Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Bennett Prescott]

Josh Evans made the unfortunate mistake of letting me know that he's working for TC Group now, so I browbeat him until he shipped me one of the new LM26 processors. Or whatever they want to call it... Lake? Lab.Gruppen? Processor? In any case, I imagine this will be of interest to those of us who already have an investment in Lake processing, which has been in limbo for over a year now unless you want to make an investment in Lab.Gruppen amplifiers. Since I represent a powered loudspeaker manufacturer, I have no way to make that work for me, and have been eagerly awaiting the release of this standalone processor.

The unit arrived yesterday, packed in this attractive box, which contained the processor, ethernet cable, power cable, software CD, quick start manual, and AES breakout which I thought was a nice touch.



Functionally, the unit appears to be a return to the Lake processor's origin: a 2x6 processor that networks, updated to the 21st century. You can tell that Lab.Gruppen has redesigned it after the slick look of their amplifiers, but unless I'm missing it nowhere on the unit does it actually say "Lab.Gruppen". Every effort has been made to make this a Lake processor, not a "Lake Processor by X". The most noticeable updates are 1000-T ethernet on the back for running audio and control down, a lockable IEC inlet, and front panel controls. They claim that LCD display is "daylight viewable"... we'll see about that.



Fired up, it's got some pretty lights, which seems to be one of the continuing selling points for the Lake processor. At least these lights make some more sense than the DLP... those front panel meters were pretty, but told me almost nothing about my actual drive levels. Oh, and here's a peek inside.



The LCD display is a little cramped, but conveys plenty of useful information. You'll just have to get close to read it. Unfortunately, the tradition of useless front panel controls is continued from the DLP. It is possible to reset the processor, set delays, levels, and polarity, set input sources and priorities... but not possible to change EQ parameters, crossovers, etc. This was somewhat possible on the DLP, but the interface sucked. This interface is somewhat better, but now you can't change any of those parameters. I'd just like to have an emergency backup in case my laptop goes down, and I have to rough in settings the hard way.

Otherwise, it seems by all accounts to be... a Lake processor. With Dante built in the 2 input limitation doesn't really bother me. It certainly doesn't bother me anywhere near as much as the three outputs to a processing block limitation. If I need to process one input to four outputs, I have to burn an entire module. This isn't Lab.Gruppen's fault, but an annoying inflexibility that was inherited from the original Contour and hasn't changed. If I wanted to process passive boxes, I could just buy your amps, people! This would be less annoying if the software made it easier to group sets of outputs, but it doesn't. While the Lake is a fantastic processor with the killer app interface that lets me get more done in less time, I sure have to fight with it to get it to do what I want.

Bitchfest over, I'm going to rack this puppy this weekend and use it for some additional subwoofer processing.

[Evan Kirkendall]

Hey Bennett,
You don't need that junky thing. Send it to me so I can put it to real use.




Evan

[Ryan Garnett]

Looking at that PCB, one has to wonder if they couldn't make the thing a quarter of the size?

[TJ (Tom) Cornish]

Ryan Garnett wrote on Thu, 22 July 2010 08:48

Looking at that PCB, one has to wonder if they couldn't make the thing a quarter of the size?

The 4.75" rack format was never very popular.

[Ryan Garnett]

TJ (Tom) Cornish wrote on Thu, 22 July 2010 09:50

Ryan Garnett wrote on Thu, 22 July 2010 08:48 Looking at that PCB, one has to wonder if they couldn't make the thing a quarter of the size? The 4.75" rack format was never very popular.

Depth...which means a smaller chassis which means lighter weight which means my rack might be a couple pounds less of a pain in the ass.

[TJ (Tom) Cornish]

Ryan Garnett wrote on Thu, 22 July 2010 08:52

TJ (Tom) Cornish wrote on Thu, 22 July 2010 09:50 Ryan Garnett wrote on Thu, 22 July 2010 08:48 Looking at that PCB, one has to wonder if they couldn't make the thing a quarter of the size? The 4.75" rack format was never very popular. Depth...which means a smaller chassis which means lighter weight which means my rack might be a couple pounds less of a pain in the ass.

It's a joke, son.

[Silas Pradetto]

Ryan Garnett wrote on Thu, 22 July 2010 09:52

TJ (Tom) Cornish wrote on Thu, 22 July 2010 09:50 Ryan Garnett wrote on Thu, 22 July 2010 08:48 Looking at that PCB, one has to wonder if they couldn't make the thing a quarter of the size? The 4.75" rack format was never very popular. Depth...which means a smaller chassis which means lighter weight which means my rack might be a couple pounds less of a pain in the ass.

I HATE shallow rack gear. When it's racked with a bunch of deeper stuff, like 16" deep amps, you can no longer reach the XLRs to plug or unplug anything.

I'd prefer a standard depth like 16" even if it's full of air. Maybe put in a trap door for "storage"

[Randy Pence]

+1, it is quite annoying to have to repatch shallow controllers

[Bennett Prescott]

I only need about four more of these and I'll really be able to have some fun.



Spent some quality time with the LM26 this weekend, and I'm all smiles. As I've said before... it's a Lake Processor. It does what a Lake Processor does. If you are familiar with a Lake Processor, you will be able to use it to do Lake Processor sorts of stuff.

I am using Dante for the first time. Very cool, and it answers all my questions about input routing. I have nothing but an Ethernet cable going between the DLP and the LM26, and I have it set up so the LM26 can pick up any of the four inputs going into the DLP and process them. Since the LM26 has 6 input routers, I can still process (for example) a mono mix of L&R for my fills as well as a subwoofer feed. I seem to have incurred .8 milliseconds of latency for this privilege, which is inconsequential for my application.

Aside from one small setup hiccup on my part, configuration was extremely easy. The new zero-conf auto configuration just works. Everything talks together, talks to my Cisco access point, and talks to the laptop.

The LM26 has a few new features that I don't need to mess with, like GPIO for contact closures.

Otherwise, I'll spend some more time kicking it around and listening starting Thursday, when I'll really be able to make use of the extra processing available to me. I'm also going to try and take some measurements to confirm that the latency matching is working properly, since in that case .8ms could be a very big difference!

So far my verdict is: Buy 12.

[Andy Peters]

Ryan Garnett wrote on Thu, 22 July 2010 06:48

Looking at that PCB, one has to wonder if they couldn't make the thing a quarter of the size?

Because if the PCB was a quarter of the size, it'd likely have eight layers, and that's a lot more expensive than a larger four-layer board.

A larger PCB with the parts spread out doesn't get as hot as one with everything crammed all together.

Also, the stuff in the lower center/right are all parts of a switch-mode power supply, and they kept that stuff away from the analog electronics (top to the right). All of the digital stuff (the DSP, and FPGA, the Ethernet, their associated power supplies) and the mains power supply are on the other side of the PCB from the analog. Keeps everything quiet.

-a

[Mac Kerr]

Andy Peters wrote on Fri, 06 August 2010 00:22

Ryan Garnett wrote on Thu, 22 July 2010 06:48 Looking at that PCB, one has to wonder if they couldn't make the thing a quarter of the size? Because if the PCB was a quarter of the size, it'd likely have eight layers, and that's a lot more expensive than a larger four-layer board. A larger PCB with the parts spread out doesn't get as hot as one with everything crammed all together. Also, the stuff in the lower center/right are all parts of a switch-mode power supply, and they kept that stuff away from the analog electronics (top to the right). All of the digital stuff (the DSP, and FPGA, the Ethernet, their associated power supplies) and the mains power supply are on the other side of the PCB from the analog. Keeps everything quiet. -a

I addition to Andy's very real technical reasons, what would be the point? It is already a 1RU device, the box isn't going to get any smaller. Making the box shallow makes it hard to reach the connectors when it is mounted is a rack with other gear.

Mac

[John Roberts {JR}]

Mac Kerr wrote on Fri, 06 August 2010 11:45

Andy Peters wrote on Fri, 06 August 2010 00:22 Ryan Garnett wrote on Thu, 22 July 2010 06:48 Looking at that PCB, one has to wonder if they couldn't make the thing a quarter of the size? Because if the PCB was a quarter of the size, it'd likely have eight layers, and that's a lot more expensive than a larger four-layer board. A larger PCB with the parts spread out doesn't get as hot as one with everything crammed all together. Also, the stuff in the lower center/right are all parts of a switch-mode power supply, and they kept that stuff away from the analog electronics (top to the right). All of the digital stuff (the DSP, and FPGA, the Ethernet, their associated power supplies) and the mains power supply are on the other side of the PCB from the analog. Keeps everything quiet. -a I addition to Andy's very real technical reasons, what would be the point? It is already a 1RU device, the box isn't going to get any smaller. Making the box shallow makes it hard to reach the connectors when it is mounted is a rack with other gear. Mac

Yup there is no single best answer, and it would probably be somewhat cheaper to use 2 sided PCB for rear jacks, maybe as many as 4 separate different technology and physically smaller PCB, connected together with a bunch of ribbon cables... What could possibly go wrong with ribbon cables?  

While a single large board may cost a little more, it pays dividends in reducing point to point wiring, increased reliability, etc.  

I doubt this sku is that much of a sharp pencil exercise and there is perceived value in the deeper/heavier chassis.

JR

[Bennett Prescott]

Over the weekend I had a chance to run some critical audio through the LM26. Long story short, I used it to run one half of the PA system, and I was unable to hear any sonic difference one side or the other, which is excellent.

About that daylight viewable screen... seems about at good as the DLP, which is to say passable. Mostly it's just so small that you have to get real close to tell what it's saying, but the meaningful front panel level indicators tell you most of what you need to know and can be seen easily except in direct sunlight.



Before I ran a show on it, however, I measured it.

[Bennett Prescott]

Here is a screen shot of a measurement of the processing applied to what happens to be the middle section of my line array. The LM26 is the purple trace. As you can see, the two processor provide exactly the same processing for identical settings. The deviation in the phase trace is due to half a sample of difference in the delay times and doesn't actually exist.

This is pretty interesting given that the audio comes into the DLP over analog, arrives at the LM26 over an ethernet cable via the Dante protocol, and comes out just the same! Modern technology never ceases to amaze me.



P.S. Smaart 7 really is fantastic, I've used it to make several measurements over the last few days and it makes getting the data I need to see much easier.

[John Roberts {JR}]

Bennett Prescott wrote on Wed, 11 August 2010 16:18

This is pretty interesting given that the audio comes into the DLP over analog, arrives at the LM26 over an ethernet cable via the Dante protocol, and comes out just the same! Modern technology never ceases to amaze me. P.S. Smaart 7 really is fantastic, I've used it to make several measurements over the last few days and it makes getting the data I need to see much easier.

Repeatability is an under appreciated benefit of DSP. I recall the blood sweat and tears of trying to make a 4 pole analog crossover adjustable over a reasonable range and accurate enough between HP and LP to be usable.

Even using high quality parts (alps potentiometers) we had to manually measure and tweak the ganged pots at mid rotation.  I don't miss those good old days.


JR

[Bennett Prescott]

What I found more interesting is that the latency through these two processors is not the same.

Base latency, analog in to analog out, of the DLP is 1.1ms. For the LM26 it is 1.02ms.
If I turn on "Latency Matching" both processors exhibit a latency of 2.02ms analog in to analog out.

Things get tricky when I use Dante, however. "Latency Matching" does not account for the latency of this interface, so even with it off the DLP is still exhibiting 1.1ms while audio from LM26's analog outputs arrives after 2.69ms. Activate latency matching on both processors and the spread seems to increase to 1.65ms. That may be meaningless in some circumstances, but in many you must be very careful to account for it, and interestingly enough it is twice what Dante indicates to me its latency should be... hopefully it is not variable! I am not using any FIR filters in either processor, so I do not know if latency matching accounts for them as well.

For an exciting visual example of this, here are two traces. This is the same magnitude response as in my post above. The purpleish trace is the phase shift due to the differing delay through each processor, analog in to analog out on both (.08ms). The greenish trace is the phase shift due to latency induced through the Dante interface (about 1.6ms). It gets much worse if latency matching is enabled on the LM26 and not on the DLP, but that would be a stupid thing to do so I certainly didn't do it.



The good news is I had a great conversation with Josh Evans this morning about these issues, and he has promised to talk to the head honchos about allowing latency matching over Dante, as well as allowing aux outputs to be grouped, and perhaps some other things that would be handy. Until a solution appears, he has told me to "use AES to link Lake processors". I therefore pass this advice on to you.

Oh, also he taught me about how to create modules with variable numbers of aux outputs, which eliminates half of my problems with the Lake processing. I should have figured this out, but this is why it's good to have friends in high places.

[Bennett Prescott]

 One final point of interest, here is a screenshot of what the Lake control software says it is doing to my audio. Close... but not quite. I suppose using it as a guesstimate of the -6dB points on my butterworth crossovers isn't as reasonable as I'd hoped. What it's doing in the HF as well is a mystery to me, since the measurement is reasonable while the display is not. Perhaps I need to measure at 96K to really see what's going on in the HF, but I don't have an interface that can do that, and the Lake display sure looks like it's behaving funny before that point.

[Dan Brown]

I should have a pair of them in tomorrow.
Looking forward to it.

I loved my original Lake Contour Pro26 and am sure the LM26 will be every bit as good.
I am glad LabGruppen has picked these up and is supporting the product

The new digital input options are certainly going to be a plus over the original

db

[Franz Francis]

There are also some training videos on theis website, which can only be accesed by a registered user.

http://training.labgruppen.com/videos/

Franz

[Moritz Nasterlack]

Hello Bennett  

At first, big "Thank You" for your postings about the Lake products! They are VERY interesting and informative!

Short, easy question about the Lake Control software.
Which tablet-PC you are using to control the parameters?
I'm looking for a cheap one  
You have new informations about your HF-"problem"?

Greetings from Germany,
Moritz