XA 100
GL-I think you are on to something related to the dc offset vs AC voltage variation in the house mains. My house mains during the winter are around 122-125 volts with a 60hz +/- 3hz frequency.
During the summer this can drop to 115 volts easily and I am not sure what the frequency does. I have a monster ac power supply filter which measures these ac variations. Now I am wondering if I should build a humongus regulator for the power supply similar to what NP recommends. I can't think of anyother way to cure this problem with DC offset (if it indeed contributes).
I still will carefully match all the parts. Thanks for keeping these ideas flowing. dave
GL-I think you are on to something related to the dc offset vs AC voltage variation in the house mains. My house mains during the winter are around 122-125 volts with a 60hz +/- 3hz frequency.
During the summer this can drop to 115 volts easily and I am not sure what the frequency does. I have a monster ac power supply filter which measures these ac variations. Now I am wondering if I should build a humongus regulator for the power supply similar to what NP recommends. I can't think of anyother way to cure this problem with DC offset (if it indeed contributes).
I still will carefully match all the parts. Thanks for keeping these ideas flowing. dave
Hi Dave,
Thank you for the comments. This started out as a "hey look what I built" type of thread and has evolved into a pretty constructive sharing of experiments and ideas.
Hi William,
I redid the AC gain measurements and found that increasing the bias from 500ma to 540ma did not change the AC gain of the Aleph CCS at all. I did the test at 100Hz and 75Hz.
I am next going to use my variac to "adjust" the AC mains voltage to the amp and see what it does to the abs. DC offset. I am not sure I completely understand why the two are related. The voltage across the 392 ohm resistor is set by the 10ma from the CCS. The 10ma from the CCS is essentially set by the zener. Adjusting the mains voltage slowly should not affect these values. However, it will change the voltage across the diff pair.
In the Salt Lake Shootout thread I noticed that another member tried ZVP3310 parts as the input to an Aleph with very favourable sonic results. I recall that Cheff also had this idea. I am now thinking about building new driver boards with ZVP3310's cascoded with ZTX550's as Cheff suggested. I have all the parts. I am going to buil new boards because I am using the amps regularly and I don't want to perform major surgery on the existing boards.
Enough has changed in the "as built" AX100 and enough interest has been shown by other members that I have decided to post an updated schematic and circuit description.
Cheers,
Graeme
Thank you for the comments. This started out as a "hey look what I built" type of thread and has evolved into a pretty constructive sharing of experiments and ideas.
Hi William,
I redid the AC gain measurements and found that increasing the bias from 500ma to 540ma did not change the AC gain of the Aleph CCS at all. I did the test at 100Hz and 75Hz.
I am next going to use my variac to "adjust" the AC mains voltage to the amp and see what it does to the abs. DC offset. I am not sure I completely understand why the two are related. The voltage across the 392 ohm resistor is set by the 10ma from the CCS. The 10ma from the CCS is essentially set by the zener. Adjusting the mains voltage slowly should not affect these values. However, it will change the voltage across the diff pair.
In the Salt Lake Shootout thread I noticed that another member tried ZVP3310 parts as the input to an Aleph with very favourable sonic results. I recall that Cheff also had this idea. I am now thinking about building new driver boards with ZVP3310's cascoded with ZTX550's as Cheff suggested. I have all the parts. I am going to buil new boards because I am using the amps regularly and I don't want to perform major surgery on the existing boards.
Enough has changed in the "as built" AX100 and enough interest has been shown by other members that I have decided to post an updated schematic and circuit description.
Cheers,
Graeme
Hi Jacco,
The purpose in creating the AX100 was to take Grey Rollins original circuit and 1) scale it up to 100W, 2) move the design closer to the factory XA amps by incorporating all the crumbs dropped by NP over the past 4 years.
Ideas like cascodes, new CCS designs, servos and JFETs create new designs that I think should be subject of new threads. That is why I used the term "as built".
Cheers,
Graeme
The purpose in creating the AX100 was to take Grey Rollins original circuit and 1) scale it up to 100W, 2) move the design closer to the factory XA amps by incorporating all the crumbs dropped by NP over the past 4 years.
Ideas like cascodes, new CCS designs, servos and JFETs create new designs that I think should be subject of new threads. That is why I used the term "as built".
Cheers,
Graeme
Graeme,
as far as I understand the current through the diff pair is not affected by the power supply voltage / mains voltage.
Also the voltage between the gate of the output transistors and the negative rail stays the same as the current through the drain resistors (392R) stays the same.
What is affected is the voltage between the gate of the output transistors and ground. This changes the same amount as the neg rail voltage.
If Vdg stays the same (wich it probably will not) the absolute dc will change the same amount as the negative rail.
mmmm, maybe said another way:
you set the absolute offset by setting Vgs on the output fets and thus changing the drain-gate voltage. If this reaches rail voltage minus Vgs then your absolute offset is zero.
Jacco,
it is a standard servo with an OP77GP, 2x 1uF, 2x 330k and an output resistor. It is connected to ground and to the gate of the current source fet. I would feed it with +18V and -6V with a few zeners connected to the supply rails. The output resistor must be determined by trying but could be between 2 and 4k.
William
as far as I understand the current through the diff pair is not affected by the power supply voltage / mains voltage.
Also the voltage between the gate of the output transistors and the negative rail stays the same as the current through the drain resistors (392R) stays the same.
What is affected is the voltage between the gate of the output transistors and ground. This changes the same amount as the neg rail voltage.
If Vdg stays the same (wich it probably will not) the absolute dc will change the same amount as the negative rail.
mmmm, maybe said another way:
you set the absolute offset by setting Vgs on the output fets and thus changing the drain-gate voltage. If this reaches rail voltage minus Vgs then your absolute offset is zero.
Jacco,
it is a standard servo with an OP77GP, 2x 1uF, 2x 330k and an output resistor. It is connected to ground and to the gate of the current source fet. I would feed it with +18V and -6V with a few zeners connected to the supply rails. The output resistor must be determined by trying but could be between 2 and 4k.
William
Well,
the question is more like:
will putting in a servo in this way hurt the sound?
-it is not really in the signal path as it feeds the current source
- caps and reistores can be set for a very slow following of the output dc voltage
-you can probably leave out the McMillan resistors
-you can raise the value of the output resistors to ground
I think it should be tried before saying it´ll hurt the sound.
William
the question is more like:
will putting in a servo in this way hurt the sound?
-it is not really in the signal path as it feeds the current source
- caps and reistores can be set for a very slow following of the output dc voltage
-you can probably leave out the McMillan resistors
-you can raise the value of the output resistors to ground
I think it should be tried before saying it´ll hurt the sound.
William
Here is an updated schematic for the AX100.
Although the schematic shown on page 1 of this thread is still valid this updated version accurately reflects the current "as-built" amps. The changes shown include tweeks done over the past few weeks some of which are discussed in this thread.
These are:
1) The value for the MacMillan resistors R46 and R47 are changed from 4.75K to 10K. I may try 22K again but not for while.
2) The protection diode scheme for the inputs has changed. I did this after examining the close-up photos of the XA200 PCB that NP published on pages 58 and 59 of the big Aleph-X thread. This scheme was previously used by NP on the Aleph 0. I don't believe it affects the sound but it does delete two parts and two ground connections.
3) R11 and R33 are now shown as 100K which gives 540ma of bias per output transistor on the Aleph CCS. I am going to stick with this value for now. I have not tried 600ma or higher yet. The AC current gain of the Aleph CCS maesures the same for either 500ma or 540ma of bias.
4) I have now decided that the PS inductors are not saturating.
5) The notes have been cleaned up.
I am not going to publish another schematic for this particular amplifier because I don't foresee any significant changes. Besides I am having too much fun listening to these so I am moving on to other projects. The ideas coming out of the on-going discussions here will go into a new project rather than being retrofit into this one. The only thing I might try in these amps at this point is a set of "Hansen Snubbers" as described recently over in the Chip Amps forum.
Cheers to all,
Graeme L.
Although the schematic shown on page 1 of this thread is still valid this updated version accurately reflects the current "as-built" amps. The changes shown include tweeks done over the past few weeks some of which are discussed in this thread.
These are:
1) The value for the MacMillan resistors R46 and R47 are changed from 4.75K to 10K. I may try 22K again but not for while.
2) The protection diode scheme for the inputs has changed. I did this after examining the close-up photos of the XA200 PCB that NP published on pages 58 and 59 of the big Aleph-X thread. This scheme was previously used by NP on the Aleph 0. I don't believe it affects the sound but it does delete two parts and two ground connections.
3) R11 and R33 are now shown as 100K which gives 540ma of bias per output transistor on the Aleph CCS. I am going to stick with this value for now. I have not tried 600ma or higher yet. The AC current gain of the Aleph CCS maesures the same for either 500ma or 540ma of bias.
4) I have now decided that the PS inductors are not saturating.
5) The notes have been cleaned up.
I am not going to publish another schematic for this particular amplifier because I don't foresee any significant changes. Besides I am having too much fun listening to these so I am moving on to other projects. The ideas coming out of the on-going discussions here will go into a new project rather than being retrofit into this one. The only thing I might try in these amps at this point is a set of "Hansen Snubbers" as described recently over in the Chip Amps forum.
Cheers to all,
Graeme L.
100xa
GL, again thanks for sharing, I would be interested to know from where you got the new j-fets (3310's and 550's)? I didn't see them in my digikey catalog but maybe I missed them. I am just beginning to start my proto type boards. Also, in the past I have had issues with PS xformers making noise (Aleph 1.2's using 2kva tranny's). What KVA did you settle on for your xa100's and do you get any buzzing. I may just spend the bucks to get potted xfromers. I want to design in enough headroom so that the xformers are coasting and not roasting. thanks, dave
GL, again thanks for sharing, I would be interested to know from where you got the new j-fets (3310's and 550's)? I didn't see them in my digikey catalog but maybe I missed them. I am just beginning to start my proto type boards. Also, in the past I have had issues with PS xformers making noise (Aleph 1.2's using 2kva tranny's). What KVA did you settle on for your xa100's and do you get any buzzing. I may just spend the bucks to get potted xfromers. I want to design in enough headroom so that the xformers are coasting and not roasting. thanks, dave
Hi Dave,
Just go to www.digikey.com and type "ZVP3310A" or "ZTX550" into the search box. If I know what I want then this is better than using the catalog. It also tells you whats in stock before you call - a big plus for some stuff like electrolytics.
The transformers are 88 volts times 8 amps = 704 VA. I originally used used them to build A40's. A 750VA - 800VA transformer is fine in my view. Others on this site prefer to overspec the transformer. They claim that for the extra money they get better sound. They could be right. If I was was buying new transformers today I would probably buy 1000VA units. I hope this helps
Cheers,
Graeme
Just go to www.digikey.com and type "ZVP3310A" or "ZTX550" into the search box. If I know what I want then this is better than using the catalog. It also tells you whats in stock before you call - a big plus for some stuff like electrolytics.
The transformers are 88 volts times 8 amps = 704 VA. I originally used used them to build A40's. A 750VA - 800VA transformer is fine in my view. Others on this site prefer to overspec the transformer. They claim that for the extra money they get better sound. They could be right. If I was was buying new transformers today I would probably buy 1000VA units. I hope this helps
Cheers,
Graeme
100xa performance
GL, now that you have been listening to the XA100 amp for awhile how is it doing. Have all the DC offset issues been solved?
How does it sound??
I have a copy of the schematic from this thread that you used, any changes?? I want to biamp some speakers and of course the tweeters can't take any dc from the amps since they will not be protected by the crossovers and I would like 4 amps the same to avoid differences in gain etc. I was thinking about building 4 of the xa 100's this fall. I am not sure if I will fan cool or use giant heatsinks since I want to keep the weight down.
My two Aleph 1.2's our on top of four wheel dolleys so I can move them around (wife has banned them from the family room and i have them in my basement listening room.
Did you use a CLC power supply to keep out ripple? I found this necessary with the Aleph's to keep low level hum out of the speakers?
GL, now that you have been listening to the XA100 amp for awhile how is it doing. Have all the DC offset issues been solved?
How does it sound??
I have a copy of the schematic from this thread that you used, any changes?? I want to biamp some speakers and of course the tweeters can't take any dc from the amps since they will not be protected by the crossovers and I would like 4 amps the same to avoid differences in gain etc. I was thinking about building 4 of the xa 100's this fall. I am not sure if I will fan cool or use giant heatsinks since I want to keep the weight down.
My two Aleph 1.2's our on top of four wheel dolleys so I can move them around (wife has banned them from the family room and i have them in my basement listening room.
Did you use a CLC power supply to keep out ripple? I found this necessary with the Aleph's to keep low level hum out of the speakers?
Hi Dave,
The AX100's are doing fine. They broke in quite quickly but have continued to improve slightly. I had to take them out of service for a couple of months looking for a problem that was causing a sine wave test signal to look "stunted" on one side of one amp. It was hard to localize and fix. The problem must have been there from the start and was inaudible !!!!!!!!. I only found it because I was running some general tests on the system. Anyway, things have been back up and running for a couple of weeks.
I have made no further improvements.
It is my plan to have 4 AX100's to biamp a pair of MG3.6's. I am just about to start building the second pair. I am convinced that a few tens of millivolts are not harmful virtually any loudspeaker. But I do understand your concern. I have spent quite a bit of time thinking about this.
The AX100 has a gain of 26 dB balanced and 20 dB unbalanced. IIRC that is the same as the Aleph 1.2. If not you could change the gain easily. You might need to build only 2 amps then.
Boy do I hear you regarding the WAF thing.
I used the CLC supply because of all the positive discussion regarding that approach. I did not try the amp without the LC. I am very happy with the results. There is NO hum or noise evident even with my ear pressed to the speaker (zero).
The current schedule for AX100 upgrades and "experiments under consideration" are as follows:
1) Hanson snubbers as detailed in Charlies thread in the Chip Amps section. These go in within a week or so.
2) Construction on the second pair of amps has started. I hope to be finished by November 1. The initial build will use IRF610's on the outputs cascoded with 240's. Since this would require less current to drive I am using a 2SJ109 cascoded with ZTX550's running at 5ma per leg on the front end. The voltage rails are increased to 30V to deal with the extra voltage required by the cascodes. If this doesn't work out it will be easy to convert the amps back into AX100's.
3) I have come up with an integrated xover line amp based using a simplified version of the Aleph P on the outputs for gain and unbalanced to balanced conversion. That will be this winters project.
4) I have thought about substituting a pair of ZVP3310A's cascoded with MPSA92's in the front end but so far I haven't. I'd rather perform risky experiments with the new amps.
I hope this helps. Good luck with your project and keep me posted.
Cheers,
Graeme
The AX100's are doing fine. They broke in quite quickly but have continued to improve slightly. I had to take them out of service for a couple of months looking for a problem that was causing a sine wave test signal to look "stunted" on one side of one amp. It was hard to localize and fix. The problem must have been there from the start and was inaudible !!!!!!!!. I only found it because I was running some general tests on the system. Anyway, things have been back up and running for a couple of weeks.
I have made no further improvements.
It is my plan to have 4 AX100's to biamp a pair of MG3.6's. I am just about to start building the second pair. I am convinced that a few tens of millivolts are not harmful virtually any loudspeaker. But I do understand your concern. I have spent quite a bit of time thinking about this.
The AX100 has a gain of 26 dB balanced and 20 dB unbalanced. IIRC that is the same as the Aleph 1.2. If not you could change the gain easily. You might need to build only 2 amps then.
Boy do I hear you regarding the WAF thing.
I used the CLC supply because of all the positive discussion regarding that approach. I did not try the amp without the LC. I am very happy with the results. There is NO hum or noise evident even with my ear pressed to the speaker (zero).
The current schedule for AX100 upgrades and "experiments under consideration" are as follows:
1) Hanson snubbers as detailed in Charlies thread in the Chip Amps section. These go in within a week or so.
2) Construction on the second pair of amps has started. I hope to be finished by November 1. The initial build will use IRF610's on the outputs cascoded with 240's. Since this would require less current to drive I am using a 2SJ109 cascoded with ZTX550's running at 5ma per leg on the front end. The voltage rails are increased to 30V to deal with the extra voltage required by the cascodes. If this doesn't work out it will be easy to convert the amps back into AX100's.
3) I have come up with an integrated xover line amp based using a simplified version of the Aleph P on the outputs for gain and unbalanced to balanced conversion. That will be this winters project.
4) I have thought about substituting a pair of ZVP3310A's cascoded with MPSA92's in the front end but so far I haven't. I'd rather perform risky experiments with the new amps.
I hope this helps. Good luck with your project and keep me posted.
Cheers,
Graeme
AX100
I plan to turn on the soldering iron for this project this fall. I am finishing up a Phono pre-amp and then I will begin the XA 100 driver boards out of parts left over from my Aleph projects.
I have to still think through the heat sink design and I am DEFINETLY PUTTING THIS AMP ON STOUT MACHINE ROLLERS!
I plan to use a split power supply, one tranny for the + an one for the minus rails as these are more readily available than some monster 2kva or 3kva unit per mono amp.
I will breadboard your design and test it using my variac and just several fan cooled outputs until I get it right and then proceed from there. I will let you know. By the way to get it to look good I will chrome plate it and do some metallic coating. Wether I do a vertical design (like computers) or the typical horizontal box design will depend on what type of heat sinks that I can find.
dave
I plan to turn on the soldering iron for this project this fall. I am finishing up a Phono pre-amp and then I will begin the XA 100 driver boards out of parts left over from my Aleph projects.
I have to still think through the heat sink design and I am DEFINETLY PUTTING THIS AMP ON STOUT MACHINE ROLLERS!
I plan to use a split power supply, one tranny for the + an one for the minus rails as these are more readily available than some monster 2kva or 3kva unit per mono amp.
I will breadboard your design and test it using my variac and just several fan cooled outputs until I get it right and then proceed from there. I will let you know. By the way to get it to look good I will chrome plate it and do some metallic coating. Wether I do a vertical design (like computers) or the typical horizontal box design will depend on what type of heat sinks that I can find.
dave
Hi Dave,
The AX100 isn't that much of monster really. Each amp weighs 64 lbs. Not a lightweight for sure. but not in the "dolly required" category either.
The transformer only needs to supply 18-20VAC @ 20-22A times 2. That's about 750 - 800VA. You have some leeway as well. There are several suppliers you can tap. You can even buy something close to the original transformer if you want to. Go to the Belfuse website and look under Signal Transformers for the 80-8 or 80-12. They're expensive but the company is still in Brooklyn or thereabouts last I checked. That might be close to you.
The design was worked out on unit 1. Unit 2 came right up the same. Presuming you don't make any significant value changes and provide adequate heatsinking your results should be the same too.
If you have any questions along the way just post them. A lot people here helped me out. I'm happy to pass it forward.
Cheers,
Graeme
(P.S. IMO blue LED pilot lights would complement the chrome just Jack.)
The AX100 isn't that much of monster really. Each amp weighs 64 lbs. Not a lightweight for sure. but not in the "dolly required" category either.
The transformer only needs to supply 18-20VAC @ 20-22A times 2. That's about 750 - 800VA. You have some leeway as well. There are several suppliers you can tap. You can even buy something close to the original transformer if you want to. Go to the Belfuse website and look under Signal Transformers for the 80-8 or 80-12. They're expensive but the company is still in Brooklyn or thereabouts last I checked. That might be close to you.
The design was worked out on unit 1. Unit 2 came right up the same. Presuming you don't make any significant value changes and provide adequate heatsinking your results should be the same too.
If you have any questions along the way just post them. A lot people here helped me out. I'm happy to pass it forward.
Cheers,
Graeme
(P.S. IMO blue LED pilot lights would complement the chrome just Jack.)
AX100
GL, I just pulled out my Aleph 1.2 schematic and compared it against your XA100 and it looks simple enough as it is just a lower voltage mirrored image version of the Aleph 1.2 but looking a your drawing notes, I have some questions:
1. You state that R14 and R31 are 100k for 115 watts at 8 ohms.
Do you really mean R 11 and R33 since you show R14&31 as 1.5k on your as built?
2. The only symetry feedback resistors are the 10K you show to the sources of the input pair. Was this your final value? I know there has been a lot of discussion in the threads regarding the value of R46 & R47. The ac feeback to these resistors attached to the source is before the normal open loop feedback to the gates so this must be the super symetry feedback that NP discusses in his white paper?
thanks, dave
GL, I just pulled out my Aleph 1.2 schematic and compared it against your XA100 and it looks simple enough as it is just a lower voltage mirrored image version of the Aleph 1.2 but looking a your drawing notes, I have some questions:
1. You state that R14 and R31 are 100k for 115 watts at 8 ohms.
Do you really mean R 11 and R33 since you show R14&31 as 1.5k on your as built?
2. The only symetry feedback resistors are the 10K you show to the sources of the input pair. Was this your final value? I know there has been a lot of discussion in the threads regarding the value of R46 & R47. The ac feeback to these resistors attached to the source is before the normal open loop feedback to the gates so this must be the super symetry feedback that NP discusses in his white paper?
thanks, dave
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.