• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Review and Measurements of Acoustic Power APL1 DSP

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
20,048
Likes
23,034
Location
Seattle Area
#1
This is a review and detailed measurements of the Acoustic Power Labs APL1 digital audio signal processor. It is on kind loan from a member. The APL1 comes in a number of formats. The one under review is the automotive version and costs 510€ or USD $570. This is for the bare hardware. Software that let you program its internal "FIR" filters costs 262€ or USD $295. I am surprised that given the cost of the hardware, the software is not bundled in for free. For this testing, I only focused on the measurements of the hardware as the software seems non-trivial to use.

The automatic configuration of APL1 is nice and slim with phoenix connectors for power:
Acoustic Power Labs APL1 Digital Signal Processor Filter and EQ Audio Review.jpg

The LEDs change color depending on where you leave the slider switch. In my case I set the switch to bypass.

Here are the back connectors:
Acoustic Power Labs APL1 Digital Signal Processor Filter and EQ Inputs and Outputs Audio Review.jpg

We have analog RCA input and output. In addition there is also S/PDIF and Toslink for input and output. There were no obvious way to change inputs and I spent good bit of time searching and googling but could find nothing. There is no manual either that I could find. Finally I saw the teardown pictures of another model and noticed that one had a label inside the case on how to change the input with jumpers. I opened this unit but there were no such labels in it but I did find the same jumper. It would select between Toslink and S/PDIF. I could not get the Toslink to work in either setting but that turned out to be a broken Toslink input (the unit was purchased used and has been loaned out for competition). I select S/PDIF and that worked and that is how I conducted my tests.

While I had the case open, I took a picture of the guts (click on the image for larger size):

Acoustic Power Labs APL1 Digital Signal Processor Filter and EQ Teardown.jpg


The heart of the unit is the Spartan FPGA in the middle. FPGAs are a very good fit for large number of parallel operations which is what we have in audio signal processing. To the top left there is an Arm M Cortex microprocessor for general housekeeping and communication with the host. Top right is a switching power supply which should allow the unit to run on wide range of voltages.

A cirrus 4270 handles ADC and DAC functionality. Bur-brown (TI) silicon handles S/PDIF and Toslink functionality.

A set of beefy diodes protect the input RCA jacks against static electricity and voltage surges which is nice to see.

Audio Measurements
I first treated the APL1 as a DAC and measured its analog output performance using S/PDIF as the input:

Acoustic Power Labs APL1 Digital Signal Processor Filter and EQ Audio Measurements.png


The output voltage is healthy at 2.4 volts which comes in handy in automotive applications to provide better noise immunity. The distortion+noise ratings though are disappointing by our standards. SINAD of 83 dB is way short of 95 dB that Cirrus Logic advertises for the CS4290 "codec" chip. For automotive applications this is probably fine but the same hardware is sold for other applications and there, one would like to have the full capabilities of the CS4290.

The low SINAD lands the APL1 in the middle of our fourth tier of performance:
Acoustic Power Labs APL1 Digital Signal Processor Filter and EQ Audio SINAD Measurements.png


Many users may opt to use the analog RCA input instead of digital so let's test that using our dashboard:
Acoustic Power Labs APL1 Digital Signal Processor Filter and EQ RCA In Audio Measurements.png


This is almost identical to using digital input! In other words, the ADC is transparent with respect to how low the performance of the DAC is.

Cirrus Logic rates the CS4290 at 105 dB dynamic range although that is with a-weighting. Here is what I got:

Acoustic Power Labs APL1 Digital Signal Processor Filter and EQ Dynamic Range Audio Measurements.png


So we fall short again when using the digital input to the tune of 6 dB. Analog input (on the right) shows 3.5 dB of penalty which is reasonable.

Measuring the analog input frequency response, we see confirmation of what is implied in the datasheet which is the unit running internally at 48 kHz sampling:

Acoustic Power Labs APL1 Digital Signal Processor Filter and EQ Frequency Response Audio Measu...png


48 kHz sampling naturally limits the response to half that at 24 kHz. Response is down -1.16 dB at 20 kHz but you can compensate for that using the DSP I am sure.

Intermodulation distortion versus level shows both the noise level and distortion trailing our desktop audio products (at much lower cost):
Acoustic Power Labs APL1 Digital Signal Processor Filter and EQ Intermodulation Distortion Aud...png


We see that distortion rises at -18 dB or so. I dialed down the level in the dashboard and indeed SINAD improved by a few dBs as shown in the original graph.

THD+N versus frequency but with much wider bandwidth than the dashboard showed very poor results (I have increased the resolution versus the reference in red and hence the smoother graph):

Acoustic Power Labs APL1 Digital Signal Processor Filter and EQ THD+N versus Frequency Audio M...png


So I ran a 1 kHz test to see what is going on above audible band:

Acoustic Power Labs APL1 Digital Signal Processor Filter and EQ 1 kHz FFT Audio Measurements.png


The 50 Khz twin tones are surely contributing to much worse performance as they are taller than the distortion products! I wonder if they are the switching frequency of the dc to dc converter.

The distortion rose with frequency though so let's run a 10 kHz test:
Acoustic Power Labs APL1 Digital Signal Processor Filter and EQ 10 kHz FFT Audio Measurements.png


The larger graph is at 176 kHz sampling and shows a lot of harmonic distortion. Switching to low sampling rate of 48 khz left most of them there which is surprising. Filtering of the DAC should have sharply reduced their levels. So I ran the square wave test to see what is going on there:
Acoustic Power Labs APL1 Digital Signal Processor Filter and EQ Square Wave Audio Measurements.png


I don't think I have ever seen such a distorted square wave in all the times I have run this test. I wonder what is going on.

Back to THD+N versus frequency, here is what happens with analog input:
Acoustic Power Labs APL1 Digital Signal Processor Filter and EQ THD+N versus Frequency RCA Inp...png


So essentially the same showing the issue is in the DAC, not ADC.

Linearity is a mess indicating a) the unit is running in 16 bit mode and b) the input 24 samples are being truncated:

Acoustic Power Labs APL1 Digital Signal Processor Filter and EQ Linearity Audio Measurements.png


The reason I say it is truncating is due to zigzag positive and negative error depending on input level.

I expected jitter to be poor but it was pretty clean:
Acoustic Power Labs APL1 Digital Signal Processor Filter and EQ Jitter Audio Measurements.png


Conclusions
For a company that is in signal processing business, I thought the performance would be better than what we have. Low-end consumer DAC chip limits performance by itself, there was no reason to add to it on top of that. Things like lack of dither/conversion to 16 bits is unfortunate to see. For the use the owner will get out of it, the APL1 is fine but the cost new is too high in my opinion. Charging so much for the software in consumer applications doesn't make sense to me either. At least give me rudimentary capability to program the filters for free.

Overall, while I was excited to see an automotive DSP which I may deploy myself, I am turned off by the performance and pricing.

Needless to say, I can't recommend the APL1.

------------
As always, questions, comments, recommendations, etc. are welcome.

I am running out of ideas for jokes for this section so thinking about hiring someone to write them for me. I would need money for that so please donate generously using:
Patreon: https://www.patreon.com/audiosciencereview), or
upgrading your membership here though Paypal (https://audiosciencereview.com/foru...eview-and-measurements.2164/page-3#post-59054).
 
Last edited:

AresHarvest

Active Member
Patreon Donor
Joined
Sep 5, 2018
Messages
114
Likes
54
#2
Hey, just something I spotted: text says Response is down -1.6 dB at 20 kHz, but the image shows it to be -1.16 at 20 kHz (and -1.6 at 23.5 kHz)
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
20,048
Likes
23,034
Location
Seattle Area
#5
Joined
Dec 31, 2018
Messages
12
Likes
14
#6
Full disclosure: I've been fairly active in the APL community helping people figure out how to use these. The founder of APL(Raimonds Skurals) is Latvian and his english is VERY rough, so I'm happy to help when I can. The only benefit I've received from this is a free piece of software(for hardware I don't own) to demo.

-----

I’m disappointed by the measurements, but happy with the tuning results I get with it, so the measurements don't really bother me too much. I'll still probably buy a trio of APL1012's for the fully active 5.2 home system I'm building.

The hardware was originally designed for the pro audio world, that is why the software and hardware are two separate purchases. In that world you may have dozens of unit bouncing around the country at any given time, but only a few actual techs running shows, so you only need a few software installs. It makes sense for the pro audio world, but has been an issue in the consumer world.

The other issue with the software is that it has a steep learning curve. Again, it was designed for the pro world, where simple and pretty user interfaces aren’t needed, or expected. This has also been an issue with consumers, and after reading Amir’s MiniDSP review, I’m not surprised he didn’t wish to test the tuning functions. :) I work in the pro video (LED) world and software like this is just a daily part of life, it took me no time to figure it out.

This particular unit has been to more states that most people. I’ve sent it off to a handful of people to demo it, and well as help out in a pinch. It also bounced around in a Peterbilt for several months, and before sending it to Amir it was in my bedroom system at home. It was in perfect condition when I bought it, I don’t know when the optical ports got damaged, sigh…

Here are the tuning results from my bedroom:


(No smoothing)

Yellow = Before
Purple = After
Green = Target

These measurements are taken from APL Workshop and imported into REW.

The unit I originally purchased a few years ago is in the dash of my Miata. The signal chain is: iPhone —> Lightning to HDMI dongle —> Kanex HDMI Audio de-embedder —> Toslink to APL1 —> Toslink to JL Audio VX800/8i —> Analog Out —> Analog into JL VX800/8i —> Speakers. The JL amp powers the whole system and does volume control. I loop out of the amp and back into it so I can use the extra bank of PEQ’s on the line out channels. Overall it works well and is very reliable.

Here is the REW tuning file for the Miata: http://lumbermansvo.com/APL/Miata/MiataSPL.mdat I made some mistakes while tuning that it’d like to correct, mainly better level matching between the subs/mids/tweets before doing the second round of PEQ.

That's more to my liking although it doesn't have digital input.
I have an unused JL TwK D8 sitting around, I'm curious if it measures up... :D
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
20,048
Likes
23,034
Location
Seattle Area
#7

Tks

Senior Member
Joined
Apr 1, 2019
Messages
382
Likes
327
Location
NYC
#9
Am I the only one who took a glance at the PCB and braced themselves before any of the graphs appeared?
 
Top Bottom