Overrun conditions detected--please stretch to a Quad parallel
In regards to the photo in post #71, please add another LM1875 chip to the center, stretching it out to a Quad Parallel.
This is necessary to enhance the usefulness of the amplifier. Too many of the applications put triple LM1875 into overrun condition. It must be a Quad Parallel, to be more worthy of PCB design efforts and support a wider variety of speakers.
With post 71, you may need to size most of the resistor pads for 1/2w resistors (especially those that jump over traces); however, it should be fairly easy to pop another chip right in at the middle. Considering the current, the power rails and 0v need to be thickened when possible.
In regards to the photo in post #71, please add another LM1875 chip to the center, stretching it out to a Quad Parallel.
This is necessary to enhance the usefulness of the amplifier. Too many of the applications put triple LM1875 into overrun condition. It must be a Quad Parallel, to be more worthy of PCB design efforts and support a wider variety of speakers.
With post 71, you may need to size most of the resistor pads for 1/2w resistors (especially those that jump over traces); however, it should be fairly easy to pop another chip right in at the middle. Considering the current, the power rails and 0v need to be thickened when possible.
Last edited:
you're going to need a fairly beefy trafo to run that (600va springs to mind)
i am not sure i can fit 4 as the boards i can do are limited to a 100x100 size
i will see what i can figure out as i need a bit of down time to get my brain doing a bit of thinking. has been nice being on holidays.
i am not sure i can fit 4 as the boards i can do are limited to a 100x100 size
i will see what i can figure out as i need a bit of down time to get my brain doing a bit of thinking. has been nice being on holidays.
It seems to work to calculate at 48 watts per board times 1.5.
For a bridged monobloc, that's a low cost 4 ampere transformer.
This seems workable for a nice big stereo amp: http://www.antekinc.com/details.php?p=614
For a bridged monobloc, that's a low cost 4 ampere transformer.
This seems workable for a nice big stereo amp: http://www.antekinc.com/details.php?p=614
Last edited:
think i've answered my first vopeak question, as 17v vs 35v seems somewhat more consistent
2.2a vs 4.4a
i guess where i'm headed is i calculate each of the LM power requirements individually, then, multiply by a factor of (x) being the number of modules used (which for 4 would be 8.8 and 3, 6.6 respectively.
the voltage requirement though I would expect to remain at 17v across all modules
so a 300va supply would suffice here - ie 8.8x17/0.6 = 250va
2.2a vs 4.4a
i guess where i'm headed is i calculate each of the LM power requirements individually, then, multiply by a factor of (x) being the number of modules used (which for 4 would be 8.8 and 3, 6.6 respectively.
the voltage requirement though I would expect to remain at 17v across all modules
so a 300va supply would suffice here - ie 8.8x17/0.6 = 250va
ah daniel-san
i didn't think this was possible
some caveats ;
in your original post theres a couple of errors which i have worked around - however there are still things to be added to this so, its by no means complete.
i have omitted the 100nf 'filter' caps - they can be soldered direct under the board
there are resistors on the feed to the line out --> these would be wire 'bridges'
most of the resistors - are still to be determined values.
NOTE the overlapping parts when the board is build would NOT overlap - they are 'oversized' in eagle for larger pin spacing.
please note this is a first cut - and needs bits and bobs before it is in any way shape or form usable - its taken me MANY hours to get it sorted though.
oh i reckon another pair of capacitors would balance the board out too
i didn't think this was possible
some caveats ;
in your original post theres a couple of errors which i have worked around - however there are still things to be added to this so, its by no means complete.
i have omitted the 100nf 'filter' caps - they can be soldered direct under the board
there are resistors on the feed to the line out --> these would be wire 'bridges'
most of the resistors - are still to be determined values.
NOTE the overlapping parts when the board is build would NOT overlap - they are 'oversized' in eagle for larger pin spacing.
please note this is a first cut - and needs bits and bobs before it is in any way shape or form usable - its taken me MANY hours to get it sorted though.
oh i reckon another pair of capacitors would balance the board out too
ok so went a bit mad
- managed to get the snubbing capacitors (100nf) on the leads of the radial capacitors - tight fit but there.
- also managed to provision space on the input for a DC blocking cap from the line input - with multiple vias for different types of capacitors
- re-arranged things and moved radial capacitors around to accomodate spacing
- added an additional 2 x radial capacitors (so 6 x 220uf) instead of 4 x 330uf
i'll put the board images up later today, need to get some shuteye.
oh - we also need to decide on some values for resistors. plus i remember andrewT suggesting diodes inverse on the resistor joining the audio ground from the power ground. the resistor is bottom left next to the line out
- managed to get the snubbing capacitors (100nf) on the leads of the radial capacitors - tight fit but there.
- also managed to provision space on the input for a DC blocking cap from the line input - with multiple vias for different types of capacitors
- re-arranged things and moved radial capacitors around to accomodate spacing
- added an additional 2 x radial capacitors (so 6 x 220uf) instead of 4 x 330uf
i'll put the board images up later today, need to get some shuteye.
oh - we also need to decide on some values for resistors. plus i remember andrewT suggesting diodes inverse on the resistor joining the audio ground from the power ground. the resistor is bottom left next to the line out
The board needs to show correct size 330uF power caps--they are the same width as LM1875 chips.
The power caps need to be closer to the chips via more compact layout.
The feedback resistors need relocated to under-board direct connect
The NFB cap does not share the groundlift because resistance on both sides of a cap turns high efficiency caps into cheap lossy caps--the series resistor already installed as the feedback-shunt resistor is sufficient for any necessary current drop.
The NFB cap needs to share groundpoint with the 330uF caps which are already at small signal and do serve as a noise shield.
Only the + input needs the groundlift feature due to the possibility of a grounded source ground loop and other unwanted currents.
The intermittent connectivity terminal blocks need removed from the board to avoid promoting randomized unexpected operation.
The power caps need more even spacing such as: chip, caps, chip, chip, caps, chip at least.
The speaker return needs direct connect via shortest possible trace to the 0v input tap of the board, to avoid accidentally applying speaker return INTO the source device, which would be a "hard sound" noise promotion error, and if possible keep speaker return away from audio input by locating at opposite ends of the board.
The 100x100mm board needs center perforated to fit TWO of quad parallel amplifiers into 100x50mm space each, which will help avoid wallpapering the board with signal degrading space filling convolutions and we could literally cut PCB costs in half instead. It should be an interesting challenge, although perfboard might be magnitudes faster than computer software. . .
The power caps need to be closer to the chips via more compact layout.
The feedback resistors need relocated to under-board direct connect
The NFB cap does not share the groundlift because resistance on both sides of a cap turns high efficiency caps into cheap lossy caps--the series resistor already installed as the feedback-shunt resistor is sufficient for any necessary current drop.
The NFB cap needs to share groundpoint with the 330uF caps which are already at small signal and do serve as a noise shield.
Only the + input needs the groundlift feature due to the possibility of a grounded source ground loop and other unwanted currents.
The intermittent connectivity terminal blocks need removed from the board to avoid promoting randomized unexpected operation.
The power caps need more even spacing such as: chip, caps, chip, chip, caps, chip at least.
The speaker return needs direct connect via shortest possible trace to the 0v input tap of the board, to avoid accidentally applying speaker return INTO the source device, which would be a "hard sound" noise promotion error, and if possible keep speaker return away from audio input by locating at opposite ends of the board.
The 100x100mm board needs center perforated to fit TWO of quad parallel amplifiers into 100x50mm space each, which will help avoid wallpapering the board with signal degrading space filling convolutions and we could literally cut PCB costs in half instead. It should be an interesting challenge, although perfboard might be magnitudes faster than computer software. . .
Attachments
Last edited:
$84? Check out 1 pair of 36vct (18,18) amper transformers, which is a very common 144va transformer for both EI core and Toroid as well. They are $30ea at most. So, $60 for dual mono or likewise monoblocs. That's an awful price for a stereo build, but a very good price for a dual mono build.
Also check out 1 pair adjustable SMPS derated to run no harder than 50% advertised current (derating is same as for computer power supply SMPS).
As far as I know, you can sometimes source 24v (adjustable a bit via little orange dial in most of their photos) SMPS for $20 each. It would take two for split rail. The resulting ~$40 for a stereo amp power source is reasonable although not easy to find. SMPS used for split rail have to be isolated from each other as documented at Decibel Dungeon. However, the clean DC output is a great convenience.
in order ;
done - i have added 2 ; so 3 x 220uf
radial or NP 100nf?
easy to do under board
not sure i get that on the NFB cap - i think what you're saying is it needs to be part of the 0v ground
+ input and groundlift - what is groundlift?
OK will make them wire pads instead - will buy board space
see the 6 x 220uf
speaker return do you mean speaker negative, and, if so, does it return back to 0v ground?
100x100 board - if i read you right you want 2 lots of quad setups (so 8 LM1875 on there 1 x 100x100 board - can't do without serious removal of components and using SMD resistors and caps. best I got my non SMD down to was 50x50 (2" x 2") - what we could do is something similar to the BWRX modules where he got the simple down to 1" x 1" (25 x 25mm) or a quarter in your school thinking I think - but he used SMD components
trafo's are shocking here DWB
next down at 18-0-18 in 160va is 55 bucks AUD
EI they come down to about 30 bucks but the amperage is lucky to be 2A
SMPS - don't even go there. HIDEOUSLY expensive here. 20 bucks for an SMPS here - i wish
will post up the updated board in a tic with the single compact i did.
done - i have added 2 ; so 3 x 220uf
radial or NP 100nf?
easy to do under board
not sure i get that on the NFB cap - i think what you're saying is it needs to be part of the 0v ground
+ input and groundlift - what is groundlift?
OK will make them wire pads instead - will buy board space
see the 6 x 220uf
speaker return do you mean speaker negative, and, if so, does it return back to 0v ground?
100x100 board - if i read you right you want 2 lots of quad setups (so 8 LM1875 on there 1 x 100x100 board - can't do without serious removal of components and using SMD resistors and caps. best I got my non SMD down to was 50x50 (2" x 2") - what we could do is something similar to the BWRX modules where he got the simple down to 1" x 1" (25 x 25mm) or a quarter in your school thinking I think - but he used SMD components
trafo's are shocking here DWB
next down at 18-0-18 in 160va is 55 bucks AUD
EI they come down to about 30 bucks but the amperage is lucky to be 2A
SMPS - don't even go there. HIDEOUSLY expensive here. 20 bucks for an SMPS here - i wish
will post up the updated board in a tic with the single compact i did.
That's beginning to look quite good.
A few suggestions.
Break the ground link below the 4th terminal PIN (labeled -24V)
Add in two pads above the row of terminals for a ground link. Space the pads apart so that you can fit a resistor here, instead of a link.
Add RF attenuator (1k&680pF)after input cap C1.
Delete terminals 3 and 7 (now only 3prs of terminals) leaving space to make the PCB narrower.
A few suggestions.
Break the ground link below the 4th terminal PIN (labeled -24V)
Add in two pads above the row of terminals for a ground link. Space the pads apart so that you can fit a resistor here, instead of a link.
Add RF attenuator (1k&680pF)after input cap C1.
Delete terminals 3 and 7 (now only 3prs of terminals) leaving space to make the PCB narrower.
Last edited:
Previously seen as a rather bright blue 2.2 ohm resistor.
Here is for installing it on the 1-chip board:
thanks AndrewT
i built it for something to do and it works really well
i'll make those mods and report back
yep - i remember the one DWB.
IIRC AndrewT suggest we add inverse diodes too.
can someone explain why we do this please - just want to understand why i am making these changes - i understand that we need to seperate (lift) grounds between input ground and 0v - when i previously made the above board i didn't have the link between the two ground planes and ended up with +26vdc on the speaker out.
i built it for something to do and it works really well
i'll make those mods and report back
yep - i remember the one DWB.
IIRC AndrewT suggest we add inverse diodes too.
can someone explain why we do this please - just want to understand why i am making these changes - i understand that we need to seperate (lift) grounds between input ground and 0v - when i previously made the above board i didn't have the link between the two ground planes and ended up with +26vdc on the speaker out.
locating the Main Audio Ground on the PCB works well in a monoblock arrangement, if the trace layout is done correctly.
As soon as one introduces extra channels, one ends up with extra MAG/s.
At some stage those MAG have to be connected back to chassis. That's where the problem rears it's hum & buzz.
By separating the MAG from the PCB one has a choice of where to locate it and how the various PCBs attach to it.
As soon as one introduces extra channels, one ends up with extra MAG/s.
At some stage those MAG have to be connected back to chassis. That's where the problem rears it's hum & buzz.
By separating the MAG from the PCB one has a choice of where to locate it and how the various PCBs attach to it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.