AX100
GL, I have been reading your thread with the idea of building the AX100 but trying to get a wee more power. I have some 2kva transformers at around 28vots ct. In reading this thread is the high initial dc offset 6-12Volts being measured at the 30ohm resistor to ground on the the +/- outputs? Or is it being measured directly between the +/- speakrer outputs? I built the A-75 and the the Aleph 1.2's an the dc offset cold may be 20mv at best and it goes down to less than 10mv after warm up at the speaker outputs. I wouldn't want 12 volts dc across the speaker.
I will use perf board to construct the driver board but who is selling the circuit boards? I would just llike to see how they are laid out.
thanks, dave
GL, I have been reading your thread with the idea of building the AX100 but trying to get a wee more power. I have some 2kva transformers at around 28vots ct. In reading this thread is the high initial dc offset 6-12Volts being measured at the 30ohm resistor to ground on the the +/- outputs? Or is it being measured directly between the +/- speakrer outputs? I built the A-75 and the the Aleph 1.2's an the dc offset cold may be 20mv at best and it goes down to less than 10mv after warm up at the speaker outputs. I wouldn't want 12 volts dc across the speaker.
I will use perf board to construct the driver board but who is selling the circuit boards? I would just llike to see how they are laid out.
thanks, dave
Hi William,
Thank you for the thread reference. Yes , now I remeber your amps. A beautiful job by the way. Afte reading through the thread carefully I found that your experience with absolute DC offset is similar to mine. The biggest dynamic influence I have found is the temperature of the output transistors (and heat sinks). If I blow on the heat sinks or place my hands on them the DC offset will start changing. I can make it move two or three hundred millivolts in less than a minute quite easily. Larger value MacMillan resistors produce a larger move and more rapid response to heat changes. Even a few minutes of loud music can cause heat changes and a shift in the absolute DC offset. Again, I think that as long as this activity stays within plus or minus 1 volt and settles to this range within one hour after turn-on then all is well.
I measured one channel yesterday after making some changes. Absolute DC offset started at approximately 8V after a few seconds. After 60s it was 5.5V, 5m = 3.8V, 15m = 2.45V, 30m = 1.45V, 45m 1.11V, 60m = .93V. After 90 minutes and after playing the amp at high volume for a while this number would vary over a range of .7V to 1.3V so I will adjust the bias to center this and check it again in a week. This is with 10K MacMillan resistors.
I agree with you: it would be nice to know if this DC offset performance is what one would see with a factory XA amp.
Hi Dave,
I will answer your questions in order (as best I can). First, how much more power is "a wee more". The AX100 will easily give you 125W with just a few adjustments. 28V secondaries will give you pretty high power rail voltages. You'll need big heatsinks and lots of output devices. If you've built the Aleph 1.2 then you know the territory. It sounds do-able to me.
William has already answered your absolute DC offset question. I wouldn't worry about it.
I do the perf board thing every time. It gives me total control over the parts placement. It lets me choose the parts I want to use without worrying about fit and physical interference problems. It also lets me control the grounding scheme and the amount of copper in the "traces". And finally it lets me change things if I run into a problem with stability or if another person discovers a beneficial mod.
O.K. I don't want to sound negative but here goes: Buying a board means buying someone elses potential problems. Good board layouts are produced by experienced people and even then they usually take a couple of "turns" or revisions to get things right. Most of the boards seen here are designed by inexperienced amateurs having fun (nothing wrong with that) and have no turns or at most one turn. You often see people lining up to buy boards that haven't even been built-up or tested. This is a hobby for all of us which means we get to do whatever we each want to do. For me, perf boards are more fun than laid out boards, provide more flexability, and perform just as well. I guess I should state that I am an engineer and have designed, and debugged more than my share of boards. I have also had professional layout "artistes" lay out my designs. I consider this whole arena to be one of those "fools rush in" kind of deals. I apologize if I've offended anyone. End of rant!
Cheers,
Graeme
Thank you for the thread reference. Yes , now I remeber your amps. A beautiful job by the way. Afte reading through the thread carefully I found that your experience with absolute DC offset is similar to mine. The biggest dynamic influence I have found is the temperature of the output transistors (and heat sinks). If I blow on the heat sinks or place my hands on them the DC offset will start changing. I can make it move two or three hundred millivolts in less than a minute quite easily. Larger value MacMillan resistors produce a larger move and more rapid response to heat changes. Even a few minutes of loud music can cause heat changes and a shift in the absolute DC offset. Again, I think that as long as this activity stays within plus or minus 1 volt and settles to this range within one hour after turn-on then all is well.
I measured one channel yesterday after making some changes. Absolute DC offset started at approximately 8V after a few seconds. After 60s it was 5.5V, 5m = 3.8V, 15m = 2.45V, 30m = 1.45V, 45m 1.11V, 60m = .93V. After 90 minutes and after playing the amp at high volume for a while this number would vary over a range of .7V to 1.3V so I will adjust the bias to center this and check it again in a week. This is with 10K MacMillan resistors.
I agree with you: it would be nice to know if this DC offset performance is what one would see with a factory XA amp.
Hi Dave,
I will answer your questions in order (as best I can). First, how much more power is "a wee more". The AX100 will easily give you 125W with just a few adjustments. 28V secondaries will give you pretty high power rail voltages. You'll need big heatsinks and lots of output devices. If you've built the Aleph 1.2 then you know the territory. It sounds do-able to me.
William has already answered your absolute DC offset question. I wouldn't worry about it.
I do the perf board thing every time. It gives me total control over the parts placement. It lets me choose the parts I want to use without worrying about fit and physical interference problems. It also lets me control the grounding scheme and the amount of copper in the "traces". And finally it lets me change things if I run into a problem with stability or if another person discovers a beneficial mod.
O.K. I don't want to sound negative but here goes: Buying a board means buying someone elses potential problems. Good board layouts are produced by experienced people and even then they usually take a couple of "turns" or revisions to get things right. Most of the boards seen here are designed by inexperienced amateurs having fun (nothing wrong with that) and have no turns or at most one turn. You often see people lining up to buy boards that haven't even been built-up or tested. This is a hobby for all of us which means we get to do whatever we each want to do. For me, perf boards are more fun than laid out boards, provide more flexability, and perform just as well. I guess I should state that I am an engineer and have designed, and debugged more than my share of boards. I have also had professional layout "artistes" lay out my designs. I consider this whole arena to be one of those "fools rush in" kind of deals. I apologize if I've offended anyone. End of rant!
Cheers,
Graeme
Hi Graeme. You are right about the board thing, indeed. I am ofcourse not the right one to say so (I am buying a set of all the boards being designed here at the Pass department
) But having seen all the board GB's with corrections added afterwards, one must agree with your point. Making an amp with perfboards is a very nice experience. I made the ZenV8/9 on perfboard and I have a feeling about it, like its a child of mine But a nice printed circuit-board sure looks neat and tidy So I guess I have a foot in each camp.
Steen
) But having seen all the board GB's with corrections added afterwards, one must agree with your point. Making an amp with perfboards is a very nice experience. I made the ZenV8/9 on perfboard and I have a feeling about it, like its a child of mine But a nice printed circuit-board sure looks neat and tidy So I guess I have a foot in each camp.Steen
Hi Steen,
I agree with you that an etched board gives a project a polished look. My main point is that the hobby needs to be fun. We each should be doing what we need to do to make it so.
I just wanted to caution newer DIYers that if you buy a PCB you are really buying the designers skill (or lack thereof). I might make a comparison with tying your own flies or buying them.
Graeme
I agree with you that an etched board gives a project a polished look. My main point is that the hobby needs to be fun. We each should be doing what we need to do to make it so.
I just wanted to caution newer DIYers that if you buy a PCB you are really buying the designers skill (or lack thereof). I might make a comparison with tying your own flies or buying them.
Graeme
Hey Graeme, now you have me nailed down. I would never buy a single fly
I better think this over properly Thanks for putting it in those clear words Steen
BTW My youngest son and I have just agreed to go to Norway this summer and tempt the Salmon with our flies. Norway is a great country for that particular interest
If you care to send me an email, Graeme, I would be happy to send you a couple of my own flies!
My two cents:
I built my Aleph-X monoblocks using perfboard and I fully agree with what Graeme has said. I've never had a first version layout be perfectly correct, and it is nice to have the flexibility to change things along the way. Personally I didn't find that creating a perfboard layout for the Aleph-X that difficult, and I wonder why more people don't go this way.
Secondly, I thought I'd emphasise the importance of component matching when building the Aleph-X. I matched the input differenctial pair Vgs to less than 1mV, and I matched all the output FET Vgs's to within 1mV, and then matched all the output FET source resistors to within 0.05 ohm. The relative offset is less than 50mV at all times. On a cold startup the absolute offset is around 4V and settles to less than 0.5V within a half hour, and settles to less than 0.3V at equilibrium. I have 50 ohm resistors from the outputs to ground, and no MacMillan resistors. Now the downside is that I had to buy 3 times the number of Fet's and source resistors that were actually needed by the circuit, and I spent a fair bit of time building matching circuits that exactly replicated the amps operating parameters, and then spent even more time matching the components, but in the end I feel the additional effort and cost was easily justified by the end results.
Cheers, Terry
I built my Aleph-X monoblocks using perfboard and I fully agree with what Graeme has said. I've never had a first version layout be perfectly correct, and it is nice to have the flexibility to change things along the way. Personally I didn't find that creating a perfboard layout for the Aleph-X that difficult, and I wonder why more people don't go this way.
Secondly, I thought I'd emphasise the importance of component matching when building the Aleph-X. I matched the input differenctial pair Vgs to less than 1mV, and I matched all the output FET Vgs's to within 1mV, and then matched all the output FET source resistors to within 0.05 ohm. The relative offset is less than 50mV at all times. On a cold startup the absolute offset is around 4V and settles to less than 0.5V within a half hour, and settles to less than 0.3V at equilibrium. I have 50 ohm resistors from the outputs to ground, and no MacMillan resistors. Now the downside is that I had to buy 3 times the number of Fet's and source resistors that were actually needed by the circuit, and I spent a fair bit of time building matching circuits that exactly replicated the amps operating parameters, and then spent even more time matching the components, but in the end I feel the additional effort and cost was easily justified by the end results.
Cheers, Terry
understanding Aleph-X absolute offset
Hi,
the last few days I´ve thought a lot about absolute dc offset and measured a few things.
The main cause for the big variation (in my Aleph-X) is the changing of the bias current for the diff pair. (cost me quite some ice-spray......)
The Vgs of the current source fet changes from 3.78V (cold) to 3.58V (73°C). This changes the current from 22.06 to 23.89mA and dc offset from 12.1 to -1V.
So I thought it would be nice to have a current source wich put´s out the same current from 20 to 70°C.
Operating the fet at the zero Tc point was no option cause this is somewhere near 400mA so I thought of putting in a compensating element. The best place for this would be the zener but unfortunately all 9,1V zener have positive Tc´s instead of the negative that´s needed. So I looked at NTC´s, JFET´s etc. but without much luck.
After looking a bit longer I found out that smaller values of zeners have negative Tc´s and these can (theoretically) compensate for the changing Vgs.
So I put in two 3.3V zeners in series and changed R24 to 330R.
With 6k8 McMillan resistors this changed the startup value from 12V to 7V, but still had a tendency to change (down) even after hours of playing.
The ambient temp. of the zeners is around 58°C after warmup, the fet (no cooling) is then at 73°C. So the next thing I did was connect the fet to the case to limit the temp to 60°C. (This alone would also have helped reduce the change in abs. offset)
So now the amp starts at around 6V and gets to zero in around an hour. After that it varies between +1.4 and -1.0 for the next few hours.
There are a few more things to play with like the zener voltage (3 x 3,3V =9,9V will give Vgs change less influence because R24 is bigger) or combinations of zeners with different Tc´s. The current through the zeners will also have an influence.....
I will put in the 10k McMillan resistors back in tomorrow and change the current through the zeners from 3mA to 1mA tomorrow (this will keep the zener nearer to the ambient temp and minimise the change in current even further, I hope). Maybe it´s possible to up the resistors to 15k or 20k???
William
Hi,
the last few days I´ve thought a lot about absolute dc offset and measured a few things.
The main cause for the big variation (in my Aleph-X) is the changing of the bias current for the diff pair. (cost me quite some ice-spray......)
The Vgs of the current source fet changes from 3.78V (cold) to 3.58V (73°C). This changes the current from 22.06 to 23.89mA and dc offset from 12.1 to -1V.
So I thought it would be nice to have a current source wich put´s out the same current from 20 to 70°C.
Operating the fet at the zero Tc point was no option cause this is somewhere near 400mA so I thought of putting in a compensating element. The best place for this would be the zener but unfortunately all 9,1V zener have positive Tc´s instead of the negative that´s needed. So I looked at NTC´s, JFET´s etc. but without much luck.
After looking a bit longer I found out that smaller values of zeners have negative Tc´s
and these can (theoretically) compensate for the changing Vgs.So I put in two 3.3V zeners in series and changed R24 to 330R.
With 6k8 McMillan resistors this changed the startup value from 12V to 7V, but still had a tendency to change (down) even after hours of playing.
The ambient temp. of the zeners is around 58°C after warmup, the fet (no cooling) is then at 73°C. So the next thing I did was connect the fet to the case to limit the temp to 60°C. (This alone would also have helped reduce the change in abs. offset)
So now the amp starts at around 6V and gets to zero in around an hour. After that it varies between +1.4 and -1.0 for the next few hours.
There are a few more things to play with like the zener voltage (3 x 3,3V =9,9V will give Vgs change less influence because R24 is bigger) or combinations of zeners with different Tc´s. The current through the zeners will also have an influence.....
I will put in the 10k McMillan resistors back in tomorrow and change the current through the zeners from 3mA to 1mA tomorrow (this will keep the zener nearer to the ambient temp and minimise the change in current even further, I hope). Maybe it´s possible to up the resistors to 15k or 20k???
William
Hi Graeme & William,
Quite interesting findings of yours. Thanks for sharing
With the AX I built, I didn't had much problems with offset, either absolute or relative. At cold start, the absolute offset is at 1.5V, and rapidly (less than 30 minutes) settles under 100mV. Relative one stays below 10mV all the time. I used the TL431 current source used by Hugo (Netlist) for the input diff pair. But I'm cheating a little since the amps are always on (at low bias in standby mode) But the input mosfets are thermally coupled, and matched at two different currents, just like the output mosfets.
The input pair current source topology sure has an influence. May be the one I use has a better thermal behaviour, but I won't bet a horse on it. Something to try could be a LED + BJT current source, with thermal coupling between the led and the bjt. They roughly both have the same thermal dependance, so this could lead to some improvements in stability. I didn't see an AX built with such a CCS... Anyway, thanks William for the deep look on Tc point. Will read your experiments with a great interest.
And Graeme, I partly agree on the breadboard thing I also have a lot a fun, both aesthetic and electrical, designing a PCB. For me, that's part of the journey. Obviously, I make some perfboard tests, but quite generally, my first PCB iteration is the right one. May be not the best, but I can cope with it. I have fun finding the best routing, imagining all the traps I may fall in, and avoiding them. Of course, it's a matter of time, and it takes me a very long time to achieve it. But it helps waiting for all the parts be bought (money is, alas, tight)
Cheers
Quite interesting findings of yours. Thanks for sharing
With the AX I built, I didn't had much problems with offset, either absolute or relative. At cold start, the absolute offset is at 1.5V, and rapidly (less than 30 minutes) settles under 100mV. Relative one stays below 10mV all the time. I used the TL431 current source used by Hugo (Netlist) for the input diff pair. But I'm cheating a little since the amps are always on (at low bias in standby mode)
But the input mosfets are thermally coupled, and matched at two different currents, just like the output mosfets. The input pair current source topology sure has an influence. May be the one I use has a better thermal behaviour, but I won't bet a horse on it
. Something to try could be a LED + BJT current source, with thermal coupling between the led and the bjt. They roughly both have the same thermal dependance, so this could lead to some improvements in stability. I didn't see an AX built with such a CCS... Anyway, thanks William for the deep look on Tc point. Will read your experiments with a great interest. And Graeme, I partly agree on the breadboard thing
I also have a lot a fun, both aesthetic and electrical, designing a PCB. For me, that's part of the journey. Obviously, I make some perfboard tests, but quite generally, my first PCB iteration is the right one. May be not the best, but I can cope with it. I have fun finding the best routing, imagining all the traps I may fall in, and avoiding them. Of course, it's a matter of time, and it takes me a very long time to achieve it. But it helps waiting for all the parts be bought (money is, alas, tight) Cheers
Hi Steen,
I thought the fly analogy would would make an impression. Yes indeed, I would very much love to have a couple of your award winning flies. I will email you. I live very close to an excellent salmon and steelhead river. By the way your trip to Norway sounds wonderful.
Hi Terry,
I agree totally on the matching. I worked hard to achieve the same sort of matching as you did. I forgot to mention it in the opening posts of this thread but I also matched the small resistors R16 & R30, R18 & R28, R19 & R29, and R46 & R47. If you have a bag of 100 pieces @ 1% I found it was really quick and easy to get pairs that are within .1%.
Hi William,
You are doing excellent research and getting great results. I am paying very close attention. I will be interested to see how well the zeners operate at 1 ma in this circuit. That's a pretty low current value.
I have just realized that changing the bias in the output transistors to 540ma has probably changed the AC gain of the Aleph CCS. I need to re-check this.
Hi Cheff,
Let us know if you try any experiments with the MacMillan resistors.
Cheers,
Graeme
I thought the fly analogy would would make an impression. Yes indeed, I would very much love to have a couple of your award winning flies. I will email you. I live very close to an excellent salmon and steelhead river. By the way your trip to Norway sounds wonderful.
Hi Terry,
I agree totally on the matching. I worked hard to achieve the same sort of matching as you did. I forgot to mention it in the opening posts of this thread but I also matched the small resistors R16 & R30, R18 & R28, R19 & R29, and R46 & R47. If you have a bag of 100 pieces @ 1% I found it was really quick and easy to get pairs that are within .1%.
Hi William,
You are doing excellent research and getting great results. I am paying very close attention. I will be interested to see how well the zeners operate at 1 ma in this circuit. That's a pretty low current value.
I have just realized that changing the bias in the output transistors to 540ma has probably changed the AC gain of the Aleph CCS. I need to re-check this.
Hi Cheff,
Let us know if you try any experiments with the MacMillan resistors.
Cheers,
Graeme
Hi Graeme,
checking ac-current-gain would be usefull.
I just turned on my X. Now with 3 x 3V3 zeners and 10k Mc Millan resistors. Will report tomorrow after adjusting to 0V tonight.
William
BTW startup voltage today ( with 2 x 3v3 and 6k8) was 4V, it then went up to 5V and down again to zero (in less than an hour).
checking ac-current-gain would be usefull.
I just turned on my X. Now with 3 x 3V3 zeners and 10k Mc Millan resistors. Will report tomorrow after adjusting to 0V tonight.
William
BTW startup voltage today ( with 2 x 3v3 and 6k8) was 4V, it then went up to 5V and down again to zero (in less than an hour).
Hi William,
Very nice work indeed. Which 3V3 zeners are you using - 1N5226B's?
When you do your tests please try a couple of things for me. First, after playing the amps fairly hard for ten minutes or so, check the absolute DC offset for about 2 to 3 minutes to see if it drifts and by how much. Second, once the amps are up to temperature put a small fan on the heatsinks to see if the DC offset can be changed much by heatsink temperature. If you don't have a fan then just blow on the sinks for 15 to 30 seconds. I am curious to know if you see the same things I described in a previous post. I want to know if perhaps the drift I am seeing is an artifact of my heat sink design. I am not bothered by the drifting, I just want to know if someone else sees the same behaviours I do.
Cheers,
Graeme
Very nice work indeed. Which 3V3 zeners are you using - 1N5226B's?
When you do your tests please try a couple of things for me. First, after playing the amps fairly hard for ten minutes or so, check the absolute DC offset for about 2 to 3 minutes to see if it drifts and by how much. Second, once the amps are up to temperature put a small fan on the heatsinks to see if the DC offset can be changed much by heatsink temperature. If you don't have a fan then just blow on the sinks for 15 to 30 seconds. I am curious to know if you see the same things I described in a previous post. I want to know if perhaps the drift I am seeing is an artifact of my heat sink design. I am not bothered by the drifting, I just want to know if someone else sees the same behaviours I do.
Cheers,
Graeme
Hi,
during use the offset (abs.) varies by aprox. 2V.
Cooling one heatsink with ice spray had no effect on abs. dc offset, cooling the diff pair also no effect.....
Started the amp up this morning, abs. dc was around 8V (after 10 seconds) going down. (this with 10k Mc Millan and 3 x 3V3 zeners)
The zeners are 0,5W 3.3V zeners but I don´t know what make they are. Looking at different spec sheets shows the same behaviour for different makes.
William
during use the offset (abs.) varies by aprox. 2V.
Cooling one heatsink with ice spray had no effect on abs. dc offset, cooling the diff pair also no effect.....
Started the amp up this morning, abs. dc was around 8V (after 10 seconds) going down. (this with 10k Mc Millan and 3 x 3V3 zeners)
The zeners are 0,5W 3.3V zeners but I don´t know what make they are. Looking at different spec sheets shows the same behaviour for different makes.
William
Hi William,
Thank you for running the tests I requested. The 2 volts of drifting is probably acceptable as long as an amplifier has an extra volt of headroom on each voltage rail. Your results look very promising so far
I have a couple of questions:
1) You changed the zeners and the MacMillen resistors at the same time so it's difficult for me to tell if you have improved things and by how much. Are you happy with the three 3.3V zeners and will you leave them in place? Do you recommend others (like me) to make the same mod? Does the V2 adjustment pot still have enough range?
2) Are you using 1 ma of bias on the zeners? If so then why did you make this change and is it working out OK?
Thanks William.
Graeme
Thank you for running the tests I requested. The 2 volts of drifting is probably acceptable as long as an amplifier has an extra volt of headroom on each voltage rail. Your results look very promising so far
I have a couple of questions:
1) You changed the zeners and the MacMillen resistors at the same time so it's difficult for me to tell if you have improved things and by how much. Are you happy with the three 3.3V zeners and will you leave them in place? Do you recommend others (like me) to make the same mod? Does the V2 adjustment pot still have enough range?
2) Are you using 1 ma of bias on the zeners? If so then why did you make this change and is it working out OK?
Thanks William.
Graeme
Hi Graeme,
1. I already tried the 10k resistors with the 1N825A ref diode and had apr. 16V at startup going to -1V and lower (stopped measuring)
Now (with the 3x 3.3V zeners) it starts at 8V (after 20 seconds or so) and is below 2V after 50 minutes. After that it slowly gets to around zero going up and down a volt or so. I think this is OK. R17 is still 2k7 to ground so around 5mA through the zeners.
I had to change R24 and R26 to get the range right (both are 470R now)
The version with 6k8 resistors and 2x3.3V zeners was a bit better starting at 4V and being zero within an hour. I will have to listen a bit now (and change the other channel.....)
2. By trying 1-2mA I wanted to reduce the heating of the zeners by their own bias current. Haven´t done this yet because I´m still not shure if this is the way to go. It could also be that a higher current will give you more compensation. I will probably just have to try something like 2 and 10mA........
I would like to try a few more things before recommending these modifications but a few things are already clear:
1. cooling the current source fet reduces the temperature range it is going through from cold to warm. In my case from 53°C to 40°C, this reduces the current change and therefore the voltage change. The drawback is that it will take longer before it reaches it´s final temperature.
2. choosing a reference voltage around 9-12V is a lot better than around 6-7 volts because it reduces the influence of the changing Vgs on bias current by a factor 4.
3. choosing zeners with a negative temperature coefficient will help to compensate for the changing Vgs
I think that the results are a bit dependent on used voltages and case layout so everyone has to find the optimum for himself.
As soon as I know more I´ll let you know.
William
1. I already tried the 10k resistors with the 1N825A ref diode and had apr. 16V at startup going to -1V and lower (stopped measuring)
Now (with the 3x 3.3V zeners) it starts at 8V (after 20 seconds or so) and is below 2V after 50 minutes. After that it slowly gets to around zero going up and down a volt or so. I think this is OK. R17 is still 2k7 to ground so around 5mA through the zeners.
I had to change R24 and R26 to get the range right (both are 470R now)
The version with 6k8 resistors and 2x3.3V zeners was a bit better starting at 4V and being zero within an hour. I will have to listen a bit now (and change the other channel.....)
2. By trying 1-2mA I wanted to reduce the heating of the zeners by their own bias current. Haven´t done this yet because I´m still not shure if this is the way to go. It could also be that a higher current will give you more compensation. I will probably just have to try something like 2 and 10mA........
I would like to try a few more things before recommending these modifications but a few things are already clear:
1. cooling the current source fet reduces the temperature range it is going through from cold to warm. In my case from 53°C to 40°C, this reduces the current change and therefore the voltage change. The drawback is that it will take longer before it reaches it´s final temperature.
2. choosing a reference voltage around 9-12V is a lot better than around 6-7 volts because it reduces the influence of the changing Vgs on bias current by a factor 4.
3. choosing zeners with a negative temperature coefficient will help to compensate for the changing Vgs
I think that the results are a bit dependent on used voltages and case layout so everyone has to find the optimum for himself.
As soon as I know more I´ll let you know.
William
Hi William,
I am very impressed with the results of your tests. Thank you for answering all my questions.
1. Very good results! I have decided to try some 3.0V diodes so I've ordered some.
2. I don't understand why your zeners are heating up above the ambient in your amp. They are not dissipating much power and the leads should easily be able to carry that away. Have you considered glueing the zeners to the plastic case of the CCS transistor in order to take maximum advantage of the opposing tempcos?
I agree that everyone will have a different implementation and environment and so may need to possibly make their own modifications.
Graeme
I am very impressed with the results of your tests. Thank you for answering all my questions.
1. Very good results! I have decided to try some 3.0V diodes so I've ordered some.
2. I don't understand why your zeners are heating up above the ambient in your amp. They are not dissipating much power and the leads should easily be able to carry that away. Have you considered glueing the zeners to the plastic case of the CCS transistor in order to take maximum advantage of the opposing tempcos?
I agree that everyone will have a different implementation and environment and so may need to possibly make their own modifications.
Graeme
Hi Graeme,
I didn´t measure the zener temps but since they are dissipating power they must be hotter than the ambient temperature.
Noticed something else yesterday:
When the mains voltage changes, absolute dc offset changes! This was to be expected since the drain resistors of the diff pair are connected to the negative rail.
I think I will try a nice dc-servo today. This way I can probably leave out the Mc Millan resistors and up the output resistors.
William
I didn´t measure the zener temps but since they are dissipating power they must be hotter than the ambient temperature.
Noticed something else yesterday:
When the mains voltage changes, absolute dc offset changes! This was to be expected since the drain resistors of the diff pair are connected to the negative rail.
I think I will try a nice dc-servo today. This way I can probably leave out the Mc Millan resistors and up the output resistors.
William
Hi William,
I like your observation about the line voltage affecting the abs DC offset. I think you're right - it's an unregulated supply and any change across the diff pair drain resistors would be amplified by the output stage. I think I'll put a dc meter on the output and another on the minus rail (or an AC meter on the line) and see how the two values track. I live in a rural area and the line voltage regulation here can be poor so this could be a significant issue for me.
Back in post #48 you give the temp of your zener (58C). The value seemed high but maybe it's what it is because of the ambient where it's located.
Graeme
I like your observation about the line voltage affecting the abs DC offset. I think you're right - it's an unregulated supply and any change across the diff pair drain resistors would be amplified by the output stage. I think I'll put a dc meter on the output and another on the minus rail (or an AC meter on the line) and see how the two values track. I live in a rural area and the line voltage regulation here can be poor so this could be a significant issue for me.
Back in post #48 you give the temp of your zener (58C). The value seemed high but maybe it's what it is because of the ambient where it's located.
Graeme
Hi,
today I changed the second amp:
3x 3v3 zeners R24/26 470R, R46/47 10k, R17 2k7
Did a bit of measuring to see if anything has changed:
-3dB 8 Ohm: 142 kHz
-3dB 4 Ohm: 126 kHz
This is more or less the same as befor so the difference in sound is not explained by the frequency responce (as to be expected)
I will leave them like this for the near future.
I had a look at the possibility to put in a dc servo. Should be pretty simple:
I would put in a second gate resistor 220R for Q3 in series with the first one and connect the output of the servo between both. Input for the servo would be ground and output.
Since the voltage at the gate of Q3 is about 12V (supply voltage minus zener voltage) The servo could be connected to +20V and -10V to have enough headroom. Will try this sometime......
William
today I changed the second amp:
3x 3v3 zeners R24/26 470R, R46/47 10k, R17 2k7
Did a bit of measuring to see if anything has changed:
-3dB 8 Ohm: 142 kHz
-3dB 4 Ohm: 126 kHz
This is more or less the same as befor so the difference in sound is not explained by the frequency responce (as to be expected)
I will leave them like this for the near future.
I had a look at the possibility to put in a dc servo. Should be pretty simple:
I would put in a second gate resistor 220R for Q3 in series with the first one and connect the output of the servo between both. Input for the servo would be ground and output.
Since the voltage at the gate of Q3 is about 12V (supply voltage minus zener voltage) The servo could be connected to +20V and -10V to have enough headroom. Will try this sometime......
William
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.