Need help with curcuits for 5.1 wireless surround system

When you say "burned through" do you mean you can see burn marks as if something got hot?

From what you're describing it sounds like a short circuit. The ADAU 1401/1701 chips are 3.3V power, not 5V. I don't have Matt's drawings though and it's quite possible that 5V is the correct supply voltage.

Might sound like a silly question but how did you hook up your power? That sort of thing can be caused by accidentally connecting the positive and negative terminals the wrong way around and once it's done, it's too late. Most modules have an overload cut-out which will protect the regulator but nothing downstream of it. So if you destroy a board after the converter and it goes short, the short will cause the PSU to shut down until you remove it and when the short is removed, the cutout will reset, usually this is done automatically as the output is constantly monitored.

Can you do some photographs of your set up with particular attention to how everything is hooked up. 

@marcdraco

Thanks for the fast reply!

Unfortunately I can't upload any more attatchments today but I am sure I have not connected anything wrong, since it was all seperatly working. The ADAU input pin sais 5V and the individual DACs and ADCs also run on 5V, so I thought that was right.

The burned stuff I thought because I heared something and saw some smoke but I actually think now, that was just the fluxwater I used sizzling away with power running threw. However one ADAU actually did burn threw I think since it also has burn marks on some condensators but I'll check each bord again later. I think this could have also happened because the DC-DC converter maybe spiked if it failed.

Another suspicion could be the cheap ADCs. Maybe they failed? I had a similar problem with the cheap DACs since they also burned threw but after reading this forum and getting more expensive ones from linkfor from Amazon, they did work chained together.

It may be that you don't have enough posts to post images. We put everyone in a moderation queue for 5 posts to make sure they're not posting anything untoward. This happens a lot behind the scenes and poor Matt has enough to deal with without having to mess around with automated bots advertising everything from those "blue pills" through to spectacles. You can't make it up!

To get your post count up you could perhaps measure the voltage at the regulator while it's connected and see if it drops to zero when it's wired up. I assume you have a basic multimeter or this is going to be nigh-on impossible.

Smoke is never a good sign though. We don't call it "magic smoke" for nothing - it's the in joke for admitting we really got something wrong. Wiring a circuit with the wrong polarity is a sure-fire way to kill transistors and once they go short (which is often the way) they become nothing more than small resistors and get hot - very, very hot enough to make the resin break down and hence the smoke.

Ohh, ok that makes sense, cause I did post some images in my first post and I was wondering, where they went. I did some more testing and am utterly confused to where the fault is.

I put everything together again and by magic it all lit up and worked together in the center channel speaker, so lasers firing, dcdc lit up and ADAU blinked. Nice I thought, so I went to Sigma Studio, connected the USBi with GND, SDA and SCL, it compiled but upon clicking compile download it said compiled: comms failed. So I tried multiple things with this new error but eventually I gave up and unplugged.

Later I plugged it back in the socket and damn, ADAU was dark again :/

I tested some more using all new parts on a surround speaker. I wired up dcdc to 5v then that to ADAU and DAC, connected DAG output to ADAU input and it worked. Then I took my AMP and connected it with power and ADAU output to AMP input, also worked. Then I put everything together into the speaker enclosure and connected the 3 speaker wires and again, ADAU dark. I do know, that the speakers are not the problem, so I am very confused now.

Maybe I am too quick with testing with power and the ADAU/DC-DC have heat protection after soldering the wires? Which is also weird though, since I kept meassuring 5V output from the DC-DC to the ADAU even after the lights were off.

I think the smoked ADAU I have was actually indeed my mistake by accidentally connecting a wire wrong, since no smoke or anything came after my testing just now^^

Sorry, I'm a little unfamiliar with that build. Did you say you're driving speakers with an ADAU1701 module? I'm a little slow today.

The ADAU (we should just call it a DSP which is easier to type, LOL) will get warm when it's executing code but it should never get hot.

The key to any testing is to start in the middle and work out. It's called "divide a conquer" and is the way any professional repair guy will do stuff, unless we know the problem is in a particular module. D&C is about removing half of the issue (the faulty half) and then splitting that into two until it can no longer be subdivided.

Easier said than done with today's LSI.

If you're using a budget programmer (which, if I'm honest, most of us are) it could be the I2C lines are messing up. Keep them as short as practical - duponts are notorious for coming lose or going high impedance and that will mess it all up for you. That's as good as guaranteed.

V2 of the ultimate USB-C mic was sated to use a 1701 some time ago but I've found a lower-cost and simpler solution since that does what we need without costing the earth. Like most of us, I'm time constrained so I've got to play catch-up every time I get new boards back from the fab. 🙂

Ok, lets just focus on the center channel then. The DSP I'm using is the ADAU 1401/1701. In general I sticked pretty much to exactly the parts used in Matts video. Using D&C I think the problem must either be the DSP, the DC-DC or the Amplifier itself then or the wiring to it, since the laser block itself does work and the speakers should also be fine.

What exactly do you mean with I2C lines messing up?

I think I can upload an image now, so I attatched a horrible ms paint sketch of my Center Channel Speaker setup.

Great I'll take a look when I'm a bit more awake! 🙂

At least you understand using red for +Ve and black for -Ve. That's a help anyway. 

I2C is a low-speed Inter-IC bus that sends commands from the USBi to the AUDU1701 (or the EEPROM on the board). It can be finicky though. It's very like Arduino in this regard although the chip is a dedicated DSP, at it's heart it's still a custom processor much like a good ol' Arduino. I2S has some error checking built in and random noise (cough) because clone DuPont cables are just bits of poor quality wire (cough) a little bit of crosstalk might throw it. Ribbon cables are really a better option, particularly because it keeps all the lengths the same so there's no "skew" caused by the speed of the signals moving through the cable and arriving too early or too late. This is worse for higher speed lines like SPI of course because the timings are so much more critical. Even lowly low-speed USB can be flummoxed if you use a couple of random bits of wire to hook it up. Anyhoo...

Perhaps photos would be better at this stage now you can upload images and you won't get stuck in the moderation queue?

If the module is toast there's nothing really to be done other than replace the whole shooting match at this stage. Accidentally wiring the power lines is deadly to a chip; although it's quite straightforward to protect against these days it's often not used to keep costs down even though it's only a single P-type MOSFET. The idea is you ground the gate of a P-type power MOSFET and if the power is applied the wrong way around, the transistor remains "off". The old method was to use a diode but that causes a voltage drop of anything up to a volt - (usually 0.6V or 0.2V for a Schottky) but that's usually more than can be tolerated. Imagine dropping 0.2V from a 3,3 volt supply... or worse! The MOSFET, when power is applied correctly, acts like a tiny resistor of a few fractions of an ohm even if a large amount of current is being drawn

Most of the time when we summon Magic Smoke that's curtains for the chip(s) in short order (microseconds). The problem is only compounded when there are other things in the circuit like a programmer for example - that might get hit too depending on where the developing shorts propagate to.

If the DSP got hot enough to smoke flux off the board it's probably gone to silicon heaven. This would also explain why it's not accepting commands from Sigma studio.

The best way to check this out is to remove it the main harness and hook it up to JUST a power supply and the programmer and see if SigmaStudio can communicate.

It's also important to check UNLOADED output from your buck converter. Many modules have random set points so you need to make sure it's actually regulating the required voltage out. Another way to blow something sky high is to use too much voltage, particularly where there's a lot of current available - most 5V boards cannot tolerate more than about 5,5 volts and while 10% doesn't seem like much to us (a lightbulb won't go "poof" with 10% too much voltage) it's deadly to a chip. 

The unloaded voltage - that's the raw voltage as measured at the output terminals will confirm this, so you should probably check that first. 

I set the output voltage to 4.98V to be save, without anything connected to the step down converter and I still measure that now with and without anything soldered to the output -/+. Unfortunately I still can not upload more images since it tells me no more attachments are allowed today but I have some images prepared as soon as I can send them.

Yeah that image limit can be a bind sometimes. I don't have access to that side of it so I can't help. I'm just another user like you (only with moderation privileged to keep us ticking over while Matt is busy).

Voltage is spot on but the reverse polarity is probably the issue, it's almost certainly sent the DSP to silicon heaven. 

The thing is however, that I tested this with a new dsp bord, after I soldered the polarity correctly and the dsp was working. Then after plugging it in another time it randomly just stayed black. I bet if I would hook up the last one of the 5 dsp's I got now, it would light up and work again. So I am a bit afraid to do that even though polarity and voltage is correct, I don't want the last one's lifeforce to also randomly drain away 😬 

Ah, now that's interesting (sorry, I've been chin deep in FreeRTOS and I hate C++) but I digress.

So the boards are conking on you - right - it's unlikely you got a bad batch because these things are very reliable. Now a reverse polarity is a sure fire way to kill an I/C but now you've got that fettled, and the voltage at the regulator is OK.

When you say the 1701s (OK, I know those boards are 1401 but it's the same chip) are going dark something is wrong elsewhere in the chain.

I'm always a little leery when something like that goes south for no apparent reason. Are you soldering the board and then powering up? It's possible there is a voltage spike during startup until the PSU settles. Another weird one is when people power cycle something too quickly. In this event, the remaining voltage on the input capacitor can cause an over-voltage to the downstream circuits. 

What sort of test gear do you have? An oscilloscope would be really helpful here - just because a board is dark, doesn't automatically mean the whole thing has blown up. I did some digitisers a couple of months back and although the board was designed correctly and operated as expected, they were - in effect- blank.

I wonder if your uploads from Sigma Studio are putting the DSP into a race condition where it would appear to be dead.

I sadly do not have an oscilloscope, just a multimeter, multi purpose component tester and thermal cam. How could I buffer a voltage spike after the step down? With a condensator right? What uF value should that one have?

If I remember correctly I think I have tested bords before which were dark where I thought they were dead but then they lit up. What essentially could cause a bord to not show a light but then some other time yes 🤔 

I think I'll test all 5 bords again tomorrow when I find time just hooked up to a fresh step down converter

Great Scott (the YouTuber) has tested those little scope kits from Ali Express - about £30 or less. I wasn't impressed but it was my first digital scope after using (and still owning) some old analogue ones but he's a good guy and if they're good enough for him...

They're a really handy piece of test gear.

Anyway, the capacitor that causes the problem (potentially) is in the output end of the buck converter probably. It might be holding a little bit of charge sufficient that when things start up again, it goes over 5V for a small fraction of a second, but enough to send things do-lally-tap.

However, my guess is that its more likely to be something to do with Sigma if the board is holding its programming. DSPs are just highly specialised computers and they are just as prone to crashing as Windows 8.

Any complex software can crash of course and debugging a DSP isn't that easy. When I designed a 1701 board recently, the reference I worked with was quite specific about the reset timing and that could be an issue too. These low-cost boards don't have that and seem to rely on power-up conditions which are, naturally, random.

Here is a image of my center cannel setup.

I get an error , when trying to upload a new higher res image, so I can only send this one which was already in my attatchments.

In the image I disconnected the output wires to the amp and the input wires from the aux plug to the DSP

I know the laser block looks like a mess but it works and is well isolated from contacts^^

Oh don't worry about the spaghetti, you should see my workbench. I can't - it's covered in so much stuff.

Anyway, here's the likely culprit:

It's difficult to tell exactly what's going on without a scope. Simple voltage measurements only get you so far. However this is the EEPROM - the memory where the 1701 stores its program. I'll have to look back at the datasheet to see exactly how this functions but it does load the DSP from that on boot so it's possible (if not guaranteed) that an error in the code - and it only has to be 1 bit, that could crash the chip. 

I don't know anything about the internal workings (I haven't read the datasheet fully and not for a while) but it presumably should still respond to I2C instructions send over the USBi. Of course, this is assuming that the USBi didn't get dinged with the reversed supplies (it's powered by USB but it might also connect its 5V to the board which could have exposed it to an over-current, perhaps through the I2C interface. That might explain why you're unable to send commands to the board.

Ok, thanks for the help so far!

I think my problem shifted from hardware to Software now, since I have the center channel and a surround speaker set up and seaming to work.

Now my problem persists in SigmaStudio. When I connect the DSP via my USBi Jtag and open the channel tester, the USBi is recognised and green and I can also sucessfully connect to the DSP with Ready:Compiled -> Active:Downloaded. However I can't hear anything comming from my speakers. Not sure if I am doing something wrong, since I absolutely have no idea how SigmaStudio works except the basics but I unmuted all 4 DAC outputs and clicked the beep button in the workflow without anything happening audibly.

Am I missing something here? This happened on both Center Channel and Surround speakers

Ach. I wish I could be more help here but I don't a have a working model to test at the moment. It's been something on my ToDo list for a while. If you can post your Sigma project (I have the studio) I might be able to figure it out from that. No promises though as I don't have the hardware to test it on.

I'm still deep in development on the last bits of the microphone that I promised (cough) a couple of years back... EEK. It's reached the point where I can't improve the specs any more, it's already way beyond what we originally intended but juggling prototypes and sorting the last 20% of the issues is taking 80%  of the time and 100% of my beer fund. 

Can I just post stuff of the SigmaProject? I mean since I payed for the project's 3d- and sigmafiles via the link on Matts video, if I would post them here they would be public

The code? You can just send it to me in a PM if you prefer. I'll have a look as soon as I get a moment.

Ok, after some chatting and problem fixing, I managed to get everything fully assembled and working, so thanks so much for the help! For anyone reading this, my issues were:

-One time connected polarity wrong accidentally, which fried one DSP....oof

-Thinking the missing led on my DSPs after programming in SigmaStudio were dead DSPs but they all work like they should (still not sure, why they sometimes just did not show an led but the do work). Connecting everything up is very fiddely though and kept disconnecting, so I soldered on the wires for programming.

-Not noticing, that I also need to connect 12V to the REM port of the amplifier causing it to stay in standby mode the whole time.

I hope this helps someone 👍