madox

:) well put.

Hmm, I am in Oz too. I just assumed that postage would be OK, at the appropriate price. Postage from Germany to Australia is not the cheapest, though I have had no difficulty with this before. I also have had little luck purchasing these in Aus. I did manage to wrangle a couple of samples, but the local distributer was really not too pleased that they were not linked to a bulk manufactured product.

Hi dstamand and Sasha, Thanks for the thanks! Glad its appreciated. The flux I use is made by a company called Interflux. This is just the brand which my employer can readily purchase in Perth, Australia. Interflux was originally a Belgian company, though their flux manufacturing seems to be based in Asia now. I guess it would be available worldwide. The product I use is Repair Flux IF 8300. Here is a link to the retail front of the supplier which my employer uses. It is not really cheap for hobby use, but it is very useful. Even at the fairly high price, it is still really worth using flux for SMD ICs. As seen, this flux comes as a syringe of fairly thick gel/paste, at room temperature. The flux thins very quickly with heat. The problem with this type of flux is that one generally ends up applying more than is really needed. Also, one must be careful not to allow the flux to continue flowing out from the syringe. I usually suck the flux back into the syringe, and also recap the tip if I am not going to use it for a while. It does work well though. The flux which is used in the video that Thorsten posted, is a liquid flux. I think the liquid form allows greater ease in managing the small quantities which would usually be used. Unfortunately, I have only used the flux which my employer provides, so I can't offer a comparison of how they are to work with. The soldering in the video looks pretty good though, so I expect the liquid flux should work well. For hobby use, I think liquid flux will probably give better value for money, as using the gel/paste type usually involves quite a bit more waste than what is demonstrated in the video. I have not heard of diluting flux before, though it may be feasible. I use isopropyl alcohol for cleaning the flux from my PCBs once soldering is complete. If one has a nail-polish style brush dispenser, as used in the video, then I guess one could always have a try diluting some gel flux. When I use flux for work, the cost of materials, and the additional waste is not as significant a cost factor as the cost of my time, so I have not investigated alternative flux types at all. Sorry about that. I do want to try the liquid flux for home use, some time, though I don't have a strong motivation to do so while I am still in my current employment. I have also heard of water soluble flux, though I have never used it. I know there is water soluble flux core solder (not sufficient flux for SMD though). This allows for neat soldering, and PCBs to be washed in distilled water rather than alcohol. Another thing to note, is that Interflux claim that the IF 8300 I use, is "no clean". This is not true. The used flux probably could be left on the board, without harm (unlike acid flux), though it makes an unattractive scabrous looking mess. I use a large plastic fast food box as an alcohol bath, and a cut down cheap old paint brush to clean the finished PCBs. It should be noted that some components shouldn't be immersed in liquid, such as unsealed (or poorly sealed) pots, open wound transformers, etc. These devices may be able to handle an aerosole based cleaner, and good drying. Aerosoles cost more, of course, and also don't clean as thoroughly as the alcohol bath. My alcohol bath is less than an inch deep, and can be used for many PCBs. Hope that helps.

Hi dstamand:Just watched this video. Although this technique looks very careful, and neat, it is really too slow, and actually far from ideal. He has used too much solder paste, and then had to apply heat for a lot longer than is necessary. I believe that he has only used the microscope for the youtube video. This is not a practical way to solder an entire PCB. You will notice that he used the air gun at a very low angle of incidence. This blows hot air over the adjacent parts. With SMDs, this is very easy to blow other parts which you have already soldered, right off the board. Probably, he would not usually use the air gun at that angle, but had no choice due to the microscope apparatus. Also, rather than syringing solder paste onto the pads, it is very quick, cheap, and easy to pre-load the pads with solder from the tip of your iron. This leaves the pads with solid, rather than paste solder, but as soon as the heat gun blows on the pads, the solder liquifies. You can then easily place your chip into the pads, and hold the heat on for about a further 0.5 to 1.5 seconds (usually this is plenty). The nature of heat damage should be understood for this kind of thing. A very high temperature, applied for a very short time, is much less likely to cause deep overheating of a device, than a low temperature applied for a longer time. This is the same principle with performing laser cutting, or laser surgery. One only wants to heat the very surface of the materials being soldered. Deep penetration of the heat can cause damage. With soldering, there is a practical limit to how hot one can go, as material start to de-nature or burn at some point, which dissalows solder bonding. Well, I hope I am not coming off as a solder-control-freak here. Just want to share my experience. Cheers.

If I remember correctly, this has also been discussed in a previous thread (maybe regarding the OPL3 chips?). I do surface mount soldering in my work every day. It is really very quick, with a little experience. I strongly recommend it to anyone who has a bit of hobby experience, and will do more than a few PCBs. If you can paint a lead miniature, you can definitely learn SMD soldering. The real key to good SMD soldering, is using the liquid properties of the flux and hot (wet) solder. The flux reduces the surface tension on the wet solder, and allows it to flow, and bead very easily. Again, the key here is the improved flow that the flux lends the solder. The solder will tend to wick to the exposed metal surfaces. The physics of the behaviour, is that the wet solder is in a lower energy state when it is beading on the metal, than if it is bridging accross the FR4 of the PCB. It requires less elastic energy in the surface of the solder to be in contact with metal. In practice, the flux reduces the bridging behaviour, and like beads of water, the bridges can be moved with more flux and reheating. Flux is not cheap, but one only needs a really small amount in each application. So use it, but use it wisely. The difficulty with self etched PCBs is that there is generally no solder mask/solder resist to separate the pads from the tracks. These are a little more difficult than manufactured PCBs. They can still be done though. I found that without solder mask, one has to work a bit faster, when running the tip over the pads, as the solder will flow away. Also, re-applying solder has the risk of overloading pads, with potential overflow. I wouldn't recommend this as a first attempt at SMD soldering. SMDs only require a very small amount of solder. Everyone would have seen wave or oven soldered PCBs from an automated factory. These are done with a silk screened solder paste, and either EM wave heating, or oven heating. One can almost not see the solder. Another thing to be aware of, is the way surface tension pulls an SMD. This pull can sometimes be used to snap a device to sit nicely on its pads, but not always. Usually, any device should be held well with fine tipped tweezers, as the amount of surface tension pull can be really surprising. One technique which has not been mentioned here, is the use of hot air rework stations (cheapish ones from ebay are fine, pro brands are expensive though). These are concentrated hot air guns, with adjustable air flow and temperature. These are THE way to desolder large SMDs, and are also very good for soldering small devices. I use these stations a bit at work, and can solder a resistor in around a second in total. The heat is only applied for less than a second. In the same way, I can solder a dual op-amp (SO8) in around a second. Anything larger than an SO8 (SMD equivalent of DIP/DIL8) requires more skill to use a rework station, as there is the risk of overheating a device. Desoldering large SMDs, such as FPGAs, will generally compromise the reliability of the chip, which would generally just be disposed of. Of course, desoldering shouldn't be performed unnecessarily. I almost never use solder wick, and would not recommend it to beginners. Depending on how the PCBs are manufactured, some small surface mount pads are very weakly held to the PCB FR4 substrate. The pads can adhere to the wick, and tear away from the PCB. This can happen to the pads adjacent to the ones one is trying to desolder. I also wouldn't recommend a flat or sharp iron tip, as these can also damage pads/tracks when wiping solder accross the pads. Usually, to desolder a relatively small SMD, I will use two soldering irons at once, and heat all pads together. This avoids any damage to the PCB, and is also very fast, so heat will not need to be applied for an extensive duration. Some of the difficulties with SMD soldering are somewhat eleviated by using good quality PCBs. Cheap ones (which I use a lot of) do not bond the copper as well to substrate. For thicker track/larger pad through hole designs, this is generally less apparent. High pin density SMD PCBs may have tracks of only 5 thousandths of an inch, which have only minimal bond strength to a substrate. One reason I would really encourage hobbiests to learn SMD soldering, is that smaller devices and smaller PCBs reduce the cost of DIY electronics. Of course, when boards are not being made for a profit, the cost is paramount. Anyone could practive SMD soldering/desoldering with something like a broken PCI card. Just remove brackets first, possibly hold the card in a vice, and start removing/replacing devices. Hope this helps. -madox P.S. I also recommend watching the videos, as they are a great introduction, which I never had.

Hm, been there. Even DIY music costs. Oh well, an expensive, but satisfying addiction. -edit: spelling

That was my impression. It's a great chip, great MIDIBox project. Come on people, sign on, sign on! A pro quality MIDI-CV box is something which just creates so many opportunities and options in a studio.

Any chance the deadline can be extended? I would have thought more people would be interested in getting their hands on the Max525's. They're not so easy to find, if one doesn't have access to samples.

Hi Sonicwarrior, Would you please mark me down for 2 sets? Thanks, madox

No probs Chris. Gnerally with this sort of thing, most digital inputs will have a high input impedance, and digital outputs will have pretty low output impedance. FET inputs have some capacitance, which requires a small switching current during digital transitions, and only a very small quiescent current. Although many IC outputs won't have a large current sourcing capability, the input stages they are driving are so low as to allow the connection of multiple gates. Midi frequencies are also relatively low compared to what modern IC's are often designed to handle, which simplifies such connections. To evaluate specific design ideas, you can look at device data sheets to see what current drive and sinking to expect. Connecting multiple gates to an output places the input impedances in parallel. For this kind of application, very simple linear circuit theorems should be fine to see what is happening electrically. I will also have a look at the "In/Out indicator circuit of the MBHP_LTC", as TK suggested, as I am planning to design my own MIDI interface with these GM5 chips. I'm guessing this is just some CMOS gates on an IC, which will be performing the same function as the FET I described. Using IC's like this, will usually simplify the connection of multiple LED's, such as one for each input/output.

Does this have to be cheap? Are you planning to build only 1 device?

OK, I have probably just misunderstood the situation with the MIDIBoxCV. I thought there had been variations on the design for driving filters, etc, or different output ranges, MUX/sample and hold? I haven't read the project documentation for at least a year, so probably I am the one who is confused. Sorry for any misunderstanding. madox

Hi Chris, You definitely don't need any additional optocouplers to do this. The optocoupler is really just to 'couple' the input. You can think of the optocoupler as an LED and a photodiode/phototransistor within an enclosed package. The input drives the LED, which is optically coupled to the photodiode/phototransistor. When current runs in the input line, the LED transmits light to the photodiode/phototransistor, which is then connected to the UART input. Optocoupled inputs have some useful properties: -Isolation of potentially damaging electrical energy -Elimination of ground loops -Rejection of common mode noise (have I missed anything?) Depending on the circuit, there may be somewhere you can directly connect a resistor & LED. Otherwise, it is quite likely possible that you could tap a FET onto the signal you are monitoring, which drives an LED with a current limiting resistor. The gate of a FET provides a high impedance input, so it generally won't create any appreciable loading on a digital signal line.

I seem to have added myself on the list twice. Just removed my second entry, and reduced my order to one set of panels. edit- appologies for messing with the list, Doug.

Hi Thorsten, Thanks for your encouragement/vote of support. madox

Has anyone got a lead on a supplier for SSM2044? These chips go for about $6 ($7 w shipping) on ebay. Not too bad, really, but I guess a bulk order from a larger quantity supplier could be better.

Hi widdly. Thanks for the question. The merge component of the device would indeed be limited to use with the PC. This is deliberate, as I have quite limited space, and don't want to use additional cable connections either. Also, I almost exclusively use MIDI with my PC in the setup. So far, I have never gotten far enough with my music to consider gigging, etc. Merging in my DAW would probably be fine, though I would also like to be able to handle this with low latency dedicated hardware, and also provide more inputs for future controllers. I am aiming at flexibility. In the past, when I have worked with other people in my home studio, there have always been issues with limited MIDI inputs to my PC. I am aiming at allowing for multiple concurrent controllers to be available to each user, for MIDI controlled jamming. The 5 inputs from the GM5 would probably be sufficient, though I am trying to be future proof with this interface. This would allow me to have four or five additional inputs. This may be overkill. I generally try not to load my PC with anything which I can handle in dedicated hardware. My PC is getting long in the tooth, and I like to free its resources for sequencing, soft synths, and hopefully real time soft effects. To date, I have run into issues with my PC's performance, either with audio glitches, or timing errors. Another motivation for me, is to cut my teeth with MIDI process programming with this project. I think it's a fairly simple one for me to familiarise myself with the requirements. Cheers, madox

Thanks, I hadn't checked the list for a while. Added a couple of PCBs for myself, and corrected the totals.

Hi folk, I am planning to design a PIC based midi merge unit, for use with a GM5 based PC midi interface, and am just wondering if others may be interested, and possibly have some suggestions. The project would depend on the outcome of TK's assessment of the GM5 chip, and subsequent bulk order. Following on from this thread, I am planning to build a stand alone PC midi interface, with 5 banks of midi i/o, plus a custom midi merge unit which could be set to merge with the input of any one of the 5 banks, plus a midibox midimon, which could be set to monitor either the midi input or output of any of the 5 banks. I may possibly consider incorporating a switch to route the fifth bank to an internal midi-CV converter, though this option may be left to a later date, and possibly just sit on a secondary PCB, and be connected with a hardware toggle switch. The design would be realised with the existing GM5 solution, and a 4 UART PIC24Fxxxx based midi merge unit. My motivation is to be able to use a combination of midi controllers (namely 2 keyboards + midibox64, + rack unit x?), all merged into a midi stream into my PC DAW, and then echoed out as necessary, to control softsynths, or other midi hardware. My proposed unit would not perform the role of a midi patch bay, as my DAW would perform the data echo function. The 4 UARTs of the PIC24F could conceivably be used to merge midi streams to more than one midi input (eg, merge inputs 1+2 merged to interface input 1, merge inputs 3+4 merged to interface input 4). However, I'm not sure how feasible this would be in practice. My aim for the prototype is to use only hardware based UARTs (no software emulated UARTs), and have all inputs merged to a single output. Later I will try the more complex scheme, as I expect the processing shouldn't place too much demand on this PIC. The merge unit would not be MIOS based (not supported by the OS), though I will have a look at TK's code for his midimerge unit to get some ideas. This will be my first attempt at programming a midi processing unit, and though I don't expect excessive difficulty, I'm really not sure what problems I will face. I have written several PC and PIC based serial comms programs before. The unit will be a single PCB, predominantly surface mount components. I know this is not in keeping with the midibox way, but this is really the only way that I can afford to design the unit. I use almost exclusively SMD for my work, so this is what I have ready access to, and I really need to keep the PCB size small, to minimise the prototyping costs. As a side note, I really like using SMD components. Although the initial learning curve can be daunting, I would never go back to predominantly through hole. Too much of my work just couldn't be done without SMD. I plan to fit the unit in a 2U rack enclosure, with one or two LCDs, and one or two 7 segment BCD displays. All source material would be publicly released, and all external IP would of course be correctly credited. If anyone has any suggestions, recommendations, or criticisms, I would really appreciate your feedback. I'm still brainstorming with regards to the features and implementation at this stage. Cheers folk, madox

Damn good demo. I guess when I biuld my SID box, I'll leave some room to contain such a setup. Most impressive.

Yes, have to agree. Buffer size is not likely to be a real problem, but I guess it depends what one plans to use the interface for. Personally, whenever I am going to send much sysex data (bulk dump, firmware update, etc), I will make sure that there is no other midi trafic at that time. I am currently planning to make a midi merger with a few more I/Os, to go with the USB midi chips currently being ordered in the bulk orders forum. I will most likely use one PIC24Fxxx for each merger unit. These PICs have 4 hardware UARTs, and I will probably use one or two software driven UARTs as well. I don't imagine RAM limitations will be an issue, as midi is a rather low data rate, and being real time, one should not be buffering a great deal of data anyway. If/when I finish my design, I will make it available online. However, I'm pretty sure that existing solutions would be able to meet your requirements, unless you are planning a different kind of usage from what I am expecting to be the way that most people use midi.

I've done a very crude sampler of sorts, using a PIC. It was actually for voice annunciation on a radio coms system. I did mine with a MSSP of the 18F4520, connected to a 1 Meg flash chip. For my design, the audio quality was not an issue, so I just used a filtered PWM output to reproduce the sound. The same thing could be done with a cheap 16 bit audio DAC. Having said this, I would have to agree with stryd_one. Software sampling, or second hand hardware samplers will offer a much more rewarding result. I would say the difficult part of doing a DIY sampler is not so much the sample storage, reproduction, or midi event triggering, but in providing a convenient interface. Cheap hardware samplers can have reasonable audio manipulation/effects on board, and some have really good filters as well. This kind of project would probably be a pretty satisfying hobby project, but would not be likely to give as good a result as a commercial solution. So, I guess it depends on what you want. Personally, I would probably put the effort into a weekend job, and buy something good on ebay. If I were to make a sampler, I would probably look at different hardware, such as a DSP Spartan, or a more fully featured microprocessor, with external Alesis audio DSPs. Hope that's not too disparaging. Regards, madox

Personally, I wouldn't mind too much if the release date is staggered. As the timing and cost are both unkowns, it is easier to fit these outlays into a budget if they do not occur at the same time. I don't know if other people would share my perspective on this.

I imagine you could use a cheap USB hub to do that. edit- P.S. With 5 midi banks per USB chip, and 16 midi channels per midi bank, you are talking about 2 x 5 x 16 = 160 midi channels, right? Do you actually have so many midi devices? I think 5x16 is already many more than I can use, for one PC midi interface.

Sorry, didn't mean to rush things. Of course the circuit would be simple, but it is handy to have the datasheet. I'll definitely be making my own PCB for 5 bank midi interface, probably rack mounted. If I can get around to it, I may incorporate a fixed midimerger, and a switchable midimon. Looking forward to it. :)