I'm not sure the capacitance multiplier would work - when the rail falls the transistor has no power source. If the caps end up being too small, you just need bigger ones. In any case, 1000uF ought to be plenty if 220 isn't. Simpler is better, because there will be less bugs to work out.
I'm running more calculations on the dropping resistor. If we assume the input voltage stays between 35 and 30V, ignoring it dropping below 30V, then a 1k 1/2watt resistor can do the job and the zener doesn't need to be bigger than 400mW.
The main issue is when the rails swing and the dropping resistor sees that incoming voltage drop a lot, the zener current will also drop below the minimum and there won't be regulation. Only the big cap can compensate for this and the diode preventing it from discharging back into the rail can also help make it last longer.
How about having an other diode between the cap and zener? perhaps we can even split the cap into 2, having one on the zener and the other after a diode, going to power the ic. Gets more complicated. I agree simpler is better.
The main issue is when the rails swing and the dropping resistor sees that incoming voltage drop a lot, the zener current will also drop below the minimum and there won't be regulation. Only the big cap can compensate for this and the diode preventing it from discharging back into the rail can also help make it last longer.
How about having an other diode between the cap and zener? perhaps we can even split the cap into 2, having one on the zener and the other after a diode, going to power the ic. Gets more complicated. I agree simpler is better.
I would use a 2 watt resistor and a 1 watt zener - because I always keep a stock of those on hand. For things like shunt regulators. I always want overkill on resistors that just sit there and dissipate.
Also, you will want to find space on the board for a standard VI limiter. The values can be worked out later. I would also suggest proto-ing this on something expendable before committing to a real PCB. Maybe a single sided layout in magic marker and etch at home or perf board. If you have trouble with part of the circuit and it needs to change, it's better to modify something you don't intend to be the end product. There are plenty of opportuinities for things to go wrong yet.
Also, you will want to find space on the board for a standard VI limiter. The values can be worked out later. I would also suggest proto-ing this on something expendable before committing to a real PCB. Maybe a single sided layout in magic marker and etch at home or perf board. If you have trouble with part of the circuit and it needs to change, it's better to modify something you don't intend to be the end product. There are plenty of opportuinities for things to go wrong yet.
A 2watt resistor I can find, and I was sizing up the footprint on the board for 2 or 3watts, however I don't think I have anything big enough in my zeners.
I'll have to work on this later on. I don't think we can act on the drivers easily, so this should act on the pre-drivers. Won't that be a little more tricky for a VI with that gnd floating?
I agree there should be a proto before actually making that pcb. However although I do have most of the parts on hand, I'm not equipped enough to make a pcb myself, I don't have the lab equipment to test it, plus my eyesight and dexterity have gone seriously down in the pas few years and I would have troubles working on this.
We want the pcb bullet proof, so we do need to make sure all the bugs have been worked out first.
I can do all the virtual stuff, I have what I need for this, but for a proto someone with more tools would be needed. This is far too much for a breadboard.
I am laying out simple single sided boards for prototyping. Trying to keep them small. Making 2 boards, one for each side. I discarded the psu schematic as well as the extra psu for the opamp, which is also removed, as it's not needed for testing. I also removed the extra decoupling caps and the zobels. If needed, then they can be added externally. Same as on the real layout, I left out the output coil.
One detail comes to mind though, the high side doesn't have any references to the main ground, only the signal ground.
Even while prototyping, the TO3s need cooling, so obviously they need to be layed out for that, I could leave them all off board, but what a mess of wires! So I'm making it so there is a corner profile, that can be attached to whatever bigger heatsink may be available.
One detail comes to mind though, the high side doesn't have any references to the main ground, only the signal ground.
Even while prototyping, the TO3s need cooling, so obviously they need to be layed out for that, I could leave them all off board, but what a mess of wires! So I'm making it so there is a corner profile, that can be attached to whatever bigger heatsink may be available.
Here it is! A single sided layout for testing the prototype.
I wanted to have a minimum trace width of 25mils (1mm) and a clearance of 20mils, so the space got quite tight in most places due to the wide traces and wide spacings. Plus I wanted as much as possible to avoid straps. So I had to make some compromises, with a little smaller traces in the small signal areas to get everything closer together so it could all fit. I wanted also to keep the board sizes as small as possible, so since I started with 100mm x 110mm and having all the TO3s on it, which avoids wires as much as possible, the parts placement was nicer in the beginning but then I was forced to move things around a bit to allow those wide traces to pass.
I kept the psu off the boards and made on board for each amp side, so here are 2 boards, same size 100x110mm, and single sided.
This should be good enough for testing and proof of concept.
I will look into adding the protection networks now.
I wanted to have a minimum trace width of 25mils (1mm) and a clearance of 20mils, so the space got quite tight in most places due to the wide traces and wide spacings. Plus I wanted as much as possible to avoid straps. So I had to make some compromises, with a little smaller traces in the small signal areas to get everything closer together so it could all fit. I wanted also to keep the board sizes as small as possible, so since I started with 100mm x 110mm and having all the TO3s on it, which avoids wires as much as possible, the parts placement was nicer in the beginning but then I was forced to move things around a bit to allow those wide traces to pass.
I kept the psu off the boards and made on board for each amp side, so here are 2 boards, same size 100x110mm, and single sided.
This should be good enough for testing and proof of concept.
I will look into adding the protection networks now.
Attachments
For the main pcb design, here is the current version without the protections.
Board is 200x200mm, double side plated through.
I hope there is enough room to squeeze the protection networks in there, otherwise we have to increase the board size.
Board is 200x200mm, double side plated through.
I hope there is enough room to squeeze the protection networks in there, otherwise we have to increase the board size.
Attachments
Member
Joined 2009
Paid Member
Hi,
Regarding single sided prototype:
I'm going to suggest a slightly different approach that you may like - but it will be a bit more work for you before your design is ready.
Design for double sided board - much more flexibility in the layout. Use much wider traces for high current paths than you have now. Restrict the board size to 100mm x 100mm. Maybe you only need two pairs of power devices for the prototype.
Then you can use this service and get a set of boards for not a lot of money:
ITEAD Studio 2Layer Green PCB 10cm x 10cm Max
I've used this service myself and found it very good. The only downside is that the boards come on the boat from China so shipping is slow.
I would suggest putting the design of the main board on the shelf for now, finalize it only after the prototyping as you will likely end up seeing the need for further changes.
Regarding single sided prototype:
I'm going to suggest a slightly different approach that you may like - but it will be a bit more work for you before your design is ready.
Design for double sided board - much more flexibility in the layout. Use much wider traces for high current paths than you have now. Restrict the board size to 100mm x 100mm. Maybe you only need two pairs of power devices for the prototype.
Then you can use this service and get a set of boards for not a lot of money:
ITEAD Studio 2Layer Green PCB 10cm x 10cm Max
I've used this service myself and found it very good. The only downside is that the boards come on the boat from China so shipping is slow.
I would suggest putting the design of the main board on the shelf for now, finalize it only after the prototyping as you will likely end up seeing the need for further changes.
Last edited:
This wouldn't be too much extra work, as turning on the top layer would allow bringing things together tighter and there would no longer be issues with the wider traces. This could be done quickly.
And I'm thinking it might be a good idea to add the protections so they can be tested along with it.
I took a look at them before and didn't see any pricing advantage really. Once we include the cost of shipping and customs, we're back to what we would pay inside europe. We'll have to compare, everything included.
Well, shipping delays aren't a big issue, as long as the cost is really low. After all, we're not making a new product for the mass market with the usual tight schedule for product release.
There would probably be changes, although most might just be a matter of part value, and some perhaps part footprint if we find that some parts needed a larger space. Plus we really need to consider the protections. I was trying to find out if having 3 pairs outputs would be beefy enough to withstand a short of the output long enough to let the main fuse blow...
My SOA plots, if correct, show that we're well within the SOA, even at 2ohms loads, which is 4 ohms on the output, and that's already a nice bunch of watts for a 3055 little amp with rails of no more than 35V.
I will look at the double sided layout now, and we'll see if we can add protections later on too.
The board size choice was driven mostly by having the TO3s right on it, and 3 TO3s in a row is about 100mm, but since I wanted to slap the drivers on the main heatsinks as well, then I had to go 110mm long.
Perhaps going double sided can help bring down the 100mm width somewhat, but then again, if we add protections, the room spared by going double sided might just be enough to include those without going bigger.
Do they have special prices for such size pcbs??
I was only suggesting a single sided because you can draw it up with a magic marker and dunk it in FeCl3 at home. And get a board in an afternoon (not on a boat from China) - cheap. Being able to do 2 turns in one day is a definite plus. This is the only way I do throw aways and one of a kinds if they aren't complicated. Blank panels are so freaking cheap it's not funny - especially if they are a little oxidized but that's nothing steel wool won't fix. Once a design is finalized, then I spend the couple hundred bucks on a board run.
That's a good idea, if you can make the boards yourself with the proper tools. But since I can't do that, the only way is to farm out.
Maybe someone could use my single sided design and make one quickly...
Working right now on double sided layout, starting from the single sided already done, this might come out quick.
I could probably throw one together, but in all likelihood I won't be able to get to it till T-day week. The board would be a one day project, but realistically it's 3 or 4 solid days from blank board to playing music. These things always take longer than expected. The last 3055 project for fun (100W 2 ohms from an 18-0-18 trafo) took about 3 days on a perfboard, working off and on which was all I had.
This project is definitely on the radar still - and wheels are turning for even more power. I may try to cascode it or use up that batch of old RCA 100V hometaxials I've got on hand. SOA? We've got plenty of THAT....
The projects that are in progress are class H. One is a little 125W using TIP35/36's that I want for the living room (no fan or big heatsink), and the other is a 4kW monster that's been ongoing since '09. Somewhere, this has to fit in.
This project is definitely on the radar still - and wheels are turning for even more power. I may try to cascode it or use up that batch of old RCA 100V hometaxials I've got on hand. SOA? We've got plenty of THAT....
The projects that are in progress are class H. One is a little 125W using TIP35/36's that I want for the living room (no fan or big heatsink), and the other is a 4kW monster that's been ongoing since '09. Somewhere, this has to fit in.
Member
Joined 2009
Paid Member
The link I posted shows the price as US$24.90 and for me, shipping is about $5. You will get 10 boards. You may even like them enough to use them for the final project and keep the power supply separate. Or use extra boards to add more output pairs. It could be a flexible solution.
If you add protection against shorts, you could look at my TGM5 project (thread somewhere around here). In this scheme I added capacitors so that the protection will ignore short duration spikes so as to reduce the potential impact on the music - I copied this approach from Bryston but it's not unique to them.
WoW! That would be great! Whatever you can do will be nice and there is no rush.
Oh Man! you ARE having a load of fun. What we're doing this time is a little different, mostly because of that grounded bridge. We're pushing into uncharted territory, especially for the 3055.
I wouldn't mind having a few of the 100V ones to play with, but I have plenty and we can make a few amps.
Now the cascode thing is also something I'd like to look into. Cascoding 3055s can really get more out of those old things...
Now can you imagine the grounded bridge with 100V 3055s cascoded?!?!?
I could even picture a good old super leach amp with 3055s..... Although it's not a quasi.
FRANKEN-FREAKazoid!!!
I am a little wary of the class H topo, because although it helps with certain things, there is sometimes the crossover issue in the rail switching...
I have a bunch of TIP142/147 that I was planning on using for a low tim amp that I wanted to build (not the leach one). I have a good part of that one's layout done as well, but the 3055 project is interesting and I'll get back to that one later. Plus I also have to finish the big project, a leach amp with all the trimmings...
4kW??? not in 8ohms I assume! A bridge?
In the mean time, I just got done laying out the high side board for this project's proto. It's 100% routed but I still have a little work on it, and the other side is 30-40% done too.
I'm posting that first board layout now and I'll post the other views later.
I used ground planes and copper floods for the power and other things.
I also added an extra ground connection on the high side board because it makes no reference to it at all and the ground plane needed to be hooked up, so that connection would go to the central ground, which I think the best place must be the ground ouput on the low side.
That board was easier to route, although not quite so easy in some areas and I tried keeping wide traces and a good clearance, so although it's looking much better than the single sided, being strapped for space still caused a few issues in a few spots. Still it's not a bad looking board now. I'll make some 3D outputs when it's finished.
Attachments
Do you mean it's 10 boards for $24.90 total or each?
If the shipping is only 5 bucks, then it's not bad.
I really like the all-in-one way, with as little wiring as humanly feasible. We'll see how this turns out.
That proto that I'm laying out right now will be done shortly and I'm trying to not make a half way job. It's a double sided and the boards do costs some money, so let's make them worth something. I'm doing all I can to make them nice and robust.
If you can point me to that, that'll be nice, so maybe I can use this for the project.
I'm not so sure how this will work right on this quasi, with the 3 transistors "pseudo-darlington" (not the 3 pairs), I need to find something that works.
Here they are!
I layed them out together, so both boards are on the side panel/picture.
I don't include the components names on the silkscreen or it's overcrowded.
This should be scrutinized, just in case I left something out or made a mistake.
I layed them out together, so both boards are on the side panel/picture.
I don't include the components names on the silkscreen or it's overcrowded.
This should be scrutinized, just in case I left something out or made a mistake.
