Hi.
This is my first post on this forum, I found it very useful while doing first research about the Gainclone, and now I've registered so that I can share with you my design ideas. What I intend to do is build a Gainclone amp with LM4780 that can supply at least 100W, for both 4 ohm and 8 ohm speakers. It shall have the best sound possible under these circumstances, but I am NO hi-fi "freak" ;-) , so there is no need for special considerations, very expensive componentes, etc. What I need is power (as I'm going to use this amp for small public actuations of our band - like in a restaurant, bar etc...), and quality (because I'm also going to use it at home for my "hi-fi" system, when we are not "on tour"
). So, what I did was research the most I could about the gainclones, and have found quite a few different design approaches. I also consulted the datasheet and searched through this forum. The conclusions I've taken, are that I want two LM4780 chips, one for each channel (stereo), and have each chip working in parallel mode (for 4 ohm speakers) OR bridged mode (for 8 ohm speakers), so that I always get around 100W, regardless what speakers (I think I wont stumble across any 2 ohm speakers...). This shall be possible, by selecting the operation mode with one or two switches while - obviously - the amp is powered off.
Now, the problem is: how do I implement that? I don't want to build two amps... lol... But luckily I found Mark Henessy's A4 Power amp, which has a very interesting solution for the switching, but it is actually a bridged-parallel design (each channel consists always of a 4780 in parallel mode), so you only get a mono if working in bridged: http://www.mhennessy2.f9.co.uk/a4/analogue.htm
Anyway, his ideas gave me quite a good start for trying to make my own design, and here it is (to avoid confusion: this is showing only ONE channel, as they are identical, obviously). I've attached it to this post as a GIF image file. Hope it gets the correctly (as this is my first post in this forum).
Attention, there are a few small descriptional errors in the schematic (like incorrect pin number indications of missing pin number indications, and stuff like this. Thought I should mention this, so that you don't have to bother about it. They are irrelevant at this stage, and I'll correct it when the design is finished. Just didn't have the time to do eveything 100%, you know what I mean... I put my attention into doing the schematic right, not to make it pretty or perfect)
For all those of you having difficulties to understand my idea (I shouldn't think that this will be the case, as I've seen most people here seem really smart, some actually much more than me, lolol, but one never knows...):
I have both "halves" of the 4780 in inverting mode, and use (I borrowed this from Mark Henessey hehe) a OPA2134 opamp as an input buffer AND inverter, so that the signal actually is not inverted. This is talking about parallel mode, with the switches in parallel position. When changing both switches to bridged, we'll have the right output config (now one amp driving +, one driving -), and the input now is as following: one of the opa's gets the same signal as before, while the other gets the inverted signal, so that the 4780 connected to it also gets an inverted signal, so that the bridged mode is correctly implemented (at least I hope so lol...).
So, what I hope to hear from you guys is, WILL THIS WORK ??? I think it should
Or do I have to find a different solution, and if yes, any ideas?
And in case that it seems to be ok, please give your 5 cents to some of the most critical points in gainclone design that are always at discussion, so that I can choose the best for my application priorities (at least I've seen about three different approaches). Here they go:
1. I've mostly used the minimal filtering solution, like "inherited" from the gaincard and very nicely implemented by Brian and Peter Daniel and other great people . But I've seen different approaches, and now I'm wondering: does this really have such impact on the sound? Because I kind of like the idea of doing more filtering, especially to avoid HF and other interferences (keep in mind that I'm going to use this amp in public spaces, one never knows what one can encoutner there). I give you three examples:
http://www.tech-diy.com/paraclone.htm
http://www.tech-diy.com/LM4780_bridgedamp.htm
http://www.shine7.com/audio/pa100.htm
They have, for example, a filtering network from the input of the 4780 to GND, and a capacitor between + and - inputs of the amp, a capacitor parallel to the feedback resistor, a capacitor from the mute pin to GND, and other stuff I don't really see what it is for, and which I don't see in the "original" minimalistic gainclones, or in the most popular kits on here. So I didn't implement that stuff into my design. Now, what I'd like to know: maybe it would be a good idea to do it, though? Any real experiences on this area? Advantages/disadvantages? Alternatives?
Also, I have doubts about some components values:
The 2,2K from + input to GND. Somebody said: the higher the value, the better the DC offset, but the worse the noise generated by that resistor. So he set it to 1K... But I compared that to the other designs (and thought about the risks associated with DC) and so I set it to 2,2. Is that ok? or should it be even higher?
The 4,7uf cap in input (between opa's and the amps). I've seen people using 47uF... Any importance here? And would it be wiser to move the cap to the opa's input instead?? To keep out eventual DC signals completely?
The Gain setting resistors: 154k and 6,8k, giving us about 22 (hope my math's still right lol). Is that ok? Or shall I use smaller values, to get the same gain? Because i've seen like 22 k and 1 k, etc...
The OPA2134... I've heard it is one of the best opamps for audio. Is that true? So is it ok to use it, in order to get as small alteration of the sound as possible? Or have you better ideas?
Now, to the powersupply issues (which I didn't have time to do the design and the schematic yet):
I'm going to use a 500VA 25-0-25 Toroid transformer which I can get for a very nice 32 Euros from Farnell, that's ok for what I'm doing, isn't it?
Now, is it best to do a "snubberized" (didn't figure out yet what that means, but anyway) supply with big caps (20000 or above) or do what they did with the gaincard and in all minimalistic gainclones: something like 1000uF
I've heard that the sound of the amp is affected, that the bigger the caps, the more the sound gets a little more "cold", "electronic", like typical transistor amp, while when using small caps, it is warmer and better??? is this only a myth?? Because on the other hand, I've read that when using bigger caps, the amp will have a fuller and more powerful sound, especially at higher volumes. And what about regulated power supply? One of the sites I stated above, has a nice idea for that, and he actually states it improved sound. Is that valid? Or only if the transformer is kind of weak, so that voltages drops when higher power (which won't be the case here, I think...)?
Well, for now I think that's all... Would be great to have some feedback here and some good hints. At least for the design phase... later on I'll need also some practical tips (as for heat dissipation, case, etc...) but I think it's best to keep things separated and clear (too much confusion if asking all stuff at the same time, isn't it). So, thanks in advance for your answers, and sorry for the "testament", but I couldn't figure how to ask so many things and give enoiugh details on less space lol...
This is my first post on this forum, I found it very useful while doing first research about the Gainclone, and now I've registered so that I can share with you my design ideas. What I intend to do is build a Gainclone amp with LM4780 that can supply at least 100W, for both 4 ohm and 8 ohm speakers. It shall have the best sound possible under these circumstances, but I am NO hi-fi "freak" ;-) , so there is no need for special considerations, very expensive componentes, etc. What I need is power (as I'm going to use this amp for small public actuations of our band - like in a restaurant, bar etc...), and quality (because I'm also going to use it at home for my "hi-fi" system, when we are not "on tour"
). So, what I did was research the most I could about the gainclones, and have found quite a few different design approaches. I also consulted the datasheet and searched through this forum. The conclusions I've taken, are that I want two LM4780 chips, one for each channel (stereo), and have each chip working in parallel mode (for 4 ohm speakers) OR bridged mode (for 8 ohm speakers), so that I always get around 100W, regardless what speakers (I think I wont stumble across any 2 ohm speakers...). This shall be possible, by selecting the operation mode with one or two switches while - obviously - the amp is powered off.Now, the problem is: how do I implement that? I don't want to build two amps... lol... But luckily I found Mark Henessy's A4 Power amp, which has a very interesting solution for the switching, but it is actually a bridged-parallel design (each channel consists always of a 4780 in parallel mode), so you only get a mono if working in bridged: http://www.mhennessy2.f9.co.uk/a4/analogue.htm
Anyway, his ideas gave me quite a good start for trying to make my own design, and here it is (to avoid confusion: this is showing only ONE channel, as they are identical, obviously). I've attached it to this post as a GIF image file. Hope it gets the correctly (as this is my first post in this forum).
Attention, there are a few small descriptional errors in the schematic (like incorrect pin number indications of missing pin number indications, and stuff like this. Thought I should mention this, so that you don't have to bother about it. They are irrelevant at this stage, and I'll correct it when the design is finished. Just didn't have the time to do eveything 100%, you know what I mean... I put my attention into doing the schematic right, not to make it pretty or perfect)
For all those of you having difficulties to understand my idea (I shouldn't think that this will be the case, as I've seen most people here seem really smart, some actually much more than me, lolol, but one never knows...):
I have both "halves" of the 4780 in inverting mode, and use (I borrowed this from Mark Henessey hehe) a OPA2134 opamp as an input buffer AND inverter, so that the signal actually is not inverted. This is talking about parallel mode, with the switches in parallel position. When changing both switches to bridged, we'll have the right output config (now one amp driving +, one driving -), and the input now is as following: one of the opa's gets the same signal as before, while the other gets the inverted signal, so that the 4780 connected to it also gets an inverted signal, so that the bridged mode is correctly implemented (at least I hope so lol...).
So, what I hope to hear from you guys is, WILL THIS WORK ??? I think it should
Or do I have to find a different solution, and if yes, any ideas?
And in case that it seems to be ok, please give your 5 cents to some of the most critical points in gainclone design that are always at discussion, so that I can choose the best for my application priorities (at least I've seen about three different approaches). Here they go:
1. I've mostly used the minimal filtering solution, like "inherited" from the gaincard and very nicely implemented by Brian and Peter Daniel and other great people
. But I've seen different approaches, and now I'm wondering: does this really have such impact on the sound? Because I kind of like the idea of doing more filtering, especially to avoid HF and other interferences (keep in mind that I'm going to use this amp in public spaces, one never knows what one can encoutner there). I give you three examples: http://www.tech-diy.com/paraclone.htm
http://www.tech-diy.com/LM4780_bridgedamp.htm
http://www.shine7.com/audio/pa100.htm
They have, for example, a filtering network from the input of the 4780 to GND, and a capacitor between + and - inputs of the amp, a capacitor parallel to the feedback resistor, a capacitor from the mute pin to GND, and other stuff I don't really see what it is for, and which I don't see in the "original" minimalistic gainclones, or in the most popular kits on here. So I didn't implement that stuff into my design. Now, what I'd like to know: maybe it would be a good idea to do it, though? Any real experiences on this area? Advantages/disadvantages? Alternatives?
Also, I have doubts about some components values:
The 2,2K from + input to GND. Somebody said: the higher the value, the better the DC offset, but the worse the noise generated by that resistor. So he set it to 1K... But I compared that to the other designs (and thought about the risks associated with DC) and so I set it to 2,2. Is that ok? or should it be even higher?
The 4,7uf cap in input (between opa's and the amps). I've seen people using 47uF... Any importance here? And would it be wiser to move the cap to the opa's input instead?? To keep out eventual DC signals completely?
The Gain setting resistors: 154k and 6,8k, giving us about 22 (hope my math's still right lol). Is that ok? Or shall I use smaller values, to get the same gain? Because i've seen like 22 k and 1 k, etc...
The OPA2134... I've heard it is one of the best opamps for audio. Is that true? So is it ok to use it, in order to get as small alteration of the sound as possible? Or have you better ideas?
Now, to the powersupply issues (which I didn't have time to do the design and the schematic yet):
I'm going to use a 500VA 25-0-25 Toroid transformer which I can get for a very nice 32 Euros from Farnell
, that's ok for what I'm doing, isn't it?Now, is it best to do a "snubberized" (didn't figure out yet what that means, but anyway) supply with big caps (20000 or above) or do what they did with the gaincard and in all minimalistic gainclones: something like 1000uF
I've heard that the sound of the amp is affected, that the bigger the caps, the more the sound gets a little more "cold", "electronic", like typical transistor amp, while when using small caps, it is warmer and better??? is this only a myth?? Because on the other hand, I've read that when using bigger caps, the amp will have a fuller and more powerful sound, especially at higher volumes. And what about regulated power supply? One of the sites I stated above, has a nice idea for that, and he actually states it improved sound. Is that valid? Or only if the transformer is kind of weak, so that voltages drops when higher power (which won't be the case here, I think...)?
Well, for now I think that's all... Would be great to have some feedback here and some good hints. At least for the design phase... later on I'll need also some practical tips (as for heat dissipation, case, etc...) but I think it's best to keep things separated and clear (too much confusion if asking all stuff at the same time, isn't it). So, thanks in advance for your answers, and sorry for the "testament", but I couldn't figure how to ask so many things and give enoiugh details on less space lol...
Attachments
Since the power stages will remain the same regardless of configuration, I think it would be beneficial to put the input circuitry on a separate PCB (like a daughtercard) that can be swapped out depending on what configuration you want. That will eliminate any need for switches. Of course you'd still have to power the amp down to swap out the input cards, but it would be more flexible and allow you to create different input circuits if you want different effects.
Nice idea, BWRX. But actually I don't see any inconvenience in using a switch. And making the board removable would mean to have to use some kind of socket, which could get connection problems over time from repeated removal... but thanks for the idea nethertheless, because now I'm going to really put the input stages on a separated board, not to allow it to be removed (there will still be the switch instead), but to be easier to change components or make modifications later, if I want to improve sound or whatever. Because in the case of one board, it would be hard to do mods once it is installed (as the amp chips are mounted on the heatsink, etc...). Looking forward to more comments. What I really need now is feedback, from those of you more experienced in the matter, so that I can know if my switchable design has chances to work...
Next thing to do will be the design of PSU. I have to seriously think about implementing a regulated one, because I just read somewhere that for bridged mode, the voltage should be about +-28V, not +-35 what I get with the 25 V transformer. So, what I was thinking was to still use the 25V trannie but get the voltage regulated to +-30V with a power transistor circuit (actually I think one of the example designs which I provided URLs of has one like that). That would be ok for parallel AND bridged mode. What do you think about this?
Next thing to do will be the design of PSU. I have to seriously think about implementing a regulated one, because I just read somewhere that for bridged mode, the voltage should be about +-28V, not +-35 what I get with the 25 V transformer. So, what I was thinking was to still use the 25V trannie but get the voltage regulated to +-30V with a power transistor circuit (actually I think one of the example designs which I provided URLs of has one like that). That would be ok for parallel AND bridged mode. What do you think about this?
Well ppl, maybe I've asked too many questions, maybe I even answered them myself in the same thread, maybe what I'm doing is nothing new for you, maybe it's because I am not known to this community yet, or maybe it's just that you guys have enough things to do, but one thing is for sure: I'd really appreciate some feedback... I feel a little disappointed about this forum...
I didn't expect so little interest or at least feedback from a forum with so many members... anyway, lets go on...
I've done the schematic for the PSU. It's annexed to this post. And I did a lot more research in order to answer some of my own questions (as nobody helped me here at least until now.
Here come my conclusions:
Gain setting for the amps seems to be ok (22). I've read it's usual to use 220k or more resistor for inverting amp, and 20k or more for non-inverting, so I think 154k is ok.
Exact value of resistor from positive input to gnd seems to have little importance, so I think 2.2k is ok (I even saw some people not using any resistor at all, but then I read that could cause instability).
The 4.7 uF between opamps and poweramp should be ok, but now I am in doubt if I should put decoupling caps also in front of the opamps?? And if yes, does it have to be special ones?
The OPA2134 should be ok also (although I might mount a socket in order to try other pin-compatible opamps and compare their sound...).
About the filtering issue, I will be using 1000uF caps close to the lm4780, and in the PSU section I use 10000uF together with smaller caps, power transistor bypassed voltage regulators and snubbers, following Carlos' design. So for what I've read, sound should not be compromised like that, and it has the advantage of delivering steady voltage and being able to deliver high power.
I also got assured that a 500VA transformer is more than enough. I then use MUR1520 diodes (did anybody try the MUR1540 ?? They are much cheaper... do they really sound worse?).
Now, I do have a few more things about the PSU:
I configured the voltage regulators to give me +-30V. I think that's a good compromise for bridged mode (preferring +-25V ) and parallel mode (preferring +-37V).
Is 3W enough for the resistors ? (2.2k discharge for the caps, and the 2.2R series resistor before the regulators)
I suppose the power transistors will require heatsinks, right? Could I use the same heatsinks as for the power amps ? (provided the sinks are big, and isolated properly from the power amps).
Is it better to use 2x 10000uF as I have in schematic, or 4x 4700uF (2 in parallel, giving more or less the same value)? or even more smaller caps in parallel to get same capacity? I've read that has advantages for sound and is cheaper... although it has big disadvantage: taking lots of space and being more solder/pcb work...
I implemented a "proper" earth connection like it was recommended on a website, connecting chassis to mains earth and then with a few components (to avoid ground loop) to the circuit ground. Is that alright?
So now, please give me some feedback about the whole thing... answers to my questions, but especially the power supply and also about my switching method and way to implement both bridged and parallel mode (because I have not yet found anybody doing it, so I don't know if it will work ok, and I'd like your opinion about it, because if it is positive I won't take such a risk than when building it without knowing if it will work... Please, ok? It won't take that much time to answer those questions if you know the answer And I promise, after I (hopefully) have built with success this amp, that I'll be there to help and assist anybody of the forum wanting to build a similar amp
P.S.: I didn't draw the regulator section for the supply of OPA2134 yet (providing +-15V), that will be next (actually I will copy it shamelessly from Mark Henessy's page)
I didn't expect so little interest or at least feedback from a forum with so many members... anyway, lets go on... I've done the schematic for the PSU. It's annexed to this post. And I did a lot more research in order to answer some of my own questions (as nobody helped me here at least until now
.Here come my conclusions:
Gain setting for the amps seems to be ok (22). I've read it's usual to use 220k or more resistor for inverting amp, and 20k or more for non-inverting, so I think 154k is ok.
Exact value of resistor from positive input to gnd seems to have little importance, so I think 2.2k is ok (I even saw some people not using any resistor at all, but then I read that could cause instability).
The 4.7 uF between opamps and poweramp should be ok, but now I am in doubt if I should put decoupling caps also in front of the opamps?? And if yes, does it have to be special ones?
The OPA2134 should be ok also (although I might mount a socket in order to try other pin-compatible opamps and compare their sound...).
About the filtering issue, I will be using 1000uF caps close to the lm4780, and in the PSU section I use 10000uF together with smaller caps, power transistor bypassed voltage regulators and snubbers, following Carlos' design. So for what I've read, sound should not be compromised like that, and it has the advantage of delivering steady voltage and being able to deliver high power.
I also got assured that a 500VA transformer is more than enough. I then use MUR1520 diodes (did anybody try the MUR1540 ?? They are much cheaper... do they really sound worse?).
Now, I do have a few more things about the PSU:
I configured the voltage regulators to give me +-30V. I think that's a good compromise for bridged mode (preferring +-25V ) and parallel mode (preferring +-37V).
Is 3W enough for the resistors ? (2.2k discharge for the caps, and the 2.2R series resistor before the regulators)
I suppose the power transistors will require heatsinks, right? Could I use the same heatsinks as for the power amps ? (provided the sinks are big, and isolated properly from the power amps).
Is it better to use 2x 10000uF as I have in schematic, or 4x 4700uF (2 in parallel, giving more or less the same value)? or even more smaller caps in parallel to get same capacity? I've read that has advantages for sound and is cheaper... although it has big disadvantage: taking lots of space and being more solder/pcb work...
I implemented a "proper" earth connection like it was recommended on a website, connecting chassis to mains earth and then with a few components (to avoid ground loop) to the circuit ground. Is that alright?
So now, please give me some feedback about the whole thing... answers to my questions, but especially the power supply and also about my switching method and way to implement both bridged and parallel mode (because I have not yet found anybody doing it, so I don't know if it will work ok, and I'd like your opinion about it, because if it is positive I won't take such a risk than when building it without knowing if it will work... Please, ok? It won't take that much time to answer those questions if you know the answer
And I promise, after I (hopefully) have built with success this amp, that I'll be there to help and assist anybody of the forum wanting to build a similar amp P.S.: I didn't draw the regulator section for the supply of OPA2134 yet (providing +-15V), that will be next (actually I will copy it shamelessly from Mark Henessy's page
)Attachments
I've just made up my mind: I won't be using 2x LM4780, but 4x LM3886. That's because of two reasons: they are much easier to mount in p2p because of less tight pin spacing, and they will have better heat dissipation, so I won't have so much trouble with heat sink size, etc.... And regarding the price, it actually turns out more or less the same...
Still no feedback?
Come on guys, I can't wait to have all issues solved and questions answered so that I can make the definitive schematics, and then start ordering all the parts....
Still no feedback?
Come on guys, I can't wait to have all issues solved and questions answered so that I can make the definitive schematics, and then start ordering all the parts....
If you stick to the basic implementations your chances of success will be very high. Adding switches to change the topology of the amp (stereo or bridged operation) shouldn't hurt as long as you power the amp down before changing the switches. You'd have to physically change the speaker wiring anyway so you should shut it off.
Don't use such high value resistors to set the gain. Stick to values below 100k. Most use resistors in the 1k-50k range.
The OPA2134 is a fine op amp. You can also cheaper ones. A socket is your best bet so you can easily change chips.
I would personally go for a design that does away with coupling caps but if you have to use them I'd put them in front of the OPA2134s.
Don't use such high value resistors to set the gain. Stick to values below 100k. Most use resistors in the 1k-50k range.
The OPA2134 is a fine op amp. You can also cheaper ones. A socket is your best bet so you can easily change chips.
I would personally go for a design that does away with coupling caps but if you have to use them I'd put them in front of the OPA2134s.
Thanks for answering
Lower value for the gain resistors? I chose 154k because I saw it in Mark Henessy's design, and in National's datasheet they have 220k... are you sure that it should be lower? Most designs I've seen had high values in inverting mode. Only in non-inverting, they use something like 20k, etc...
Regarding the decoupling caps, I will do like you suggested, no caps between opa2134 and amp, and only one capacitor at the input before the mode switch.
Now, should it be a 4,7uF high quality polarized electrolyt capacitor, or should it be a non-polarized but still electrolyt, or a non-polarized film-type as some people use? What would be best in your experience?
Now, anybody out there who would try to answer my questions regarding the PSU?
P.S.: there is one thing I didn't mention in last post and that I think is not right in my schematic: the 7A fuse... I am not quite sure how to calculate the value I saw 3 A in many design, but they had smaller transformers, AFAIK. So would 5A be a good compromise? I guess I'll have to see: if it blows, increase it's value
Lower value for the gain resistors?
I chose 154k because I saw it in Mark Henessy's design, and in National's datasheet they have 220k... are you sure that it should be lower? Most designs I've seen had high values in inverting mode. Only in non-inverting, they use something like 20k, etc...Regarding the decoupling caps, I will do like you suggested, no caps between opa2134 and amp, and only one capacitor at the input before the mode switch.
Now, should it be a 4,7uF high quality polarized electrolyt capacitor, or should it be a non-polarized but still electrolyt, or a non-polarized film-type as some people use? What would be best in your experience?
Now, anybody out there who would try to answer my questions regarding the PSU? P.S.: there is one thing I didn't mention in last post and that I think is not right in my schematic: the 7A fuse... I am not quite sure how to calculate the value
I saw 3 A in many design, but they had smaller transformers, AFAIK. So would 5A be a good compromise? I guess I'll have to see: if it blows, increase it's value
You can use lower value resistors no problem. Have a look at National's LM4780 datasheet. On page 5 they have a bridged amp schematic with resistor values of 4.7k and 51.1k for the inverted half.
As far as the PSU goes, since you'll be using a transformer with separate secondaries it would probably be more economical to use two identical positive regulator circuits instead of a positive and a negative circuit.
As far as the PSU goes, since you'll be using a transformer with separate secondaries it would probably be more economical to use two identical positive regulator circuits instead of a positive and a negative circuit.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.