Spooky, Perhaps you should start a build thread to discuss your design choices. My questions: (intended as food for thought)
Why add an op amp to accept balanced input when you already have a differential pair input stage? Lift the feedback network ground and there's your inverted input. See the Pass A75 article for discussion about impedance matching the two legs.
Why TO-3 output devices? They may be more rugged, but mounting them on an angle adds another thermal junction, losing some of the advantage. You could fairly easily substitute 5 sets of MJL0302/1381 and be as rugged and simplify construction. MJ150034/35 have higher hfe and would be a better choice with higher numbers of output devices - more current available to drive the outputs.
Have you recalculated the output protection circuit resistor values? Those look like Leach V4.5 values, but things change with more output devices. You may find that you want to reduce R28 and R30 and/or increase the values of R41-43 and R60-62.
Have you considered higher hfe input devices? At 60V rails you could use BC546C and BC556C. The more available B types have higher hfe than the MPSA42/92. Not sure if they are the only reason my new Leach (Jens' 5 pair boards) sounds better than my V4.2 Leach, but it probably contributed.
Why add an op amp to accept balanced input when you already have a differential pair input stage? Lift the feedback network ground and there's your inverted input. See the Pass A75 article for discussion about impedance matching the two legs.
Why TO-3 output devices? They may be more rugged, but mounting them on an angle adds another thermal junction, losing some of the advantage. You could fairly easily substitute 5 sets of MJL0302/1381 and be as rugged and simplify construction. MJ150034/35 have higher hfe and would be a better choice with higher numbers of output devices - more current available to drive the outputs.
Have you recalculated the output protection circuit resistor values? Those look like Leach V4.5 values, but things change with more output devices. You may find that you want to reduce R28 and R30 and/or increase the values of R41-43 and R60-62.
Have you considered higher hfe input devices? At 60V rails you could use BC546C and BC556C. The more available B types have higher hfe than the MPSA42/92. Not sure if they are the only reason my new Leach (Jens' 5 pair boards) sounds better than my V4.2 Leach, but it probably contributed.
Hi,
I'm glad you're dropping in.
I considered this and I might end up doing it, as this thread may not be the most appropriate and I may take time away from others.
I didn't know how to make proper use of that other side of the diff amp, so I didn't mess with it, but of course if we can rig it that way, why not? But then how do we handle the case when the line doesn't come in balanced on the XLR? There would be some need for a switch or jumper then, wouldn't it?
I know, and I made that choice only because I have plenty of those devices on hand and I wanted to make use of them instead of buying even more devices while leaving the old stock alone. The fact that I added an extra thermal junction bothered me, but since I was out to overbuild and I chose rather large sinks, I preferred using my existing devices.
When I make the other versions that I planned to make, I will use other devices, but I want to use those in stock for now.
Not yet, and that's one of the big things that I still need to do now. I was also considering the V-I limiting, so I provided for the extra resistors on the pcb in case I was going in that direction.
I also made use of those non inductive MPC722 double resistors, very nice and space saving, plus since they have a common pin, that suits very well the output stage configuration.
I also have those MPC722 in 0.22 + 0.22 ohms, so the protection scheme definitely needs to be properly calculated.
I plan on doing a serious matching on the transistors, all of them, including the input ones, since I have all those in stock and sufficiently large numbers, no problem with making a selection.
With the matched transistors, the lower emitter resistor values will be less of a concern and will mostly serve for the protection.
True, and I was actually considering using the 2N5401/5551 types, since I also have those and they have a slightly better hfe than the mpsa. I also am reluctant, may wrongly, in using the BC546 because of their lower Vce0.
This current design with the MJ15003/4 is only planned for 60V rails (toroids provide around 43Vx2), since I do have the transformers on hand as well.
Later on other versions, I will use beefier transformers, actually I'm planning for 2 toroids per amp, which explains how the amp module is prepared for this with the room below the pcb support plate, with the mounting holes already there now.
I also have on hand a bunch of MJ15024/5, so that's going into the next upper version, keeping the same case design. And finally, I also have a few MJL21193/4 but not enough to allow matching. Those will make for an other version later.
Could be, as the transistors do make such a difference.
For now I want to make use of the ones on hands, and later I will get the better ones and alter the design for them.
I need a good number of amps anyway, many for myself, but also for friends around, so they will all find a good home.
I am continuing on my plans for a 4 ways active system. That's a bunch of amps!!!
I'm glad you're dropping in.
I considered this and I might end up doing it, as this thread may not be the most appropriate and I may take time away from others.
I didn't know how to make proper use of that other side of the diff amp, so I didn't mess with it, but of course if we can rig it that way, why not? But then how do we handle the case when the line doesn't come in balanced on the XLR? There would be some need for a switch or jumper then, wouldn't it?
I know, and I made that choice only because I have plenty of those devices on hand and I wanted to make use of them instead of buying even more devices while leaving the old stock alone. The fact that I added an extra thermal junction bothered me, but since I was out to overbuild and I chose rather large sinks, I preferred using my existing devices.
When I make the other versions that I planned to make, I will use other devices, but I want to use those in stock for now.
Not yet, and that's one of the big things that I still need to do now. I was also considering the V-I limiting, so I provided for the extra resistors on the pcb in case I was going in that direction.
I also made use of those non inductive MPC722 double resistors, very nice and space saving, plus since they have a common pin, that suits very well the output stage configuration.
I also have those MPC722 in 0.22 + 0.22 ohms, so the protection scheme definitely needs to be properly calculated.
I plan on doing a serious matching on the transistors, all of them, including the input ones, since I have all those in stock and sufficiently large numbers, no problem with making a selection.
With the matched transistors, the lower emitter resistor values will be less of a concern and will mostly serve for the protection.
True, and I was actually considering using the 2N5401/5551 types, since I also have those and they have a slightly better hfe than the mpsa. I also am reluctant, may wrongly, in using the BC546 because of their lower Vce0.
This current design with the MJ15003/4 is only planned for 60V rails (toroids provide around 43Vx2), since I do have the transformers on hand as well.
Later on other versions, I will use beefier transformers, actually I'm planning for 2 toroids per amp, which explains how the amp module is prepared for this with the room below the pcb support plate, with the mounting holes already there now.
I also have on hand a bunch of MJ15024/5, so that's going into the next upper version, keeping the same case design. And finally, I also have a few MJL21193/4 but not enough to allow matching. Those will make for an other version later.
Could be, as the transistors do make such a difference.
For now I want to make use of the ones on hands, and later I will get the better ones and alter the design for them.
I need a good number of amps anyway, many for myself, but also for friends around, so they will all find a good home.
I am continuing on my plans for a 4 ways active system. That's a bunch of amps!!!
btw: I'll go for that, but what is that article you're referring to? Very interesting I'm sure.
If I can properly make full use of the diff amp's inputs for a balanced input on the xlrs, then I would definitely consider removing that sym line receiver, which would remove quite a few parts at the same time, since I could do away with the extra power supply for it...
Thanks for dropping in!
Now I think it's time to take a close look at the protection and all the calculations.
The fact that there are more than one emitter resistor with the "sensor" resistors on each of them, makes it more complex to calculate and I'm not comfortable enough with this. So that's what I'd like to work on.
Plus there is the option of doing it as V/I limiting, which I also want to explore.
This can be interesting for diyers who like me want to know how to do it and not count on others to do it for them without understanding how it's really done.
I have access to a mathcad setup for now, so I am trying to use this to make some graphs and calculations, but I'll need some help figuring this out properly, as what I'm doing may not be quite right. I'm on Mac, so no MathCad for me, but I'll touch that windows thingy anyway (yuk!) just to get the work done.
Here is a graph I managed to make with it. I plotted the SOA from datasheets for the MJ15003/4 in black. I think I may have this right, since it looks like it's matching the datasheet's graph. Then I plotted 2 sets of load lines, one for Vcc at 60V for the 60V rails that I'm using on this current design, and the second for 70V.
I don't know how to add captions on the graphs, so I'm just going to provide the legend in text.
For each set, I plotted for 8 ohms and 4 ohms for a pure resistive load, which I may have gotten right, and I also tried plotting a 100% reactive trace.
The green traces are the 60V Vcc for 8ohms resistive and that possible 100% reactive load situation. I need to make sure I get this right, as it's quite near the limits. The red trace is just the 4ohms resistive on 60V and the orange and brown are the 70V Vcc versions for the 8 and 4 ohms.
Then there is that blue trace, which exceeds the SOA, which is supposed to be the 100% reactive condition for the 70V Vcc on 4 ohms.
All these traces are for 3 pairs of MJ, and that is also something that I need to double check to make sure I got it right.
If this is correct, then the 8ohms 100% reactive situation on 60V rails passes with no problem, but the 70V rails would pose a problem.
On pure resistive there are no problems anywhere...
The fact that there are more than one emitter resistor with the "sensor" resistors on each of them, makes it more complex to calculate and I'm not comfortable enough with this. So that's what I'd like to work on.
Plus there is the option of doing it as V/I limiting, which I also want to explore.
This can be interesting for diyers who like me want to know how to do it and not count on others to do it for them without understanding how it's really done.
I have access to a mathcad setup for now, so I am trying to use this to make some graphs and calculations, but I'll need some help figuring this out properly, as what I'm doing may not be quite right. I'm on Mac, so no MathCad for me, but I'll touch that windows thingy anyway (yuk!) just to get the work done.
Here is a graph I managed to make with it. I plotted the SOA from datasheets for the MJ15003/4 in black. I think I may have this right, since it looks like it's matching the datasheet's graph. Then I plotted 2 sets of load lines, one for Vcc at 60V for the 60V rails that I'm using on this current design, and the second for 70V.
I don't know how to add captions on the graphs, so I'm just going to provide the legend in text.
For each set, I plotted for 8 ohms and 4 ohms for a pure resistive load, which I may have gotten right, and I also tried plotting a 100% reactive trace.
The green traces are the 60V Vcc for 8ohms resistive and that possible 100% reactive load situation. I need to make sure I get this right, as it's quite near the limits. The red trace is just the 4ohms resistive on 60V and the orange and brown are the 70V Vcc versions for the 8 and 4 ohms.
Then there is that blue trace, which exceeds the SOA, which is supposed to be the 100% reactive condition for the 70V Vcc on 4 ohms.
All these traces are for 3 pairs of MJ, and that is also something that I need to double check to make sure I got it right.
If this is correct, then the 8ohms 100% reactive situation on 60V rails passes with no problem, but the 70V rails would pose a problem.
On pure resistive there are no problems anywhere...
Attachments
Of course, that's why I'm doing all this, it's not just to fool around, I have many of the parts already, which is the main reason behind some of the choices, plus I am right now working with a contact of mine in china who can handle the sheet metal work, they have the equipment, including the laser cutting table.
This will be nice, and I plan on other versions, but keeping a good part of the case the same to limit costs.
Very nice. So your planning of selling them!
I just see a lot of projects that never get off the ground.
I have 2 toroid transformers for this amp, 470VA and about 43V with no load, so my rails should end up close to 61V unloaded. I don't know how far they'll sag under load, but the toroids are pretty efficient and don't sag that much, so I'm expecting a power somewhere in the 180-200watts on 8ohms. We'll see how that goes when it's built.
The main heatsinks are Fischer SK479, the smaller heatsink for the pre-drivers is an SK546 and the one on the rectifier bridges is an SK597. The angled profile for the TO3s to link up to the main heatsinks is also from fischer, SWP55 anodized black.
The plans I'm making aren't just for show, they are precise and ready for manufacturing. I just forwarded the cad drawings to my metal worker in china and we're chatting about it as I'm writing this...
The main heatsinks are Fischer SK479, the smaller heatsink for the pre-drivers is an SK546 and the one on the rectifier bridges is an SK597. The angled profile for the TO3s to link up to the main heatsinks is also from fischer, SWP55 anodized black.
The plans I'm making aren't just for show, they are precise and ready for manufacturing. I just forwarded the cad drawings to my metal worker in china and we're chatting about it as I'm writing this...
I'm not doing this to make money. But if there is interest, then a group buy on some of it can be arranged to make it more affordable.
True, some projects never get to the end, but not all.
I'm arranging to have all the case parts made in china right now, so I will make this, and others. A group buy would be good to keep a handle on costs, as a single unit costs a lot more to make.
I already have a close friend who wants to get one from me, and there are a few others who might go for it as well.
If we can group buy for 8, 10 or 15 units on the case, the shipping costs would be drastically reduced and the manufacturing as well. I have supplies shipping from china shortly (to us in europe) and will have more later, so I plan on having this shipped along with other supplies, because the biggest cost really isn't the supplies, it's the shipping, as the shippers charge for 1 metric ton / 1 cubic meter minimum, regardless of what's actually shipped, so shipping something small and cheap turns out impossible because of huge costs, but when we fill up as much as possible that cubic meter and keep it below the 1 metric ton (easy) then shipping costs overall become much more reasonable.
All the heatsinks would come from Germany (Fischer) and there are many electronics parts that will probably come also from china.
I have planned to use the MPC722 power resistors (non inductive), because I already have plenty of those on hand, in 0.22ohms x 2, so that's what drove that choice on the pcb layout.
The relays will be the SongChuan 833H, with a 24V coil.
I have plenty of the MJ15003/4 on hand, which is the reason behind this choice on the layout, and that will allow me to make some matching.
Plus I have plenty of the MPSA42/92 and also 2N5401/5551 on hand, for more matching. I don't have any BC546/556, so that may be for later versions...
The next version will probably substitute the MJ15003/4 with the MJ15024/25, because I also have a bunch in stock. And that one will go up to 70V (or maybe 75) rails, which explains the SOA plot for this option.
On the pcb layout and the renderings, I used a 5W axial power resistor as the base for the output coil, and the drawing shows a coil of about 11-12 turns on a roughly 6mm inner diameter.
That takes some pcb real estate and perhaps there is a better solution that would give more room around them for evacuating the heat while providing a little more space for connection to the output. Perhaps a radial power resistor, but then the coil would have to change a bit.
What's the optimal value for that coil? If I have to make it a little bigger, how does that affect its value and how can I know what to wind???
That takes some pcb real estate and perhaps there is a better solution that would give more room around them for evacuating the heat while providing a little more space for connection to the output. Perhaps a radial power resistor, but then the coil would have to change a bit.
What's the optimal value for that coil? If I have to make it a little bigger, how does that affect its value and how can I know what to wind???
btw: I increased the clearance in the design rules from 10mils to 12mils and fixed the few dozen issues so the board now passes for this clearance.
The higher clearance will require some more serious rearranging, especially changing the TO92/126 and 220 footprints to spread their legs more.
I'm not sure this is worth the extra efforts. Can someone share their thoughts on this?
The higher clearance will require some more serious rearranging, especially changing the TO92/126 and 220 footprints to spread their legs more.
I'm not sure this is worth the extra efforts. Can someone share their thoughts on this?
Not sure what software you are using, but Eagle standard pads are on the small side. You may want to check for manual solderability.
Consider adding a second set of output device pads to allow using TO-3 or plastic output devices. They could be covered by the TO-3 insulators if those devices are used. Although MJ150xx are reasonably available, plastic devices are more so. That would make it more universally buildable.
MPSA to BC is easy - just reverse the orientation. No need for a board change. That's what I did on my Jens Rasmussen boards.
Don't worry about heat dissipation of the output resistor//coil. The only reason that resistor will see any power is if you have a high frequency oscillation. Smoke means a problem elsewhere. The coil is there to keep EMI/RF out of the feedback path. It's value isn't terribly critical. I'd lean a towards the low inductance side to keep the bandwidth up.
I agree with earlier posters, a finer mesh over the circuit board would be a good idea to keep flotsam and jetsam off the board and shouldn't add too much to the cost if you just add rectangles of finer mesh. Perhaps better looking would be a solid top with cutouts for the heat sink areas of open mesh.
In answer to your earlier question about using single ended signals, the A75 article addresses that. You could also avoid the issue using RCA to XLR adapter cables that ground the negative input.
In the interest of cost saving, why male and female XLRs? The only reason I can see for that is daisy chaining amps for PA use. Just stick with convention and use male XLR jacks the board space for an RCA and a switch.
Consider adding a second set of output device pads to allow using TO-3 or plastic output devices. They could be covered by the TO-3 insulators if those devices are used. Although MJ150xx are reasonably available, plastic devices are more so. That would make it more universally buildable.
MPSA to BC is easy - just reverse the orientation. No need for a board change. That's what I did on my Jens Rasmussen boards.
Don't worry about heat dissipation of the output resistor//coil. The only reason that resistor will see any power is if you have a high frequency oscillation. Smoke means a problem elsewhere. The coil is there to keep EMI/RF out of the feedback path. It's value isn't terribly critical. I'd lean a towards the low inductance side to keep the bandwidth up.
I agree with earlier posters, a finer mesh over the circuit board would be a good idea to keep flotsam and jetsam off the board and shouldn't add too much to the cost if you just add rectangles of finer mesh. Perhaps better looking would be a solid top with cutouts for the heat sink areas of open mesh.
In answer to your earlier question about using single ended signals, the A75 article addresses that. You could also avoid the issue using RCA to XLR adapter cables that ground the negative input.
In the interest of cost saving, why male and female XLRs? The only reason I can see for that is daisy chaining amps for PA use. Just stick with convention and use male XLR jacks the board space for an RCA and a switch.
I am using Eagle.
Which pads are you referring to? I made quite a few of my own footprints, mostly because either I didn't like the smaller pads and other details from eagles' built-in libs, and also because they were not there. So when I make my own footprints, I tend to be heavy on the pad size by default.
I need to double check, but if you can point me to the ones you have in mind, I'll definitely recheck them.
I have to make sure it all can be done by hands, especially for me now that my eyesight has gone downhill big time in the past few years and my dexterity is also a disaster, so I do need to make it as easy as possible.
That requires making new footprints and I would have to move tracks around to leave the room for them. I planned to switch to the flat packs later in any case, so I didn't think it was important to make it that versatile right now, since I planned other versions.
But if that can be done easily enough, why not?
And perhaps more "group buy ready" and make it more affordable...
I was giving some thought about making those footprints more of a staggered instead of straight line type, so the pads could possibly be a little bigger and be pulled apart a little more, which would probably allow for a wider clearance in the design rules. This is where most of the clearance errors occur if I try to increase the clearance width, so perhaps going for more spread out pads on those cases could allow for a higher clearance on the whole board...
btw: I still want in a group buy for Jen's boards, at least the 10 outputs, or even better the 12. He's the master!! and his work is superb!
Well, I'm thinking I can make a slightly different version of the top case cover, so it has a different and tighter mesh right above the pcb. This would only be a change on one part and that would not change anything in the costs, since those will be laser cut on a table. I'll work on that and post a possible option real soon...
The way I had thought of doing it with the current design is that the XLR cables would be the ones making the difference in the way they're wired, so if the amp is hooked up to a non balanced device, the cable would be wired to handle that negative input tied to ground, so the amp would never know about it and there would be no need for any switches or config of any kind on the amp itself. That of course requires XLR cables wired differently and have them on hand to use on their intended devices.
I was doing some reading on that A75 article and couldn't find the part that I need to read, however I went on reading some really interesting stuff.
I do need to learn more about making proper use of that other side of the diff amp though. Where in that article is the part I need to read first??
Exactly, that's because I use this often and people around me as well, so I wanted this feature. And the cost difference isn't much at all, as those plugs cost in the order of about 1euro each. No big expense and I do make use of them. I also have sometimes a need to use an "unconventional" XLR cable, that is not the right sex on its ends, and that allows plugging anything in there, as long as they're wired correctly inside...
This first version of the design isn't only going to be for me, as I have at least one friend already who wants one, and most likely others will too. And those guys have such weird "home made" cables. We've been doing sound stuff since the 70s and we still have old equipment around. I have a couple of amps that need serious restoration before they can be plugged in (a Power APK4500 and a Double 300). Those date from way back and they do have that same setup with dual XLRs on their inputs, so we're spoiled and we are used to making use of them.
This whole design won't be cheap in any case, as is any good leach amp or other type when done properly. So do we need to worry about saving a couple of bucks on them?
Ok, for the wary of spills and other types of intrusions into the amp, I made a different version of the top case cover with a finer mesh only on top of the pcbs.
I made holes only 2.5mm, and there are many...
Is this more satisfying?
We can still see the innards, mostly... It still looks rather nice I think (any comments on that?) and it should prevent many intrusions, although nothing can prevent the liquids intrusions, as only a water tight case would do that and that is the last thing we'd want on the amp.
I don't think there is anything we can do against a full beer can spill right on the amp's top, only a case with no holes can help that somewhat.
But when it's time to make the case, my version will be the first one, at least for me, but the choices exist. We can even make those holes smaller if that still scares some...
I made holes only 2.5mm, and there are many...
Is this more satisfying?
We can still see the innards, mostly... It still looks rather nice I think (any comments on that?) and it should prevent many intrusions, although nothing can prevent the liquids intrusions, as only a water tight case would do that and that is the last thing we'd want on the amp.
I don't think there is anything we can do against a full beer can spill right on the amp's top, only a case with no holes can help that somewhat.
But when it's time to make the case, my version will be the first one, at least for me, but the choices exist. We can even make those holes smaller if that still scares some...
Attachments
-
Amp rack full Assem imphy 470va + VU bi-color led_no spill top perspective.JPG -
Amp rack full Assem imphy 470va + VU bi-color led_no spill top view.JPG -
Amp rack full Assem imphy 470va + VU bi-color led_no spill front corner detail.JPG -
Amp rack full Assem imphy 470va + VU bi-color led_no spill side rear.JPG -
Amp rack full Assem imphy 470va + VU bi-color led_no spill top front.JPG
Last edited:
It is not only safer but looks even better too.
The electrical safety standards have details for maximum hole size, would be clever to comply with these.
Very nice work, what software did you use to render?
Best wishes
David
Last edited:
What are the standards for that? and for where? Because that can be done in many countries and although some don't have such standards, those that do will not have the same from one country to an other. Difficult to comply, and to find all the info.
Would that 2.5mm that I used be a safe bet?
Actually, those last few snapshots aren't renderings, just plain picture outputs from solidworks. (I have access to one right now, so better make good use of it)
I only have looked at the standard for Australia and I don't remember the details.
But I assume the different national standards are fairly similar.
You could pick the minimum from the major standards.
The Australian Standard is pay to view but you may be able to find some others on-line.
Perhaps UL? There is probably a Euro-standard available somewhere too.
The IP standard should be available.
Best wishes
David
OK. 2.5mm is IP3
1mm is IP4
Last edited:
I already have things manufactured (for business) right now in china. And that metal worker is well equipped. They can do laser cutting, which allows any design to be made (more or less), so this case design can be easily made there. And since they are shipping other supplies to me for business already, adding a few more amp cases would not add any cost to the existing shipments.
So I can have my exact case design made there with no problems and have them economically shipped to us in Europe.
Of course having more than one case made would drastically reduce the per unit cost, which is why a design must be finalized in a way that suits more than one user.
The heatsinks are from Germany and I don't make use of any exotic parts, so all the electronics can be found easily. I already have quite a bit of the electronics parts, including the toroids, most semis, some resitors (the MPC722 included)... I don't have the big filter caps yet, and I allowed for 40mm diameter. Those are snapins, but ordinary radials should go in there as well, with a lead spacing of 10mm (10.16).
I have the XLR plugs in stock as well (at least 25 of each sex), the rectifier bridges, fuse supports...
I don't need to locate too many parts before building, so all I need is to get the pcb finalized and get them made. There are still some calculations to perform, mostly on the protection network. The rest should be more or less ready.
I have troubles finding the right PTC, which is what the protection circuit calls for (I didn't re-invent the wheel, I adapted Elektor's proven one), but that can easily be solved, even without altering the pcb.
I also show a thermostat screwed on one of the angled profiles near the TO3s, which I would use with the power-on delay for extreme cases of overheating. That plus the PTC sensor on the heatsink with the protection circuit should make for a robust amp.
Each amp having its own power switch, the only electrical thing in common is the power cord, which I would make as thick as possible.
The toroids that I have on hand are older ones from Metalimphy, with a solid reputation for quality, although they have disappeared now. I have an other supplier for good toroids that can easily make them to order, so my next build will use those, and I planned for 2 toroids below the pcb support plate, so we can pack a lot of power into this.
As any good diy build, this is not aimed at weight saving and manufacturing cost issues, so the case is a 4U rack mount and my early total weight estimates for the current design with the imphy toroids is at least 25kg.
The new toroids are usually a little slimmer than older ones, so my existing design for those imphy toroids should be quite sufficient when switching to other makes.
I already had conversations with the transformer manufacturer and I have colleagues who already design high end hifi gear who get their toroids from them as well. So the quirks have been worked out on that, with proper choice for the core material and other considerations to avoid early saturation, good regulation and power handling for the audio usage.
If I can order several toroids at once, their prices drop nicely. We can get 750VA for about 75euros each. And buying 10 units at a time drops the unit price below 45euros. A 1500VA is about 100euros (2 x 56V) and 10 of them drops below 60euros. Not too bad.
So I can have my exact case design made there with no problems and have them economically shipped to us in Europe.
Of course having more than one case made would drastically reduce the per unit cost, which is why a design must be finalized in a way that suits more than one user.
The heatsinks are from Germany and I don't make use of any exotic parts, so all the electronics can be found easily. I already have quite a bit of the electronics parts, including the toroids, most semis, some resitors (the MPC722 included)... I don't have the big filter caps yet, and I allowed for 40mm diameter. Those are snapins, but ordinary radials should go in there as well, with a lead spacing of 10mm (10.16).
I have the XLR plugs in stock as well (at least 25 of each sex), the rectifier bridges, fuse supports...
I don't need to locate too many parts before building, so all I need is to get the pcb finalized and get them made. There are still some calculations to perform, mostly on the protection network. The rest should be more or less ready.
I have troubles finding the right PTC, which is what the protection circuit calls for (I didn't re-invent the wheel, I adapted Elektor's proven one), but that can easily be solved, even without altering the pcb.
I also show a thermostat screwed on one of the angled profiles near the TO3s, which I would use with the power-on delay for extreme cases of overheating. That plus the PTC sensor on the heatsink with the protection circuit should make for a robust amp.
Each amp having its own power switch, the only electrical thing in common is the power cord, which I would make as thick as possible.
The toroids that I have on hand are older ones from Metalimphy, with a solid reputation for quality, although they have disappeared now. I have an other supplier for good toroids that can easily make them to order, so my next build will use those, and I planned for 2 toroids below the pcb support plate, so we can pack a lot of power into this.
As any good diy build, this is not aimed at weight saving and manufacturing cost issues, so the case is a 4U rack mount and my early total weight estimates for the current design with the imphy toroids is at least 25kg.
The new toroids are usually a little slimmer than older ones, so my existing design for those imphy toroids should be quite sufficient when switching to other makes.
I already had conversations with the transformer manufacturer and I have colleagues who already design high end hifi gear who get their toroids from them as well. So the quirks have been worked out on that, with proper choice for the core material and other considerations to avoid early saturation, good regulation and power handling for the audio usage.
If I can order several toroids at once, their prices drop nicely. We can get 750VA for about 75euros each. And buying 10 units at a time drops the unit price below 45euros. A 1500VA is about 100euros (2 x 56V) and 10 of them drops below 60euros. Not too bad.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.