Fake class A
Heres a quick drawing of an idea for "fake class A" using this kit in bridged mode. it does not waste power all the time or need an output capacitor.
I did not calculate the values of the resistors yet, because i was too lazy too look up the gain of the amp, and it may take some experimentation to get it "just right".
I am sure stuff like this has been done before, but i thought i should post my idea anyways.
if the exact same offset is applied to both the non inverted and inverted inputs then both amp outputs should have the same potential relative to each other, so there will be no DC on the speaker. That means no output capacitor. and no current passing.
of course it will have to be very carefully adjusted with the trim to get zero DC on output relative to each other.
I am not an EE , i just make things for DSP, which will not let the smoke go if you make a mistake
so if there is any bad flaws in this design (besides failure modes, which i believe can be looked after in non obtrusive manner)
if the offset is half of the voltage that would give maximum output, (depending on gain) then you would have "pure class a" but at half of the maximum power of the amp.
I was thinking as a "failsafe" in case of failure of an amp or other bad thing . DC detector , but it just disconnects the +v on the offset circuit and it reverts to"normal" function.
another bonus of this is that there should be good power noise rejection, because the offset voltage will both change at the same time if there is any noise, so the relative difference will be almost nothing. that is why i did not go for a voltage reference rather than just the +v (also if the output power drops on loud passage the offset voltage will drop correspondingly so it is still half)
you might be better off with a EI type transformer If you try this, because the bottom half of the rail will hardly be used, and apparently if you load toroids asymmetrically it can lead to some problem (magnetizing of the core, something bad i forgot)
i could be comletely wrong on this though now i think about it, the rectifier supplies both halves of the power to both rails if it is center tapped, so if you have dual out transformer it might be better to set it up as center tapped, and use 1 bridge.
Heres a quick drawing of an idea for "fake class A" using this kit in bridged mode. it does not waste power all the time or need an output capacitor.
I did not calculate the values of the resistors yet, because i was too lazy too look up the gain of the amp, and it may take some experimentation to get it "just right".
I am sure stuff like this has been done before, but i thought i should post my idea anyways.
if the exact same offset is applied to both the non inverted and inverted inputs then both amp outputs should have the same potential relative to each other, so there will be no DC on the speaker. That means no output capacitor. and no current passing.
of course it will have to be very carefully adjusted with the trim to get zero DC on output relative to each other.
I am not an EE , i just make things for DSP, which will not let the smoke go if you make a mistake
so if there is any bad flaws in this design (besides failure modes, which i believe can be looked after in non obtrusive manner)if the offset is half of the voltage that would give maximum output, (depending on gain) then you would have "pure class a" but at half of the maximum power of the amp.
I was thinking as a "failsafe" in case of failure of an amp or other bad thing . DC detector , but it just disconnects the +v on the offset circuit and it reverts to"normal" function.
another bonus of this is that there should be good power noise rejection, because the offset voltage will both change at the same time if there is any noise, so the relative difference will be almost nothing. that is why i did not go for a voltage reference rather than just the +v (also if the output power drops on loud passage the offset voltage will drop correspondingly so it is still half)
you might be better off with a EI type transformer If you try this, because the bottom half of the rail will hardly be used, and apparently if you load toroids asymmetrically it can lead to some problem (magnetizing of the core, something bad i forgot)
i could be comletely wrong on this though now i think about it, the rectifier supplies both halves of the power to both rails if it is center tapped, so if you have dual out transformer it might be better to set it up as center tapped, and use 1 bridge.
Attachments
Peter & Brian,
I know it might not mean much from a board lurker, and not to drag things on, but thank you. It definitely shows that you look out for the best interests of the group.
You guys are class acts, and as a beginning audio do-it-yourselfer, your projects make it possible for someone who isn't completely confident in their skills to try out some great stuff. I'm not one for kits, but the PCB's definitely help me out.
Rest assured that I'm sure I'm not the only one on this board that appreciates both of your efforts!
Mike
I know it might not mean much from a board lurker, and not to drag things on, but thank you. It definitely shows that you look out for the best interests of the group.
You guys are class acts, and as a beginning audio do-it-yourselfer, your projects make it possible for someone who isn't completely confident in their skills to try out some great stuff. I'm not one for kits, but the PCB's definitely help me out.
Rest assured that I'm sure I'm not the only one on this board that appreciates both of your efforts!
Mike
Re: Zobel pads for LM3875 boards, too? Or just LM4780?
I don't feel that it is nearly as big of an issue for the LM3875, since it is only a single channel, not 2 channels being paralleled. Since there are no output resistors, the Zobel could quite easily be attached to the binding posts for the LM3875, something that cannot be effectively with the current LM4780.
It is possible that a future version of the LM3875 will include the Zobel network, but not for now.
--
Brian
KT said:
I don't feel that it is nearly as big of an issue for the LM3875, since it is only a single channel, not 2 channels being paralleled. Since there are no output resistors, the Zobel could quite easily be attached to the binding posts for the LM3875, something that cannot be effectively with the current LM4780.
It is possible that a future version of the LM3875 will include the Zobel network, but not for now.
--
Brian
For the recent orders, all of the US ones are shipping out tomorrow (technically today), and half of the international ones are shipping out. I still need to fill out the customs forms and address labels for the other half, but they should go out the next day. US orders should all have received a Delivery Confirmation number.
As for this group order, it seems to have been quite a success with several finished amplifiers so far (see gallery ). I have been donating portions of the proceeds to diyaudio throughout the process, and have donated $500 thus far. I would like to encourage all group orders to be run like this, in order to help out diyAudio. Thank you to all who have participated.
For those who think that they missed out, I am still taking orders for PCB sets, and basic kits on my website:
http://www.BrianGT.com/order
I am hoping to get the rest of the components for the premium kits soon, but it has been quite a task, since I bought out all small inventories from multiple vendors to fill the first batch of orders, leaving everyone to face the 5-8 week leadtimes from the manufacturers.
--
Brian
As for this group order, it seems to have been quite a success with several finished amplifiers so far (see gallery ). I have been donating portions of the proceeds to diyaudio throughout the process, and have donated $500 thus far. I would like to encourage all group orders to be run like this, in order to help out diyAudio. Thank you to all who have participated.
For those who think that they missed out, I am still taking orders for PCB sets, and basic kits on my website:
http://www.BrianGT.com/order
I am hoping to get the rest of the components for the premium kits soon, but it has been quite a task, since I bought out all small inventories from multiple vendors to fill the first batch of orders, leaving everyone to face the 5-8 week leadtimes from the manufacturers.
--
Brian
BrianGT said:
This was supposed to be placed in the LM3875 thread, but I replied to the wrong thread. Sorry for the confusion. I would removed the previous post if I had caught it earlier. This is a reference to the latest batch of LM3875 kits that shipped out. Hopefully, by the end of next month, the first batch of LM4780 kits will have been shipped out. Currently, I am preparing to get prototype pcbs made to test out the design before production. I will post pictures when I get prototype pcbs.
--
Brian
Hi,
I emailed BrianGT about this already, but maybe some other people here can give their opinion on this too
For a first time (chip-)amp I would like to build one of the kits BrianGT distributes. They will be driving my 87dB diy speakers.
What would you choose, being most educational and foolproof? This new 4780 kit or the 3875 kit?
I do have some soldering experience and recently built my own speakers, but no experience whatsoever on amps yet.
Thanks Razorfist
I emailed BrianGT about this already, but maybe some other people here can give their opinion on this too
For a first time (chip-)amp I would like to build one of the kits BrianGT distributes. They will be driving my 87dB diy speakers.
What would you choose, being most educational and foolproof
? This new 4780 kit or the 3875 kit?I do have some soldering experience and recently built my own speakers, but no experience whatsoever on amps yet.
Thanks Razorfist
Razorfist said:I emailed BrianGT about this already, but maybe some other people here can give their opinion on this too
For a first time (chip-)amp I would like to build one of the kits BrianGT distributes. They will be driving my 87dB diy speakers.
What would you choose, being most educational and foolproof
I do have some soldering experience and recently built my own speakers, but no experience whatsoever on amps yet.
Let me reiterate my response here for others. I am using the LM3875 gainclone amplifier with my < 90 dB efficient Seas Thor TL speakers (I think they are around 87dB), and I get plenty of power with them. I can turn it up to the point where it makes my ears hurt, without the amplifier clipping. (320VA 2x24vac transformer) It is also a much more proven design than the LM4780. The only hard part about the kit is creating a chassis, and wiring up the power supply (be careful with the AC). I have a manual for the LM3875 kit on my website:
http://www.BrianGT.com (I will add more to it someday when I am not busy)
I also haven't built a LM4780 amplifier yet, and am ordering prototypes soon to try it out, before doing a large order of pcbs.
It will be a month or so before the LM4780 kits start shipping out, since it will take a week to get prototype boards made, some time to test it, then another week to get production boards made.
My intention for the LM4780 is to drive the bottom end of an active crossover system, as this will allow me to use a more inefficient woofer and still match my mids and tweeter.
--
Brian
about the zobels and ect...
every feature requires you place components and then route stuff. granted the board uses 2 sides but doesn't use ALL of both sides...
adding components beyond the caps will prevent clamp-mounting.
this board appears to be a minimalist board -- there are no 10uf or 0.1uf bypass caps, and no zobels.
also, can you phycially move a zobel and maintain good effectiveness? if you move the zobels too far, wouldn't you just defeat the purpose of them becuase there would be too much inductance.
every feature requires you place components and then route stuff. granted the board uses 2 sides but doesn't use ALL of both sides...
adding components beyond the caps will prevent clamp-mounting.
this board appears to be a minimalist board -- there are no 10uf or 0.1uf bypass caps, and no zobels.
also, can you phycially move a zobel and maintain good effectiveness? if you move the zobels too far, wouldn't you just defeat the purpose of them becuase there would be too much inductance.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.