TOPIC:

I recently purchased three new (black case) OpenDRC-DA machines, which arrived last week. I’ve owned an original (silver case) OpenDRC-DI digital machine for a few years, so I’m already familiar with these devices, (not a newbie!) and my first job was to run some basic test signals through the new machines to check they worked okay, and hopefully measure how good the built-in D/A converters were…



Very early on, I discovered several annoying problems with them NOT working properly, and I’ve now spent several further days exhaustively testing and re-testing, checking, double-checking, and triple-checking my results through each machine, (before contacting MiniDSP or posting on forum), and trying everything possible to isolate the cause of each problem, which I now feel certain is faulty DA-FP boards. Looks like I’ll probably need to send them all back to Hong Kong…!

In summary…



I’m posting here on forum in case any others have experienced similar issues?
I’d be especially interested to know serial numbers of faulty devices, and how common these faults are?

In greater detail, and scrutiny, explaining my test methodology…

1st PROBLEM
Machine #204-1017 does not recognize any input from AES/EBU source.

And yet that machine does happily recognize S/PDIF coaxial or optical inputs, and displays live signal levels on plug-in’s audio meters, and gives strong analogue output from D/A converters. Convolution FIR process also works okay, and front panel controls work. Therefore I don’t think there’s any problem with MiniSHARC board or VOL-FP board, as they can be shown to work as normal.

Yet switch back to AES/EBU source input and the sound goes silent… a very faint hiss flickering between –125dBFS and –126dBFS is all that shows on the plug-in’s input and output level meters, and that’s merely the dither noise floor from the input asynchronous Sample Rate Converter, which is always there anyway. But no audio signal getting through.



The source digital signal (1kHz sinewave generated from digital mixer) is routed through my Apogee PSX100, so I know I’m feeding a very stable, rock-solid (<22 picosec jitter) re-clocked 24 bit 48kHz signal into each OpenDRC, which should be easy to handle, and I can see “48kHz” sample rate “lock” LED and level meters on the Apogee telling me there’s healthy signal going through, and it’s audible from Apogee’s D/A outputs too. I’ve owned this Apogee about 15 years and so I know it very well to trust what it’s doing. It can output three simultaneous AES/EBU stereo connections, plus simultaneous coaxial S/PDIF and optical S/PDIF (or ADAT) formats so I can quickly swap between all three OpenDRC inputs, whichever I want without interrupting source.

Plugging the very same XLR cable (AES/EBU) source into my two other OpenDRC-DA machines, both do instantly recognize that signal when set to AES/EBU input, so it’s only one machine (#204-1017) that suffers this specific problem.

I’ve done total factory reset on all four configs on every machine, using them as a blank D/A converter only. I tried re-installing the v2.17 firmware, etc. but that didn’t make any difference.

I then unscrewed the case of #204-1017, checked the internal wiring (all fine) with multimeter, and very closely looked at the XLR pins from the AES/EBU chassis connector through to the circuit board, but they’re all okay, good contacts, and signal is getting through from the XLR cable to the circuit board pins okay (checked with multimeter) and there’s no dry solder joints or anything physically wrong I can see.

I then disconnected the faulty DA-FP board, and connected up one of my spare DIGI-FP boards to machine #204-1017 and sure enough, the AES/EBU input coming from that DIGI board works fine with the machine, further proving that the MiniSHARC main board and VOL-FP boards and internal wiring are all okay. The fault must be with the DA-FP board only. I reassembled it all again, with original DA-FP, and it still doesn’t work, as before. So I continued my D/A audio quality testing using the Coaxial S/PDIF input instead for that machine.

2nd PROBLEM
For all three machines, analogue XLR output polarity is wrong.

Analogue D/A polarity is inverted / out-of-phase / reversed / opposite with respect to the original polarity of the source digital input on AES/EBU or S/PDIF or Optical.

This problem affects ALL THREE of my brand new machines! – and rather implies to me that it’s likely all DA-FP boards have been manufactured with this error at the factory by MiniDSP. That’s quite alarming on a machine whose sole purpose is to deliver correct impulse response / phase and freq response behaviour for loudspeaker systems!

Having your main outputs of OpenDRC-DA machine hardwired wrong is really bad, because it introduces a systematic error on that machine which is inconsistent with the OpenDRC-DI machine whose output polarity does remain correct. (Or inconsistent with the very same machine if you swap its output boards – because in fact, I also purchased three DIGI-FP boards along with these three machines so I could swap between analogue or digital outputs as needed – awkward, but the only possible way at present with OpenDRC hardware.)
Whilst obviously not everybody owns both versions, it’s a problem. It means any setup I’ve saved on one machine version is not the same thing when loaded into the other machine version! If I swap boards over, I’d have to reprogram all my presets too.

Some clever people might say, “You can just click INVERT button in plug-in to fix that”, which maybe true, but things can get way too complicated…
I’ve got 3x machines, with 2x channels each, and 4x config presets each (total 24 button settings I need to double-check!) but it gets really confusing if I’m already using those “Invert” button settings as integral part of my actual preset. eg. Running tests where I want to deliberately polarity-invert and cancel one channel against another to null-test if two filters, IIR vs FIR are giving the same response or not?
Or if creating certain crossover splits across channels which require specific inversion to work (eg. Butterworth 18dB/oct crossover will sum to flat amplitude either in-phase or out-of-phase (but with different allpass curve) whereas most won’t, and many crossovers, eg. Linkwitz-Riley 12dB/oct require out-of-phase summation to get flat amplitude.)

Going painstakingly through every preset in my library and manually editing them one-by-one is time-consuming and increases the chances for human error and wrong settings by mistake. And I don’t want to change them at all, since they’re “correct”.

Also, besides pressing the “Invert” button, you could equally decide to generate your FIR filter with inverted polarity, or even solder up some special XLR cables to swap polarity (I’ve got some cables already with ground-lift) but it’s all a pain in the neck to remember everything and adds to the mental stress of knowing for certain if it’s right or wrong. Anything extra that casts doubt is bad news to me…

Anyway, I really shouldn’t need to justify why having polarity correct is better - it is and that’s that! The output board ought to be wired correctly, and it isn’t. I think MiniDSP ought to do something to remedy this!

Polarity Test Methodology.

How can I know, beyond doubt, that I am measuring it correctly, and the DA-FP polarity is “wrong” – maybe it’s my test equipment that’s wrong?!

Hmm… This is a good philosophical question. Also, before I purchased in Nov 2015, surely lots of other customers before me must have purchased DA-FP boards, or OpenDRC-DA machines, or DiracDDRC-22DA machines? They all share the same DA-FP board design. Why haven’t these other people reported output polarity faults on the forum already? Does this issue affect every DA-FP board ever made? Is it just a recent batch (serial numbers???) or is it too difficult or rare for people to bother checking absolute polarity, and statistically more likely that I’d receive three brand new machines in a row, with consecutive serial numbers, all with exactly the same fault? No, I believe it’s not a glitch, it’s likely an inherent product flaw and my three boards and probably everybody else’s all suffer the same problem.

I use a simple test WAV file generated by software. It looks like this…



“3 up, 2 down.”
Three positive spikes (each spike is one single sample at 100% fullscale) and two negative spikes (single sample at –100% fullscale) with two seconds of silence after each, making them easy to see. Upwards positive spike above zero axis of waveform are positive polarity in digital WAV file format, by definition, as with first 3 impulse spikes, and the 2 downwards spikes following are negative polarity, by definition, in WAV file format, as seen here.

There’s various ways to play it back, besides using computer USB interface…
16 bit 44.1kHz version of wav can be burned to CDR for easy playback.
24 bit 48kHz (or 96kHz) version could be burned to DVD+R disc in DVD-Video format with 24 bit 48kHz linear PCM audio soundtrack using LPLEX , and it should playback 24 bit 48kHz bit-for-bit perfect from nearly any DVD player or Blu-ray player with S/PDIF output (but check your audio settings in menu!)
Personally, I’m using Tascam MX2424 multitrack 24 bit recorder to playback and record the test signals in question.

I played the same exact test WAV (3 up 2 down) digital file through numerous different hardware machines with D/A converter to analogue out (preferably balanced XLR, or otherwise jack or phono RCA, via Neutrik XLR adaptor) and recorded single channel outputs into the same exact A/D converter input for reference. (Apogee PSX100 converter A/D recorded 24 bit 48kHz to Tascam MX2424 recorder.)
Compare all the recorded WAV files on-screen in software afterwards.
If polarity is correct they should look the same as original “3 up 2 down”, and if polarity were inverted they would appear upside down. ie. “3 down 2 up” which is very easy to spot as being wrong polarity.



The evidence of proof comes from the overwhelming majority consensus of output polarity, among numerous other machines, from numerous different manufacturers, from different countries all over the world, from American manufacturers (Apogee, Alesis…), from European manufacturers (Behringer, Focusrite…), from Japanese manufacturers (Tascam, Yamaha, Sony, Denon, Panasonic, Pioneer…), where they all collectively agree, re-affirm and corroborate each other’s positive results. They cover the cross-section from professional studio to consumer hi-fi gear, from expensive to cheap, from 20 year old machines to very modern equipment, etc. They can’t all be wrong. They represent an objective universal logic.
It seems in fact every piece of studio equipment I own is wired to the international standard XLR pin-2 hot convention of the AES society, and this is true not just for digital gear with converters, but for pure analogue devices like microphones, preamps and compressors too.

Given this overwhelming consensus, under identical test conditions the OpenDRC-DA tests negative, going against the majority convention, which proves beyond doubt to me, that it’s output polarity is incorrect.

3rd PROBLEM

Left channel of machine #204-1019 is approx –4.47dB quieter than Right channel due to gross waveform corruption.

This is a particularly tricky problem! – I still don’t understand any reason it occurs…

I’ve measured various test signal sinewave sweeps, pink noise, etc. through each machine, and often noticed the peak level on left channel was weaker than right channel on one particular machine (#204-1019) only, by several dB, whereas the other two machines were channel matched very well, +/-0.05dB difference typically.

Because most of my test signals were fullscale sine sweeps, etc. at first, I thought there was a fixed overall level difference, which is bad enough(!), but it seemed to show more or less difference with some signals than others, and I wondered if it was an intermittent fault? Upon close inspection, where the discrepancy is most obvious, the Left channel waveform appears as intact, correct matched volume, only for lower levels, up to about two thirds of full height and then at louder levels, the waveform hits a weird ceiling threshold, where it won’t go any higher (like a limiter) and changes direction to a crudely distorted “mirror image” of what the peak crest should look like!!! Very strange indeed ?!! A picture is worth a thousand words here to better expain…



Upper trace is the left channel XLR analogue output, lower trace is the right channel XLR analogue output (not affected) and the test signal was a computer generated 1kHz sinewave, pre-edited with abrupt 6dB changes in amplitude, on purpose, so you can observe the strange divide in behaviour of test machine’s output between the louder section on left half of screen where left channel recorded waveform is corrupted, and the 6dB quieter section on right half of screen which is reproduced okay on both left and right channels. Although strictly, both left and right traces show absolute polarity inverted compared to my original source WAV test file (not pictured here.)

This was recorded directly from XLR outputs of OpenDRC-DA machine #204-1019 after total factory bypass and Sync, all settings blank, using the box as a straightforward D/A converter. The left and right channels were recorded separately in turn, into the exact same A/D input channel of Apogee converter via same identical XLR cable, nothing changed at all. The recordings are digitally synchronized together because OpenDRC was the 48kHz master clock to the Tascam MX2424 slave recorder and playback, and both WAV files have been opened side by side on the same screen for comparison.

I have repeated tests on this machine several times, using various types of audio signal - pink noise, square wave, sine wave, steady tones, sweep tones, impulse spikes, music program, and seen these same corrupted characteristics in many examples, but only from the left channel of machine #204-1019. The sine wave is the most obvious shape to show the fault clearly.
Measuring overall level Left vs Right channel balance with fullscale 1kHz sinewave the RMS level difference is –4.47dB. But you get different dB values using different types of music / audio signals depending on their duty cycle of how often they go above the “corruption threshold.”

If anybody can explain what’s causing this phenomenon, I’d love to know???!!!

Seems to me, whatever the reason, there’s certainly another gross fault with this DA-FP board, besides being inverted polarity too!

Swapping out this board for the DIGI-FP board into this machine #204-1019 totally eliminated the problem. Correct linear output, in correct polarity. The fault is probably something to do with the DA-FP board, and occurs through either AES/EBU input or S/PDIF coaxial input or Optical input. The 2x2 Plug-in’s level meters were showing fullscale healthy level meters even when the output is –4.47dB down on left channel, so fault must be occurring later, within the output D/A converter chip section, I think.

CONCLUSION

Obviously, so far I’m not very impressed with the DA-FP boards, because I’ve got three different kinds of serious fault present, two boards each with two faults, and a 100% reject failure rate overall. However, my confidence in MiniDSP as a company and the OpenDRC platform as a great value solution isn’t really any worse, and I believe MiniDSP will solve these issues it in due course and I’ll soon get some new boards that work. The OpenDRC machines work fine with the DIGI-FP boards I’ve tried.

For what it’s worth, I did try to measure basic D/A converter performance spec…

OpenDRC-DA machine..................... #204-1018
digital I/O noise floor........................... –126.07dBFS due to asynchronous S.R.C. dither noise.
analogue XLR noise floor.................. –96.23dBFS average RMS below fullscale*
*as recorded through Apogee A/D converter!!!
Subtracting Apogee’s noise floor of –116dBFS would leave a calculated –97.17dBFS residual noise floor due to the OpenDRC alone.

Transfer function, amplitude and phase frequency response of D/A converter (right channel shown) in HOLM. Remember DA-FP output polarity is inverted, and this fault is clearly visible in HOLM’s dotted phase trace and impulse response shapes.

Thank you, Richard, for your revealing investigation.

I'm also an owner of a DA version of the openDRC, and also suffered from the noise floor of the analogue section of DA-FP, which sounds like a little radio noise - independently from any power source upgrade.

Another annoying topic is the polarity inversion when you use two different lines (one DA and one other FP board) in parallel. The fault drove me mad for some hours ...
As an ambitious customer I would expect Devteam or miniDSP to provide some solutions since this is clearly a manufacturing fault, e.g. sending the DA-owners an hardware polarity changing adapter? Or change the DA-FP board through their distributors? Or at least allowing and providing a "how to" advice where to change and solder the corrected connection?

(Changing the polarity in the plug-in software or doing nothing is only for single box users, but that is not the point).

Hi Planetti,

Ah! So you've got exactly the same polarity inversion fault with your DA-FP board? That's very interesting. At least I'm not alone then.

May I ask how old is your device, or what serial number? Mine were all ordered very recently Nov 2015.
If you purchased yours over a year ago, for instance, then I guess probably all of them have the same manufacturing fault, rather than just a recent batch.

Did you also report the polarity fault to DevTeam when you discovered it yourself previously?

DevTeam have told me they are investigating the polarity fault, and it may be possible they could fix it with a firmware update? but since this fault affects both OpenDRC series and Dirac Series with the same DA-FP board, hopefully it's a high priority.
Maybe any other customers who own DA-FP could also please reply here if they've noticed the polarity fault or not.

I totally agree with you that it's annoying, and a workaround solution isn't really as good as a proper fix where you know you can rely on the output phase being correct - especially for a machine that's designed to correct phase response!

Anyway, hopefully MiniDSP will publish a firmware update fix fairly soon. Watch this space...

Richard wrote:

May I ask how old is your device, or what serial number?

My openDRCs are from Dec 2014, but I'm too lazy to look all numbers up ...

Richard wrote:

Did you also report the polarity fault to DevTeam when you discovered it yourself previously?

No, I didn't realise that it was the DA-FP. I accepted the surreal results and adapted the delay to the real XO situation.

Guys, very interesting discussion (although somewhat alarming form a quality control perspective) !

The manual for the Open DRC-AN clearly states that:

Analog Audio inputs
Balanced Audio connectivity on XLR Neutrik connector
Pin 1 = Shield / Pin 2 = Hot / Pin 3 = Cold ADC performance: 114dB SNR
Analog Audio outputs
Balanced Audio connectivity on XLR Neutrik connector
Pin 1 = Shield / Pin 2 = Hot / Pin 3 = Cold DAC performance: 114dB SNR

This is consistent with the AES-convention, although the term hot generally describes for the non-inverting input, and cold the inverting input.

So as I read your post the polarities have been switched: pin 2 in inverting and pin 3 is non-inverting (which is is the exact opposite from the AES-convention). This very annoying even though the polarity can inverted by software.

It seems to me that a mixed-up has occurred in either the design and/or the the layout of the boards, since both new and old boards suffer from it. I know from my own experiences with designing PCB-layouts that you're not allowed to connect components together that aren't connected in the schematic. So my guess is there's been a mix-up in the schematic.

But that makes me wonder about the quality control of the designs. A company capable of designing DSP systems should be able to design a relatively simple balanced I/O board. Anyway, mix-ups do happen and we're all human (last time I checked).

Let's hope that devteam comes up with a solution and a proper reply to this thread. I would like to purchase a DRC-AN really soon for one of my projects and I would feel more re-assured if MiniDSP proved that they were handling this problem with the care and attention it deserves.

Xcel

Hi Xcel,

Yes, I totally agree.
I'm sure MiniDSP will want to fix any problems for us all ASAP, but in the meantime it would be nice if they replied to offically acknowledge the polarity fault's existence. Surely they must have a DA-FP board in Hong Kong factory they can test this with? Just to know they've found the same thing too.

Richard tested...
1x DIGI-FP board inside OpenDRC-DI serial no #202-2050 purchased Jan 2013 = POLARITY CORRECT
3x DIGI-FP digital I/O boards purchased separately Nov 2015 = POLARITY CORRECT
3x DA-FP analogue out boards inside 3x OpenDRC-DA serial no #204-1017,-1018,-1019, purchased Nov 2015 = POLARITY INVERTED

Planetti tested...
1x DA-FP analogue out board inside OpenDRC-DA, purchased Dec 2014 = POLARITY INVERTED

Clearly both old and new DIGI-FP digital boards are okay. No worries there.
Clearly both old and new DA-FP analogue out boards are faulty, with the same polarity error.

Don't know about the AN-FP board (I don't own any) but as long as both input AND output, are both pin-2 hot, or both pin-3 hot, then it should work consistently.

Nobody really wants to have to post their boards back to Hong Kong if a simple firmware update can be download to fix this issue in a few minutes. I'm sure MiniDSP doesn't really want that either!

The miniSHARC board hex firmware, or the 2x2 plugin software is all we can upgrade as end users, so that's the best place to start.

If the firmware can somehow automatically check if it's attached to a DIGI-FP or a DA-FP board, then it could hopefully reset the polarity default appropriately, BUT if it has no way of automatically detecting which attached board exists and making any intelligent choice, then MiniDSP could simply publish a second alternate version of the firmware with the "DA polarity correction" suffix and you simply install that firmware version if you're using DA-FP board or install the original firmware version if you have DIGI-FP boards. If you swap boards at any point, you just need to reinstall the firmware you need, and it says "v2.7DA" or "v2.7DI" in the window to distinguish them.
They'll probably need to release two versions likewise for any future updates as well, unless they implement an even better solution.

The manual does state pin-2 hot, you're right.
The 114dB signal to noise ratio is most likely from the chipset spec. I measured about -97dBFS noise floor on my DA board, as I wrote in an earlier post here somewhere.

Some companies do deviate from the pin-2 hot convention, but very very few these days.
JBL loudspeakers use to move cone backwards with +ve voltage on the red terminal! Everyone else goes forwards. I think they refused to swap it for a while because of their proud tradition, but eventually changed it to avoid problems.
Alesis ADAT recorders from the 1990's famously had polarity inversion on analogue jacks but not on digital I/O which later machines didn't, causing confusion when switching old tapes. Any other famous examples people can think of...?

I expect with the Dirac series OpenDDRC-22DA, the software is automatically generating the required polarity correction as part of the output filter. I wonder how many people are looking at their filters and thinking that looks a bit weird and investigating...?

Dear all,

Sorry for the lack of reply on this, as we've mentioned to Richard in his tech support ticket, the whole team has been slammed with work hence didn't get a chance to look into this story for the past few days. We had it on our todo list, but anybody who's worked in an engineering department for 1day might be able to relate that sometimes, things aren't just a 2min task to understand the situation. So as much as we would have like to hypothesize and write long explanations/ideas, for us to answer this post, we needed to take the time to understand what was going on The QC issue is obviously a concern and we've talked to Richard here. We're trying to understand what may have happened.

Today, few engineers were able to finalize one large project and did have a look into the cause of that polarity inversion. The bad news is that we can indeed see it and that we missed it. We're human and sometimes S*&$#t happens unfortunately, especially at a bigger chance if you have many boards, many configuration, many people each wanting something different and miniDSP pushing the enveloppe to make it happen. Bug free world would be great, sure, but maybe not as fun.. The good news is that it's something that we can fix with a firmware for the miniSHARC series. So no need to worry, the world is not going to end..

We will keep everybody updated when we have something in terms of firmware + plugin.
For the time being, whoever believe that they can throwaway their DA-FP/OpenDRC-DA maybe worth to remember that it's a simple click of the invert button as a work around to make it work again for the time being as we're working to finalize the fix.. just saying ..

DevTeam

Can you point me to the invert button? I haven't found any mention of it in the manual...

Thanks!

@devteam,

Thanks for the reply, as a technical project manager I totally understand your situation ! Glad to hear though the team is on top of this thing, I trust in your abilities to fix this.

This won't keep me from ordering a OpenDRC-AN anymore !

Looking forward to hearing from you guys (and galls) again.

Xcel

Griffith, you can find it on the opening screen of the OpenDRC 2x2 plugin, in the outputs section (right side of the window). It's beneath the yellow Xover button and the Mute button.

Greetings,

Xcel

Thank you! I might have gotten confused here, the unit I ordered is the DDRC-22DA with Dirac Live. Is that unit also affected by the reverse polarity?

Cheers
Griffith