Here is a modified layout, following some of the suggestions. I worked on it with Peter last night. I tried putting the resistors above the capacitor, but it didn't look good. Instead, the R6 and R7 were swapped to equalize the trace lengths a bit.
As for the T, I think that it is unavoidable, and shouldn't be an issue. I increased the width of the output trace from the IC chip, which should help. I had tried using both layers, but came back to this solution.
The heat relief traces are removed as well.
Any comments?
--
Brian
As for the T, I think that it is unavoidable, and shouldn't be an issue. I increased the width of the output trace from the IC chip, which should help. I had tried using both layers, but came back to this solution.
The heat relief traces are removed as well.
Any comments?
--
Brian
Attachments
mhennessy said:
I think that we're all in agreement (from here, and the previous thread about this board) that Nationals board is not an example of an optimised layout
It might not be optimised (for nice appearance), but it definitely is an example of a proper layout.
It's just that some small issues you might think affect the performance, in real life may have no importance at all
Now I really like the layout. 
Maybe even a certain person will like it also. One thing though, how about the hole size for the speaker cable? ..and what about decoupling, like 100 nF close the the IC? Have you forgot those? I advice you to put in 2 psc. 100 nF 63 V polyester.
http://www.national.com/ds.cgi/LM/LM4780.pdf
The datasheet in page 5 suggests 100 nF+ 10 uF + 1000 uF. I think you at least should have 100 nF.
I think the pcb is real good looking now and far above in looks and probably also in performance compared to many commercial boards (LM3875/3886).
Maybe even a certain person will like it also. One thing though, how about the hole size for the speaker cable? ..and what about decoupling, like 100 nF close the the IC? Have you forgot those? I advice you to put in 2 psc. 100 nF 63 V polyester.http://www.national.com/ds.cgi/LM/LM4780.pdf
The datasheet in page 5 suggests 100 nF+ 10 uF + 1000 uF. I think you at least should have 100 nF.
I think the pcb is real good looking now and far above in looks and probably also in performance compared to many commercial boards (LM3875/3886).
Vertical resistors
As with all things it depends totally on the application.
The two end terminations of most resistor types present a thermocouple junction.
Different temperatures at each end of the resistor will generate a voltage, a small voltage in most cases, but a voltage nonetheless.
How much this matters depends upon what function those resistors perform, their power dissipation, the air flows and particularly their construction.
The way to minimise it is to keep the end terminations at the same temperature, and horizontal mounting helps this, from both a self-heating and ambient perspective.
None of this is to say that the effect is important at AC though, since the EMF generated will be inherently LF, since sudden temp changes are unlikely in most app's.
Andy.
As with all things it depends totally on the application.
The two end terminations of most resistor types present a thermocouple junction.
Different temperatures at each end of the resistor will generate a voltage, a small voltage in most cases, but a voltage nonetheless.
How much this matters depends upon what function those resistors perform, their power dissipation, the air flows and particularly their construction.
The way to minimise it is to keep the end terminations at the same temperature, and horizontal mounting helps this, from both a self-heating and ambient perspective.
None of this is to say that the effect is important at AC though, since the EMF generated will be inherently LF, since sudden temp changes are unlikely in most app's.
Andy.
Peter Daniel said:
Actually, it's neither. It's an example of a layout done by a draughtperson, with relatively little regard taken to layout or appearance, other than providing a rudimentory star earth. Traces are much longer than they need to be, compromising performance. I thought this was discussed in another thread? Anyway, even your initial paper draft was miles better than this layout in both regards
peranders said:
I agree about hole sizes - but having read previous comments about decoupling, I don't think you'll win that one. I believe Peter has found them unneccesary based on listening tests using good quality electrolytics. I will add just one comment - think about when the electrolytic caps start to age - that's when small parallel decoupling capacitors come into their own.
Brian - I like the new version of the layout. I might have some small comments, but I don't think they're worth voicing at this stage. Just one thing that I notice - you're using vias! Doesn't worry me in the slightest, but what does Peter think?
Best regards,
Mark
Parts precision
It's curious to note, as I just did now, that National seems less "demanding" on the parts requirements for paralleling their chip amps on the LM4780 datasheet.
On their first Application Notes for bridging and paralleling, they specify 0.1% parts and servos. On the 4780 parts are not specified and no servos are used.
Were the other Overture amps so different from one another to demand that tolerance on external parts was so tight? Or were they being too cautious?
Carlos
It's curious to note, as I just did now, that National seems less "demanding" on the parts requirements for paralleling their chip amps on the LM4780 datasheet.
On their first Application Notes for bridging and paralleling, they specify 0.1% parts and servos. On the 4780 parts are not specified and no servos are used.
Were the other Overture amps so different from one another to demand that tolerance on external parts was so tight? Or were they being too cautious?
Carlos
BrianGT said:
The pcb layout looks OK to me. But there are some issues that maybe were discussed in other threads, but I missed them.
If this is a NIGC, where are the capacitors that should be at the input and feedback? If I am not wrong this was the same philosophy used on the single NIGC boards, but aren't you running a certain risk by not using those caps?
Carlos
mhennessy said:
Actually, it's neither. It's an example of a layout done by a draughtperson, with relatively little regard taken to layout or appearance, other than providing a rudimentory star earth. Traces are much longer than they need to be, compromising performance. I thought this was discussed in another thread? Anyway, even your initial paper draft was miles better than this layout in both regards
My dictionary hasn't got the word draughtperson and I can't really figure out the word
Brian, when I think of it, you have unnecessary long traces going in long loops. Since you are going to have this pcb professionally made why don't you shorten all signal traces and uses 15 mil traces between the IC pins? I should also go for more groundplane. Consider also that rather much of your copper areas aren't very usefull. Think electrial field inside these areas.
Not having proper decoupling is like not having your hands washed before cooking dinner. I vote definitely for proper high frequency decoupling. You can bet on that some people which are building this amp will encounter RF/EMII trouble. It's easier not to solder in a cap compared to patch it in.
As a comment about mhennessy's remark: I think it's really good of you designing these pcb's but having eagle control of all parameters yourself is nearly an impossible task, even for a pro. It takes years of practice, maybe a lifetime! I wonder why Mr. D hasn't comment this project?
mhennessy said:Brian - I like the new version of the layout. I might have some small comments, but I don't think they're worth voicing at this stage. Just one thing that I notice - you're using vias! Doesn't worry me in the slightest, but what does Peter think?
As for using vias, I assume you are referring to the two traces on the top layer going to the feedback resistors. I actually forgot to change them back to the bottom layer, after playing around with them earlier. I put them back on the bottom layer now. I doubt that it would make much difference, but I would rather have them away from the power traces a bit more.
What small comments do you have? I am curious. You can e-mail them to me if you don't want to post them here
--
Brian
If you have 0.1 ohms resistors it's probably not very important to have thick wires but I imagine that it looks nicer if you can choose such cables if you want to.carlmart said:
Also Brian, my remarks are probably not very important. I gather the amp will work alright but wouldn't it be a kick to have done everything to make it right?
peranders said:
The pins on the LM4780 are quite a bit closer than the LM3875, so I don't like the clearance issues with running traces in-between them. I really don't see any reason to need to do this. The only long traces are the input traces.
As for decoupling, I was going to just try soldering decoupling capacitors on the bottom of the board, and see if it makes a difference. I don't really see an optimal place to put them, except on the bottom, on the leads coming from the electrolytics, or the extra mounting holes (the unused set)
--
Brian
peranders said:
I initially tried, but it didn't seem to come out as nicely. I had the power inputs on the far left, and the inputs on the far right, with the output and grounds in the middle. The board size increases a lot also, since the BG 1000uF STD caps have large footprints. After trying this, I decided that the small tradeoffs of the current layout are greatly outweighed by the symmetry and form factor obtained by the current layout.
--
Brian
Who is Mr. D?
"I wonder why Mr. D hasn't comment this project?" - P/A
How much can one say about yet another chip amp............
I think the things that are important to consider for the layout are in outlined pretty well in the data sheet . Don't confuse ground planes and single point grounding, they are not the same thing. The data sheet mentions the desirability of single point grounding several times I believe.
"I wonder why Mr. D hasn't comment this project?" - P/A
How much can one say about yet another chip amp............
I think the things that are important to consider for the layout are in outlined pretty well in the data sheet . Don't confuse ground planes and single point grounding, they are not the same thing. The data sheet mentions the desirability of single point grounding several times I believe.
Newbie
Hi all,
I have just registered on this forum but i've been following this thread for some time now. And i've also read the last thread about the LM3875 amp.
I'm very interested in building a amp like this but there are a couple of questions,
- How much power does this amp put out with a 8 ohm load?
- How hard will it be it to put volume control on it ?
- What kind of power-supply will it need? (how much Va)
This will be my fist attempt at building an amp, but i allready have experience with building speakers and I have read a lot of documentation about different types of amps.
Most of them are a little to expensive for me at the moment and something like this would be ideal for now.
tnx.
Hi all,
I have just registered on this forum but i've been following this thread for some time now. And i've also read the last thread about the LM3875 amp.
I'm very interested in building a amp like this but there are a couple of questions,
- How much power does this amp put out with a 8 ohm load?
- How hard will it be it to put volume control on it ?
- What kind of power-supply will it need? (how much Va)
This will be my fist attempt at building an amp, but i allready have experience with building speakers and I have read a lot of documentation about different types of amps.
Most of them are a little to expensive for me at the moment and something like this would be ideal for now.
tnx.
peranders said:
That size decoupling cap is best soldered underneath, detectly to electrolytic, if you really needed. Providing special spots for those caps on a board would be compromising simplicity of a ground plane. I got used to attaching components to both sides of PCB (whenever more convenient) and it shouldn't even present clearance problem with such small value. I have my protoype working withaout any extra bypass caps and it prforms fine.
I looked at National's layout again and I think a lot of thought went into creating it. If tight packing of components wouldn't be my goal here, I don't think I could improve on that (National's) layout. It is very good for what it was intended. Maybe they just wanted to show that the chip performs even when not every trace on PCB is analysed in depth and optimised to perfection?mhennessy said:
Actually, it's neither. It's an example of a layout done by a draughtperson, with relatively little regard taken to layout or appearance, other than providing a rudimentory star earth. Traces are much longer than they need to be, compromising performance. I thought this was discussed in another thread? Anyway, even your initial paper draft was miles better than this layout in both regards
carlmart said:
You definitely don't need input caps if your source is protected against excessive offset. The two parallel channels I've built presented much lower offset than what I got with LM3875 (it was about 24mV). As to the feedback capacitor, I never required it with LM3875 chip either. It's hard to predict if its really needed here, that's why we go with a smaller initial run for testing purposes. After that we will know for sure.
Hi Brian,
It occured to me that it might be good to have the 2 rows
of diodes in the power supply facing in opposite directions or
some other idea so that they could be all be attached to
heatsinks. MAybe this would only be important if the bridge were used for 6 100w channels- but many people are probably going to make bigger ampps with this kit. Also would be very useful if someone wanted to use just the power board for a Son of Zen or other big class A amp. Then they would be like Big Parsnips diode boards? (but without snubbers) So only an idea if it doesn't compromise something. What is the capacity in amps for the assembled bridge? Would small heatsinks double the rating?
Second- thank you for my latest order. Excellent counting!!!
BUT (I hope this doesn't drive you over the edge) It's missing the 2 small BlackGate caps for the power supply. They are supposed to be part of the premium bridge kit , right?
Mark
It occured to me that it might be good to have the 2 rows
of diodes in the power supply facing in opposite directions or
some other idea so that they could be all be attached to
heatsinks. MAybe this would only be important if the bridge were used for 6 100w channels- but many people are probably going to make bigger ampps with this kit. Also would be very useful if someone wanted to use just the power board for a Son of Zen or other big class A amp. Then they would be like Big Parsnips diode boards? (but without snubbers) So only an idea if it doesn't compromise something. What is the capacity in amps for the assembled bridge? Would small heatsinks double the rating?
Second- thank you for my latest order. Excellent counting!!!
BUT (I hope this doesn't drive you over the edge) It's missing the 2 small BlackGate caps for the power supply. They are supposed to be part of the premium bridge kit , right?
Mark
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.