Lectrosonics Quadra Digital IEM Review

[Langston Holland]

Conclusion:

The best sounding wireless transmission system of any type I've ever heard. Like the Line 6 digital wireless, it's stupid simple to operate and somewhat limited in maximum channel count (16 in this case). Unlike the Line 6, this is digital without dropouts.*

Usage:

I just happened to have the pleasure of reinforcing the same band last week as was alluded to in my PSM900 review. The front man had the Quadra and the remaining muso's used the PSM900's. Once I got the lead comfortable with the four knobs, we were set (we ran two stereo mixes to him, band on one pair and his keys and vox on the other pair). He loved it. The bassist fussed about the PSM900's ducking effect on the keys with snare hits, but the remainder of the band were quite happy with the Shure's. Happily, it's rare for folks to fuss at me about the compression effect of the Shure's.

Of note is the fact that I drove all 4 channels of the Quadra transmitter digitally at a 48kHz sample rate via two AES3 outputs of my digital snake rack. Sources on stage were converted to 48kHz/24bit by Yamaha AD8HR's, sent via EtherSound to an M7CL for mixing and an AuviTran AVM500 for multitrack recording, sent back through the same CAT5 cable to the digital snake and converted from EtherSound to AES3 by a Yamaha NAI48-ES.

Particulars:

This is a 900MHz digital wireless IEM system capable of configurations in 1, 2, or 4 channels in various arrangements per beltpack. One thing I'd like to see is the ability to have the same mix on channels 3 and 4 go to all beltpacks. That way everyone could get the lead's vox and instrument with independent control while using only 2 of the available 16 maximum transmitter channels. The 4 transmitter input channels can accept all analog, all digital or two analog and two digital. The beltpack holds 3 AA batteries thus is a little larger than most, has a Mercedes feel and the membrane switches feel solid and as if they'll last longer than you will.

The transmitter headphone out volume knob is one of those types you can press in flush with the face of the unit. The nice thing about that is you can set the volume the way you like it, press it in and it'll stay at that way. The unit I had rubbed against its hole a bit which made it difficult to adjust since the knob is not serrated. There are 4 membrane switches along the bottom of the display that make it painless to choose which channel you'd like to monitor. The headphone output jack itself is 1/4" instead of 1/8". That'll bug some monitor engineers I'm sure, but I use console headphones for this kind of thing so it's good news to me.

HF Shelf is unusual in that it not only allows up to 9dB of boost in 3dB increments, but allows you to choose what appears to be 2nd order shelf corners at 5kHz or 7kHz. Continuing with the unusual theme, the indicated adjustments equal the actual adjustments.

I'd prefer the beltpacks' external gain pots be harder to turn and have a strongly detented feel to them. This did not occur in use, but it seems to me that the easy to adjust knobs could provide quite a surprise to a performer that has a belt or shirt dragged quickly across the knobs - such as the Chili Peppers used to do when they ripped their clothes off and painted themselves orange during their sets. I'm surprised Karl didn't think of this because he loved those shows. Anyway, 3 of the pots have a detent 1/2 way through their range - same as you'd have on an console pan pot. This is unnecessary IMO because the knobs already have a nicely raised center edge to indicate position by look or feel.

The power button on the transmitter rocks. Big with a feel more appropriate for a 20 amp switch. That thing will not be turned on or off accidentally. A nice addition would be to copy the Shure PSM900 design where you have a second switch for the RF transmit that allows you to turn the unit on and make adjustments without possibly surprising someone.

At first I was annoyed that the beltpack didn't sport a battery meter - only a green LED next to the display panel. After reading the manual I found that the LED will turn red when there's only 1/3 battery life remaining and eventually start flashing red to inform the monitor engineer that this is his last show with me. Battery life is quite workable - 3 hours of use and the green battery LED is still on. One very nice aspect of having the battery LED not included with the display is that it can be seen even after the display times-out and goes dark. It would be nice to have battery meters on the transmitter as well. Digital can do anything, right?

The transmitter clip indicators are dead on though not easy to see in a live show environment. The world standard in clip indication are the Sennheiser units where the whole display panel flashes red - brilliant - too bad the metering itself is misleading. The Quadra beltpack receives its metering data from the transmitter, thus shows clipping at the identical threshold. With the transmitter adjusted to its lowest gain setting of "-20", the maximum input is +18dBu broadband or +20dBu at frequencies above about 50Hz. Also of interest is that at maximum input to the transmitter the beltpack gain controls must be at 50% or less in order to maintain distortion free output, though this is something I found only with measurement, not listening (it's minor). At drive levels with appropriately rare +20dBu peaks this system will perform flawlessly.

Driving the transmitter digitally is interesting. Low frequency under 20Hz is flatter if you feel you have a need for that (!). Something this system does that I've come across before that I don't understand is that every time you cycle the beltpack's power the latency changes by a sample or two. It's completely immaterial, but fascinating nonetheless. Based on the frequency response it appears that the Quadra system is operating at 48kHz internally.

The limiter works as indicated; 3dB gain reduction increments up to -18dB max. IEM limiters are intended for hearing protection but are rarely used. FFT measurements assume linear behavior in the DUT and limiters violate that requirement, so no (not-so) pretty traces.

Useful background on the following measurements.

Input Levels from -40dBu to +18dBu:

HF Shelves:

6.5ms Gated Sine Bursts with THD:

Analog vs Digital Transfer Functions:

Edit 1: my initial measurements using CLIO's digital output indicated increased latency when using the Quadra's digital inputs (2.42ms at 96kHz and 3.25ms at 48kHz). Latency when driving the analog inputs is about 0.95ms. After bothering to read Lectro's specs this morning, latency should drop as you would expect when bypassing the A/D section of the Quadra. I'll figure out what I'm doing wrong and report back.

Edit 2: actually, CLIO's latency results above should be even longer because the measurement hardware's D/A converter is removed from the loop. See Further Observations post for further observations.

Edit 3: Don Boomer made an good point in a PM to me today that I either forgot about or don't remember - the Line 6 demo I had was a beta unit. If the production units prove as dropout-free as the analog RF stuff I'm used to, it'll be an amazing value.

 

[Langston Holland]

Further Observations

When buying a new measurement program, I compare it's output to everything else I have ever had access to. My tests show CLIO's analog I/O measurements identical to Smaart, MLSSA, SysTune, EASERA Pro and ARTA. Here is an example comparison between Smaart and CLIO. Digital I/O testing is new to me and part of the reason I'm scratching my head a bit on the Quadra's latency measurements when driven digitally.

Quadra HF Phase at 48kHz:

It turns out that the wiggles in the phase trace when the Quadra was driven by CLIO's digital output were due to issues with the latter. The same thing can be seen in the measurement of the near perfect Benchmark DAC1:

Latency Measurements:

I also have the amazing Benchmark ADC1 and inserting this with the DAC1 in a measurement loopback yields a state of the art result.

The measurements also yield a combined throughput latency slightly longer than spec'd by Benchmark:

ADC1 + DAC1 48kHz Spec: 3.10ms, Measured: 3.17ms
ADC1 + DAC1 96kHz Spec: 2.18ms, Measured: 2.27ms

When deriving a digital drive signal from the ADC1 to feed the Quadra the 48kHz phase wiggles disappear. It also provides more insight on just how well designed the Quadra is.

At this point I'm looking forward to clarification on the Quadra's <0.5ms latency spec when driven digitally. Inserting the ADC1 and Quadra in the measurement loop produces latencies of 4.69ms at 48kHz and 3.26ms at 96kHz. If Benchmark's latency specs for the ADC1 are close to reality, which they are as shown earlier, then the approx. actual latency of the Quadra system from digital input to earphone output on the beltpack is 3.60ms at 48kHz and 2.59ms at 96kHz.

BTW, the Quadra had no trouble accepting digital inputs from sample rates of 44.1, 48, 88, 96, 176 and 192kHz.

 

[Karl Winkler]

Re: Lectrosonics Quadra Digital IEM Review

Langston, thank you for this detailed and informative review. As I mentioned to you by email, we have been looking at the latency question when feeding the transmitter with digital sources. It wasn't until today that we had the time to set up the test, after gathering the necessary bits.

First, let me point out what we think is a simple error in your math. I've eliminated the 96 kHz specs for simplicity and show only the 48 kHz figures.

You state: "ADC1 + DAC1 48kHz Spec: 3.10ms, Measured: 3.17ms", OK - makes sense. Then here: "Inserting the ADC1 and Quadra in the measurement loop produces latencies of 4.69ms at 48kHz. If Benchmark's latency specs for the ADC1 are close to reality, which they are as shown earlier, then the approx. actual latency of the Quadra system from digital input to earphone output on the beltpack is 3.60ms at 48kHz"

We think what you meant was that if you subtract the ADC1+DAC1 latency of 3.17 ms from the total latency of ADC1+Quadra latency of 4.69 ms, you will reveal the latency of the Quadra wireless transport itself, which would thus be 1.52 ms by your measurements.

We were able to corroborate a similar figure on the bench: we used an A/D converter known to have 3 ms of latency, so: A/D + Quadra = 4.25 ms, less the A/D latency of 3 ms = 1.25 ms.

Interestingly, we tried the analog input to output of the Quadra and came up with a hair under 1.0 ms. (?!?)

Presently, we do not know why this is the case - it clearly should not be, since by eliminating the A/D conversion, we should see a reduction of approximately 0.4 to 0.5 ms. Indeed, this is the difference we see when we compare the analog in to out vs. the digital in to analog out of the D4 system: there is a reduction of .4 ms in bypassing the A/D conversion process.

We are not yet certain if this is a hardware or software bug, but the investigations begin.

Thank you again for the review - sometimes this is how we find bugs in our products. Somewhere between the proto stage and the final product, we perhaps made a mistake of some sort. We'll get to the bottom of it!

 

[Langston Holland]

Re: Lectrosonics Quadra Digital IEM Review

First, let me point out what we think is a simple error in your math. I've eliminated the 96 kHz specs for simplicity and show only the 48 kHz figures.

You state: "ADC1 + DAC1 48kHz Spec: 3.10ms, Measured: 3.17ms", OK - makes sense. Then here: "Inserting the ADC1 and Quadra in the measurement loop produces latencies of 4.69ms at 48kHz. If Benchmark's latency specs for the ADC1 are close to reality, which they are as shown earlier, then the approx. actual latency of the Quadra system from digital input to earphone output on the beltpack is 3.60ms at 48kHz"

We think what you meant was that if you subtract the ADC1+DAC1 latency of 3.17 ms from the total latency of ADC1+Quadra latency of 4.69 ms, you will reveal the latency of the Quadra wireless transport itself, which would thus be 1.52 ms by your measurements.

We were able to corroborate a similar figure on the bench: we used an A/D converter known to have 3 ms of latency, so: A/D + Quadra = 4.25 ms, less the A/D latency of 3 ms = 1.25 ms.

Whoever "we" is is too kind! I really did mean to omit the DAC1 digital to analog converter in my latency spec for the Quadra system - and I was wrong to do so. When I removed the DAC1 from the test rig I just went on autopilot and removed it's contribution to the latency spec, which is fine if human beings have AES3 inputs, but, alas, they do not. To keep apples to apples, I needed to subtract the latency of converting the output of my digital source (console, etc) into analog in order to make a realistic comparison of the Quadra's digital and analog input options. AFAIK, all competing systems only have analog inputs at this time. That means that those systems also experience the added latency of a D/A conversion stage when a digital console is involved. Once corrected, the Quadra should have the ability to achieve a latency less than half that of any other analog IEM when driven by a digital console.

Interestingly, we tried the analog input to output of the Quadra and came up with a hair under 1.0 ms. (?!?)

If you mean analog input of the Quadra transmitter to analog output of the Quadra beltpack receiver, I measured 0.95ms. Precisely (1) hair.

PS: Please make 2 channel beltpacks with a single knob that are compatible with this system as a less expensive option to the 4 channel beltpacks. That would be very, very persuasive.

 

[Langston Holland]

Re: Lectrosonics Quadra Digital IEM Review

we used an A/D converter known to have 3 ms of latency...

Be very careful with that kind of thing. I've been communicating with the designer of the Benchmark products recently and he remeasured the DAC1-ADC1 latency and found their specs to be off a small amount: "+/- 0.1ms". They based their published specs on the OEM provided specs of the individual parts used in the units.

It was enough to bug him, but is immaterial in this review and it still means my FFT based latency measurements may be too high. Maybe. I'm getting my dual channel scope out to figure out what the stink is going on here. CLIO, Smaart, ARTA, SysTune, EASERA Pro and MLSSA agree with one another in the time domain in my experience - so this is quite the mystery - or maybe digital stuff like this simply exhibits tolerance variations on this scale.

 

[Karl Winkler]

Re: Lectrosonics Quadra Digital IEM Review

Be very careful with that kind of thing.

Good point - in fact, we think that the A/D converter we were using actually has a measured latency at about 2.75 ms. Guess what? That means that your results and ours are almost exactly matched.

So, one thing we've figured out is that the re-sampling buffer introduces a group delay, and our initial investigation into this points to an amount of latency consistent with what we've found. The engineers are seeing if there is a way to shorten this delay by reducing the buffer size - they think they can do this. The result, interestingly, would probably be that the Quadra system would have about 1 ms of latency, whether it is fed from an analog OR digital source (right now, it has ~ 1.5 ms when fed digitally).

Once we nail this down, we will revise our spec sheet.

 

[TJ Cornish]

Re: Lectrosonics Quadra Digital IEM Review

Conclusion:

The best sounding wireless transmission system of any type I've ever heard. Like the Line 6 digital wireless, it's stupid simple to operate and somewhat limited in maximum channel count (16 in this case). Unlike the Line 6, this is digital without dropouts.* <---

Was this where your asterisk going to point, or was there further expansion forthcoming?

Conclusion:

Edit 3: Don Boomer made an good point in a PM to me today that I either forgot about or don't remember - the Line 6 demo I had was a beta unit. If the production units prove as dropout-free as the analog RF stuff I'm used to, it'll be an amazing value.

 

[Langston Holland]

Re: Lectrosonics Quadra Digital IEM Review

Reasonable question TJ.

The "Edit 3" point you quoted is the end of it. My goal was to soften the edge of the Line 6 dropout statement without avoiding the fact of the matter. The prototype explanation makes sense to me, time will tell. Even if the dropouts I experienced aren't improved by future design changes, the thing is a gift at its price point. If I lived with those units long enough I'm pretty sure I could figure out how to reduce or more likely eliminate the dropouts - possibly at a reduced maximum channel count, or just getting the RF I/O a bit closer, or both.

 

[Langston Holland]

CLIO Cleared of Wrongdoing, Released on own Recognizance

CLIO is correct, thus the rest of the FFT world that produces the same time domain results can breathe a sigh of relief. I'm joking of course - I'd be shocked if a representative of any of the measurement packages I mentioned read this stuff...

Background:

The latency or group delay measurement in question is analog to analog of my Benchmark ADC1 to DAC1 converters operating at a 48kHz sample rate.

My CLIO FFT measurement program reported 3.172ms using a 192kHz resolution, but the phase curve at this delay step was still -10.5 degrees at 20kHz, thus the true latency per CLIO is about 3.173ms.

Proof:

Using a Fluke 192C scope (60MHz) I setup the following: Signal generator producing a 1kHz sine 5ms on / 10ms off. The signal was split so that scope channel A received a direct connection to the generator output and scope channel B got the output of the ADC1-DAC1 chain. For fun I tried both the coax and AES3 links between the Benchmark units and got identical results. Per the attached, the scope measured 3.16ms with the cursor readout and 3.158ms with phase readout. Moving either cursor one step further from the other jumped to 3.20ms, thus the scope could have very well been in perfect agreement with CLIO but unable to report it given the coarse readout.

Two perspectives of the same thing:

 

[Andrew Broughton]

Re: CLIO Cleared of Wrongdoing, Released on own Recognizance

We found the latency higher also in the Galileo when using the Digital Inputs, which seems counter-intuitive. Haven't heard an explanation from Meyer, but my assumption is the SRC adds a significant latency which is not there when running through the A/D. Does the Quadra use SRC as well?

 

[drew gandy]

Re: CLIO Cleared of Wrongdoing, Released on own Recognizance

I would expect that: the better the SRC, the longer the latency. So, getting everything on the same bit per second setting (and removing any SRC from the path) would tend to be ideal. Langston, since you are doing a fair amount of digital connection between live gear now, do you have any experiences to comment on about the latency issues between gear? I can imagine scenarios, particularly with iems, where the latency difference between different paths causes unexpected or unintuitive issues?

-drew

 

[Langston Holland]

Re: CLIO Cleared of Wrongdoing, Released on own Recognizance

...since you are doing a fair amount of digital connection between live gear now, do you have any experiences to comment on about the latency issues between gear? I can imagine scenarios, particularly with iems, where the latency difference between different paths causes unexpected or unintuitive issues?

I've yet to measure the latency differences between my M7CL 48 channel consoles with entirely analog peripherals vs. EtherSound based digital snakes. The digital snakes make a striking subjective improvement particularly with acoustic sources and full throttle rock. I very reluctantly went from large frame Crest consoles in flawless condition to the M7's for the same reason people get boats during a flood. Not a happy situation, so I made it a goal to eventually have high end analog sound in the midst of water world.

I found it and IMO bettered it slightly with AD8HR front ends without further conversions until inside the loudspeaker processor. Stupid business decision in most ways, but I'm not in this purely for money.

There are no audible latency issues with the Quadra whether fed by analog or digital sources. There are no audible latency issues with the M7 when using analog or properly setup digital snake systems (I have both AD8HR's and SB168's). Sound quality is improved either way, but the AD8HR sounds like analog to me. Ultimately, apart from the acoustic transducers, a digital console should be interfaced with digital peripherals.

 

[Langston Holland]

Re: Lectrosonics Quadra Digital IEM Review

> Once we nail this down, we will revise our spec sheet.

Found something of interest in one of the manuals to the greatest loudspeaker processor ever made. These are the various latencies seen in the Dolby Lake Processor depending on the I/O options selected when using an IIR module: