Hi Graeme.
Thanks for your report.
I have been glued to the Australian Open so nothing done.
But my matched JFets are on their way.
My relative offset (with 9610) also shifts a bit after full warm up but not much.
I also have to capacitively couple the inputs on both sides.
Even with my X2.5 connected it seems to throw out the relative offset.
A question. Are you J Fets touching each other on each side?
As long as its consistant I dont thing its a real problem.
As William say a 10 ohm pot maybe a good thing on the souce of one side.
Macka
Thanks for your report.
I have been glued to the Australian Open so nothing done.
But my matched JFets are on their way.
My relative offset (with 9610) also shifts a bit after full warm up but not much.
I also have to capacitively couple the inputs on both sides.
Even with my X2.5 connected it seems to throw out the relative offset.
A question. Are you J Fets touching each other on each side?
As long as its consistant I dont thing its a real problem.
As William say a 10 ohm pot maybe a good thing on the souce of one side.
Macka
Hi Ian,
Thank you for your thoughts. Are you waiting for your new JFET's before proceeding? I had thought you had the parts you needed for this particular project already on hand.
It has occurred to me that drift in the relative offset is being caused by current hogging in the JFETs. So your suggestion about Williams pot is a good one. I don't like using Source resistors in JFET circuits. I decided years ago I don't like the way they make a circuit sound. A highly personal and arbitrary statement I realize. In a SuSy circuit I would prefer to see the common mode feedback resistors connect directly to the Sources and not to a node that puts any Source resistors in series. I could use four Source resistors, one for each JFET, and then four MacMillan resistors of twice the current value, again one to each Source. But that's a lot of parts. What do you think?
Tonight I am going to fire up the amp, let it heat up to operating temp, then measure the temperature of each JFET package to see if they are heating up unevenly. And no the J109s are not touching each other.
Cheers,
Graeme
Thank you for your thoughts. Are you waiting for your new JFET's before proceeding? I had thought you had the parts you needed for this particular project already on hand.
It has occurred to me that drift in the relative offset is being caused by current hogging in the JFETs. So your suggestion about Williams pot is a good one. I don't like using Source resistors in JFET circuits. I decided years ago I don't like the way they make a circuit sound. A highly personal and arbitrary statement I realize. In a SuSy circuit I would prefer to see the common mode feedback resistors connect directly to the Sources and not to a node that puts any Source resistors in series. I could use four Source resistors, one for each JFET, and then four MacMillan resistors of twice the current value, again one to each Source. But that's a lot of parts. What do you think?
Tonight I am going to fire up the amp, let it heat up to operating temp, then measure the temperature of each JFET package to see if they are heating up unevenly. And no the J109s are not touching each other.
Cheers,
Graeme
Hi Graeme,
The parts I have on had are not of suitable match.
I think finding out the relative offset is happening before any modification is a good idea.
A thought is to swap sides for the Jfets.
The differecne is coming from somewhere and the output stage gain and rising temperayure is magnifying it.
Macka
The parts I have on had are not of suitable match.
I think finding out the relative offset is happening before any modification is a good idea.
A thought is to swap sides for the Jfets.
The differecne is coming from somewhere and the output stage gain and rising temperayure is magnifying it.
Macka
Hi Ian,
Yes I agree. The issue regarding the rising relative offset needs to be solved first. I've been thinking that since it is only 3 to 4 mv at turn on that the existing component matching is fine. The problem is the result of component parameters changing during warm up. The first thing I think I'll try is 10 ohms in series with each JFET source and see if that changes the relative offset performance.
Cheers,
Graeme
Yes I agree. The issue regarding the rising relative offset needs to be solved first. I've been thinking that since it is only 3 to 4 mv at turn on that the existing component matching is fine. The problem is the result of component parameters changing during warm up. The first thing I think I'll try is 10 ohms in series with each JFET source and see if that changes the relative offset performance.
Cheers,
Graeme
Hi,
just to make it clear. I´ve used a 10R trimpot. Connected the current source to the wiper and the fets to both ends,
In the middle position the fets have a 5R source resistor. Turning the pot makes one bigger and the other smaller and compensates for not perfect matching.
It starts at around 0mV and stays at around 0mV.....
In my amp both fets touch eachother.
William
just to make it clear. I´ve used a 10R trimpot. Connected the current source to the wiper and the fets to both ends,
In the middle position the fets have a 5R source resistor. Turning the pot makes one bigger and the other smaller and compensates for not perfect matching.
It starts at around 0mV and stays at around 0mV.....
In my amp both fets touch eachother.
William
aleph 1.2 to xa100
Erno Borbely in his balanced line differential tube line amp did the same thing in two different places. He put a pot between the plate resistors and between the current sources at the cathodes of the input tubes to make sure the voltage gain on both tubes was the same for +/- signals-works well. dave
Erno Borbely in his balanced line differential tube line amp did the same thing in two different places. He put a pot between the plate resistors and between the current sources at the cathodes of the input tubes to make sure the voltage gain on both tubes was the same for +/- signals-works well. dave
aleph1.2 to AX100 and the Doobie Brothers Wine
William, is there a sonic penalty to putting in a trimpot or trim resistor in series with the fet's. I don't know myself, just asking.
It seems to me to be very logical and can save a considerable amount of time and money. I think I trimmed one side of my A75 diff input pair but I can't remember becasuse I built it in 1993.
I didn't do that in my AX120 but I struggled to get a good match.
dave
William, is there a sonic penalty to putting in a trimpot or trim resistor in series with the fet's. I don't know myself, just asking.
It seems to me to be very logical and can save a considerable amount of time and money. I think I trimmed one side of my A75 diff input pair but I can't remember becasuse I built it in 1993.
I didn't do that in my AX120 but I struggled to get a good match.
dave
Hi,
I don´t know because I never tried without. It should be small as the source resistors are quite small as such.
Not using them would probably be better but I found a list of my 2sj74. It were 34 pieces and I measured Idss and Vp but they varied quite a lot.
I´m in the proces of building a CCS-CS-XBOSOZ for a friend (who was very happy with his Aleph 3´s). After that maybe I´ll have some time to try a few things on my Aleph X´s.
BTW changed speakers from Thiel CS6 to Jordan "the Wall". Quite a different concept but not sounding bad at all.
Must only find a suitable sub.
Anybody interested in a pair of Thiels
?
William
I don´t know because I never tried without. It should be small as the source resistors are quite small as such.
Not using them would probably be better but I found a list of my 2sj74. It were 34 pieces and I measured Idss and Vp but they varied quite a lot.
I´m in the proces of building a CCS-CS-XBOSOZ for a friend (who was very happy with his Aleph 3´s). After that maybe I´ll have some time to try a few things on my Aleph X´s.
BTW changed speakers from Thiel CS6 to Jordan "the Wall". Quite a different concept but not sounding bad at all.
Must only find a suitable sub.
Anybody interested in a pair of Thiels
?William
I have done some capacitor rollling.
The Panasonic FC series, well known around here.
The capacitor seem to work well when bypassed with a small value film / PPE capacitor 0.01 - 0.047. Unbypassed it was clean but otherwise lacked extension and detail.
Nichicon Gold Muse.
Not sure about this capacitor. Lush / bloom in the midrange and veiled high frequencies.
I plan to buy some Panasonic FM series and bypass with Panasonic PP 0.047. This variety / combination appears very similar to parts found in the real deal.
http://www.rsaustralia.com/cgi-bin/...cchadedeifjdikcefeceeldgkidhgm.0&cacheID=auie
The Panasonic FC series, well known around here.
The capacitor seem to work well when bypassed with a small value film / PPE capacitor 0.01 - 0.047. Unbypassed it was clean but otherwise lacked extension and detail.
Nichicon Gold Muse.
Not sure about this capacitor. Lush / bloom in the midrange and veiled high frequencies.
I plan to buy some Panasonic FM series and bypass with Panasonic PP 0.047. This variety / combination appears very similar to parts found in the real deal.
http://www.rsaustralia.com/cgi-bin/...cchadedeifjdikcefeceeldgkidhgm.0&cacheID=auie
Hi Ian,
Time for an update. I have implemented all of the AX100J mods that were detailed above. The main goal, if you recall, was to improve the relative offset performance. Here are the results.
1) JFET Source resistors - I was going to put a 10 ohm resistor in series with each JFET source. Turns out I didn't have that value so I used 15 ohms instead. Nothing changed, but I've left them in for now.
2) All of the resistors that touch the input diff pair were matched to within .1%. The relative offset improved somewhat. It now started at 68mv at turn on and rose to 90mv after one and a half hours.
3) The JFET gate to ground resistors were increased to 22K. The relative offset now became 28mv at turn on and increased to 41mv after 90 minutes. A big improvement. I changed them to 47K and the offset now started at 250mv. 100K made it ~500mv. The 22K's are back in. I may fish around for a more optimum value later. But for the moment I can live with 41mv.
OK. Now here's where I'm a bit puzzled. The amp is not DC stable when it's direct coupled. If I run it with no input connection and the XLR (-) shorted to ground the relative offset rises to several hundred millivolts. This situation remains even if a connection is made to the RCA input. If I tack in the 5uf input coupling caps everything is fine. I don't know what would happen if I connected a balanced source since I don't have one. I suspect that everything would be OK. I would appreciate any thoughts you might have on this. Either I've made a mistake somewhere or this is one more unusual behavior mode of this circuit. Perhaps there's a trick. But for now the amps will be AC coupled.
I put the modded amp into the system for a listen. After 2 hours you could here it starting to clear up as it broke in. It sounded at least as good as the unmodded unit. Making comparisons beyond this is a bit of a fools game so I won't do it. However, I will say that I'm very encouraged.
I think I'll sit back and give this situation a bit of a think. I may run some test signals through the amp and see what they look like. I'd also like to experiment with the JFET Source resistors and do some more measurements.
I'll keep you posted. What are your thoughts? Anyone else? All suggestions are welcome.
Cheers,
Graeme
Time for an update. I have implemented all of the AX100J mods that were detailed above. The main goal, if you recall, was to improve the relative offset performance. Here are the results.
1) JFET Source resistors - I was going to put a 10 ohm resistor in series with each JFET source. Turns out I didn't have that value so I used 15 ohms instead. Nothing changed, but I've left them in for now.
2) All of the resistors that touch the input diff pair were matched to within .1%. The relative offset improved somewhat. It now started at 68mv at turn on and rose to 90mv after one and a half hours.
3) The JFET gate to ground resistors were increased to 22K. The relative offset now became 28mv at turn on and increased to 41mv after 90 minutes. A big improvement. I changed them to 47K and the offset now started at 250mv. 100K made it ~500mv. The 22K's are back in. I may fish around for a more optimum value later. But for the moment I can live with 41mv.
OK. Now here's where I'm a bit puzzled. The amp is not DC stable when it's direct coupled. If I run it with no input connection and the XLR (-) shorted to ground the relative offset rises to several hundred millivolts. This situation remains even if a connection is made to the RCA input. If I tack in the 5uf input coupling caps everything is fine. I don't know what would happen if I connected a balanced source since I don't have one. I suspect that everything would be OK. I would appreciate any thoughts you might have on this. Either I've made a mistake somewhere or this is one more unusual behavior mode of this circuit. Perhaps there's a trick. But for now the amps will be AC coupled.
I put the modded amp into the system for a listen. After 2 hours you could here it starting to clear up as it broke in. It sounded at least as good as the unmodded unit. Making comparisons beyond this is a bit of a fools game so I won't do it. However, I will say that I'm very encouraged.
I think I'll sit back and give this situation a bit of a think. I may run some test signals through the amp and see what they look like. I'd also like to experiment with the JFET Source resistors and do some more measurements.
I'll keep you posted. What are your thoughts? Anyone else? All suggestions are welcome.
Cheers,
Graeme
Hi Graeme.
Good work.
What value are your McMillan resisters?
Also, how well re your dc bias settings matched on both sides?
The single ended offset issue has been identified before and is normal as far I know.
Use the best capacitor you can offord on the inputs
Macka
edit: I suspect the Jfet benefit from warming up
Good work.
What value are your McMillan resisters?
Also, how well re your dc bias settings matched on both sides?
The single ended offset issue has been identified before and is normal as far I know.
Use the best capacitor you can offord on the inputs
Macka
edit: I suspect the Jfet benefit from warming up
Hi Ian,
Thank you for your thoughts on this. If what you say is true about a MOSFET input Aleph-X with DC coupled inputs showing the same characteristics as I'm seeing with DC coupled JFET inputs (when used with a single ended input) then I'm relieved. I don't recall any discussion of this back in the old days, but then there's lots I don't remember from back then. I did see a recent set of posts on this though. I may try DC coupling my remaining MOSFET amp just to confirm the behavior. If it's confirmed then I'll convert the second amp to JFET inputs.
The MacMillan resistors are 10K for now. I am going to try 22K after the amps have broken in.
I'm using Panasonic PP film caps. They are shown in the pics at the beginning of this thread.
The JFETs seem to warm up right away according to my temp probe. As I recall what with other changes made to the amp, it had a relative offset of about 40mv already in its MOSFET incarnation. I think 40mv is fine and I'm not going to worry about it anymore.
How are your amps doing? How's the JFET mod coming along?
Cheers,
Graeme
P.S. Last night I spotted a Fred (Harry Haller) posting on the first page of the big Aleph-X thread that proposed a dual 2SJ109 front end for the Aleph-X. I don't recall seeing that post before even though I've probably read it a dozen times. It brought me up short for a moment. So let's give credit where it's due.
Thank you for your thoughts on this. If what you say is true about a MOSFET input Aleph-X with DC coupled inputs showing the same characteristics as I'm seeing with DC coupled JFET inputs (when used with a single ended input) then I'm relieved. I don't recall any discussion of this back in the old days, but then there's lots I don't remember from back then. I did see a recent set of posts on this though. I may try DC coupling my remaining MOSFET amp just to confirm the behavior. If it's confirmed then I'll convert the second amp to JFET inputs.
The MacMillan resistors are 10K for now. I am going to try 22K after the amps have broken in.
I'm using Panasonic PP film caps. They are shown in the pics at the beginning of this thread.
The JFETs seem to warm up right away according to my temp probe. As I recall what with other changes made to the amp, it had a relative offset of about 40mv already in its MOSFET incarnation. I think 40mv is fine and I'm not going to worry about it anymore.
How are your amps doing? How's the JFET mod coming along?
Cheers,
Graeme
P.S. Last night I spotted a Fred (Harry Haller) posting on the first page of the big Aleph-X thread that proposed a dual 2SJ109 front end for the Aleph-X. I don't recall seeing that post before even though I've probably read it a dozen times. It brought me up short for a moment. So let's give credit where it's due.
aleph 1.2 to xa100
GL, where did you get yor j-fets? Digikey? Also, I am running my AX100's DC coupled, and balanced on the front end. The relative offset never changes. On one amp it is a few mv's and on the other about 90mv. I did notice that when I remove the top cover from the chassis to check something, my absolute offset changes as the heat floats up out of the amp (it has been on seveal hours). I had to set my absolute dc offset with the amp totally buttoned up. I can do this since I left an opening in the side.
When setting up the amps I tested them single ended because my test equipment is single ended (sine and square wave generators).
I have ac coupled the output with 40uf caps on the +/- speaker outputs that feed my tweeter line arrays to protect them from start up thumps or an amp failure putting dc on the outputs.
Since my tweeters are crossed over at 2600hz the caps have no impact on the sound. I am using the DEQX 2.6p digital crossover and speaker correction device as the pre-amp. Their is no dc present on the outputs of this device.
I never tried ac coupling the inputs, maybe that is why I was going crazy trying to get a matched set of devices.
The amps sound so good that I haven't measured anything lately but they do run hot. I may add some cooling fans later.
I don't understand why changing the gate to ground resistors would change the relative offset if both resistors are the same?
Is it do to the gain multiplication? dave
GL, where did you get yor j-fets? Digikey? Also, I am running my AX100's DC coupled, and balanced on the front end. The relative offset never changes. On one amp it is a few mv's and on the other about 90mv. I did notice that when I remove the top cover from the chassis to check something, my absolute offset changes as the heat floats up out of the amp (it has been on seveal hours). I had to set my absolute dc offset with the amp totally buttoned up. I can do this since I left an opening in the side.
When setting up the amps I tested them single ended because my test equipment is single ended (sine and square wave generators).
I have ac coupled the output with 40uf caps on the +/- speaker outputs that feed my tweeter line arrays to protect them from start up thumps or an amp failure putting dc on the outputs.
Since my tweeters are crossed over at 2600hz the caps have no impact on the sound. I am using the DEQX 2.6p digital crossover and speaker correction device as the pre-amp. Their is no dc present on the outputs of this device.
I never tried ac coupling the inputs, maybe that is why I was going crazy trying to get a matched set of devices.
The amps sound so good that I haven't measured anything lately but they do run hot. I may add some cooling fans later.
I don't understand why changing the gate to ground resistors would change the relative offset if both resistors are the same?
Is it do to the gain multiplication? dave
Hi Graeme,
On the SE input a friend in Europe had the exact same issue so I asked him to a/c couple both +- inpuys with a capacitor and problem solved. In balance mode of course it it almost cancelled out.
No my Jfets have not arrived yet but I am going to set up one amp in anticipation.
Macka
On the SE input a friend in Europe had the exact same issue so I asked him to a/c couple both +- inpuys with a capacitor and problem solved. In balance mode of course it it almost cancelled out.
No my Jfets have not arrived yet but I am going to set up one amp in anticipation.
Macka
Hi Dave,
I bought 25 2SJ109's from MCM a couple of years ago just in case the part became unobtainable. And it did. I wish I had bought more. I've already used up a third of them. You can match up 2SJ74's for a similar result - like Ian (macka) is doing.
The absolute DC offset on my amps wanders if the thermal environment changes - just like it does on your amps. If I blow on the input diff pair or the heat sink the offset heads off and slowly returns when the breeze stops. Nature of the beast.
Frankly, the amps sound wonderful with the MOSFET inputs. I'm only changing to JFETs because I'm curious and because I can. NP's comment a few posts back is also encouraging. I'm keeping the MOSFET front end boards in case I want to change back.
Graeme
I bought 25 2SJ109's from MCM a couple of years ago just in case the part became unobtainable. And it did. I wish I had bought more. I've already used up a third of them. You can match up 2SJ74's for a similar result - like Ian (macka) is doing.
The absolute DC offset on my amps wanders if the thermal environment changes - just like it does on your amps. If I blow on the input diff pair or the heat sink the offset heads off and slowly returns when the breeze stops. Nature of the beast.
Frankly, the amps sound wonderful with the MOSFET inputs. I'm only changing to JFETs because I'm curious and because I can. NP's comment a few posts back is also encouraging. I'm keeping the MOSFET front end boards in case I want to change back.
Graeme
ax100 update
GL, thanks for the info, by the way I have had some success solving my transformer noise problem. I (I know this is against the diy credo) bought another 1kva trannie from Avel Lindberg and installed it in parallel with original 1kva trannie I had working in the amp and the noise disappeared. I can only hear buzzing and humming with my ear on the amp chasis now.
It seems that even though the original 1kva trannie was electrically fine size wize; that the steady current draw caused the core to buzz.
Now I have one amp with 2kva and no noise. I will have to get another trannie for the other amp. They cost $124 and although affordable I wanted to solve it the cheap way.
I converted thes amps for minimal cost, I already had the chasis and all the outputs. I built the boards from your schematic using parts left over from my Aleph 1.2 project and other projects long finished. So having to buy extra trannies isn't too bad.
I have no idea now close this design is to the commercial amps that are sold by HP, all I can say that is if I got within 80% I am more than happy. I couldn't afford the 18k for the XA160 (If up to me I really would and get the Prime Minister a used car but you know how that would have turned out as I would be sleeping in it).
All of HP's designs sound incredible and my 350.5 is also sounding supberb!! It is hard to watch tv now, although Hillary vs Obama is heating up. dave
GL, thanks for the info, by the way I have had some success solving my transformer noise problem. I (I know this is against the diy credo) bought another 1kva trannie from Avel Lindberg and installed it in parallel with original 1kva trannie I had working in the amp and the noise disappeared. I can only hear buzzing and humming with my ear on the amp chasis now.
It seems that even though the original 1kva trannie was electrically fine size wize; that the steady current draw caused the core to buzz.
Now I have one amp with 2kva and no noise. I will have to get another trannie for the other amp. They cost $124 and although affordable I wanted to solve it the cheap way.
I converted thes amps for minimal cost, I already had the chasis and all the outputs. I built the boards from your schematic using parts left over from my Aleph 1.2 project and other projects long finished. So having to buy extra trannies isn't too bad.
I have no idea now close this design is to the commercial amps that are sold by HP, all I can say that is if I got within 80% I am more than happy. I couldn't afford the 18k for the XA160 (If up to me I really would and get the Prime Minister a used car but you know how that would have turned out as I would be sleeping in it).
All of HP's designs sound incredible and my 350.5 is also sounding supberb!! It is hard to watch tv now, although Hillary vs Obama is heating up. dave
Re: ax100 update
Hi Dave,
I don't know why you would think that. There are, I expect, many people (myself included) who would identify with the following philosophy. First try to make things work with elegant engineering or cheaper hardware, but if that doesn't work out, the solution that works for you/me is the best one.
Also, note that you are contributing to a Class A amplifier thread in primarily a Class A bias forum. Bigger is better! For an example, see "Power Supplies" by Nelson Pass on http://www.passlabs.com/articles.htm.
Cheers,
Jeremy
daly41k said:
Hi Dave,
I don't know why you would think that. There are, I expect, many people (myself included) who would identify with the following philosophy. First try to make things work with elegant engineering or cheaper hardware, but if that doesn't work out, the solution that works for you/me is the best one.
Also, note that you are contributing to a Class A amplifier thread in primarily a Class A bias forum. Bigger is better! For an example, see "Power Supplies" by Nelson Pass on http://www.passlabs.com/articles.htm.
Cheers,
Jeremy
As a postscript, I solved my hum a very slightly different way. I bought two of the Plitron low noise transformers to replace my standard 1000VA transformers. They killed the hum, but they are likely to kill my back when I get everything assembled into proper power supply chassis. It turns out your solution may have worked better for me. I could have made four power supply chassis (plus and minus rails for each amplifier) that weigh about 30 lbs each instead of two that weigh about 60lbs each. Of course, with an extra 7V to dump using RC filters, the output is going to be very quiet.
Jeremy
PS. Anyone want two Plitron LoNo 1000VA, 30VAC transformers?
Jeremy
PS. Anyone want two Plitron LoNo 1000VA, 30VAC transformers?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.