I've built the Ad815 pre-amp based on Carlosfm's schematics
http://www.diyaudio.com/forums/showthread.php?s=&threadid=40783&perpage=10&highlight=&pagenumber=7
and thought I'd share my experiences:
Good:
1. Sound is marvelous-Soundstaging is huge.
2. Vs the 627/634 combo, highs are more resolved, smoother, more extended, more enjoyable.
3. Bass is deeper as well. The one characteristic that stands out is the impression of "drive", especially in the upper bass.
4. Construction is relatively straightforward.
5. Using a nulling circuit, DC output can be reduced to zero.
Neutral:
1. The chip is only available as a 24-pin surface mount component. Not a problem if you can fabricate your own PCB's-I simply made a SOIC-DIL adapter and it works like a charm. The SOIC footprint allows the use of SM components (caps, resistors), which could bring noise benefits because of the small feedback loop etc.
Bad/Educational!:
1. For two weeks I struggled to get the Ad815 to work consistently with my monoblocked 3875 gainclone. In my initial configuration, the ground of the AD815 pre was coupled to the Ac ground for the amp.
The pre was ENORMOUSLY sensitive to positioning, volume pot adjustments, the phase of the moon, etc. humming at the drop of a hat and seeming to pick up all sorts of inteference, including transformer magnetic fields etc-drove me insane for a while trying to understand the problem.
The BIG lesson out of all of this was: Ensure that your amp has a ground connection to chassis. I tied the amp gnd and pre-amp gnd (which were connected to Ac mains gnd) to the chassis and since then the pre has been working perfectly.
My 2c for anyone trying this fantastic little chip.
Ryan
http://www.diyaudio.com/forums/showthread.php?s=&threadid=40783&perpage=10&highlight=&pagenumber=7
and thought I'd share my experiences:
Good:
1. Sound is marvelous-Soundstaging is huge.
2. Vs the 627/634 combo, highs are more resolved, smoother, more extended, more enjoyable.
3. Bass is deeper as well. The one characteristic that stands out is the impression of "drive", especially in the upper bass.
4. Construction is relatively straightforward.
5. Using a nulling circuit, DC output can be reduced to zero.
Neutral:
1. The chip is only available as a 24-pin surface mount component. Not a problem if you can fabricate your own PCB's-I simply made a SOIC-DIL adapter and it works like a charm. The SOIC footprint allows the use of SM components (caps, resistors), which could bring noise benefits because of the small feedback loop etc.
Bad/Educational!:
1. For two weeks I struggled to get the Ad815 to work consistently with my monoblocked 3875 gainclone. In my initial configuration, the ground of the AD815 pre was coupled to the Ac ground for the amp.
The pre was ENORMOUSLY sensitive to positioning, volume pot adjustments, the phase of the moon, etc. humming at the drop of a hat and seeming to pick up all sorts of inteference, including transformer magnetic fields etc-drove me insane for a while trying to understand the problem.
The BIG lesson out of all of this was: Ensure that your amp has a ground connection to chassis. I tied the amp gnd and pre-amp gnd (which were connected to Ac mains gnd) to the chassis and since then the pre has been working perfectly.
My 2c for anyone trying this fantastic little chip.
Ryan
Revitalizing an old thread here - came as a result of seeing a mention of the AD815 over on 6moons as its mentioned in a review of a 'Clones Audio' preamp. I read a longish thread on DIYHifi first where CarlosFM presents his design and helps punters get theirs going.
Seems the part needs very careful layout - oscillation is indicated when the DC offsets can't be nulled out successfully or one channel is markedly different from the other. Many of the original builders were using the through-hole part which has now been discontinued, the SMT package should in theory have fewer problems as the parasitics can be lower.
I shall order up some of these and have a play - one idea is to build a paralleled array of them (say 8 to 10) to get enough current to drive an 8R speaker to reasonable levels. This seems to be the ADSL driver with the highest usable voltage rails and also the lowest LF noise corner (a current hobby-horse of mine). I'm envisioning a kind of audiophile version of Doug Self's 'beast of many 5532s'
Seems the part needs very careful layout - oscillation is indicated when the DC offsets can't be nulled out successfully or one channel is markedly different from the other. Many of the original builders were using the through-hole part which has now been discontinued, the SMT package should in theory have fewer problems as the parasitics can be lower.
I shall order up some of these and have a play - one idea is to build a paralleled array of them (say 8 to 10) to get enough current to drive an 8R speaker to reasonable levels. This seems to be the ADSL driver with the highest usable voltage rails and also the lowest LF noise corner (a current hobby-horse of mine). I'm envisioning a kind of audiophile version of Doug Self's 'beast of many 5532s'
Intersting part
seems you want a ground plane for power dissipation to the PCB.
don't most all current mode opamps need many MHz layout techniques
maybe separate power and signal returns still can be done using a ground plane?
ground plane using reverse voltage PS would be my approach "HEAT TABS ARE CONNECTED TO THE POSITIVE SUPPLY"
seems you want a ground plane for power dissipation to the PCB.
don't most all current mode opamps need many MHz layout techniques
maybe separate power and signal returns still can be done using a ground plane?
ground plane using reverse voltage PS would be my approach "HEAT TABS ARE CONNECTED TO THE POSITIVE SUPPLY"
Last edited:
In that case, highly recommended reading http://www.diyaudio.com/forums/chip-amps/226437-optimizing-tda7294-output-15.html#post3312141
nah I always use a ground plane even for PTP wiring.
dead bug ICs on a unused FR4 PCB using lil islands of cut up PCB chip pieces soldered on top of blank.
BTW the data sheet says
If a ground plane is to be used on the same side of the board as
the signal traces, a space (5 mm min) should be left around the
signal lines to minimize coupling..
my translation = keep gain and input R's close and isolated from ground ( hanging in air)
dead bug ICs on a unused FR4 PCB using lil islands of cut up PCB chip pieces soldered on top of blank.
BTW the data sheet says
If a ground plane is to be used on the same side of the board as
the signal traces, a space (5 mm min) should be left around the
signal lines to minimize coupling..
my translation = keep gain and input R's close and isolated from ground ( hanging in air)
Last edited:
Thanks - I had been following that thread a while ago, I see I'm well out of date. KSTR often has very valuable insights into chipamp stuff. I see he's showing a single-ended topology, whereas I prefer bridged to reduce the number of caps. In my experience, caps are the limiting factor for SQ more so than the silicon. When the silicon's above a certain quality level that is.
Ti's version http://www.ti.com/lit/ug/slou034/slou034.pdf
2 layer PCB has good layout notes
they recommend strip line style interconnects with grounds on top and bottom shows clearance around inputs / outputs.
back layer is ground / power / heat plane
thermally s/b be much better than ADI
2 layer PCB has good layout notes
they recommend strip line style interconnects with grounds on top and bottom shows clearance around inputs / outputs.
back layer is ground / power / heat plane
thermally s/b be much better than ADI
Last edited:
Another AD815 User
All the praise about how good the AD815 sounds is true!
I made a PCB for this chip to use as a 3.5X gain/buffer stage in my Fanfare Tuner. Stunningly good sound from an FM tuner.
All the challenges and complaints about the AD815 are also true.
The DC offset is large and shifts with temperature and is worse with more gain. These chips run hot to the touch. I bread-boarded a servo circuit to correct for DC offset and expect to receive the PCBs with this improvement tomorrow. I should know in about a week how this turns out.
Three of these boards from OSHPARK cost $28USD including shipping to the USA. I use two of these boards for dual mono mode. One AD815 user claims that you get the best sound if you only use one channel for line out application with the second channel set at unity gain and no load. I will share the Eagle CAD PCB files with anyone who is interested. I can also post the schematic if anyone is interested.
All the praise about how good the AD815 sounds is true!
I made a PCB for this chip to use as a 3.5X gain/buffer stage in my Fanfare Tuner. Stunningly good sound from an FM tuner.
All the challenges and complaints about the AD815 are also true.
The DC offset is large and shifts with temperature and is worse with more gain. These chips run hot to the touch. I bread-boarded a servo circuit to correct for DC offset and expect to receive the PCBs with this improvement tomorrow. I should know in about a week how this turns out.
Three of these boards from OSHPARK cost $28USD including shipping to the USA. I use two of these boards for dual mono mode. One AD815 user claims that you get the best sound if you only use one channel for line out application with the second channel set at unity gain and no load. I will share the Eagle CAD PCB files with anyone who is interested. I can also post the schematic if anyone is interested.
I build a 30W power amp, each channel has 6 AD815 chips in bridged parallel. Works like a charm.
Jan
Oh well, that reads as a true quest, hopefully shortly with a satisfactory ending for you.
Apparently, you did not experience oscillation as some others have had.
I would be interested in your schematic, and Eagle files. You might post them here, or I could pm you.
@Jan: In April, I probably will be near your place. Will let you know timely. Arjen
Apparently, you did not experience oscillation as some others have had.
I would be interested in your schematic, and Eagle files. You might post them here, or I could pm you.
@Jan: In April, I probably will be near your place. Will let you know timely. Arjen
Last edited:
After a little more experience with the AD815 - in particular as the output stage for a DAC design - I decided not to go ahead with the multiple paralleled amp. I did though experiment with the AD815 as a unity gain bootstrapped buffer, a quick listen on headphones shows this has promise.
A couple of reasons for using bootstrapped supplies (emitter followers on the two rails) - firstly the quiescent consumption rises significantly at higher supplies. The bootstrapping I used kept to +5/-5V. Another reason was to improve the PSRR.
A couple of reasons for using bootstrapped supplies (emitter followers on the two rails) - firstly the quiescent consumption rises significantly at higher supplies. The bootstrapping I used kept to +5/-5V. Another reason was to improve the PSRR.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.