Hi all,
after comparing my GC with quality commercial amps I've confirmed that my build sounds "bloated". By that I mean that there is more than enough bass and mid-bass, but mids and highs seem recessed. The amp obviously has more dynamics compared to others, lacks in HF transparency, but it sounds "bloated".
I'm just wondering if anyone else had a problem like this, before I test it without the regs and play around to see if I should change some components?
These are my changes to a stock chipamp:
Removed R1 and Ci
Lowered R2 to 10k to reduce DC offset
Increased Rz to 5R
Added 220pf across pins 9 and 10
Added carlosfm regulators and snubbers
I don't use a decoupling input cap at all
Trafo is a toroid 160VA, 2x29V sec, regs are fine tuned to 30V
after comparing my GC with quality commercial amps I've confirmed that my build sounds "bloated". By that I mean that there is more than enough bass and mid-bass, but mids and highs seem recessed. The amp obviously has more dynamics compared to others, lacks in HF transparency, but it sounds "bloated".
I'm just wondering if anyone else had a problem like this, before I test it without the regs and play around to see if I should change some components?
These are my changes to a stock chipamp:
Removed R1 and Ci
Lowered R2 to 10k to reduce DC offset
Increased Rz to 5R
Added 220pf across pins 9 and 10
Added carlosfm regulators and snubbers
I don't use a decoupling input cap at all
Trafo is a toroid 160VA, 2x29V sec, regs are fine tuned to 30V
Ok, here is the original amp scheme with my changes:
http://db.tt/VfzG5lL5
And here is a link to the regulated snubberized PSU:
http://www.diyaudio.com/forums/chip-amps/56106-lm338-regulated-snubberized-psu-audio-amplifiers.html
Can you guys comment on the overall sound of your GC's?
I read somewhere that "a GC is known to lack authority in the bass" (compared to a F5, I think), which is exactly the opposite of what I'm experiencing. I guess that HF transparency will be achieved with better quality caps, like Silmic II, but for now I just want to understand if there is something wrong here.
http://db.tt/VfzG5lL5
And here is a link to the regulated snubberized PSU:
http://www.diyaudio.com/forums/chip-amps/56106-lm338-regulated-snubberized-psu-audio-amplifiers.html
Can you guys comment on the overall sound of your GC's?
I read somewhere that "a GC is known to lack authority in the bass" (compared to a F5, I think), which is exactly the opposite of what I'm experiencing. I guess that HF transparency will be achieved with better quality caps, like Silmic II, but for now I just want to understand if there is something wrong here.
Go back and make this an AC coupled amplifier.
Add in all the "optional" components listed by National
Ensure the HF decoupling has a very short route. The two ceramic caps should be soldered to each other. The other ends get soldered to the power pins.
The MF decoupling can be a little further away. BUT and this is very important: The link between the two MF decoupling caps MUST be very close to the junction between the two HF decoupler caps. Connect these two points together to create your Power Ground.
Connect the output Zobel to this Power Ground.
This minimises the route that the HF spikes must follow around the output stage. This really is very important and 99% of the PCB layouts (discrete and chipamp) that get posted on this Forum completely ignore this HF route requirement.
Ensure the power input leads and traces and the Power Ground all follow each other very closely until they reach the power pins. Again this is ignored in almost all posted layouts.
Check the bandwidth of your two input filters. Are the F-3dB and F-1dB frequencies sensible to the audio signal you want to pass?
Add in all the "optional" components listed by National
Ensure the HF decoupling has a very short route. The two ceramic caps should be soldered to each other. The other ends get soldered to the power pins.
The MF decoupling can be a little further away. BUT and this is very important: The link between the two MF decoupling caps MUST be very close to the junction between the two HF decoupler caps. Connect these two points together to create your Power Ground.
Connect the output Zobel to this Power Ground.
This minimises the route that the HF spikes must follow around the output stage. This really is very important and 99% of the PCB layouts (discrete and chipamp) that get posted on this Forum completely ignore this HF route requirement.
Ensure the power input leads and traces and the Power Ground all follow each other very closely until they reach the power pins. Again this is ignored in almost all posted layouts.
Check the bandwidth of your two input filters. Are the F-3dB and F-1dB frequencies sensible to the audio signal you want to pass?
@djoffe - yes, I removed the Ci completely after a lot of reading here on the forum. That's actually an option with a Chipamp, you just need to resolder the R3 to another hole on PCB. I wanted minimalism, especially after reading about Peter Daniel's 3875 based amp.
Here is his scheme: http://db.tt/dXPEJbkB
@AndrewT - by now I know you are very sensitive to "improperly implemented" stuff, but would you please comment on PD's minimalistic approach then? I hear only praises about it. I think you are right, and I will try to properly implement the chipamp one of these days, but you are aware that using coupling caps and Ci means I have to buy Mundorfs or such - any cheap cap (even Panasonic FC here) sounds bad, and I'm much happier with a "no cap" right now.
Will post a photo of my build soon, it will be easier to comment then...
Here is his scheme: http://db.tt/dXPEJbkB
@AndrewT - by now I know you are very sensitive to "improperly implemented" stuff, but would you please comment on PD's minimalistic approach then? I hear only praises about it. I think you are right, and I will try to properly implement the chipamp one of these days, but you are aware that using coupling caps and Ci means I have to buy Mundorfs or such - any cheap cap (even Panasonic FC here) sounds bad, and I'm much happier with a "no cap" right now.
Will post a photo of my build soon, it will be easier to comment then...
I have been castigated for interfering (commenting) in Peter's/Audiosector's Threads.
I pointed out that some of Peter's philosophy should never be followed by beginners.
I pointed out that DC coupled increases the risk of damaged speakers.
I pointed out that Peter's implementation was developed in conjunction with very efficient speakers and thus could not be transferred directly across to normal efficiency, nor low efficiency speakers.
I made various other comments (too long ago to remember) that tried to forewarn Members of the dangers in Peter's advice.
If you really are interested you could search out my posts.
I do not believe that a properly sized NFB capacitor that imposes no AC signal on the signal current can add to the distortion of an amplifier.
That alone allows one to use a "normal" commercial quality capacitor and NOT be able to hear added distortion compared to an expensive capacitor.
If the expensive capacitor does "sound" different then it is faulty, or it is the wrong size.
If you want a DC coupled amplifier then my advice is learn how to protect your speakers. And learn why DC detection and DC servo and speaker isolation and maybe input muting, should all be implemented. That is a lot of learning and a lot of board space and a lot of cost to avoid AC coupling that can give very good sound.
I pointed out that some of Peter's philosophy should never be followed by beginners.
I pointed out that DC coupled increases the risk of damaged speakers.
I pointed out that Peter's implementation was developed in conjunction with very efficient speakers and thus could not be transferred directly across to normal efficiency, nor low efficiency speakers.
I made various other comments (too long ago to remember) that tried to forewarn Members of the dangers in Peter's advice.
If you really are interested you could search out my posts.
I do not believe that a properly sized NFB capacitor that imposes no AC signal on the signal current can add to the distortion of an amplifier.
That alone allows one to use a "normal" commercial quality capacitor and NOT be able to hear added distortion compared to an expensive capacitor.
If the expensive capacitor does "sound" different then it is faulty, or it is the wrong size.
If you want a DC coupled amplifier then my advice is learn how to protect your speakers. And learn why DC detection and DC servo and speaker isolation and maybe input muting, should all be implemented. That is a lot of learning and a lot of board space and a lot of cost to avoid AC coupling that can give very good sound.
Last edited:
Ok, a few questions:
What happens if you hook up an amp like PD's to a normal or low efficiency speakers? How does that sound or what problems come up?
And to return to the topic - could that be the problem with my amp? My speakers are 89dB (Tannoy Eyris 3).
About speaker protection - I have a pair of muting relays connected to a switch on the front panel, so every time I turn the amp on, it powers up muted, so then I unmute it. And before powering down, I mute it first. Absolutely no pops or clicks, I know I could do it automatically but manually is fine with me. Of course, I mute it first before powering the cd-player or PC on or off, so I think I won't get the DC surge (is that the word?) amplified through to my speakers.
PS could you post a schematic and/or a photo of all the things you mentioned in post #5? I can't figure out what you mean by soldering the ceramic caps together first, because they are each on a different amp board.
And also how do I check the bandwidth of my input filters? Are you refering to the coupling caps or Ci + R3?
What happens if you hook up an amp like PD's to a normal or low efficiency speakers? How does that sound or what problems come up?
And to return to the topic - could that be the problem with my amp? My speakers are 89dB (Tannoy Eyris 3).
About speaker protection - I have a pair of muting relays connected to a switch on the front panel, so every time I turn the amp on, it powers up muted, so then I unmute it. And before powering down, I mute it first. Absolutely no pops or clicks, I know I could do it automatically but manually is fine with me. Of course, I mute it first before powering the cd-player or PC on or off, so I think I won't get the DC surge (is that the word?) amplified through to my speakers.
PS could you post a schematic and/or a photo of all the things you mentioned in post #5? I can't figure out what you mean by soldering the ceramic caps together first, because they are each on a different amp board.
And also how do I check the bandwidth of my input filters? Are you refering to the coupling caps or Ci + R3?
So many questions I do not have time to detail all you ask for.
A pair of radial 0.1" xr7 100nF 100V caps make good HF decoupling.
Cut a lead on each very short. Solder these short ends together.
Cut the other two leads to about half length. This soldered junction is a Power Ground
Solder one half length lead to +ve pin.
Solder the other half length lead to the -ve pin.
You now have the power pins connected to an HF power ground.
Keep going, build an MF Power Ground using electrolytics.
A pair of radial 0.1" xr7 100nF 100V caps make good HF decoupling.
Cut a lead on each very short. Solder these short ends together.
Cut the other two leads to about half length. This soldered junction is a Power Ground
Solder one half length lead to +ve pin.
Solder the other half length lead to the -ve pin.
You now have the power pins connected to an HF power ground.
Keep going, build an MF Power Ground using electrolytics.
Thanks, I will try this in a week or so, I'm moving so everything is a mess...
I will first put back all the components I took out or changed, and remove the regs so I can properly test the amp with and without them, and post the conclusions.
In the meantime, anyone who has the "bloated" sound problem or similar, or has implemented the decoupling AndrewT mentions, please post here.
I will first put back all the components I took out or changed, and remove the regs so I can properly test the amp with and without them, and post the conclusions.
In the meantime, anyone who has the "bloated" sound problem or similar, or has implemented the decoupling AndrewT mentions, please post here.
I see a few errors and omissions.
A pair of radial 0.1" xr7 100nF 100V caps make good HF decoupling.
Cut a lead on each very short. Solder these short ends together. This soldered junction is a Power Ground, the HF PG.
Cut the other two leads to about half length.
Solder one half length lead to +ve pin.
Solder the other half length lead to the -ve pin.
You now have the power pins connected to an HF power ground.
Keep going, build an MF Power Ground using electrolytics.
Connect the HFPG to the MF PG with a short wire or a short trace.
A short trace that is wide has much less impedance than an equal length wire, no matter how thick you make the wire.
At HF impedance rules. Everything that can be done to reduce inductance is important.
The electros are far too big to fit in close to allow their MF PG to be attached to the HF PG.
You need to assemble the electros in a location where they can "fit". Then connect the MF PG to the HF PG.
That connection should be of low impedance to get great performance from your decoupling.
The copper foil from a coax screen may well be a good trace substitute. Nice idea. Shame most of mine are aluminium metalised onto plastic film.
That foil strip could also be used to solder/bolt an IEC filter to a chassis. Wide thin strip connection is far better than a wire or a bolt connection. I have been reading Ott, does it show?
You need to assemble the electros in a location where they can "fit". Then connect the MF PG to the HF PG.
That connection should be of low impedance to get great performance from your decoupling.
The copper foil from a coax screen may well be a good trace substitute. Nice idea. Shame most of mine are aluminium metalised onto plastic film.
That foil strip could also be used to solder/bolt an IEC filter to a chassis. Wide thin strip connection is far better than a wire or a bolt connection. I have been reading Ott, does it show?
Last edited:
Ok, settled to the new appartment now... 
A few questions, AndrewT, if you're still following this:
1) what would be the best value for the MF decoupling caps?
2) would 100-330pf ceramic caps be a good choice for the HF decoupling?
I'm trying to do this with stuff I have at home, so it's either that or some really small xfk caps, which are multifoil if I'm not mistaken.
3) shouldn't I connect this HF + MF + Zobel ground to the star ground? You didn't mention it, so I have to ask.
Unfortunately I still have to find that copper shielded satellite cable again, so no copper trace for now. But a visit to a specialized satellite shop should solve that...
PS someone suggested that the 47uF of Ci should be increased as high as 330uF because of the -3dB point being too high. I calculated it to be 4.98Hz for 47uF (with 680R), 2.34Hz for 100uF, and 0.7Hz for 330uF. I doubt I would hear any difference, but what would this change achieve?
A few questions, AndrewT, if you're still following this:
1) what would be the best value for the MF decoupling caps?
2) would 100-330pf ceramic caps be a good choice for the HF decoupling?
I'm trying to do this with stuff I have at home, so it's either that or some really small xfk caps, which are multifoil if I'm not mistaken.
3) shouldn't I connect this HF + MF + Zobel ground to the star ground? You didn't mention it, so I have to ask.
Unfortunately I still have to find that copper shielded satellite cable again, so no copper trace for now. But a visit to a specialized satellite shop should solve that...
PS someone suggested that the 47uF of Ci should be increased as high as 330uF because of the -3dB point being too high. I calculated it to be 4.98Hz for 47uF (with 680R), 2.34Hz for 100uF, and 0.7Hz for 330uF. I doubt I would hear any difference, but what would this change achieve?
MF decoupling is usually from 47uF to 1500uF. A big spread to play with.
HF decoupling is usually 100nF. I see very little variation around this value.
I standardised on 100nF 50V or 100V x7r 0.2"pin pitch.
I have recently bought 100nF smd for lower inductance where a circuit layout suits.
The Zobel Ground + both decoupling grounds are the amplifier's Power Ground. Yes, this goes to the Main Audio Ground whether it is on board, or near the board.
I did quite a few experiments with that input cap RC and found that very approximately when near 90ms I could not hear any further changes.
This was for a couple of amplifiers that had a very wide pass band. Soon after I did this, I discovered on THIS FORUM that the amplifier pass band should ALWAYS be wider than the signal being applied. The ratio suggested was at least sqrt(2).
Combining the 90ms input filter and the sqrt(2) factor gave the NFB RC >=130ms.
I tried this and sure enough I could not hear any difference compared to the wide pass band.
But, I then tried the narrow passband that is very often shown by builders and found I could hear the bass change again. I moved it up another octave above what I thought I could now hear and the bass became unnatural. That then confirmed my conclusion that the suggestion was right. I have stuck with this ever since.
It gives deep pleasant bass.
HF decoupling is usually 100nF. I see very little variation around this value.
I standardised on 100nF 50V or 100V x7r 0.2"pin pitch.
I have recently bought 100nF smd for lower inductance where a circuit layout suits.
The Zobel Ground + both decoupling grounds are the amplifier's Power Ground. Yes, this goes to the Main Audio Ground whether it is on board, or near the board.
I did quite a few experiments with that input cap RC and found that very approximately when near 90ms I could not hear any further changes.
This was for a couple of amplifiers that had a very wide pass band. Soon after I did this, I discovered on THIS FORUM that the amplifier pass band should ALWAYS be wider than the signal being applied. The ratio suggested was at least sqrt(2).
Combining the 90ms input filter and the sqrt(2) factor gave the NFB RC >=130ms.
I tried this and sure enough I could not hear any difference compared to the wide pass band.
But, I then tried the narrow passband that is very often shown by builders and found I could hear the bass change again. I moved it up another octave above what I thought I could now hear and the bass became unnatural. That then confirmed my conclusion that the suggestion was right. I have stuck with this ever since.
It gives deep pleasant bass.
AndrewT;3473684...If you want a DC coupled amplifier then my advice is learn how to protect your speakers. And learn why DC detection and DC servo and speaker isolation and maybe input muting said:I too blew up a pair of speakers testing my stripped down AudioSector LM3875 amp in the first few hours of its life. I assume I had DC at the input from my Microsoft Zune MP3 player that was amplified and sent to the speakers, which blew them up. This happened a second time to another pair of speakers, but they were more resistant to blowing up and allowed me to study the problem for a few seconds before I shut the amp off. So far my NAD 1600 Tuner-Preamp has been a good match with this amp.
I don't plan to risk my good speakers on this amp though until I replace R1 (shown on the amp board here http://www.audiosector.com/images/lm3875_se_pcb.gif) with a capacitor and add that Zobel network, Rz and Cz.
Iggy111, with all of that said, my amp does sound fantastic - no "bloat" in the audio at all. I have run it through 4ohm and 8ohm speakers and aside from being a little hot with the 4ohm speakers, it sounds great, when it is not amplifying DC that is
-Glen
Last edited:
Before Andrew freaks out:
http://www.diyaudio.com/forums/power-supplies/167579-light-bulb-tester.html
I think there's a more thorough description, technical explanation and schematic of the Light Bulb Tester ("light bulb limiter")in one of the recent threads if someone with more google-fu can find it.
Last edited:
Before Andrew freaks out:
http://www.diyaudio.com/forums/power-supplies/167579-light-bulb-tester.html
I think there's a more thorough description, technical explanation and schematic of the Light Bulb Tester ("light bulb limiter")in one of the recent threads if someone with more google-fu can find it.
When properly used, it would have saved your speakers.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.