RME UFX modifications

There’s a lot of mystique, anxiety, snake oil, and misdirection with audio modifications. I’ve purchased a few modified units over the years, and had some gear modified (most of which I’ve now sold) by the very reputable companies that do this work, and I have to say, I’m always a bit irked by the rather euphemistic language used to proclaim proprietary knowledge about methods, ‘special’ techniques, secret knowledge, and so on. None of this stuff is really so secret, just trawl DIY Audio and gearslutz for a few minutes (okay, more than a few minutes––euphemisms and snake oil are big over there too). Big companies (like RME, M-Audio, Presonus, etc.) cut corners so they can stay competitive in an overwhelmingly pro-sumer dominated market, and the subtle differences which make one interface sound better than another mostly go unnoticed by the average bedroom singer-songwriter, or the FOH engineer who wants to be certain something, anything comes out the PA system when the music starts. These companies spend their money on ergonomics and driver/software development, and shave off costs on components so they can sell this stuff at a reasonable price point, make a profit, and invest in new lines of products which suit the mode-du-jour. It’s sad and wasteful, but pure capitalistic euphoria. And for the most part, all this stuff is pretty good and decently designed––the EE knowledge base is enormous, as is the industry, and very smart people are designing this equipment. So if someone says audio modification will do nothing more than simply void your warrantee, they are 95% correct, considering the taste of the marketplace, hence the mystique, anxiety, snake oil, and misdirection to convince you to dish out a few hundred more bucks for a major audio modification. Plus, it’s pretty hard to convince someone to send their $2000 piece of gear to a stranger to have its build quality compromised for some vague promise of ‘better sound’ which no end-user listener with crappy earbuds and a smartphone DAC streaming sub-MP3 quality audio will even hear, let alone appreciate.

But I digress, since I was about to demonstrate how I went about willfully compromising the build quality of a perfectly functional RME UFX for ‘better sound’. I got this used on Ebay relatively cheap with a misbehaving PSU. Here’s a before photo, having joyfully broken the warranty void seal:

IMG_0103
Clockwise from top: input section, output section, headphone drivers, and microphone preamps

Including my iPad notes on cap values, proposed modifications, etc:

IMG_0106
Two mic preamp sections and inst. input (using everyone’s favorite, the TL072)

I was doing some sort of color-coding scheme which I later gave up, but couldn’t revert the image. You’ll notice there are capacitors between the output/input of the DAC/ADC chips and their buffer stages, and that’s because the PCM4104 and CS5536 operate with a +2.5V bias. Can’t really eliminate caps altogether, unfortunately, since that’s what I would rather do…

IMG_0105
Output stage, with my stupid color coding. All NJM4580.
IMG_0107
Headphone driver stage, with my modification plans
IMG_0104
Input stage: takes differential input at TRS plugs and creates single-ended signal, then there’s a gain stage, then that single-ended signal is inverted in NJM2043, and that inverted signal is inverted again in NJM4580, producing differential input to CS5536.
IMG_0101
Digital board and capacitor values. Found out later the Reverb/Echo isn’t working (EQ and Dynamics are fine, however), which is most likely a problem with RAM chip (oblong chip next to blue capacitors). I replaced it (that was fun! [not]), to no avail, so probably DSP chip is also wonky, or bad connection… above my pay grade, for now. Also means DUREC is also not working, which is the real bummer.
IMG_0100
The PSU, which often die in these units. Here are the values. ALSO, 4.7uF, 50V is wrong! It’s actually 47uF. PSU wouldn’t start up until I realized I had a wrong decimal point…
IMG_0110
The whole shebang!

 

So now for the modifications. I had a ton of 100uF 63V Panasonic FR on hand. Yes, that is too big for inter-stage coupling, and barely fits, plus I bypassed them with WIMA MKP 0.1uF caps like so:

IMG_20181113_172617848.jpg

But it all fits… barely. I would use smaller voltage values if I did this again (which I won’t), but it’s what I had on hand.

DSC01774
Output section generalized view. Panasonic FR 100uF/63V bypassed with WIMA MKP 0.1uF
DSC01782
Output stage: OPA1612 for diff. DAC buffer, followed by OPA1602. DC offset was low enough I could remove the DC-blocking caps at the output.
DSC01778
Headphone output stage: again OPA1612, followed by high-current OPA1688 with output impedance changed to 1ohm.
DSC01783
Close-up of headphone output.
DSC01781
Input stage: Nichicon MUSE 100uF/25V bypassed with WIMA MKP 0.1uF, OPA1642, and OPA1602 for diff. stage. Panasonic FR again.
DSC01784
Close-up of input diff. stage, OPA1602.
DSC01775
ADC decoupling: 560uF/16V UCC organic polymer capacitors, lowest ESR I could find in that value range.
DSC01776
Closeup of ADC and one DAC
DSC01779
Mic preamps and inst. inputs: not a lot of room in there!
DSC01773
The whole analog board!
DSC01777
UFX psu, Panasonic FR in its day job…
img_20181122_011809111.jpg
Digital board with new caps. I also modified the clock. Yes, I did that. Did it matter? Probably not that much, and I don’t have a way to measure––RME’s SteadyClock is already very effective and bringing clock jitter down to the pico-second range. But the low-jitter (0.5pS RMS) 50MHz Crystek oscillator was $1.50, 5x7mm package fits perfectly on underside of board (beneath Spartan FPGA which generates the clock signal) in place of old oscillator, everything works fine, why not… Does that mean I now have the world’s first “UFX fs”? Doubtful (if jitter reduction was so easy, many engineers would be out of a job), but a tantalizing thought…
IMG_20181119_140238649.jpg
In progress: There were lots of leftovers… to save time, I used the ‘twist-off’ technique to remove SMD caps. I did lose one pad because I got sloppy, but I was able to bridge it. The twist-off method is where you grab the can with pliers and very gently twist it side to side (horizontal plane) until the leads break off from over-working the metal, still soldered to the board. Generally very effective and safe if you’re gentle. I’d rather that than a pad flaking off due to melted adhesive from too much hot air.
IMG_20181118_161139622.jpg
More leftovers… I think there was even more than this?
IMG_20181122_014750826.jpg
It’s alive!!!

Did I have any problems? Yes, a couple. First, phantom power to the mic preamps decided to stop working one day. Turned out I had weakened a trace on the digital board when putting in the main 48V decoupling capacitor, it eventually succumbed to my sloppiness, but was just a matter of bridging with some lead clippings. Also, there was a lot of low-frequency noise in the main ADC. Turned out I hadn’t properly soldered one of the leads of one decoupling cap to the SMD pad. Decoupling caps really do matter, because there were some WEIRD low-frequency peaks causing interference across the entire frequency bandwidth. But otherwise it’s been great! I mentioned Reverb/Echo aren’t working. This is indeed a hardware problem, was probably already an issue when I got the unit, but I never use that stuff so I didn’t think to check before diving in. As I said, replacing RAM didn’t change anything, probably an issue with solder balls under the DSP chip, either needs a rework or replacement. Replacing a BGA-type DSP chip, at $26/ea and probably requiring a proprietary firmware flash, is a bit beyond me.

How does it sound? It doesn’t. I mean, it doesn’t sound like anything, it’s extremely transparent, almost to a fault. Where there is a slight graininess in the RME ‘sound’ (which is admittedly already very much a ‘non-sound’), this is very clear and smooth, I hardly notice it. I’ve been using it for mixing, and everything sounds beautiful, clear, spacious, everything mixes well, good separation, without that in-the-box pileup I sometimes notice. I guess that’s the Sisyphean pay-off here––you do all this work to remove any semblance of personality from your gear, besides that of the performers you’re tracking, only to sit back to admire the complete and utter quiet nothingness of your labor.

So do I recommend this as a DIY mod? Let me lock arms (in spirit) with Matthias Carstens by proclaiming, nope! Not at all. I’m not going to provide a parts list or else it will seem like I’m endorsing this project (well, if ye wish, ye shall have a link to a Mouser project for ye’r reference). It was a pain in the ass. It sounds great, but I’m never doing this again. I mean, I probably will, but as a figure of speech I’m ‘never doing this again’. This is why you pay someone at BLA $750 to do this. Parts cost me $200ish, including some stock I already had, but this took me months of occasional labor. Again, this is why you pay someone––not for secrets (there aren’t any), not for snake oil (this is literally for lower noise floors and distortion, better current delivery to analog circuits and lower current demand from said circuits… it’s all measurable), not for a ‘better’ sound, but for a ‘no sound’ and a trust that if they broke something, they’ll fix it for you. Remember that mods like this, as you can CLEARLY see, and as evidenced by my dying 48V connections and weird failed decoupling noise, delight in opening the door for compromises to build quality––this is why this particular warranty is now void (it already was, but whatever). I took every precaution––I hot-glued parts together and reworked every joint with flux, going through with a magnifying glass to check for shorts or little lost solder beads that might eventually become a short, yet still things happen. If you want to take something on the road, leave it stock, or else bring: a soldering iron, solder, flux, a screw-driver, a magnifying glass, and a bucket for all the vomit you’ll discharge when something weird happens 10 minutes before you need this thing to be bulletproof.

But if you want a reference converter that’ll live in a studio, go ahead and mod the shit out of it. This thing is awesome. I’m sure it’ll serve me for years to come.

Until I have to repair something in one of the mic preamps, in which case I’m *******.

I’ll make some comparison mixes with a stock 802 (technically an upgrade to the UFX) sometime this summer, for those who are curious.

Advertisement

5 thoughts on “RME UFX modifications

    1. Sadly I’ve been too busy to do a direct side-by-side comparison of a stock model with my modded version––obviously I am biased to hear improvements, since I spent hours upon hours modifying it, but it does sound pretty stellar…

      Anyway, doing a side-by-side is on my list of things to do, but probably won’t get around to it till this summer.

      Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s