Sounds Devices MixPre-6 vs. Zoom F4

[jcs]

I waited about a year to test the MixPre-6 (for various firmware fixes, e.g. 192kHz issue) and compare to the Zoom F4.

Short summary: I returned the MixPre-6.

Limiters: using a Schoeps CMC641 + Mogami cable, in side-by-side testing vs. the F4's limiters (regular and advanced), the MixPre-6's limiters didn't impress as being any better. I prefer the USBPre2's limiters (more on this later).

Build quality: MixPre-6 is much higher, though the plastic knobs feel cheap compared to the USBPre2.

Usability: The MixPre-6 menu system needs a complete redesign. My opinion comes from personally leading professional software design teams of far more complicated apps which are much easier to use. The F4 by comparison is vastly superior and much easier to navigate and use, despite offering a lot more features and functionality.

Safety: The MixPre-6 has a very powerful headphone amp. Due to the current design, it's possible for 'ear blasting events' when changing settings. Additionally, many state changes cause little clicks which make the device sound cheap (F4 doesn't do this). Audio mixing engines need to constantly fill a buffer which feeds the analog output. Any time that buffer will starve or otherwise cause a click, the analog output should be muted before such changes are started. Any change which can cause hearing damage on the headphone output, the output should be set to zero before the operation starts. For additional safety, a software limiter for the headphones should be added to prevent hearing damage no matter what the device or other inputs generate.

Ambisonic support: this was one of the main reasons in getting the MixPre-6 (need 4 channels of input), with the goal to improve sound over the F4. Unfortunately, ambisonic mode is completely broken- any time the 'analog' limiters engage, random clicks can be heard in both the 4-track and mixed binaural outputs. I quoted 'analog' because it doesn't make sense that clicking/distortion could occur if 100% analog. It's possible that this bug is hardware related to the single device. In the A-direct mode, the live binaural and mixed BIN recording level is way too low. This promotes turning the headphone volume up quite high to hear what's happening. Then, when playing back the recording, the next bug results in hearing damage level output: the full level A-format 4 track mix is sent to the headphone output vs. playing back the low-level binaural file and/or using the binaural filter on the 4-track recording (which is what the F4 does correctly).

Sound quality: the AKM "Velvet Sound" ADC sounds great on paper. However comparing the Schoeps CMC641 (and CMIT5U, Audix SCX-1 HC) and the Zoom F4, showed another problem. While monitoring on headphones, the MixPre-6 sounds a lot better than the F4. However playback of the files on the computer shows a different story. While the F4 recordings sound neutral, detailed, fully textured, and accurate, the MixPre-6 sounds low-mid boosted with cut highs, and worse, a major reduction in detail and texture. I verified this by looking at the FFTs in Audition (Analysis) and Logic X (Match EQ): The MixPre-6 is indeed boosting the low mids and cutting the highs, and is also smoothing the spectral variance, which removes sound texture and creates a 'spectral plastic wrap' shown on the FFT as smoothed spectral peaks (in code: convert time domain to frequency domain via FFT, perform a low pass filter on the spectral bins, and convert back to time domain via iFFT). Is this the "Velvet Sound"? If not some kind of bug, it may work OK for some sounds, however it doesn't work in the general case and an option to turn it off should be added.

Ambisonic sound quality: HRTF binaural mixing: The MixPre-6 is outstanding here- the 3D spatialization is very good, perhaps the best I've heard for a general-use HRTF, even vs. what I can do in post so far! The Sennheiser AMBEO sounds really good, other than the veiled spectrum and cut highs (perhaps can be fixed with a future firmware update?).

Noise floor: The MixPre-6 is really, really great there- very quiet: impressive. However the F4 is so far good enough.

I do like the MixPre-6's low-mid boost to some extent, so I was curious what they were doing so I can match the F4's more textured, non-veiled recording. Here's the Match EQ result that sounds pretty good:


I was a bit puzzled that the MixPre-6 didn't surpass the F4 in all aspects for sound quality. So I did further tests with my (many years old!) Sound Devices USBPre2 (computer recording). The USBPre2 was clearly better than the F4 in all aspects, limiters, detail + texture, and overall sound quality. The MixPre-6 is a step backwards unfortunately- clearly shown on the FFTs and not just via listening. This is not just a possible issue with this particular copy of the MixPre-6: https://www.gearslutz.com/board/high-end/1245013-sound-devices-702-vs-mixpre-3-a.html:

My friend uses a 788T, and also has the MixPre6 and he likes the sound better on the 788T and thinks the MP6 has a softer sound that he doesn't care for. I agree with him, the MP6 seems to be less neutral in tonal balance and somewhat rounded off in detail. I was considering getting one but now not interested.

I had a chance to try out the mixpre 3 and can confirm that the sound is not as great. I think I will just keep the 702.

The sound seemed a bit less full on the mixpre 3. I'm not sure what the proper terminology is...

The USBPre2 has the same audio topology of the 702. So it's looking like one needs to get into the $6K range of gear to match the USBPre2 for 4 channel ambisonics (and it doesn't appear those devices support ambisonic monitoring?).

Physical design: I like the MixPre-6's small size, however the power switch placement, headphone knob, and powering options could use improvement.

In summary, some of these issues could be a single-product defect (manufacturing), and perhaps the sound quality can be fixed with a firmware update (turn off whatever filtering they are doing to veil the spectrum unless it's a 'product differentiation' business decision to protect the higher end devices (even if that's true- turn it off anyway, please. Those devices have many more features)).

I can match the F4 close enough to the sound elements I like from the MixPre-6 and USBPre2 via EQ and the F4 is close enough in noise floor to not be too bummed having to return the MixPre-6. I'm looking forward to testing the new Zoom F6 with unclippable 32-bit float! If I can EQ to match close enough to the USBPre2, that will be a nice workflow. It would be nice if Zoom added some user-selectable 'coloring' options to save on post work (beyond the simple HPF)

 

[Guest]

Fantastic review. Thank you for sharing!

I’ll add my experience with the Mixpre3

As a camera operator first, i find the Mixpre3 to be a good accesory for mirrorless camera rigs. The form factor stacks in front of an anton bauer brick nicely. I have used a tascam dr701, and found that nice, but ultimately enjoyed the mixpre3 more.

After my experience with the Mixpre3, i don’t know if i would use a mixpre3 or mixpre6as a dedicated recorder/mixer for audio, as there are plenty of options that are probably better suited for the task. However, as a hybrid accesory, that can occasionally serve as a dedicsted recorder, the mixpre’s are an interesting solution.

 

[cpreston]

I've had a variety of Sound Devices products including a MixPre3. I never did a direct comparison with the MixPre3, but I am intrigued that it actually colored the sound relative to their other products to a Zoom. Sound Devices cut some corners with those Mixpre preamps to get to that price point. They always touted the low noise floor, but I am guessing it was at the expense of the frequency response. Did it have the same response when recording at line level or very low preamp levels?

 

[jcs]

Thanks James! I really looked forward to getting the MixPre-6, figured it would be better than the USBPre2 after all these years. The bugs and software design issues are surely fixable with a firmware update. The ADC* has selectable "color filters" so perhaps they'll expose those selections in the UI, including an option to turn them all off!

Here's the FFTs showing what "Kashmir / Velvet Sound" looks like recording spoken voice (Schoeps CMC641 + Mogami cable):

MixPre-6 (Kashmir / Velvet Sound):


Zoom F4:


USBPre2:


The FFTs show the MixPre-6 has lost spectral variance, texture, detail, etc., which can never be recovered with any amount of EQ. This kind of filtering might work OK for some sources, however not for the general case, and in any case, should be user switchable since it's highly destructive compared to an accurate reference capture.

The F4 has less overall spectral energy vs. the USBPre2, however with some EQ it can still sound excellent because it retains the original, natural texture, unlike the MixPre-6.

I originally purchased the USBPre2 along with a Tascam DR100 Mark II to feed the digital out of the USBPre2 (and using a USB battery) into the DR100 II along with the Canon 5D3 for camera + separate audio. While it did work, it was way too complicated and the USBPre2 found home for the intended design: as a computer audio device. It does have a major design flaw: no power switch. That would not be a big deal, however it continues to stay active when the computer is in sleep mode (doesn't turn off). I spent quite a long time to source a USB power switch (that also passes data correctly), which doesn't appear to be available anymore if this one wears out (can probably take this one apart and replicate the circuit and make a new one- not what I want to spend my time doing though).

* ADC product info: https://www.akm.com/akm/en/product/datasheet1/?partno=AK5578EN

 

[jcs]

I've had a variety of Sound Devices products including a MixPre3. I never did a direct comparison with the MixPre3, but I am intrigued that it actually colored the sound relative to their other products to a Zoom. Sound Devices cut some corners with those Mixpre preamps to get to that price point. They always touted the low noise floor, but I am guessing it was at the expense of the frequency response. Did it have the same response when recording at line level or very low preamp levels?

These tests were with mic level using the Schoeps CMC641. I heard/saw similar results with the Audix SCX-1 HC and the Ambeo (4-channel tetra mic). Different mics and cables, same results. The FFTs matched what I heard as well as those commenting on Gearslutz.

I know how to (basically) replicate what the MixPre-6 is doing to the sound in software, using the FFT + iFFT and a trivial low-pass IIR filter. However the FFT+iFFT is a very expensive operation, CPU wise (unless implemented as an ASIC etc.). The AKM ADC is a pretty kick-*ss part, not a corner cutting item! So my guess is this is an intentional design decision, vs. cost cutting. Maybe some people like it? Before comparing to the Zoom F4 and USBPre2, the MixPre-6 sounded pretty good, partly due to the very excellent headphone amp. I've only listened on really good Focal and Stax headphones/earspeakers. Will try again with Sony 7506 and AT ATHM50.

The MixPre-6 is very, very quiet- I didn't test this directly, as the Zoom F4 and USBPre2 are quiet enough to not worry about it.

As found on Gearslutz from a post in Jan 2019, others found exactly the same sound coloring, spectral smoothing/plastic wrap for the MixPre series vs. the 7xx series. Hopefully SD can turn off this filtering with a firmware update (or make it an option). At least match the old USBPre2!

In a chat with James offline, I realized that the MixPre-6 does have some mojo, and the coloring is cool to a point, which makes you want to use it (emotion), however that's mostly from the EQ, which I can replicate in post with any decent recording. If SD can fix all those bugs/design issues (firmware), and turn off the filtering (make optional), it will be a very nice product.

In the meantime, really looking forward to the Zoom F6!

 

[Throwback]

The FFTs matched what I heard as well as those commenting on Gearslutz.

There has been quite a variety of opinion on Gearslutz. As with such forums, not all comments carry the same weight, of course. For example, I was interested to read the thread started by the renowned classical music recording engineer - Tony Faulkner - who, despite having a frustrating time with some problems with the functionality of his Mixpre-10T (bugs long since fixed), thought its sound was very good: https://www.gearslutz.com/board/rem...cording/1211879-sound-devices-mixpre-10t.html

Personally, I have found my Mixpre-3 up to the mark for all I can throw at it (speech and music) during heavy use over the last two years, and have found the interface straightforward (clearly, matters of opinion only). Also, I have found the diminutive size and the ability to overdub (using the Musician plug-in) invaluable: I have been filming and recording a Playing for Change-style series with musicians adding their parts in different locations. The overdubbing is rare (unique?) among field-recorders of this form: it would be nice if Zoom (with their experience of the feature in other ranges of equipment) added this functionality to their F series (certainly I would think seriously about a Zoom F8n - or a similar successor - with such functionality).

Returning to your FFT analysis, presumably you have discussed your findings with SD: what was their response? I have always found them very approachable and, indeed, very responsive to an early bug I picked up in relation to MS and stereo-linking.

Cheers,

Roland

 

[Noiz2]

Interesting thread. The filtering makes some sense for dialog especially for a product aimed at low budget filmmaking. It would tend to minimise some of the problems inexperienced sound people often make, but I completely agree that it should be something that can be turned off.

As a twist a good friend who is a big budget recordist got one of the earlier SD recorders and absolutely hated the sound. She thought it was too "digital and sterile", so she might actually like this EQ.

I had always disliked the SD menu system. I find it frustrating to plow through and very counter intuitive.

 

[Run&Gun]

Interesting thread. The filtering makes some sense for dialog especially for a product aimed at low budget filmmaking. It would tend to minimise some of the problems inexperienced sound people often make, but I completely agree that it should be something that can be turned off.

As a twist a good friend who is a big budget recordist got one of the earlier SD recorders and absolutely hated the sound. She thought it was too "digital and sterile", so she might actually like this EQ.

I had always disliked the SD menu system. I find it frustrating to plow through and very counter intuitive.

I was thinking something similar...

 

[Throwback]

the MixPre-6 sounds low-mid boosted with cut highs

Just following up on the EQ issue, this, of course, suggests something very different from the specs provided by Sound Devices - i.e. 10Hz to 40kHz +/- 0.5dB re 1kHz @ 96kHz sample rate (Mixpre-3) and 10Hz to 80kHz +/- 0.5dB re 1kHz @ 192kHz sample rate (Mixpre-6 and 10T). Which seems a little surprising.

Cheers,

Roland

 

[jcs]

Just following up on the EQ issue, this, of course, suggests something very different from the specs provided by Sound Devices - i.e. 10Hz to 40kHz +/- 0.5dB re 1kHz @ 96kHz sample rate (Mixpre-3) and 10Hz to 80kHz +/- 0.5dB re 1kHz @ 192kHz sample rate (Mixpre-6 and 10T). Which seems a little surprising.

Cheers,

Roland

This is one of the videos that got me to purchase the MixPre-6 (figuring it would sound similar):

See the comments: same pattern- boosted low-mids and reduced highs = "sounds better". I agree in this case it does, and as noted by the topmost review, the F8 can be EQ'd to match. Ideally they'd give us the option for accuracy with color/flavor as optional EQs, even better, general EQ/filters (even plugins), on the device to save time in post (we've seen similar patterns with cameras and picture styles, LUTs).

Low noise, accurate, reliable, good power options & battery life, and easy to use as first priorities. Then options for EQ/filters, with nothing permanently baked in (as apparently with the MixPre).

 

[jcs]

Interesting thread. The filtering makes some sense for dialog especially for a product aimed at low budget filmmaking. It would tend to minimise some of the problems inexperienced sound people often make, but I completely agree that it should be something that can be turned off.

As a twist a good friend who is a big budget recordist got one of the earlier SD recorders and absolutely hated the sound. She thought it was too "digital and sterile", so she might actually like this EQ.

I had always disliked the SD menu system. I find it frustrating to plow through and very counter intuitive.

Curious what your friend prefers that sounds better? The current SD's have great headphone amps which improves perceived quality on the recorder- perhaps the earlier devices didn't? Or perhaps back then SD hadn't added their current low-mid boost high cut EQ?

From https://www.gearslutz.com/board/rem...anything-better-than-sound-devices-722-a.html (2006):

Nagra V - http://www.nagraaudio.com/pro/index.php
Zaxcom Deva - http://www.zaxcom.com/
Aaton Cantar - https://www.aaton.com/cantar-x3/overview
Sonosax MiniR82 - http://www.sonosax.ch/

 

[jcs]

Looking at the ADC spec sheet https://www.akm.com/akm/en/product/datasheet1/?partno=AK5578EN :

　● Sound Color 　Four Types Digital Low-pass Filters
　- Short Delay Sharp Roll-off, GD=5/fs
　- Short Delay Slow Roll-off, GD=5/fs
　- Sharp Roll-off
　- Slow Roll-off
Digital High-pass Filters

● VELVET SOUND Technology Low-distortion Technology

The Sound Color filters are only listed as LPF/HPF: the spectral smoothing & EQ appear to be SD features ("Kashmir"?) and not part of the "Velvet Sound".

 

[Noiz2]

Curious what your friend prefers that sounds better? The current SD's have great headphone amps which improves perceived quality on the recorder- perhaps the earlier devices didn't? Or perhaps back then SD hadn't added their current low-mid boost high cut EQ?

From https://www.gearslutz.com/board/rem...anything-better-than-sound-devices-722-a.html (2006):

Nagra V - http://www.nagraaudio.com/pro/index.php
Zaxcom Deva - http://www.zaxcom.com/
Aaton Cantar - https://www.aaton.com/cantar-x3/overview
Sonosax MiniR82 - http://www.sonosax.ch/

What she really likes is the Nagra RtR using Nagramaster and still uses one from time to time. RtR though is just not practical most of the time. She liked the sound of the Zoom H2, which she felt was more "natural" than the SD.

 

[Throwback]

the spectral smoothing & EQ appear to be SD features ("Kashmir"?) and not part of the "Velvet Sound".

I'm still interested/puzzled as to why the EQ you have noted isn't reflected in the SD spec, which shows much better linearity: Sound Devices aren't noted for publishing suspect specs, so am I missing something or what? And, as I mentioned earlier, how much discussion have you had with SD? It would interesting to understand more.

Cheers,

Roland

 

[jcs]

I'm still interested/puzzled as to why the EQ you have noted isn't reflected in the SD spec, which shows much better linearity: Sound Devices aren't noted for publishing suspect specs, so am I missing something or what? And, as I mentioned earlier, how much discussion have you had with SD? It would interesting to understand more.

Cheers,

Roland

SD was informed of all these findings. I provided the serial number of the unit in case some of the issues are isolated (manufacturing). I spent many hours debugging and trying to figure out what was going on- I look forward to hearing back from them!

Since others have noticed the EQ and reduction of detail/texture (I'd call it spectral smoothing / spectral plastic wrap since it's not just a low pass filter effect (ribbon mic sim?)), it doesn't seem like this is isolated to that particular unit. I do like the MP6 EQ vs. the Zoom F4, and the F4 sounds better when EQ'd the same way (as noted earlier in this thread). Perhaps some people like that spectral-smoothed sound; all good, just prefer to be able to turn it off since it can't be changed in post (as with EQ). Not clear why this isn't noted in their specs, seems like it's part of their "secret sauce"?

Without an accurate known reference device, it would also be fair to say that the Zoom F4/F8 have low bass-mid and accentuated highs relative to the SD MixPre; typically called thin vs. fat sound.

 

[jcs]

What she really likes is the Nagra RtR using Nagramaster and still uses one from time to time. RtR though is just not practical most of the time. She liked the sound of the Zoom H2, which she felt was more "natural" than the SD.

That's pretty cool: https://www.dandugan.com/Nagra_Master_Article.html . More discussion about velvety sound and high-rolloff. Yeah reel to reel tape has some limitations vs. the tiny recorders we have today.

Since analog systems (tape, tube, etc.) are so popular, you'd think digital recorder makers would implement accurate filters to replicate them at the recorder level. It's possible to emulate/simulate pretty much anything now, from mics, to amps, to speakers and analog effects: https://www.fractalaudio.com/ . More sales: sell 'em as plugins for the recorder, or open up to third parties and take a cut (as with phone apps).

Each time I've implemented real-time audio systems (PC/Mac/Phone), it so nice (fun!) to have real-time monitoring and recording of filters & effects. Having a fat-tube-tape sound live on an SD/Zoom type device would be very cool! Then you could feed that to the camera and if levels set correctly, not have to do any audio post. It's certainly possible to put analog fx between the output and into the camera- that's a lot of complexity (though could work for things we shoot in the studio on a tripod). For live streaming, OBS supports VST filters, so that's another option to get analog style fx into the mix (and of course anything goes in post).

Or build this kind of system into the recorders: https://townsendlabs.com/products/sphere-l22/ so it's possible to live monitor with the target sound, then either bake it in to save time (or when live streaming), or allow tweaking the raw audio in post (using the same audio process).

 

[Eric Coughlin]

I have a MixPre-6 and don't like its lack of features. I'm considering a Sound Devices 633 or Zaxcom Nova, but not sure I want to spend that kind of money since audio isn't my main gig. Would be nice to have, though.

 

[klk]

I have a MixPre-6 and don't like its lack of features. I'm considering a Sound Devices 633 or Zaxcom Nova, but not sure I want to spend that kind of money since audio isn't my main gig. Would be nice to have, though.

+1 on this Eric. I have the MixPre-6. Used to have a 552, but that was a PITA (ie: big and heavy) for what I used it for. Something in-between a 633 and MixPre-6 would fit the bill better.

 

[Noiz2]

That's pretty cool: https://www.dandugan.com/Nagra_Master_Article.html . More discussion about velvety sound and high-rolloff. Yeah reel to reel tape has some limitations vs. the tiny recorders we have today.

Since analog systems (tape, tube, etc.) are so popular, you'd think digital recorder makers would implement accurate filters to replicate them at the recorder level. It's possible to emulate/simulate pretty much anything now, from mics, to amps, to speakers and analog effects: https://www.fractalaudio.com/ . More sales: sell 'em as plugins for the recorder, or open up to third parties and take a cut (as with phone apps).
...

My initial reaction was yes I would like that, but then I would probably never use it. I can see it as a future option on consumer recorders, because of the instant wow factor though. If I am doing the recording though I generally want the sound as "pure" as my gear can record it. If that is how it's being recorded that is the way I want to hear it. That said I am not the majority of consumers, and my "customers" are not going to want preprocessed sounds, generally.

Where I can see it in the future though is when there is a "raw" audio format like there is for images. I might well record with a filter I thought would enhance the sound if down the line it was easy to non destructively undo that.

 

[jcs]

My initial reaction was yes I would like that, but then I would probably never use it. I can see it as a future option on consumer recorders, because of the instant wow factor though. If I am doing the recording though I generally want the sound as "pure" as my gear can record it. If that is how it's being recorded that is the way I want to hear it. That said I am not the majority of consumers, and my "customers" are not going to want preprocessed sounds, generally.

Where I can see it in the future though is when there is a "raw" audio format like there is for images. I might well record with a filter I thought would enhance the sound if down the line it was easy to non destructively undo that.

I thought of this after testing the MixPre-6 vs. the F4: the MP6 just sounds better live (EQ + better headphone amp). I've been doing tests on the computer with Logic X using a variety of effects which run with very low latency- I think it enhances the vocal performance when it sounds better / cooler / fatter / <insert favorite audio adjective here>. Same reason we switch mics & mic position for different speakers / singers: just sounds better live: helps the performance IMO.

To your point, they could offer live EQ + filters on the device, then record raw (or raw + cooked), and offer VST/AU/RTAS plugins for post. Even store metadata so plugins in post can read and match custom settings.

I heard back from SD tech support- they are open to working on the suggested changes; hopefully we'll see (hear!) them sooner vs. later- really think the MixPre can be a great tool for us when they do.