This Focusrite 18i20 surprised me with many faults!

[@FocusriteTV]

We appreciate your kind words regarding our customer service and apologize the issue persisted even after the initial repair. Our team will be back in touch soon regarding the next best steps for your interface.

[@tony359]

Hello and thanks for your comment! I am in touch with your customer service already, I understand some sort of official statement is being discussed.

[@walter7671]

Focusrite's service is impressive and unusual. I wish all manufacturers would support their products that way. Excellent analysis.

[@tony359]

I cannot say anything about the customer service indeed!

[@TheDefpom]

Great support from the manufacturer, very impressive.

[@tony359]

It really is!

[@Roobotics]

The mod being 4 capacitors for an 8 channel unit, is a huge indication why there is crosstalk. Those soic8 chips are likely dual opamps and they ran the even/odd channels into the A and B of the same opamp. Not ideal as the crosstalk is more or less at the mercy of the opamps noise rejection and capacitive coupling as well. A true fix is to isolate each sound chain to its own components and not share packages. Another thing you could try is tack a little ceramic 22uF right across the top of the opamp package, to bypass as close as possible.

[@tony359]

I'd be tempted to. I'd like to assume that Focusrite knows more about their own products but I'd also imagine that sometimes there is a compromise between the outcome and time spent to apply the fix. To fit 4 capacitors like that is a simple task. To fit a ceramic on the IC requires a little more skills/time :)

[@Roobotics]

@@tony359 There's definitely a trade-off, Engineering Change Notices are best kept to a form that can be readily applied by those with 'adequate' soldering experience and not be fiddly. Also fyi "NJM4565" is indeed a dual opamp +/-18v so pretty positive that's what those are to go with the +/-15 rails, also you can see little ceramic caps beside each SOIC package, so they at least tried to local bypass. If those are only 0.1 or 0.22uF I'd be tempted to beef them up on one channel just to see(for a more ECN friendly idea) I'm uncertain if they would need to be at least 15v or 30v rated. But maybe not worth faffing over.

[@tony359]

I suspect it's the usual "who is going to notice -65dB crosstalk?". Thanks, I might re-visit this 18i20 but first I need to find another one as I need this for some work! First, I need to fix the 1V issue - which has unfortunately returned....

[@Roobotics]

@@tony359 I was actually immediately concerned about the ~1.35v they use on the EN pin, I don't know what part that is but that sounded too close to a nominal threshold value for a typical logic EN signal. But every chip is a bit unique of course.

[@tony359]

someone mentioned that. But the IC says 1.2V minimum. The SOLO schematics also shows a voltage divider from 3.3V reducing it to 1.3V for the enable pin so it seems to be "by design".

[@bitsundbolts]

Wow, I am really surprised that they gave you a new power supply and addressed the issue after so long. You don't see that often these days anymore. Great video! I like your analysis and the test circuit for the chips from AliExpress. I also shouldn't blindly trust what I get from there, but so far, I don't have any bad experience.

[@tony359]

yes that was unexpected! I normally don't care too much but often I can test things before they go wrong. Think of that new controller for that Socket 7 CPU: I can test without the CPU first. But in this case if the thing just passed 3.3V out, by the time my Fluke noticed it, it might have been too late.

[@blakecasimir]

I've had excellent support from Novation who are part of Focusrite. It doesn't surprise me that Tony received that kind of support from them. Good chaps.

[@zzlg]

I've started watching your channel about 3 months ago. And it's amazing seeing all of this detailed repairing. I work fixing Air Conditioning PCBs, and one time I did your trick to heat up an area to trigger a capacitor to work (or not work), and really it was a faulty capacitor. Thanks mate. Keep up!

[@tony359]

Thanks for your kind words and for following me! The heat/cold is honestly a critical tool in up my sleeve. Sometimes I know which capacitor is bad but I just want to see if the unit works besides that so hot air it goes and that usually revives the capacitor just enough to work so I can quickly test the rest!

[@DjMarik78]

Hi Tony! C365 is used for feedback loop frequency compensation purposes, it performs a phase lead function ( it adds a "zero" in the transfer function ), adjusting the gain and phase at the XO frequency, for a more stable feedback loop.

[@tony359]

The Lord of the Power Supply :) Thank you!

[@DjMarik78]

@@tony359 i have used that trick many times myself in my designs, both on voltage and current feedback loops, and it does a great job!

[@khronscave]

Thanks for the shoutout at 37:55 and really glad my documenting has helped! While i indeed assumed those added capacitors were some third-party mod, i'm glad to stand corrected on that. To be fair, it was a surprisingly clean job if it was a third-party one, but nevermind then. But as i've mentioned on my blog as well, at the very least Focusrite's spares department is second to none!

[@tony359]

You're welcome and thanks again! Too bad the "mod" doesn't resolve the issue completely. That said, the 18i20 has been working fine - besides that 1V issue which returned first, then disappeared (??) but the I/F still refuses to power on when it's warm!

[@khronscave]

Regarding those added caps for the crosstalk - feel free to check what the positive leg connects to, but i've got strong hunch it's the Vcom of the ADC's. And each channel pair happens to go into each of the four ADC's... I wonder if the crosstalk might get even lower if those 47u's were replaced with something like 1000u/6.3v. They should only see like max 2.5v across them, so large values are still practical.

[@lapdogshuffle]

I have the same first gen 18i20 and it also has crosstalk. My 18i20 was 'donated' to me with a broken USB connector. My main interface at the time was the nearly identical Saffire Pro 40 (which didn't have bleed issues). While the bleed was disappointing when I initially tested it, it was manageable. I remember reading on a forum at the time that the solution was replacing the opamps but I was just happy to get it running. We used it for years standalone as a poor man's Octopre in our rehearsal room (settings saved with inputs routed to ADAT out), it was used to track drums on a bunch of recordings. It even had a stint driving 8 speakers in an audio installation for a year. Its outlived the Pro 40 and I still use it today on my gaming PC

[@tony359]

Good to hear it lived for so long! Now you can mod it :)

[@danpedersen55]

This is something that makes me want to buy from Focusrite, impressed by their customer service 👏

[@tony359]

yes, customer service is important!

[@obiwanjacobi]

4 blocks (with 4 caps) for 8 channels probably means that 2 channels share a dual opamp which is probably where the cross talk originates from. You could lookup the data sheet of those opamps and see what the specs are. If they'd isolated each channel to dedicated opamps I think there would be less cross talk, or at least you'd have more control over it in the pcb layout.

[@tony359]

I think the 2nd gen is similar in construction - I do know know if it has the same issue though. I doubt any op-amp came with 30dB separation between channels, I suspect the issue originates somewhere else - or anyways cannot be improved by just swapping some components.

[@wintermute740]

As long as those resistors don't become Light Emitting Resistors, you should be good ;)

[@tony359]

That only happens when moving the interface over 88 miles per hour :)

[@SianaGearz]

Resistors are going to be good, the capacitors they're baking nearby though...

[@patrickcraenen5163]

Hi there, From what I can gather from the NCP1521B datasheet, the voltage on the enable pin needs to be higher than 1.2V to activate the regulator. By default, the enable pin should be 2.0V. In your case, the enable pin was at 1.3V, which is very close to the threshold voltage, which is probably why you get the intermittent behavior... Maybe take a look at where the enable signal is coming from to see if there is anything wrong there. Have you tested the voltage on the regulator's enable pin after the repair by Focusrite? Thanks for the interesting video. Take care! Cheers

[@tony359]

That's an interesting point and thanks for looking into that. The Scarlett SOLO schematics have a voltage divider from 3.3V to make the enable pin and I did the math and comes out as 1.3V to the ENABLE pin. There is when I stopped looking into whether the enable pin was the problem. However... spoiler alert - the issue has returned!! I'll look into the datasheet - I admit the SOLO regulator is a different model so maybe the 1.3V there is just a coincidence, it works for the SOLO but not for the 18i20. Thanks again! EDIT: No, the SOLO also has the NCP1521B, the ENABLE pin is the output of a voltage divider made of a 20K and 13K resistor. That makes 1.3V exactly. I could check on my 2i2 to confirm though. It should have the same components.

[@patrickcraenen5163]

@@tony359 Then I assume 1.3V is perfectly fine to activate the regulator. I just noticed that the enable voltage was so close to the threshold. Sorry to put you on the wrong track and I hope you can find the problem. Thanks for all the time and effort you put into your videos, I really enjoy watching them.

[@tony359]

oh please do not apologies and you're not putting me on the wrong track at all! Any input is very welcome. Next I'll swap those feedback resistors. I don't think I cracked them, I think they were cracked already and I just dislodged them. Failing that... ghosts?

[@Mtaalas]

I've never liked Focusrite interfaces. I've tried MOTU, but their driver support for Windows at least is atrocious... tried RME: I'll never go back to anything else. Their interfaces are just cream of the crop. Priced as such, but they have the best drivers I've ever used, their hardware is absolutely top notch and it's a company run by engineers, not marketing people.

[@tony359]

My 4Pre is my main sound interface on my Windows PC. Yes, drivers haven't been great though I haven't had issues for quite some time now. All I know is that whenever the 2i2 or the 18i20 "misbehave" showing weird spikes or elevated noise floors or (whatever), the MOTU is always showing me a faithful representation of the sound. My budget is not so high to get to RME unfortunately! :) But thanks!

[@franciscotrigo9656]

I do appreciate your honesty!

[@tony359]

Always!

[@eddiehimself]

I've had one of these for the last 10 years and not had any problem with it. Of course, I've not had any need for an ultra-low noise floor, but also I've not had any problem with the power supply and the digital circuit dropping out.

[@tony359]

Would you be able to check the crosstalk? Even without an RTA, if you feed full scale signal on channel 1, do you see channel 2 creeping up? Everything else are unique faults with this specific interface.

[@eddiehimself]

@@tony359 Yeah, there is a bit of crosstalk. With my bass ukulele plugged into channel 1, channel 2 crept up to about -83 dB from a noise floor of about -86 dB in my DAW. I don't know if this is one of the affected units or it's just the design and to be fair, I would rather not send it in to be repaired even if it's free because I need it to hear what's coming out of my computer!

[@minombredepila1580]

Great video as always Toni. Perhaps if there were some heat sinks for resistors.... or a new ones with larger legs, allowing them to "touch" the metal case, helped perhaps by some thermal paste for better heat transference, would reduce its temperature immensely. Just an idea.

[@tony359]

Thanks - plenty of "mods" would improve this interface for sure!

[@IlBiggo]

I'm at 9:00 and you haven't checked C310 yet. Circuit generates voltage, stops after a while, there's a capacitor to ground on the output; the first thing I'd think is that the cap is faulty and shorts out when it heats up. It looks like it has some kind of crack on the top, too.

[@tony359]

I hope you kept watching :)

[@IlBiggo]

@@tony359 I did :D You killed my favorite capacitor without checking it? You monster.

[@tony359]

Tony the Capacitor Monster! :)

[@PorchBass]

Great job. Focusrite should be applauded for supporting the repairs. It's not an instrument grade interface I reckon

[@tony359]

Absolutely!

[@simontay4851]

39:55 123C with the cover off! Thats far too hot! With the cover on they could reach 200C! The resistors and the caps next to them won't last long at that temp. You should mount some 10W ceramic resistors of the same value on bare part of the metal case in front of the PCB and run copper wires from them to the holes in the PCB where the original burning hot resistors are. The metal case will act as a heatsink. For the wire you can use some old 2 core 0.75mm2 mains flex.

[@tony359]

Thanks - lots of potential improvements on this interface indeed!

[@DANILOCOCCIA]

Can you please add some info on the spectrum analyzer SW you use? Looks interesting!

[@tony359]

YT does not like links, search for Rational Acoustics Smaart. Thank you!

[@gorjy9610]

If you're paranoid about these voltage regulators why going through everything you did and not just made a simple crowbar circuit, with 1v just adding two diodes in series to ground would do the job. And that way you have permanent protection of the IC.

[@tony359]

That's a good idea. thanks!

[@anomicxtreme]

I have a Liquid Sapphire 55 that uses the same exact power pcb that died and I'm doing a component replacement on. I'd really like to find a schematic. There's a couple mosfets/voltage regulators that are no longer manufactured.

[@tony359]

The Scarlett Solo schematics are available online. The supply section has similarities - it might help a bit? The regulator I replaced is also discontinued but available on Aliexpress.Good luck :)

[@anomicxtreme]

@@tony359 Thanks for responding so quickly, seeing as how the power supplies are almost exact, after scouring the interwebs for weeks, I found a forum where they were having the same model as yours and the culprit is Q109 (I think, I don't have mine in front of me) the one on the heat sink next to the big 50v cap. I tested mine and it was bad but also fried a bunch of smd's on the reverse side. It might be worth it to test out. Edit: The one getting super hot in the thermal video.

[@tony359]

the one getting super hot is a resistor on the main PCB. But you're talking about the PSU?

[@anomicxtreme]

@@tony359 yes, right next to the monster capacitor

[@abijeetrs6522]

Are you on the latest firmware for this device?

[@tony359]

I am!

[@dipakbhalodia7155]

I love you work and information best explained

[@tony359]

Thank you so much!

[@nikelquint]

How coincidental.. My friend just had a second generation that seems to have shorted one of the data lines on the USB input.. those lines go directly to the cpu, and not a separate usb controller like that smaller one you were working on, or this one. No communication between computer and interface. Everything else seems to check out. No idea focusrite were that kind with parts. I might have to see if they’d sell me a new cpu!!

[@tony359]

Yes, it was simpler when USB was going to the small USB IC. I find a bit... unwise to have USB going straight to the CPU! I don't think FR will sell you components - those can be bought online BTW, there is nothing proprietary in their products AFAIK.

[@SidneyCritic]

If there is 4 caps for 8 channels, it must have 2 channels in a chip. Maybe the CT is in the chip, and maybe there is a new better replacement from that company. Maybe search for a latter model, assuming it doesn't have CT, and see how they did it.

[@tony359]

I cannot believe there is a sound IC with only 30dB separation between channels. I feel that the issue is deeper in the design of the circuit. Had it been a matter of replacing those ICs, Focusrite would have replaced them I guess. When it comes to sound it’s simple to mess things up. It could be some traces on the PCB.

[@SeanBZA]

@@tony359 Very likely the issue is due to running the audio with a simple ground trace, which would certainly be able to inject that amount of crosstalk. Perhaps in later revisions they went to a thicker ground trace, and stitched it to the ground plane better, instead of only at the ends. the capacitors improving things says there was also power supply crosstalk, the other channels showing harmonic distortion varying with input level from the others shows that, so the extra capacitors are an attempt to mitigate that. But hard to mitigate that sort of crosstalk without having all 4 pairs of channels being separated, and fed via separate power traces, with LC filters on the input, and a shielded box around each one. Simple test is to take all those internal wires, ans twist them, so as to reduce common mode noise they couple, followed by taking each pair, and slipping a shield over them, and a shrink sleeve over that, with only the end by the input side connected to a short braid to a chassis ground. especially those running over the sensitive input analogue circuitry, just acting like little antennas.

[@tony359]

Ha, I was wondering all those wires with no shielding, no twisting... It's such a shame to see those basic things being missed by a manufacturer. Thanks for your input!

[@MomirPeh]

@@tony359 I agree. Almost all dual OpAmps (including JRC4558) will have internal crosstalk up to -105dB in the worst case. I think the crosstalk here is coming through the power supply rails. I bet those bodged capacitors are between the positive and negative supplies for the OpAmps. You could try and measure the crosstalk on the pins directly with an oscilloscope and see where it's the strongest.

[@tony359]

as far as I know you are correct, it's what "Kron's cave" reported.

[@PileOfEmptyTapes]

As for the remaining crosstalk issue, it almost looks like this part of the circuit may be running on a single supply?! (Weird. Maybe they wanted to save power?) Then the capacitor tacked on to would be in an R/RC voltage divider for Vs/2 generation. Such a virtual ground has a finite output impedance and it shows when shared by multiple noninverting opamp circuits, depending on resistor values. For a better mitigation you would need an even bigger capacitor. For a clearer picture you would probably have to reverse-engineer one instance of the problematic circuit, including what the capacitor is involved in. Resistors running at 125°C are clearly running fairly close to their nominal power rating and should really be replaced by a higher-wattage part. Actual dissipation x2 is a good rule of thumb. I suppose they are ahead of some of the linear regulators, what are those regulating to? Don't say +/-5 V, that would be spectacularly dumb. (Maybe +/-8 or 9 V?) I guess regulator number 3 is for +3.3V? I am under the impression that the 1st-gen Scarletts still had a number of rough edges in general, e.g. people complained about limited max input limit in instrument mode (though this may have been mostly an issue for the smaller models with their +/-5 V supplies, which mind you were an upgrade over the single-supply Saffires that preceded them).

[@tony359]

thanks for the details. Let's see what FR comes back with!

[@TLang-el6sk]

If you still have the desoldered regulator you may want to check the actual threshold of the enable pin. The datasheet states 1,2V (they say min but I think they mean max), below in the text they state 0,7V typical. What if the desoldered device is a bit on the high side? You measured only about 1,3V. Problem is, On Semi says you should keep the device in shutdown until Vin is above 2,7V. With the given tolerances (we don't even have min/max values but only know the threshold is at least between 0,7V and 1,2V) and a nominal Vin of 3,3V it's not possible to meet this requirement with a simple resistor divider.

[@tony359]

There is a divider but also a zener - at least this is what the SOLO schematics say. I can check on my 2i2 which should have identical components. The issue is that... the 1V issue is not resolved. I'll publish a follow up. I suppose I need to turn on the scope :)

[@TLang-el6sk]

@@tony359 It's not a Zener but a Schottky (symbol + part number say so) as far as I can see on the circuit detail you showed in the video. I don't really see the prupose of this part here. If there was a capacitor on the shutdown to limit rise time it would make sense to enforce a fast shutdown when the 3,3V voltage goes down, but so... In general I think the idea was to sequence the two rails which is why the first regulator has EN tied to Vbus and the second gets it's EN derived from the first regulator's output voltage. If it does not start again you could try to bridge the shutdown to the 3,3V to check if the NCP starts then. I think I saw that EN can be up to Vin+0,3V.

[@tony359]

ah you're right, I forgot! I'll experiment a bit! Thank you!

[@aleksandardjurovic4773]

Thank you!

[@tony359]

You're welcome!

[@teejmiller]

Excellent video, what a weird situation. You clearly hold the 2 channel in high regard. There is a very obvious design defect in the 'better' model. The company did what they should have done and thats great to acknowledge their mistake.. but it falls short of being perfect in my opinion. If you're anything like me you'll still be looking for a different model in the back of your mind as time goes on. I wonder if a newer used model with a board revision would display the same crosstalk?

[@tony359]

ahah I am! Thinking of getting a gen3! :) I am also thinking what you're thinking: is this a "bad" PCB revision which Focusrite could not swap so it was "bodged"? Someone else seems to suggest all gen1 are like that though.

[@Sir-Dexter]

nice work

[@antonjansenvanrensburg4145]

would have liked to see if you still only had 13v instead of 15v with the new powersupply. they must share op amps if they only did 4 caps on 4 of the inputs for the op amps. can understand why they have crosstalk, maybe a open area for some mods where you can run separate op amps? nice video overall. new sub added :D

[@tony359]

Thank and welcome! Ah, I forgot to check but the +/-15V on a sound device shouldn't be too needy. If not mistaken it drives the output drivers? Oh I am sure this interface would be great for modding :)

[@willyarma_uk]

I'd be curios to see whats on those new piggy back caps with an oscilloscope and would a bigger or different kind of cap like a tant make any difference? Also did you swap out those scratched resistors? I love the detail and thoroughness in your vids.

[@tony359]

Not yet - but I'll do as...... the 1V issue has returned!!!! DO'H! I might make a short 2nd channel video to cover that! So frustrating!

[@willyarma_uk]

@@tony359 Oh no. I really hope u solve it and would definitely like to see how.

[@FireballXL55]

Hi Tony, does the cross talk go away if the level controls on the non used channels are turned down?

[@tony359]

Good idea, that would tell me where the crosstalk happens. I didn’t try that. I’ll test! Thanks.

[@GabrielZ666]

Ohh Topazcon, long lost cousin of Rubycon 😆😆 Maybe we'll see green Emeraldcon caps soon!

[@tony359]

OMG - I didn't realise that! LOOOOLLLL! Next it's going to be Pichicon and Tarasonic.

[@enric0r560]

Nice video! Want more audio gear :)

[@tony359]

Thanks! Well this video has started really badly unfortunately so we shall see for the future :)

[@za_ozero]

well, putting a 50 hz frequency filter on power supply output would be a future solution. I mean, a choke is more durable than a capacitor

[@compozic8161]

For small chemical capacitors, is it possible to replace them, for example, with higher precision capacitors, such as N-type electrolytics of the same values, or is it more complicated than that?

[@tony359]

Standard capacitors are rated 20% by design. If the designer selected a standard capacitor that means that the circuit doesn't need a precise value. You can replace capacitors with better ones - longer lifetime and higher temperature - but I don't see the purpose of installing higher precision capacitors in place of standard electrolytics.

[@compozic8161]

@@tony359 Thank you for your explanations that I understand well. The question is not about this card (which I will not buy under any circumstances), but capacitors in general. However, I notice that the more sound cards go up in game, the less they have chemical capacitors, in favor of aluminum body capacitors.

[@tony359]

My reply was not about this interface :)

[@X-101]

What Gen is that?

[@tony359]

it's a Gen 1

[@jonesconrad1]

my gen 2 has been great

[@Dutch-linux]

the 2 resistors that get hot replace them with a 5 or 8 watt resistors that will lower the temps and wont affect anything else the resistors are just a little under spec of what they need to be saving pennies is what that is !!!!

[@tony359]

Indeed - a penny here, a penny there and here we are! :) Thanks, I'll think about doing that!

[@blakecasimir]

This is the kind of thing you can expect from what is ultimately a budget audio interface. They certainly do cram in a ton of features at a reasonable price point... but this is why I say to people: buy RME. ;)

[@tony359]

I do wonder: what would have cost to have the bigger resistors and better capacitors? £5? £10? Is it really worth? Anyways, everybody does that so I am not too surprised :)

[@blakecasimir]

@@tony359 I suspect that adventurous folks like yourself could mod these and improve their performance and reliability.

[@TonnyCassidy]

that solder joint on the fix cap..... do they not know how to apply flux...

[@tony359]

yeah, I skipped on that! :)

[@k7iq]

Does the crosstalk still exist when you add a 150 Ohm R across the adjacent channel ? I would say it is unfair to look at a channel that is not loaded on its input. I would also expect that bit of THD to possibly be worse with unloaded inputs. 50Hz might have been a leaky diode in the power supply ? That might explain why cooling the CAPs down helped some.

[@tony359]

Yes, it’s the same with all channels loaded. Also consider that the 2i2 and the Motu were tested one channel at a time like the 18i20. Anyways I still have the same crosstalk with all channels loaded I’m afraid.

[@k7iq]

@@tony359 Thanks ! I just did not see those jacks with a dummy load plugged into them. That is an important little fact I would say. The Xtalk issue must be layout related then I guess. Obviously it's not much of a problem by power supply or wherever they put that CAP

[@tony359]

@@k7iq during the video I indeed tested with PN coming from the Focusrite itself. But I normally work on an 8ch device which can output PN and I see the same crosstalk when I test with all 8 in and 8 out wired to this device. (Balanced in and out). So yes must be a severe design flaw. Too bad Focusrite hasn’t come back to me on this subject!

[@maxtornogood]

When Focusrite got it Focuswrong. Sorry I had to say it!

[@tony359]

ahahaha! I didn't think about that! Yes, they tried but this is a bit disappointing. :)

[@Google_Is_Evil]

Looks like the crosstalk is happening because they are sharing power for for the opamps for 1/2, 3/4,5/6 and 7/8. Voltage drop on an undersized power supply by one opamp will result in a voltage differential on the power supply of the second opamp, knocking the output off 0 and there is your crosstalk. By adding extra capacitors the power draw by the pair will give less of a voltage drop, but not solve the problem. You could probably improve on this by either building 4 more power supplies on a separate board and lifting power legs on 4 of the opamps to connect them to the extra power supplies, or maybe add more piggyback capacitors directly on all 8 opamps legs. I totally understand if you don't want to, but in my opinion you could make a better solution and a youtube video out of it. Regarding the 50Hz, it probably originated in the 230V section of the switching power supply. However much is left, you can either work with the capacitor on the 230V side that smooths out the rectified AC, or put an inductor on the output lower voltages, possibly even counter wind it between the plus and minus so any hum would magnetically be filtered out. Did you replace that capacitor on the PSU that you recapped, or was it only the secondary side where you recapped it?

[@tony359]

No I swapped them all, including the main bulk one. That is what puzzles me! And hot air was applied on the secondary. Weird. I'm sure there would be plenty of mods for this poor interface to make it better!

[@TheDefpom]

R515 looked like it had a cracked solder joint

[@tony359]

R515 is cracked :) Reflowing did not help unfortunately and - little spoiler alert - the 1V issue has returned. Only when the interface is warm. I'll replace those feedback resistors (I suspect they were cracked before). If that does not help, then I have to assume it's an overcurrent protection.

[@TheDefpom]

@@tony359 also try soldering the nearby via, I know it sounds strange but I have fixed many things by soldering the via due to contact issues in them, try to get solder down into them, sometimes the plated holes fail.

[@tony359]

Absolutely. It's so frustrating, it MUST be in that area!!

[@qwaqwa1960]

Years ago, I bought the MOTU equivalent of this. Complete rubbish. And...one channel was DOA. I opened it up and found a "tombstoned" resistor, which I proceeded to fix. Unbelievable-they didn't even test the unit before shipping.

[@tony359]

I've had some bad experiences with (old) MOTU recently, this interface has gained some points in my mind now :)

[@rudilussana9194]

1:45 isn't that white noise? I thought pink noise was supposed to be decreasing with frequency?

[@tony359]

Pink Noise is equal energy per octave, whereas white noise is equal energy per Hz. If your pink noise is accurate, then a raw FFT display will slope down at 3 dB per octave, since the FFT's bins are all the same width in Hz. An RTA shows the energy per fractional octave band. If your pink noise source and your RTA are both accurate, then every band should have the same level. (copied from a forum! The bottom line is that Pink Noise looks flat on an RTA! Thanks for watching!)

[@rudilussana9194]

​@@tony359 Stupid me, I didn't realize the x axis is logarithmic, which makes it constant energy density per octave and not per Hz, as you wrote. Thanks for the explanation and keep up the great job :)

[@tony359]

there is no such thing as stupidity when someone is asking questions! I could not remember myself, hence the copy and paste from the internet! :)

[@gamerpaddy]

18D is 150k what. EIA96 is weird i would have guessed 18 and D (4th digit, so 4 zeros.. 180k) that explains the few unsuccessful repairs in the past

[@tony359]

oh no! I hate when SMD components have those cryptic codes which lend themselves to so many interpretations!

[@SianaGearz]

BEES! THEY PUT BEES INTO THE POWER SUPPLY! Or worse, Yellow Jackets. 105° generic caps, they're going to be OK, i'm just kidding.

[@tony359]

LOL!

[@scotshabalam2432]

I've had a lot of problem with a Gen2 Fuckusrite. Actually, I suffered with it for years, then I got a Soundblaster and switched. Soundblaster works great in fact I've never had a problem with a SB product. Fuckusrite products are all hype.

[@tony359]

I see they have a good reputation :)

[@blakecasimir]

More audio gear, more fun guaranteed. But folks: if you're doing this in any way seriously, please skip such gear and go for Audient or preferably RME. You'll be better off.

[@tony359]

It’s clear that these are budget devices. The equivalent MOTU is twice the price. ‘Topazcon’ is the consequence of that. 😂

[@blakecasimir]

@@tony359 IMHO MOTU gear does not deserve to be at the prices it is generally listed at. Though I would generally put it steps above Focusrite's interfaces. Honestly I would just steer folks towards the Audient Evo range, or - if you're getting serious - head straight for RME.

[@tony359]

Thank you. I'll definitely take a look.

[@blakecasimir]

@@tony359 Also, to be fair, I am not surprised that you received such support from Focusrite. I've had similar experiences from Novation. They seem to have good folks and an good work ethic going on there. But Focusrite could do well to step up their software driver QA as that is still an area where these interfaces can be lacking.

[@SebastianSnoeck]

Other brands are more expensive but when you see how much years it last it is worth it ... I have personally have a rme fireface uc and I had only one problèm once in 15 years and it was the external psu and drivers still work in 2024 on the latest computers ( in my case pc with the last windows and the driver for Mac is even more recent )

[@Route66Wanderer]

This is some annoying troubleshooting. I have no idea why you started off checking DC voltages? If the unit had hum, it seems that you would start measuring for ripple at the DC supplies, especially since ripple not only causes hum, it can also cause things to stop working.

[@tony359]

Because the unit was dead? I think you skipped some parts!

[@AsmodeusDeviluke]

its not a true 8 channel input. Much like AMD FX cpus, it looks like the 18i20 channels share filtering.

[@tony359]

Focusrite 9i10 😂

[@jeremybalzarini279]

Dude! Their workmanships standards are way off! Your soldering looks way better than the factory. Not only did you find the stray solder fod but thier solder joints had projections. Not to mention if they are going to MacGyver this mod, at least put some RTV or hot glue to secure the caps.

[@tony359]

ahah yes! The picture on Kron's cave did show some hot glue on the caps indeed. "Going to McGyver", LOL!

[@tigerrat8]

i just switched because there drivers are a nightmare my solo worked well on 7 works great on Linux....windows 10 forget it just trouble mic input has no head room painful just stay away

[@tony359]

Interesting, I heard about that, I didn't know it was a driver issue!

[@simontay4851]

CrapXon capacitors! Eugh! The absolute worst quality caps. No wonder there is 50Hz noise.

[@tony359]

:D

[@SianaGearz]

Hey all the Taiwanese PSU companies use them at low-midrange and when they do it properly, they hold up just fine! You can't rely on just electrolytic caps neither for reliability nor for noise mitigation, this is a whole-subsystem design issue, like in a good PSU, no matter whether equipped with cheaper caps like these or with more upmarket ones, you see a sprinkling of mitigations of various kinds all over the place. But also they would never specify 85°C caps on the secondary, no, not even in much better ventilated units.

[@Dutch-linux]

the newer generations have had a complete redesign... your gen always has that problem.. and there was never a full fix for it hence the other gens had a full redesign !!!!

[@tony359]

oh wow, are you saying that all gen 1's are that bad??? Wow!

[@Dutch-linux]

@@tony359 yes they are all that bad.... I believe they are now up to gen 4 but the gen 3 was better then the gen 4 the gen 4 lack volume its very quit compared to the gen 3 so really if you are looking for a secondhand focusrite product look for 3th gen... but i think they are harder to find as they are the best

[@tony359]

I think the current version is gen3. I wonder if gen2 doesn't have the crosstalk issue, the PCB is very similar. Gen 3 looks like a complete re-design.

[@guilhermephsor]

CapXon capacitors... these crap components along ChongX are everywhere. Focusrite should be ashamed of using those.

[@tony359]

Yes, let's use TOPAZCON everywhere :D

[@guilhermephsor]

I bet the crosstalk is because they probably used the same opamp on 2 channels

[@humbopp]

Mkłłi och

[@ste76539]

To be fair, after Focusrite addressed the design flaw AND supplied a free replacement power supply on a decade old product you aren't even the original buyer off, I might have handled things differently afterwards. For instance, I don't think I'd have uploaded a video to KZbin with a complaint of 'only mitigating' the design flaw. I'm not surprised they never sent you that follow up statement, I know I wouldn't have either. Nothing wrong with the rest of your video, but that element didn't sit very well to be brutally honest.

[@tony359]

I appreciate your honestly - but I have been honest. I have praised Focusrite for the excellent service and I made sure it was very clear. At the same time, I "discovered" that their gen1 interfaces are potentially not great and cannot really be "fixed". Why would I hide that? I'm just being very objective here. Following your logic, if Focusrite had sent me a gen3 replacement I should have cancelled the video. :)

[@ste76539]

I'm not entirely sure you fully understood what I was getting at. I would never suggest anyone "hide" or lie about anything. I just think after the company went above and beyond what anyone should expect, you did do them a bit of a disservice by complaining it wasn't good enough afterwards. Maybe not include that, or perhaps phrase things differently, as it is it comes across, no other way of putting this, 'p!ssy'. Ungrateful. Yeah sure, their work didn't make it all wonderful, but they improved it, plus gifted you a brand new power supply all for free and without actually ever having to. I'd be very grateful, not very publicly picky.

[@tony359]

Well, the 18i20 is the top of the line of the Scarlett series. Back then it was £500. Today's price list is still £500, though it can be found for less. I purchased a Focusrite Interface, thinking it was a good one. It had a manufacturing defect. Fair enough, Focusrite fixed it for free. But they couldn't fix it because... I assume it's a poor design and can only be mitigated. You as a buyer would be happy with that? I praised Focusrite for their service. But I also showed an objective issue with their top-of-the-line interface. I am sorry but I disagree with your view. Because Focusrite did their best to help, doesn't cancel the fact that the 18i20 gen 1 is a poor design with poor sonic performance.

[@jdmjesus6103]

Have you ruled out a fault downstream of the 1v generation circuit that could be pulling it down? I think I would remove the chip again and apply a current limited (from the datasheet) 1v from a supply and see if that boots the device while monitoring the current. I'm an amateur though, don't listen to me, it'll probably fry it!

[@tony359]

That's a possibility. I hope it's not that as it would be difficult to resolve. I think I can monitor the current with a scope and a resistor. But I'd still expect the regulator to come up to life briefly when that happens. So next would be to scope everything and see what's happening there. Thank you!