jjonas Posted March 23, 2014 Report Share Posted March 23, 2014 Hi, I've recently built myself an AOUT_NG which I've been calibrating. I noticed that the TLV5630 gets very hot, so that after 3 seconds you have to remove your finger off of it in order not to get burnt. I noticed this only after 20 minutes of fiddling with MBSID settings, and during that time I managed to get the frequency filter working (not yet resonance; I'm using a self-made seppoman SSM2044 PCB). So it didn't burn out the chip in that time, and it seems to be working as well, so I'm asking whether it's normal that the chip gets very hot. The AOUT_NG board draws 560mA (and the LED lights up normally), when it's attached to the MB6582's J6_CORE4 and nothing else – to me that sounds like a lot, but I'm not sure how analog synth stuff is supposed to operate :-) Quote Link to comment Share on other sites More sharing options...
Altitude Posted March 23, 2014 Report Share Posted March 23, 2014 That's not right, 0.5A current draw and the heat usually points to a short somewhere.. Quote Link to comment Share on other sites More sharing options...
jjonas Posted April 21, 2014 Author Report Share Posted April 21, 2014 Hi, I've continued to troubleshoot this problem, which still persists. I couldn't find any shorts anywhere, so I desoldered the chip and put a new one in its place, though not on the AOUT_NG board (some of the pads were damaged while desoldering) but on a daughterboard that I got from OSHpark. I soldered this onto the AOUT_NG with wires, and tried whether the new chip gets hot. It didn't, not even after a while, so I started recalibrating the AOUT_NG board. All went ok, calibration was finished successfully without the chip getting hot. I guess it took an hour or something. Now I attached the seppoman SSM2044 pcb that I got from OSHpark as well, one that I made based on the PCB picture and the schematic that are available on the wiki, and started calibrating it with MB6582. Everything seemed to work, cutoff was ok, only right channel resonance wasn't working. At this stage the chip was still ok, i.e. not hot. When I finally found what was wrong with the right channel's resonance (C20's legs were soldered to the same point), at the same time I noticed that the new TLV chip on the daughterboard was getting hot, and when I tried whether the filters were still working, they weren't (I had been inspecting the filter PCB for mistakes for some time). TLVs are kind of expensive to be fried like this while troubleshooting, so I'm not sure what the best course of action would be. I have one more chip left, it's already soldered onto a fresh AOUT_NG board, I thought I'll test it by attaching the complete and calibrated board to the MB6582 and leaving the power on for several hours, and if it's still working, maybe at least the AOUT_NG is ok. After that the decision of whether to use the DIY SSM2044 board or something else is open. If people have ideas for external filters to be used with MB6582, other than self-drawn PCBs, I'm all ears :-) (Giving up on the SSM2044 would be ok for another reason too – like I found others pointing out in another thread – namely it cuts the bass when resonance is turned up.) I don't know if anyone has the time to check the filter PCB (in case that's the culprit), but anyway here it is (I'm not really asking anyone to comb it through). It's more or less a direct copy of seppoman's board from the wiki, though it's possible I've added a few mistakes :-) Anyway for me the mystery is that for a while everything was working, and as far as I could tell, voltages were ok (on both AOUT_NG and the filter PCB) Quote Link to comment Share on other sites More sharing options...
latigid on Posted April 21, 2014 Report Share Posted April 21, 2014 Can you upload a picture with the ground plane poured? Some connections look a bit funny. Quote Link to comment Share on other sites More sharing options...
Shuriken Posted April 21, 2014 Report Share Posted April 21, 2014 Isn't the PCM1702 meant to be an Audio DAC? Quote Link to comment Share on other sites More sharing options...
jjonas Posted April 21, 2014 Author Report Share Posted April 21, 2014 Shuriken: I don't know, I just needed a component that is of the same size as TLV5630 (SOIC-20), and that's what I found in the component databases I had installed. latigid on, here's the same circuit with GND. Quote Link to comment Share on other sites More sharing options...
Imp Posted April 21, 2014 Report Share Posted April 21, 2014 (edited) Did you solder the parts that connect to wires on top and bottom layer (e.g. left leg of R16) on both sides? btw. I'd rather not troubleshoot Aout and filterboard at the same time. Better test the Aout with a multimeter and the Filter with some kind of makeshift voltage source (pot as voltage divider). This way one won't fry the other. Edited April 21, 2014 by Imp Quote Link to comment Share on other sites More sharing options...
Altitude Posted April 21, 2014 Report Share Posted April 21, 2014 Hmm. Have you run the ERC and DRC for OSHpark on that? That isolation looks really tight and recipe for shorts. Also, I cant tell if this is an artifact from the screen capture or a short.. Quote Link to comment Share on other sites More sharing options...
jjonas Posted April 21, 2014 Author Report Share Posted April 21, 2014 Altitude: Thanks for checking, that's an artifact that's visible also on Eagle at this level of zoom. I ran only DRC, it produced a few errors like this, all of which but one are visible here. Do you they matter..? I kind of thought they don't, but I might be wrong. If they do matter, is it clear to you how they would matter in this case..? Imp: Sorry I don't understand your question: "Did you solder the parts that connect to wires on top and bottom layer (e.g. left leg of R16) on both sides?" Your idea for testing SSM2044 board separately like that is definetely something I'll try! Quote Link to comment Share on other sites More sharing options...
Altitude Posted April 21, 2014 Report Share Posted April 21, 2014 (edited) what's layer 20? Those sort of errors shouldn't matter. What does the ERC say? I'd also get the DRC file from Lain, it will go beyond the stock DRC. Also: put your meter between +5 and GND, +12 and ground, -12 and ground an make sure you dont have continuity. I dont know the details of what the short circuit protection there is on the DAC but if there isnt any, a short on either of the 12V rails would cause that failure so poke around those traces Edited April 21, 2014 by Altitude Quote Link to comment Share on other sites More sharing options...
Imp Posted April 21, 2014 Report Share Posted April 21, 2014 What i meant was: If a connection between top and bottom layer is needed, you usually use a via. If you do it like at the left leg of R16, you need to solder that leg on both sides of the pcb. Quote Link to comment Share on other sites More sharing options...
Altitude Posted April 21, 2014 Report Share Posted April 21, 2014 OSHpark PCBs have plated through holes, no need to solder on both sides. Quote Link to comment Share on other sites More sharing options...
Imp Posted April 22, 2014 Report Share Posted April 22, 2014 Ah, ok. One less potential source of trouble. Quote Link to comment Share on other sites More sharing options...
jjonas Posted April 23, 2014 Author Report Share Posted April 23, 2014 Hi, re testing the filter PCB with an external PSU setup, I don't have a functioning AOUT_NG at the moment, so I can't check this, and that's why I'll ask: which voltages do I need to "emulate" AOUT_NG control voltages, both unipolar and bipolar? If I recall correctly, unipolar range is 0–10,67 VDC, and seems easy enough to emulate with a single pot, but what about bipolar? Somehow unipolar control voltage seems more intuitive to understand, but I don't really get how a bipolar CV works, and so I can't really improvise a test setup involving it on my own. Further: on the wiki page for SSM2044 PCB it says: "If you want to use this module [seppoman SSM2044 PCB] together with “non-AOUT†CV sources, you will need to add the control summer circuit externally." (Description of the "control summer" is in the 'Design' section of the entry.) Are there consequences from this to a test setup? Quote Link to comment Share on other sites More sharing options...
Imp Posted April 26, 2014 Report Share Posted April 26, 2014 That remark about "non-AOUT CV sources" means that you need another circuit, if you want to archive V/Oct scaling of the frequency. It#s not needed for testing purposes. Bipolar CV means the CV can be anything between -5V and +5V in relation to ground. If you have a bipolar PSU with a center tap, (e.g. (+12V)-0-(-12V)) you can connect a pot to +/-12V with the outer pins. Then your CV is between middle pin and ground. Quote Link to comment Share on other sites More sharing options...
jjonas Posted April 28, 2014 Author Report Share Posted April 28, 2014 Ok, I'm now testing a new AOUT_NG board with MB6582. I thought I'll do it in stages, as previously everything seemed to be operating ok for an hour or so. So now I've connected the AOUT_NG only to J6_CORE4 – the AOUT_NG led is on, chip not getting hot. I thought I'll leave it like that for a few hours, then add +12/-12 and see what happens. I tried Imp's test setup to test the SSM2044 board on its own yesterday. The test setup seemed to be working ok, though not the board's second resonance channel :-) I'll have to spend some more quality time with it. Does it btw matter – for non-audio technical reasons – if the PSU I'm using is the Meanwell RPT-60, i.e. a switching PSU..? It's the same I'm using for the MB6582, so it would be most convenient to use that for the AOUT_NG and filter board as well. Quote Link to comment Share on other sites More sharing options...
Imp Posted April 29, 2014 Report Share Posted April 29, 2014 My mbSID has an aout_lc and the SSM2044 board built in, all fed by the same supply. It works. It's no switching PSU, but i don't think that makes a difference. Quote Link to comment Share on other sites More sharing options...
jjonas Posted May 9, 2014 Author Report Share Posted May 9, 2014 Hi, I decided to give up on AOUT_NG because I've already fried two chips (maybe three, I'm not sure, and not keen on testing whether it really is broken), tried two separate AOUT_NG builds without getting it working on the long run, and I think at some point I blew the CORE4 PICs J6 as well – the PIC seems to be operating normally otherwise, only the J6 doesn't seem to send out any information. So inspired by Imp's posts in this thread I tried out two AOUT_LCs, as I figured I'd much rather fry 595s in a socket than TLVs surface mounted on a PC :-) I hadn't really considered the option before, as there was much more info around for calibrating SSM2044 with an AOUT_NG than with other options. So, I changed a few lines in the setup*.asm, recompiled etc. It works. At first it seemed it didn't work either, but then I tried it on CORE3, where it did work. So it was here that I noticed that the CORE_J6 was not working, and I reckoned it was because of my AOUT_NG tests. The SSM2044 board's problem was that for some reason the LM13700's other side was not working, which was fixed by switching in another one. Here it is, the external filter box gets all it needs via the extension port. The red LED & switch are for +/-12VDC on/off, and the green one for whether the filter is set to on or bypass (via J5 digital switch pin). Thanks for all who helped with troubleshooting! Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.