the unsuitable input cap mentioned above while being at it). The chip costs 90ct in quantities of 100 pieces.........

Quote

My guess would be that the remote is one PCB and you getting basically a new remote out of it minus the case. It is still cheaper than a new remote, but you are right, not by much.

It basically is one PCB (the other two PCBs are completely passive and only contain the smt switches, some LEDs and the stacking connector), however there's not much to it. Below you see virtually all the electronic componets of the Remote (nothing on the other side, except a through hole cap and a diode). There are no expensive components and as i mentioned above, the PCB itself is 2 Layer junk:

The only expensive part of the Remote is most likey the metal case, which would probably comply with military standards (even the colour ).

Quote

You can actually hook it to a PoE switch and hook your Kemper to it and it works fine

The problem is, you can't just hook it to any PoE switch, which would be the case if they had followed the spec. The spec clearly states that the powering device decides which mode to use when a powered device (e.g. Remote) is connected and that (for exactly that reason) it is mandatory for the powered device to support both modes of powering. Being limited to certain PoE switches is an unneccessary limitation that could have been avoided by a proper design.

But I might not have complained if they had just decided to only implement mode A, if they would at least have done that consistantly (Remote only uses mode A and Profiler delivers power according to mode A). Instead, for using the Remote directly with the Profiler, they used the wiring of mode B on the profiler but only in a completely out of spec way (which significantly reduces the possible cable length and puts higher demands on cable quality) and that's simply bad design.

Quote

This isn't an internet device but rather a remote for an amp head, so they can design it as they see fit.

Of course, but then they shouldn't have called that "Ethernet" but something like "Kemper-Link" and not advertise with a standardized interface.
If you think you need to make something proprietary then feel free to do so, but consequently you then shouldn't pretend to provide something standardized.

As for the question you were looking for in my post:

- Why are defective Remotes not really being repaired instead of replacing everything that's inside the case (we are in 2024, shouldn't sustainability be an issue?)

- Why is the replacement so expensive that you would rather trash the Remote and buy a new one?

- Why was the Remote designed in a way that, in my opinion, simply doesn't suit a company like Kemper (and that doesn't justify the price)?

The main point however, was an advice to those users who are facing the same connection problem with their remote:
Once they verified, that the cause of the problem is a defective Ethernet interface of their Remote, i recommend they save the money for the shipping, trash the Remote (or re-use it as a door stopper) and get a new one (or buy an alternative solution).

If i would have known about the "repair" cost before, i wouldn't have sent my equipment to the Kemper service.

If your Kemper Remote does no longer connect to your Profiler and you have definitely ruled out possible cable problems, and your Remote is outside the warranty period, then here is my advice:

Save yourself the hazzle and unneccessary cost for shipping the Remote (and your Profiler, as they always want you to send in both units) to the service.

They won't really repair your Remote, but simply replace all the PCBs (printed circuit board), which means almost everything except the metal case and they will charge you almost 300.- EUR for that, which is completely uneconomical (and not exactly sustainable either), considering that you "only" pay 419.- for a brand new Remote.

Of course there is a possibility, that the cause of the connection problem is located on the Profiler, but this can be easily checked by using an approptiate PoE Ethernet Switch. Connect that switch to your LAN and hook up the Profiler and the Remote to the Switch. Then check, if the Profiler connects to the Switch. If you get a link on the port, the Profiler is connected to and if you then can access the Profiler through the Rig Manager App on an Android Smartphone that is connected to your (W)LAN, you have proved, that the Ethernet interface of the Profiler basically works.

If your Remote powers up on the PoE switch but you don't even get a link on the Switch port it is connected to, then your Remote has a problem with its Ethernet interface (-> go and get a new remote). If you do get a link and the Remote now does connect to the Profiler, the cause of the problem can reside in the Profiler or in the Remote and is most likely related to the strange power supply concept that is used when the Remote is directly connected to the Profiler (however you do have a solution then, just use a PoE switch, which will also allow you to use longer cables).

For powering 10/100MBit PoE (Power over Ethernet) devices, there are basically two approaches: One (called Mode A) is to use the two Ethernet signal pairs that carry receive (RX, pair 1-2) and transmit (TX, pair 3-6) signals ALSO for delivering power to the Ethernet device. The other approach (called Mode B) is to directly use the cable pairs ( 4-5 and 7-8) which are unused for 100/100MBit Ethernet.

Unfortunately Kemper did not completely follow the corresponding IEE802.3af spec when designing Profiler and Remote and did something quite weird: While the Remote can be powered by a standard PoE switch that delivers power according to Mode A, it can't be powered by using Mode B, which is already a violation of the spec. But even worse: When you directly connect the remote to the profiler, the Profiler does NOT behave like a standard PoE Switch supporting mode A, as it does not deliver the power for the remote through the 1-2 and 3-6 signal pairs. The profiler uses the 4-5 and 7-8 cable pairs to power the remote, but in an out-of-spec way, as it only delivers 5V, while the nominal voltage would be 48V. There was a good reason for choosing that considerably high voltage, as it results in lower currents (the Ethernet interface was not designed to deliver power in the first place, hence cables and connectors are for signal transmission and not for conducting high currents) and is less sensitive to voltage drop along the cable. I really don't see the point in implementing a standardized interface (Ethernet with PoE) and then not following the corresponding specification.

The Profiler is such a great device, hence it is hard to understand why they decided to go for such bad engineering here (and unfortunately this does not only apply to Kempers PoE concept. The circuit design of the Remote itself also lets you raise your eyebrows on several places....).

By the way, Kemper describes the speaker as:

Quote

"The KEMPER Kone is a 12“ full-range speaker which is exclusively designed in cooperation with Celestion for KEMPER."

News | Kemper Amps

How exactly would you define the difference between a full-range and a broadband speaker?

Of course a single speaker system necessarily has a limited range compared to a 2- or 3- way system (that's why i put the "full range" in quotes) or at least needs appropriate EQing ( e.g. Bose 802) to achieve a proper frequency response.

Anyway, the Kone will most likely not roll off at 5kHz, like most guitar speakers and is hence not suitable for standard amps if you do not apply any filtering.

Unfortunately, the information that can be found on the corresponding product page has apparantly been collected from the marketing department and not from engineering, so there is no datasheet, containing a frequency response diagramm of the speaker.

Well, I'm not concerned about the low end, but to my understanding, the Kone is a "full range" speaker, that will hence sound terrible when being driven by a standard guitar amp that does not limit the frequency range of distorted signals.

I'm afraid it's quite unlikey that anyone has ever tried this, but i thought i might ask, just in case.

So has anyone ever hooked up a cabinet with Kemper Kone Speakers to a standard guitar amp (without cab emulation) and used a low pass filter (LC) between amp and cabinet, to limit the range of the Kone speakers to the range of a standard guitar speaker?

Why would i want to do this?

Well, i'm thinking about getting 4 Kone speakers to replace the stock speakers in my 4x12 cab, so i could use it as a FRFR cab for my Profiler.
However i still have an analog Preamp / Poweramp that i would use once in a while, but i couldn't use it along with my 4x12 cab, once it is equipped with the Kemper Kones and i don't have the space for a second 4x12 cab with standard guitar speakers. The idea is, to add a switchable 2nd order low pass filter to the cabinet, so i could use it with the Profiler (filter bypassed) and also use it with a standard guitar amp when the filter is enabled, rolling off at 4,5kHz or 5kHz, to avoid any unpleasant frequencies in overdrive settings.
Of course this is technically possible, the question is however, will such a setup provide a decent sound?

+1

When i discovered the Android App, which already connects via LAN, i was quite surprised, that the Windows Rig Manager does not yet provide that Feature.