TC Impact Twin mods

[12Bass]

Hey Everyone,

Recently I modified my TC Impact Twin FireWire audio interface.

These units are known to have problems with failing electrolytic capacitors in the power supply section of the circuit, near the voltage regulators. Although I had not yet experienced such problems, I preemptively replaced the existing electrolytics with high quality 105C Panasonic FRs which will hopefully last the life of the unit.

I also replaced the relatively small electrolytic capacitors at the inputs of the two microphone preamps with much larger Panasonic FR electrolytics. Additionally, I added polypropylene film bypass capacitors in parallel with all electrolytics in the analog signal path and replaced the original NE5532 op amps with OPA2211As with ADA4627-1s on the main 1/2 outputs.

The result is a marked increase in subjective transparency, especially in the microphone preamps. I also have a stock Saffire Pro 40 and the preamps on the modded Impact Twin are much more open sounding, with both more clarity in the highs as well as deeper lows. The Saffire Pro's preamps sound relatively low-midrange focused in comparison.

Included are pictures of the back of the PCB showing the polypropylene bypass capacitors as well as the replacement op amps on the analog outputs on the front of the board.

[Khron]

I'm a bit confused - how exactly did the NE5532's (dual opamps) get replaced with the ADA4627 (which are singles only)? Or do those SOIC-to-DIP have a second opamp on the underside?

[stj]

he said "NE5532 op amps with OPA2211As", "with ADA4627-1s on the main 1/2 outputs."

[Khron]

The main outputs also have dual opamps. I have a couple of Konnekt 8's, which use the same board Ditto with the Konnekt 24.

[stj]

found them
http://cimarrontechnology.com/single...rpn020302.aspx

[12Bass]

Originally Posted by Khron

I'm a bit confused - how exactly did the NE5532's (dual opamps) get replaced with the ADA4627 (which are singles only)? Or do those SOIC-to-DIP have a second opamp on the underside?

Correct, the DIP adapters have another ADA4627-1 on the bottom.

The ADA4627-1 is the best sounding op amp I've found thus far. Oh, those LM4562s are on the 3/4 outputs which are not likely to be used very much. I used OPA2211As to replace the NE5532s because I had them on hand and they sound better, though not as nice as the ADA4627-1; I haven't found a bipolar input op amp that I like as much as that one.

In total, the replacement op amps cost more than I paid for the interface (used).

[Khron]

Somehow, i can't say i'm shocked to hear that last part...

And at least in some ways, you're kinda lucky the TC engineers knew what they were doing, and included pF-range caps in the feedbacks of all the opamps that do any level of gain. Otherwise you'd be enjoying the lovely "serenade" of MHz-range oscillation and whatnot (with those uber-fast opamps, etc)

[12Bass]

Hey Khron,

I'm hoping you might have an idea what might be causing this noise in my Impact Twin. I suspect that these two peaks (9350Hz & 18,700Hz) may be due to the unit's internal switching power supply. These peaks were present in analog loopback testing which was done before any modifications were made, so they are not a result of the new op amps and bypass capacitors. I've tried adding some ceramic capacitors in parallel with the analog supply electrolytics with minimal improvement.

Any suggestions?

[Khron]

I guess that's a possibility, but i have no idea. I don't think i've ever scope-probed the power supplies to see what frequencies they work at (and/or it wouldn't surprise me if that was even load-dependent, since they're self-oscillating). "They", because the two are independent - the smaller transformer's for the 5v & 3.3v digital rails, while the bigger transformer's for the +/- analog rails and the 48v phantom power.

Interesting that they seem to be present on only one channel(?), though...

<bad joke> But hey, what do you care about measurements, if the sound's so much more "transparent"? Don't those peaks provide some more "air" to the sound?

[12Bass]

Originally Posted by Khron

I guess that's a possibility, but i have no idea. I don't think i've ever scope-probed the power supplies to see what frequencies they work at (and/or it wouldn't surprise me if that was even load-dependent, since they're self-oscillating). "They", because the two are independent - the smaller transformer's for the 5v & 3.3v digital rails, while the bigger transformer's for the +/- analog rails and the 48v phantom power.

Interesting that they seem to be present on only one channel(?), though...

<bad joke> But hey, what do you care about measurements, if the sound's so much more "transparent"? Don't those peaks provide some more "air" to the sound?

Thanks!

Actually, the noise is present in both channels, but stronger in the right channel. The second peak appears to be a multiple of 2 times the first. Given that they are -120dB, perhaps it doesn't matter, but they shouldn't be there. If they were audible, I doubt that they would be adding anything desirable to the sound. More air would likely be something like a gentle boost in the highs with a shelving EQ. I've looked at quite a few RMAA tests of other interfaces and most of them are pretty clean at the noise floor. That said, the noise has been decreased somewhat by the various modifications, I'm guessing that the Panasonic FR replacements in the power supply are the biggest factor.

It is unfortunate that you live so far away... otherwise I'd be happy to loan you the interface so that you can listen for yourself and compare it with your other TC interfaces. The larger caps on the microphone input noticeably deepened the bass response and the polypropylenes added some transparency. Not sure why, but TC used much smaller input caps on the preamp than recommended in the THAT 1512 design notes, which is likely the reason why they sound a tad thin stock.

[Khron]

"Much smaller"? My Konnekt 8 has 100uF caps on the mic inputs, and according to my photos, so does the Konnekt 48. These are the exception rather than the rule; more often i've seen 47uF (Focusrite, MOTU).

And even so, with a pessimistic 1kohm input impedance, 47uF input caps give a -3dB corner frequency of about 7Hz (yes, seven Hertz). 100uF pushes that down to 3.5Hz. You'll have to excuse me if i'm skeptical about the (objective / measurable) audibility of that particular difference.

Yes, i know and i'm aware, it's a roll-off which starts a bit higher in frequency (-1dB at 14Hz / 6.5Hz respectively), but even that's a good octave or two lower than most loudspeakers reach so... I dunno...

But hey, as long as you're happy...

[12Bass]

Actually, it is the electrolytic capacitor on the preamp gain potentiometer that I changed. Sorry, should have been more clear earlier. The datasheet recommends 6,800µF for THAT 1512, but IIRC TC used 470µF. The replacements are 3,300µF, which was the largest that would fit at the same voltage. I don't have the schematic, so I'm not sure of the surrounding resistance values.

[Khron]

Ah, that makes a bit more sense then. That's there in order to avoid having DC across the gain potentiometer (which, in time, would lead to premature crackles). That, and the output offset, according to the datasheet.

In older MOTU gear (specifically, two 8Pre's i had), those gain-DC-blocking caps were 6.3V - i'm quite certain that's more than enough voltage rating, which would allow you to go with even larger values (within the physical constraints).

[12Bass]

iXBT.com reviewed the Impact Twin and some similar high frequency noise is evident in their RMAA dynamic range test. Perhaps a bit of switching noise is normal for a compact audio interface with two internal switching power supplies?



If possible, I would like to reduce the switching noise as much as possible, but I'm not sure how to approach that. I did a little bit of reading about switching noise suppression and it appears that adding local ceramic supply bypass capacitors has little effect, which is what I have found. Sounds like a proper LC filter may be necessary, which may not be practical given my level of expertise.

Another thing I'm concerned about is crosstalk. There seems to be more interchannel leakage as frequency increases, as seen in the attachment below. This issue was also present before any modifications were made. Input/Output channels 1&2 seem to be worse for this, with 3&4 a bit better. I'm wondering if it might have something to do with the extra Stereo/Mono/Side circuitry on the main outputs (CH 3&4 have more direct internal routing)?

[Khron]

Just via guess-work, you can't really know whether those noise spikes are conducted (where extra filtering might help), or radiated (where filtering wouldn't do squat, but shielding would).

Also, is the mono / side stuff done in hardware (with switches), or software? My hunch would be the latter.

PS: "Similar high-frequency noise" - yeah, at -140dB(!!!). That's like 4dB above the theoretical(!) noise-floor of 24-bit audio.

[12Bass]

Not sure how the stereo/mono/side switching is done (transistors, CMOS?)... it can be controlled by the software and the button on the front. However, if you route the audio output directly to 1/2 (instead of going through the mixer),that section gets bypassed completely. Anyway, it is not directly through a hardware switch.

Yes, I realize the that noise on the iXBT test is much lower in magnitude... but it looks similar in nature. My Impact Twin was purchased used and I'm assuming that they tested a brand new one. Perhaps mine has a bit more wear on the components, particularly electrolytic capacitors in the power supply? I replaced most of the big caps in the analog supply, but not the digital supply.

Perhaps new electrolytics in the digital supply might help? Putting Panasonic FRs in the analog supply did seem to reduce the spurious noise somewhat according to before/after measurements. However, IIRC, I made my initial stock measurements with my old PC which has a noisier power supply (i.e. EMI), so I'm not sure how valid they might be.

Or maybe I should just box it up and use it as is and forget about those imperfections....

[Integrion]

Hi Everyone,

I have an Impact Twin that recently died. It won't power up or turn on. The only light I get is the internal red status led near the fire wire interface when it's plugged in with the unit turned off. I am suspecting that one of its internal switcher supplies has crapped out.

After doing a bunch of Googling around, I managed to get my hands on the schematic for the unit's dual internal switching supplies, which I've uploaded here.

I've read in numerous places that the filter caps near the internal voltage regulators such as C315 thru C318 are usually the culprits, so I shotgunned and replaced those but no joy. Neither the digital or analog PSU switcher sections appear to be switching on.

Before I invest a ton of additional energy into troubleshooting this thing, I was wondering if anyone else has experienced this kind of issue with an Impact Twin? I can see how one of the internal PSU switchers can die, but both not switching on? It makes me wonder if something else other than bad a filter cap in the PSU output section could be the culprit?

Any and all thoughts much appreciated!

[Khron]

Here's a question - is the power brick outputting what it's supposed to be?

The red LED from the Firewire, i would reckon, only turns on if there's power coming through the Firewire connection (assuming the internal power supplies aren't working).

[Integrion]

Hi Khron,

Thanks so much for the reply.

Yes, at least it would appear so. The brick's output label states 12V DC at 2 amps. With the brick plugged in, I'm measuring on the internal side - 11.73 DC at TP 881 (junction of L62 & C524), TP930 (Pin 2 of TR3 14220094-3 transformer), TP 896 (junction of L61 & C307), and TP915 (Pin 2 of TR4 14170143-1 transformer).

Looking at the schematic of these switchers, and if I understand correctly, the only thing the unit's power button does is select whether the unit is getting its power from the FW interface or externally. Other than that, it appears that each switcher is virtually identical, except for the output rail differences and the loading feedback loop via their respective opto-isolators.

That said, it's not clear to me what causes the switchers to initially oscillate and turn on, other than for the mere presence of DC power. I wish I had a running unit so that I could scope out what these things look like when they're actually running.

~JC

[Khron]

What the power switch does is enable or disable the digital power supply. It's the "S7" DPDT switch right there (and i'm looking at my mutilated Konnekt 8 board as i type).

Now, if i'm reading this correctly, in order for the digital supply to turn on, TP939 should be "high" (~12V), which would then make TP938 "low" (0V), which should then allow TP909 to float up - that looks to go to the input of the totem-pole driving the switching FET.

With the switch in the "off" position, TP939 will be "low", which makes TP938 be "high" and TP909 "low"), thereby inhibiting the digital supply from powering on.

For the analog power to turn on, TP926 should go "low", because TP902 / TP888 should be "high" (i'd guess 3.3V, since probably "ANA_PSU_ON" comes from the ATmega microcontroller). That (the ATmega) can only power on once the digital supply is active.