seppoman

well when using a center tapped transformer to get two different voltages, you can't really avoid having more load on the "lower half" of the transformer compared to the upper half, that's part of the concept. I guess as long as the transformer is strong enough to still have some "headroom", that shouldn't be a problem. When playing with values on that simulator, I noticed that when taking a lot of power from only one of the rails while the capacity is rather low, some flow back into the center tap could happen, which in turn might be bad for the capacitors. I'm not really sure how to configure the transformer part on that simulator to get realistic results (e.g. I found no way to set the transformer to other relations than 1:1). So if it works well in reality, probably everything is fine. If you wanna be on the safe side, in my opinion the solution is either using really a lot of capacity (let's say two or three 2200uFs per rail) or maybe after all changing the concept and using two bridge rectifiers, which wouldn't be a bad idea either. S

lyle, sorry I didn't want to sound rude - just in my impression you sounded still too much set on thinking of it as a regular bipolar supply cirquit. From theory I'd say it should definitely work but I didn't try it out in reality yet. I'm playing a bit with a cirquit simulator now (http://www.falstad.com/circuit/) and so far it looks like it should work that way. It seems like some larger capacitor value on the primary side wouldn't hurt if the cirquit should supply more than a few hundred mA, so maybe a second 2200uF cap could be added. S

no I don't think so. I've already explained what happens here - would you mind reading the rest of the thread before giving more comments? And if you still think it won't work, please explain why. S

Well I researched the still available S&H ICs during the design process of the AOUT_NG and there are not many options. If the replacement module should stay at least somewhat affordable and similar size, it narrows down to the AD SMP08 http://www.analog.com/static/imported-files/data_sheets/SMP08.pdf. That's the only 8 channel S&H that has 1 to 8 multiplexer, output buffers and hold capacitors already integrated (i.e. doesn't need any supporting cirquitry). It's listed as "production" at AD, available lead free (which means they don't intend to drop the model very soon or else they wouldn't have ported it to the lead free process), at least the SOIC version is available from Digikey and Farnell for 12-16 Euros. So a 64 channel module would be around 100 Euros just for the 8 ICs. I see the point of a S&H module for the 64 channel people, but I'm still not convinced people who need 24 or 32 channels wouldn't be better off with chained AOUT modules. The SMP08 is about the same price like the NG DAC and if you don't stuff the parts for the bipolar option, the rest of the module is quite affordable, without adding extra errors of S&H to the output CVs (I guess about 10 mV). The main problem that needs to be adressed first is that the MBCV firmware doesn't support more than 8 channels. Before that is changed, I don't think it would make much sense to work on a redesigned S&H module. If someone wants to work on the software side and does own more than one AOUT/AOUT_NG, chaining them is quite easy, both software and hardware-wise, so I'd suggest using chained modules for a start. When (or probably better: if) the software part is settled, I'd be glad to do a nice layout with the SMP08... :) S

I also wouldn't say that an 8 bit sampler is "under-specced". For the hifi stuff I still have my EMU 6400 sampler which does an amazing job at that and can be bought on eBay nowadays for less than any DIY sampler would probably cost. I'm really fascinated by the thought of a diy sampler, but even if a cheap i2s 16 or 24 bit codec was used, I'd say 8 bit sampling with 32 kHz and maybe 6 voices is absolutely sufficient. Better use the freed processing power for good oversampling/anti-aliasing/more voices/maybe one-shot streaming from SD/destructive functions/advanced synthesis features like granular or wavetable synthesis than for trying to achieve a commercial sampler's quality that's probably out of reach anyway. S

That's because you didn't extend the MB CV software to support that feature yet :P but seriously, I think it should be not too much of a problem to service 4 AOUTs from MBCV as this software has relatively few things to do in real time (compared to e.g. MBSID), so the PIC probably has plenty processing power for supporting more than one AOUT module. That app only isn't really in the center of focus for most people here (including TK), so nobody worked on these features yet. "It's a dirty job but someone's got to do it" ;) S

you've got the rotation of the rectifier wrong - the diodes must always poit towards the + rail :) And the upper cirquit should be the one with the 7809. But looks good otherwise - good luck :) S

the small caps around the regulator are there to prevent that the regulator starts to oscillate. Don't ask me about why exactly this could happen etc, this is just the standard implementation from the 78xx datasheet. But as long as both capacitors are in direct neighbourhood of the regulator, the order won't matter much. S

yes that's what I meant :) I'd suggest something like: 2200uF -> (maybe 47uF, although not really needed) -> 330nF -> Reg -> 100nF -> e.g. 10 or 47 uF (some small electrolytic does make sense here) -> if you like, another 100nF (also not really needed but maybe doesn't hurt). no it's still full-wave, because "the GND is full-wave" in relation to the center tap :) no problem, just keep us posted about your results/experiences :) S

But your output has only one GND pin and you're using a GND pour polygon, so in fact the "two GNDs" are connected to each other. Why's the fuse blowing? Imagine pin1 of your input is higher than pin2 which is in turn higher than pin3. In that case, the 5V rail is using pin2 as GND while the 9V rail is using pin3 as GND. So if "these GNDs" are connected to each other, you're getting a short cirquit across GND, because they don't have the same voltage potential. So what can be done about that? you need to make sure that after rectification you have three wires that always have the same polarity, just like with a bipolar supply. You could achieve this also by a correct useage of two bridge rectifiers but that introduces more problems than it helps anything. So just use one bridge like shown here (first drawing): http://www.midibox.org/dokuwiki/doku.php?id=bipolar_12v_psu. You can see there's a defined order of voltage levels, and if you just take the negative rail as GND reference, the other two are 12V and 24V above it (in that example - use the right transformer to get the right voltages). Of course the rest of the cirquit has to be different than in that bipolar drawing. On the right side of the rectifier, place your 7809 cirquit connected to "+..." and "-..." and the 7805 cirquit connected to "GND" and "-...". BTW, I guess you're having a bit too many capacitors. I dont think the 47uF primary and the 10nF and 1nF on the secondary side of the regulators do anything you would notice. If you're really keen on smoothing the output even more, you could introduce a so-called Pi filter right of the regulators, i.e. after the first electrolytic cap, put a large coil in series and afterwards a second electrolytic cap. But for that to have an impact in the audible range, the coil would need to have quite a high inductance and need to be able to endure at least 1A, i.e. it would be somewhat on the "large and expensive" side. Good luck :) S

gates? are you talking about the CV outputs on the NG or the gate outputs from core:j5? The gate outs should always be 0V or 5V, that's a logic signal. The CV outputs should be able to output either 0..11.7V in regular mode or around +/-5V in bipolar mode. S

that should be fine - I didn't measure how high the resistance across the supply is but if you'd get a continuity there you'd have a short on the supply rail and the LED wouldn't light up. Also, a current of some uA (micro-A, not milli-A, right?) between both grounds is nothing to worry about. Do you get symmetrical (analog) voltages if you connect only J4, i.e. without connection to the core? S

Core32 will provide two Midi ins and outs. I guess if that should not be sufficient you could expand that via 2c modules. no. The gm5 is a regular PC midi interface just like e.g. a M-Audio MidiSport, only DIY, cheaper and even better... S

J3 and J4 are the same, J4 is only for convenience, e.g. to supply a VCF module like the ssm2044. You can use it as an input, too. Your strange voltage levels are still about 30 V in total so I guess you've got a problem with GND reference level. Did you connect the GND of your bipolar supply, too? As long as bipolar and "regular" core GND are both connected to the NG (i.e. are interconnected), the voltage levels should be alright. Only question is why there's a difference of a few volts between the GNDs, so maybe you should make sure that there's no current flowing between the two GNDs (to check, disconnect one of the two GNDs from the NG, set your multimeter to current mode (A) and make that connection via the two probes). BTW, if you intend to use some CV outputs in bipolar mode while powered from +-15V, you should read this: http://www.midibox.org/forum/index.php/topic,10907.msg84344.html#msg84344 S

Yes Master :P First thing to say is: DON'T leave the box running for more than a few seconds while this problem exists - actually it would have been better not to switch it on again at all. Things like "turning off the box" usually mean that there's large currents flowing, like e.g. a short cirquit. If you're unlucky, you can fry everything! I just looked at the schematics of that supply cirquit and I think I've got something here: Apart from the fact that half wave rectification in conjunction with "only" 12V AC doesn't seem the best way to get a very stable voltage, and that from my experience, most switchable wall warts output DC (you're sure that it's really AC?), I think there's a problem/danger with using a mixture of half wave (MFOS) and full wave (core) rectification. If you look at both schematics assuming that the "upper wire" on the input is the same transformer output on both. On the positive half wave, WW2 (MFOS) and P2 (core) are "GND", but there's already a difference in GND level of around 0.6...0.7 V (because the return path of the bridge rectifier introduces a voltage drop of around 0.6V). Now on the negative half wave, WW2 still is GND, but the bridge on the core makes P1 become the GND. I'm not quite sure if that theory might be utter nonsense but I think this leads to the effect that on the negative half wave, there's a short cirquit between both transformer terminals that's going from WW2 through "GND" and out again through the rectifier to P1. Now, my first thought was maybe you could try to swap the wires going to the core, but I guess that only "mirrors the problem". But you could replace the bridge rectifier on the core with a wire connecting P2 to GND and a current-capable diode (like the 1N4004) connecting P1 to +. By this, you convert the core cirquit to half wave rectification and as long as the wiring is correct (i.e. WW2 and P2 are the same wire), this should fix your voltage levels. Hope you didn't damage any parts yet, good luck :) Seppoman

was soll das heißen, definiert? Das SID-Modul hat damit erstmal nix zu tun. Es wird halt verwendet, und die Software weiß nichtmal ob eins angeschlossen ist weil die Kommunikation nur in eine Richtung geht. deshalb kann also der Bootvorgang davon nicht ausgebremst werden. wieso schon wieder "erkennt"? Den Mios :D bzw. besser die SID-Applikation "launcht" das CS selbst, d.h. wenn es damit fertig ist weiß es natürlich das es fertig ist..... Mit der ID sollte das ganze eigentlich nichts zu tun haben, denn die "launching" Meldung wird nur bei ID 0 angezeigt, sonst steht "CS not enabled". die Frage ist, wobei er dann hängenbleibt - evtl Wackelkontakt zum Bankstick?

nein das kann nicht sein. CS heißt Control Surface. S

in some commercial units, the caps are glued to the pcb before soldering. you could try to melt one leg's joint and then apply brute force sideways in the direction of the other leg and so try to tilt the cap in a way that the heated leg comes up a little. then same thing in the other direction and alternate until the cap is lifted.

you could probably do that but would end up with a mono output signal :) S

question is for what? As I understood your question you want to use the keyboard to play the MBSID, right? for that you only need to connect the keyboard's midi out to the mbsid's midi in. The gm5 is just a PC midi interface like e.g. a MidiSport Uno etc., you don't need it to play the MBSID from a keyboard. If you want to have an additional PC connection via the gm5, you'll need to think about midi mergers etc. S

a GM5 is not suitable here (that's a midi->usb interface only for connection to a computer). You could buy a cheap real(!) midi keyboard (i.e. not one of these "midi" keyboards they sell nowadays that has only an USB connection) and connect the midi out (internally) to the core. SIDs were manufactured in various places all over the world, so you can't tell anything from that info. S

Nice :) Looks like SmashTV already composed a song for the AY ;D S

I guess you mixed up something with the pot wiring - if wired correctly, a normal 10k pot can't ever get hot or even explode. But if you accidentally connected Vcc or GND to the wiper (the middle terminal), you can short the PSU by turning the pot to one end. And a little bit before reaching that shorting position, you've got only a few Ohms of resistance between the supply levels resulting in a lot of power flowing through the thin resistive layer inside the pot. That would be a reasonable cause why the pots went POOF :) S

I didn't say their performance is inferior, at least it would be better than with the AOUT_LC module. They're just not really practical for an 8 channel design. For a 2 channel AOUT this DAC would probably be ok, but there's no PCB and no software support for it so you'd need to do all that on your own. For a cheap and basic 2 channel solution I'd recommend using the AOUT_LC module. S

Can't be used for what exactly? Sure they could be used for some AOUT stuff, but they're not currently supported by MIOS, they're only two channels and they're around $4 at Digikey in small numbers, so 4 of them wouldn't be really cheaper than the TLV5630 that is used on the AOUT_NG, plus they'd take up more PCB area. Of course there are dozens and hundreds of DACs that could theoretically be used to generate CV voltages, so you're free to layout a suitable PCB and write the necessary new MIOS driver for the DAC of your choice ;) S