ajst2duk:
Two points to make.
Firstly, if the DC offset was initially set at proper operating temperature the change you have mentioned seems unusually high for this particular amplifier. My amplifier has not been adjusted since I set it up months ago and the offset is still within 2mV of its initial set value. This is despite the fact that we are in a different season.
That said, the change you mentioned is totally acceptable and not unheard of. It is still very good for a non-servoed singleton input amplifier. BUT...it seems strange that this should only have happened to one channel. SO...
It might just be worth making sure that everything is running OK. In particular, make sure your output devices are still bolted down tightly, they may have loosened up due to the thermal cycling.
Finally, and point two. If all is running OK, 30mV is no problem at all!
Two points to make.
Firstly, if the DC offset was initially set at proper operating temperature the change you have mentioned seems unusually high for this particular amplifier. My amplifier has not been adjusted since I set it up months ago and the offset is still within 2mV of its initial set value. This is despite the fact that we are in a different season.
That said, the change you mentioned is totally acceptable and not unheard of. It is still very good for a non-servoed singleton input amplifier. BUT...it seems strange that this should only have happened to one channel. SO...
It might just be worth making sure that everything is running OK. In particular, make sure your output devices are still bolted down tightly, they may have loosened up due to the thermal cycling.
Finally, and point two. If all is running OK, 30mV is no problem at all!
Thanks all - I may be guilty of being a bit hasty with the setup and bias procedures as I was so keen to hear it running. I think it is nothing more sinister than that. As mentioned before - it has been running solid for around three weeks - sounds very impressive and has more punch than its nominal output rating suggests. I haven't yet had it clipping - not that I'm aware of anyway - even with movie soundtracks. Thanks - AJ
Andrew,
Greg and Andrew are right, this is not a critical adjustment however and since birth the amp has clearly settled down. Make sure the outputs are tight - case temperature will slightly affect offset - but much of the offset control comes back to the VAS mosfet which of course carries limited but entirely adequate heatsinking on the board. If this device is cooled with a freeze spray you will note the offset change abruptly. Correspondingly, until the VAS reaches equilibrium after perhaps half an hour, the offset will drift, so it might be a good idea to reset the offset after it's been running for a while, then keep an eye on it.
Oddly, offset won't hurt the speaker, up to about 500mV in fact. Offset will generate a standing current in the voice coil which heats it up, but the heating effect is trivial compared to the high rating of most voice coils - 30W continuous at least. For example, 500mV on a 6.8R voice coil will generate 37mW of heat - just enough to keep a few dust mites warm in winter! Of far more significance is the force this voltage creates on the voice coil, moving the entire cone off centre and very slightly increasing second and fourth harmonic - no bad thing.
Cheers,
Hugh
Greg and Andrew are right, this is not a critical adjustment however and since birth the amp has clearly settled down. Make sure the outputs are tight - case temperature will slightly affect offset - but much of the offset control comes back to the VAS mosfet which of course carries limited but entirely adequate heatsinking on the board. If this device is cooled with a freeze spray you will note the offset change abruptly. Correspondingly, until the VAS reaches equilibrium after perhaps half an hour, the offset will drift, so it might be a good idea to reset the offset after it's been running for a while, then keep an eye on it.
Oddly, offset won't hurt the speaker, up to about 500mV in fact. Offset will generate a standing current in the voice coil which heats it up, but the heating effect is trivial compared to the high rating of most voice coils - 30W continuous at least. For example, 500mV on a 6.8R voice coil will generate 37mW of heat - just enough to keep a few dust mites warm in winter! Of far more significance is the force this voltage creates on the voice coil, moving the entire cone off centre and very slightly increasing second and fourth harmonic - no bad thing.
Cheers,
Hugh
Andrew,
The Dartzeel is an impressive amplifier and I have chatted with Herve Delacraz by email about the design. It is highly regarded sonically, but I have heard inclined to poor quiescent control in the output stage.
In fact, one of my designs, the top drawer Maya, uses quite similar techniques to deliver wonderful sound, but there are very few of them out there.
Feel like offering a business investment?
Cheers,
Hugh
The Dartzeel is an impressive amplifier and I have chatted with Herve Delacraz by email about the design. It is highly regarded sonically, but I have heard inclined to poor quiescent control in the output stage.
In fact, one of my designs, the top drawer Maya, uses quite similar techniques to deliver wonderful sound, but there are very few of them out there.
Feel like offering a business investment?
Cheers,
Hugh
Fetzilla Power Supply
Hi Hugh and all,
I've been tearing out what's left of my hair over the last few weeks trying to get to the bottom of this:
1. Built up 4 Fetzilla boards with recommended 4700uF PSU Caps and 20A Shottky diodes. Both boards tested fine according to the instructions and extra measurements on this thread, and were set for 380mA quiescent current.
2. Tested boards individually, they were dead silent, and played music when given a signal.
3. Built up a stereo Fetzilla amp using a single 300VA transformer and 2 boards, using the same PS arrangement as used in my AKSA amp.
4. On testing, I got something that sounds like mains hum, but ONLY when both inputs are connected. It's dead silent otherwise. Tried every PS and earthing arrangement I could think of, with absolutely no difference to the hum. Tried using the other Fetzilla boards, with same results.
5. Noticed that there was 5mA ripple at the +ve and -ve fuses at idle, and the hum appears if I connect the 2 +ve inputs together without touching the earths - it seems that with both inputs connected, the PSRR of the amps disappears.
6. Tried the sledgehammer/nut approach - Added 15,000uF to +ve and -ve rails, the ripple and hum is virtually gone, but still not as quiet as in single channel mode.
There seems to be some interaction across the secondary windings or 0V connections between the 2 channels which results in the hum.
I haven't tried it, but I'm virtually certain that if there were 2 transformers (or 4 windings) the problem would go away.
Before I go and order some big-*** caps or more transformers and try to shoe-horn them into the case, does anyone have a simple solution to this "issue"?
Alternatively, does anyone have a functioning stereo Fetzilla with one tranny that I can take a look at?
Cheers,
Ron.
Hi Hugh and all,
I've been tearing out what's left of my hair over the last few weeks trying to get to the bottom of this:
1. Built up 4 Fetzilla boards with recommended 4700uF PSU Caps and 20A Shottky diodes. Both boards tested fine according to the instructions and extra measurements on this thread, and were set for 380mA quiescent current.
2. Tested boards individually, they were dead silent, and played music when given a signal.
3. Built up a stereo Fetzilla amp using a single 300VA transformer and 2 boards, using the same PS arrangement as used in my AKSA amp.
4. On testing, I got something that sounds like mains hum, but ONLY when both inputs are connected. It's dead silent otherwise. Tried every PS and earthing arrangement I could think of, with absolutely no difference to the hum. Tried using the other Fetzilla boards, with same results.
5. Noticed that there was 5mA ripple at the +ve and -ve fuses at idle, and the hum appears if I connect the 2 +ve inputs together without touching the earths - it seems that with both inputs connected, the PSRR of the amps disappears.
6. Tried the sledgehammer/nut approach - Added 15,000uF to +ve and -ve rails, the ripple and hum is virtually gone, but still not as quiet as in single channel mode.
There seems to be some interaction across the secondary windings or 0V connections between the 2 channels which results in the hum.
I haven't tried it, but I'm virtually certain that if there were 2 transformers (or 4 windings) the problem would go away.
Before I go and order some big-*** caps or more transformers and try to shoe-horn them into the case, does anyone have a simple solution to this "issue"?
Alternatively, does anyone have a functioning stereo Fetzilla with one tranny that I can take a look at?
Cheers,
Ron.
Ron,
I've been following this thread in the background... I haven't built the amps though.
Hearing you say that is something that reminded of something I struggled with, and literally for years until I sat down and thought it through. You would have to read the whole thread to get the gist of it but this dives in at the most relevant point.
http://www.diyaudio.com/forums/soli...-lin-topology-nfb-tappings-2.html#post1624677
Now I'm certainly not saying that the same issue is at work here but it's something to look at.
Also if your hum is a just that, a pure deep 50/60hz hum it could be induced currents from the stray field of the mains trx acting on a "loop" caused when the ground are connected together. In other words just a physical issue. Sometimes putting the toroid in the centre of an amp as many do isn't always the best thing apart from visual symmetry.
I've been following this thread in the background... I haven't built the amps though.
Hearing you say that is something that reminded of something I struggled with, and literally for years until I sat down and thought it through. You would have to read the whole thread to get the gist of it but this dives in at the most relevant point.
http://www.diyaudio.com/forums/soli...-lin-topology-nfb-tappings-2.html#post1624677
Now I'm certainly not saying that the same issue is at work here but it's something to look at.
Also if your hum is a just that, a pure deep 50/60hz hum it could be induced currents from the stray field of the mains trx acting on a "loop" caused when the ground are connected together. In other words just a physical issue. Sometimes putting the toroid in the centre of an amp as many do isn't always the best thing apart from visual symmetry.
Hi Ron,
First up, this is solveable!
Second, thank you for building the FetZillas, it was a leap of faith, and we will fix it.
Now, like the NAKSA, these amps are designed with fully independent rail supplies and require two separate secondaries with no centre tap or any other galvanic connection to a tap on the transformer. In fact, the grounds on the modules are fully independent as well.
Looking at the positive rail on one channel (the negative is the same) we have a full wave bridge rectifier feeding a filter cap - 4,700uF in most cases. Then there is a second filter cap next to it, with two 0.15R resistors connecting the positive electrodes to each other, and the negative electrodes to each other. The negative electrode of the second cap is connected to the opposite polarity of its counterpart on the negative rail, and then to earth. The positive electrode of the second filter cap then becomes the pos rail, and negative electrode of the counterpart cap on the other rail becomes the negative rail.
This all means that the AKSA power supply WILL NOT WORK SATISFACTORILY because there is galvanic connection between centre tap/star earth and the ground of the power amp, removing the influence of the interconnecting resistors. You know what this means...
You can use a transformer with just two secondaries. One secondary, with joined together leads, four instead of two, now powers the positive supply of both amps, while the other secondary powers the negative supply of both amps. Note that in this case too there is no centre tap. I wanted to get away from that approach because it never made sense to me to connect a quiet DC earth on an amp to an AC transformer.......
So, Ron, you can still use one transformer, but you must unfortunately ditch the AKSA power supply which is now from a bygone Aspen era, no longer applicable here. I should add that the NAKSA supply is the same topology as the FetZilla, and that they do work much better sonically than the AKSA supply - almost double component count, but far better performance.
Hope this helps, and thanks to Karl for a good suggestion,
Hugh
First up, this is solveable!
Second, thank you for building the FetZillas, it was a leap of faith, and we will fix it.
Now, like the NAKSA, these amps are designed with fully independent rail supplies and require two separate secondaries with no centre tap or any other galvanic connection to a tap on the transformer. In fact, the grounds on the modules are fully independent as well.
Looking at the positive rail on one channel (the negative is the same) we have a full wave bridge rectifier feeding a filter cap - 4,700uF in most cases. Then there is a second filter cap next to it, with two 0.15R resistors connecting the positive electrodes to each other, and the negative electrodes to each other. The negative electrode of the second cap is connected to the opposite polarity of its counterpart on the negative rail, and then to earth. The positive electrode of the second filter cap then becomes the pos rail, and negative electrode of the counterpart cap on the other rail becomes the negative rail.
This all means that the AKSA power supply WILL NOT WORK SATISFACTORILY because there is galvanic connection between centre tap/star earth and the ground of the power amp, removing the influence of the interconnecting resistors. You know what this means...
You can use a transformer with just two secondaries. One secondary, with joined together leads, four instead of two, now powers the positive supply of both amps, while the other secondary powers the negative supply of both amps. Note that in this case too there is no centre tap. I wanted to get away from that approach because it never made sense to me to connect a quiet DC earth on an amp to an AC transformer.......
So, Ron, you can still use one transformer, but you must unfortunately ditch the AKSA power supply which is now from a bygone Aspen era, no longer applicable here. I should add that the NAKSA supply is the same topology as the FetZilla, and that they do work much better sonically than the AKSA supply - almost double component count, but far better performance.
Hope this helps, and thanks to Karl for a good suggestion,
Hugh
Gaetan,
that would likely work well, however, the FetZilla supply has two full wave bridge rectifiers to each module, one for positive rail supplied from one secondary, and the other for the negative rail supplied from the other secondary. The two secondaries are independent, and can be from one or a separate transformer.
Aside from that, your picture is spot on,
Cheers,
Hugh
that would likely work well, however, the FetZilla supply has two full wave bridge rectifiers to each module, one for positive rail supplied from one secondary, and the other for the negative rail supplied from the other secondary. The two secondaries are independent, and can be from one or a separate transformer.
Aside from that, your picture is spot on,
Cheers,
Hugh
I was enjoying re-reading this and remembering many things I had already forgotten about.
That is, until Dx piped up in post50 with his usual physcobable.
How can reducing buzz and hum not improve the audio quality coming from the speakers?
I am serious read my question again if you need to.
I am glad he has almost bowed out of interfering in this Forum.
Thanks for the responses and ideas guys, I'll be following up on them as time allows.
I forgot to say in my previous post that my AKSA amp has a PSU very similar to the latest Naksas, using CRC and one bridge per rail. Sorry for the confusion.
The recommended PSU from the documentation thread was the first thing I tried, the only difference being a single transformer, with both +ve bridges fed from 1 winding, and both -ve bridges fed from the other (exactly as Hugh describes above). It hummed with both inputs connected
With the second set of boards I left off the bridges and smoothing caps and tried 2 "HackerCap" boards, using 3 x 6800uF per rail per board in CLCLC configuration, but still fed from one tranny. Slightly less hum but still
Finally I added 15000uF per rail between the bridges and HackerCap boards, this resulted in the least hum. I can only hear it from <1M from the speakers, but it's still there.
So it looks like the hum is proportional to the amount of ripple on the rails.
This weekend I think I'll go back to the recommended PSU arrangement and try it with 2 transformers, to prove that I've built the Fetzillas correctly (or not - we'll see!).
Cheers,
Ron
I forgot to say in my previous post that my AKSA amp has a PSU very similar to the latest Naksas, using CRC and one bridge per rail. Sorry for the confusion.
The recommended PSU from the documentation thread was the first thing I tried, the only difference being a single transformer, with both +ve bridges fed from 1 winding, and both -ve bridges fed from the other (exactly as Hugh describes above). It hummed with both inputs connected
With the second set of boards I left off the bridges and smoothing caps and tried 2 "HackerCap" boards, using 3 x 6800uF per rail per board in CLCLC configuration, but still fed from one tranny. Slightly less hum but still
Finally I added 15000uF per rail between the bridges and HackerCap boards, this resulted in the least hum. I can only hear it from <1M from the speakers, but it's still there.
So it looks like the hum is proportional to the amount of ripple on the rails.
This weekend I think I'll go back to the recommended PSU arrangement and try it with 2 transformers, to prove that I've built the Fetzillas correctly (or not - we'll see!).
Cheers,
Ron
My FetZillas are noisy with the inputs shorted. Probably noisier than any amplifier I have ever assembled.
I have tried a stereo pair and ran into hum problems as well.
They sound OK, but I don't describe them as great.
I will have to re-read post888 in case there is something I have got wrong or missed or not implemented.
I have tried a stereo pair and ran into hum problems as well.
They sound OK, but I don't describe them as great.
I will have to re-read post888 in case there is something I have got wrong or missed or not implemented.
Ron.
re post #886. My FETzilla amps where initially noisy when connected to my CD player (an Audiolab 8200cdq). The solution was to remove the earth lift resistor, ie the 10 ohm R26 in the attached drawing. I suspect my issue was that the Audiolab 8200cdq is earthed to the mains, so I had in effect an earth loop despite the 10 ohms. Your FETzilla may have the same problem. Just cut R26 off the pcb, the back to back diodes are sufficient. With R26 removed the are very quiet.
My FETzilla amps are build as two separate monobloc type amps (in NAKSA-70 chassis).
I hope this helps,
Paul Bysouth
re post #886. My FETzilla amps where initially noisy when connected to my CD player (an Audiolab 8200cdq). The solution was to remove the earth lift resistor, ie the 10 ohm R26 in the attached drawing. I suspect my issue was that the Audiolab 8200cdq is earthed to the mains, so I had in effect an earth loop despite the 10 ohms. Your FETzilla may have the same problem. Just cut R26 off the pcb, the back to back diodes are sufficient. With R26 removed the are very quiet.
My FETzilla amps are build as two separate monobloc type amps (in NAKSA-70 chassis).
I hope this helps,
Paul Bysouth
Spent some time last weekend putting together some "Modules" and cables, so I could test various configurations of PS.
1st experiment - 2 transformers, 4 bridges, 4 CLCLC cap banks using 6800uF / 10uH. Result - silence. No hum at all even with my ear right next to the speaker cone, so it looks like the boards are correctly assembled.
2nd experiment - 1 transformer, 4 bridges, 4 CLCLC cap banks. Result - silence with one input connected, and hum when both inputs are connected. The only change from above was to remove the 2nd transformer and connect the 2 windings to 2 bridges each.
3rd experiment - 1 transformer, 2 bridges, 4 CLCLC cap banks. Result - silence with one input connected, and hum when both inputs are connected.
4th experiment - Removed R26 on both boards. Result - no change to the hum at all.
If I connect one input and just the outer shield of the other input, it's silent. I don't think that earthing's the problem here.
In fact, the more I listen to the Hum, the more it sounds like a "Buzz".
1st experiment - 2 transformers, 4 bridges, 4 CLCLC cap banks using 6800uF / 10uH. Result - silence. No hum at all even with my ear right next to the speaker cone, so it looks like the boards are correctly assembled.
2nd experiment - 1 transformer, 4 bridges, 4 CLCLC cap banks. Result - silence with one input connected, and hum when both inputs are connected. The only change from above was to remove the 2nd transformer and connect the 2 windings to 2 bridges each.
3rd experiment - 1 transformer, 2 bridges, 4 CLCLC cap banks. Result - silence with one input connected, and hum when both inputs are connected.
4th experiment - Removed R26 on both boards. Result - no change to the hum at all.
If I connect one input and just the outer shield of the other input, it's silent. I don't think that earthing's the problem here.
In fact, the more I listen to the Hum, the more it sounds like a "Buzz".
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.