I am posting this to let people know of project under way for the last few months to create a new Leach clone. I am only posting this at this time because of the current group buy going on. I struggled with making the announcement at this time, because I didn't want to appear to be stealing buyers from the other buy. However, this project has been going on for a few months now, so before anyone spends all their budget in the other buy, I wanted to let them know there may be a choice in the future.
This potential group buy is to design a board using the ThermalTrak transistors, using at least four pair. Also, it is intended to provide an input board and an output board to allow the outputs to be spread out more on the available heat sink, and/or to make it easier, with more available options, to mount the boards in an enclosure or heat sink. Further, the boards are intended to accommodate more flexibility in parts selection and terminal spacings, especially for the acknowledged audiophile approved parts.
The board is intended to be an experimenters board that will accommodate several choices of compensation, including 2-pole, phase lead, and others, yet still allow one to make the amp using Leach's scheme(s)
The design is intended to have a more sophisticated protection circuit, yet still accommodate Leach's or Jens' scheme, if desired.
There remains the issue of getting a good thermal compensation scheme going with the use of the ThermalTrak diodes. I have been told that connecting the diodes in the Leach arrangement works, but we are holding out for a possibly better scheme, if possible/practical. We really don't know how long this will take to settle this issue.
There is no definite time line here. There is no established cost yet. I think I can state that the plural boards will not be as inexpensive as the other group buy. There will not be any schematics or board layouts posted here in the near future. If they are posted at all, it will not be until a pretty firm design is near final.
Let it be noted that just about every desired change posted in the other group buy thread has been considered or accommodated in this design before the desired changes were posted.
In sum, if you are in a hurry and don't want any changes, by all means buy the boards in the other group buy. If you want boards with the changes, you may want to hedge your bets and buy a few of the other boards to make sure you get some while available, in case the noise in this thread builds up to the breaking point and the project is abandoned.
I don't imagine these boards being available before the new year at the earliest. Aside from finalizing a design, the time for prototyping and testing is a very firm unknown. I really regret having to make this announcement at this time, because I really don't want to be spending time answering questions about this project right now, or people demanding to know when it will be done.
However, I did want to let people know that there is a potential option in the works so they won't kick themselves later. Sorry, that is all I can report at this time. We are not a commercial enterprize, and don't feel obligated to answer every post made to stay in business. Also, we do not intend to radically depart from Leach's established topology if Leach's basic topology cannot be built with these boards as a firm option, with the possible exception of getting the ThermalTraks involved.
This potential group buy is to design a board using the ThermalTrak transistors, using at least four pair. Also, it is intended to provide an input board and an output board to allow the outputs to be spread out more on the available heat sink, and/or to make it easier, with more available options, to mount the boards in an enclosure or heat sink. Further, the boards are intended to accommodate more flexibility in parts selection and terminal spacings, especially for the acknowledged audiophile approved parts.
The board is intended to be an experimenters board that will accommodate several choices of compensation, including 2-pole, phase lead, and others, yet still allow one to make the amp using Leach's scheme(s)
The design is intended to have a more sophisticated protection circuit, yet still accommodate Leach's or Jens' scheme, if desired.
There remains the issue of getting a good thermal compensation scheme going with the use of the ThermalTrak diodes. I have been told that connecting the diodes in the Leach arrangement works, but we are holding out for a possibly better scheme, if possible/practical. We really don't know how long this will take to settle this issue.
There is no definite time line here. There is no established cost yet. I think I can state that the plural boards will not be as inexpensive as the other group buy. There will not be any schematics or board layouts posted here in the near future. If they are posted at all, it will not be until a pretty firm design is near final.
Let it be noted that just about every desired change posted in the other group buy thread has been considered or accommodated in this design before the desired changes were posted.
In sum, if you are in a hurry and don't want any changes, by all means buy the boards in the other group buy. If you want boards with the changes, you may want to hedge your bets and buy a few of the other boards to make sure you get some while available, in case the noise in this thread builds up to the breaking point and the project is abandoned.
I don't imagine these boards being available before the new year at the earliest. Aside from finalizing a design, the time for prototyping and testing is a very firm unknown. I really regret having to make this announcement at this time, because I really don't want to be spending time answering questions about this project right now, or people demanding to know when it will be done.
However, I did want to let people know that there is a potential option in the works so they won't kick themselves later. Sorry, that is all I can report at this time. We are not a commercial enterprize, and don't feel obligated to answer every post made to stay in business. Also, we do not intend to radically depart from Leach's established topology if Leach's basic topology cannot be built with these boards as a firm option, with the possible exception of getting the ThermalTraks involved.
Split Board
I am interested in both the super amp redesign and the current group buy of Jens's latest redesign of the low TIM amp. However, I am also interested in more designs that allow a separation of the first stages from the output stage as suggested by Jens for the double barreled super amp. I need to be able to use both metal case and plastic case outputs and would like to develop optimized designs for both the low TIM and cascoded super amp front ends. Would you have an interest in this?
I am interested in both the super amp redesign and the current group buy of Jens's latest redesign of the low TIM amp. However, I am also interested in more designs that allow a separation of the first stages from the output stage as suggested by Jens for the double barreled super amp. I need to be able to use both metal case and plastic case outputs and would like to develop optimized designs for both the low TIM and cascoded super amp front ends. Would you have an interest in this?
Pooge,
To let you know, I'm watching with interest. I think there will be "more" GB's of this topology. I am nagged by my own fear of missing "the last one".
Any updates?
To let you know, I'm watching with interest. I think there will be "more" GB's of this topology. I am nagged by my own fear of missing "the last one".
Any updates?
Project got put aside for the last couple of months due to holidays, and other scheduling conflicts, etc. Getting together with an experienced board designer soon to get the ideas down. This may take a few weeks.
We've pretty much abandoned the ThermalTrac idea. Such boards would limit selection of transistors to just two types, and possibly limit interest in the boards for those having other output favorites. Further, the complications discussed in the ThermalTrac thread put a damper on the idea. Also, I'm not convinced that they would make the amp any better. The original idea was to make it better and easier to construct. However, after reading Self's articles on thermal compensation, I'm not sure that an immediate thermal response is a desired goal.
Leach's original design seems to work, although seemingly overcompensated with four diodes, and these are spaced a long thermal distance away from the output transistors. What I would like to do is mount the thermal sensors, in the form of flat pack transistors, on the back/top of the output transistors; either directly via the same bolt hole, or on a thermally conductive plate used as a clamp on the back/top of the output transistors. These could then be directly soldered to the board in the same manner as the output transistors.
This would have a much faster response than the Leach arrangement, and closer to the response of the ThermalTraks, without being too fast, and/or requiring the necessary circuit design to accommodate them to seemingly no thermal advantage.
In Self's articles (as well as I could interpret some of his British slang and manner of delivery), there appears to be a somewhat optimum time delay between the output junction heating and the thermal sensor, according to his models. He found the optimum by modeling a thermal resistance between the output junction and the thermal sensor. This thermal resistance appears to be very close to the thermal resistance of the output transistor package on the back side. Therefore, mounting a sensor on the back of the output transistors does not appear to be such a disadvantage over the ThermalTraks, if any. In any event, it would be faster than Leach's.
A suggested mounting arrangment for a flat pack output transistor is using a clamp on it to avoid warping the transistor's heat plate by over torquing the mounting bolt, and therby losing some contact area with the heat sink. If a strip of copper or stiff aluminum plate was placed across the backs of the output transistors, the sensors could be mounted on this plate, and have a faster thermal response than on the main heat sink. The plate could also serve the dual purpose of keeping the outputs flat against the heat sink. If the output transistors were adequately spaced, the sensors could be placed between pairs of output transistors so that they can be separately mounted on the plate, and not via the same bolt holes as the output transistors. This may make assembly a little easier, as well as disassembly for any reason.
I still anticipate an input board and an output board. I don't want to scrunch all the output transistors on a limited area of the heat sink, and off to one end due to the length of the board. I think that giving them some space would be a benefit to heat removal. Removing input circuitry to another board that requires no relationship to the heat sink would help this. Also, removal of the Zobel and output inductor to another board mounted on the output posts would put them in their ideal location and save real estate on the output board for such other large components, such as rail-to-rail caps suggested by Bob Cordell for limiting the current loop area of the pulse currents in the output stage, by bypassing the loop back to the power supply and ground leads.
I'd like more room for a larger electrolytic cap in the feedback, for a variety of choices for this sonically important cap, and to be able to make it larger in case one wants to reduce the value of the feedback resistors, to help lower noise or to match different values of resistors at the input.
I think that minimizing the ground plane may also be of benefit. Although good in theory, it could be a negative if not done properly. And properly may involve many prototypes. Capacitive coupling from the negative input to the ground can screw up the amp's compensation and cause oscillation, for example. Leach used a ground plane in his original amp and abandoned, it more for practical reasons than anything, and didn't find any detriment. This is further supported by others
here. Therefore, it may be better left out rather than get it wrong.
Instead, I'd like to work to minimize current loops. I'm familier with how to do that in the output stage. I recall reading a post or article somewhere about the loops in the input stages, which is more complicated by such things as compensation paths, etc. If anyone knows of such a post or article, please inform me.
My goals of both making it easier to thermally couple the input transistor pairs and accomodate the Toshiba low noise transistor lead pattern still apply. Also, my goal of making this an experimenter's board to try different compensation schemes suggested by Self and other in some of the other amplifier threads still holds.
I'd still also like to make the input and output boards on one board that can be separated into separate boards if one desires.
We've pretty much abandoned the ThermalTrac idea. Such boards would limit selection of transistors to just two types, and possibly limit interest in the boards for those having other output favorites. Further, the complications discussed in the ThermalTrac thread put a damper on the idea. Also, I'm not convinced that they would make the amp any better. The original idea was to make it better and easier to construct. However, after reading Self's articles on thermal compensation, I'm not sure that an immediate thermal response is a desired goal.
Leach's original design seems to work, although seemingly overcompensated with four diodes, and these are spaced a long thermal distance away from the output transistors. What I would like to do is mount the thermal sensors, in the form of flat pack transistors, on the back/top of the output transistors; either directly via the same bolt hole, or on a thermally conductive plate used as a clamp on the back/top of the output transistors. These could then be directly soldered to the board in the same manner as the output transistors.
This would have a much faster response than the Leach arrangement, and closer to the response of the ThermalTraks, without being too fast, and/or requiring the necessary circuit design to accommodate them to seemingly no thermal advantage.
In Self's articles (as well as I could interpret some of his British slang and manner of delivery), there appears to be a somewhat optimum time delay between the output junction heating and the thermal sensor, according to his models. He found the optimum by modeling a thermal resistance between the output junction and the thermal sensor. This thermal resistance appears to be very close to the thermal resistance of the output transistor package on the back side. Therefore, mounting a sensor on the back of the output transistors does not appear to be such a disadvantage over the ThermalTraks, if any. In any event, it would be faster than Leach's.
A suggested mounting arrangment for a flat pack output transistor is using a clamp on it to avoid warping the transistor's heat plate by over torquing the mounting bolt, and therby losing some contact area with the heat sink. If a strip of copper or stiff aluminum plate was placed across the backs of the output transistors, the sensors could be mounted on this plate, and have a faster thermal response than on the main heat sink. The plate could also serve the dual purpose of keeping the outputs flat against the heat sink. If the output transistors were adequately spaced, the sensors could be placed between pairs of output transistors so that they can be separately mounted on the plate, and not via the same bolt holes as the output transistors. This may make assembly a little easier, as well as disassembly for any reason.
I still anticipate an input board and an output board. I don't want to scrunch all the output transistors on a limited area of the heat sink, and off to one end due to the length of the board. I think that giving them some space would be a benefit to heat removal. Removing input circuitry to another board that requires no relationship to the heat sink would help this. Also, removal of the Zobel and output inductor to another board mounted on the output posts would put them in their ideal location and save real estate on the output board for such other large components, such as rail-to-rail caps suggested by Bob Cordell for limiting the current loop area of the pulse currents in the output stage, by bypassing the loop back to the power supply and ground leads.
I'd like more room for a larger electrolytic cap in the feedback, for a variety of choices for this sonically important cap, and to be able to make it larger in case one wants to reduce the value of the feedback resistors, to help lower noise or to match different values of resistors at the input.
I think that minimizing the ground plane may also be of benefit. Although good in theory, it could be a negative if not done properly. And properly may involve many prototypes. Capacitive coupling from the negative input to the ground can screw up the amp's compensation and cause oscillation, for example. Leach used a ground plane in his original amp and abandoned, it more for practical reasons than anything, and didn't find any detriment. This is further supported by others
here. Therefore, it may be better left out rather than get it wrong.
Instead, I'd like to work to minimize current loops. I'm familier with how to do that in the output stage. I recall reading a post or article somewhere about the loops in the input stages, which is more complicated by such things as compensation paths, etc. If anyone knows of such a post or article, please inform me.
My goals of both making it easier to thermally couple the input transistor pairs and accomodate the Toshiba low noise transistor lead pattern still apply. Also, my goal of making this an experimenter's board to try different compensation schemes suggested by Self and other in some of the other amplifier threads still holds.
I'd still also like to make the input and output boards on one board that can be separated into separate boards if one desires.
Pooge, It seems that you have put a great deal of thought into this project. Have you given any thought to using a separate front end power supply. Constructing the front end board with regulated supply and using a separate output board might have other design advantages in line with your thermal compensation scheme. When you finish this design phase I will definately be in line to purchase 4 boards.
Have you read any of Roender's thread about his Thermal Trak amplifier. He seems to address several excellent points concerning the thermal compensation.
I will be awaiting results of your work Tad
Have you read any of Roender's thread about his Thermal Trak amplifier. He seems to address several excellent points concerning the thermal compensation.
I will be awaiting results of your work Tad
I've given a great deal of thought to a lot of things, but as more of these thoughts end up being incorporated in a competing thread that may end up drying up the pool of group buyers, I think I'll hold those thoughts for now. Suffice it to say I've considered most of the requests that have been brought up, and more. We will build our prototypes to our liking. If there's then enough interest for a GB, we'll do it. If not, I'll still have the boards that I want.
The McIntosh MC2KW uses these transistors. The amplifier comes in three chassis. The two power chassis provide 1KW each. They send current to an autoformer housed in a third chassis. Each power chassis has 48 thermal trak devices. The total for each monaural amplifier is 96.
Following is a link to this on the McIntosh site.
http://www.mcintoshlabs.com/mcprod/...at=Power+Amplifiers&prodid=1115&product=MC2KW
--HK
Following is a link to this on the McIntosh site.
http://www.mcintoshlabs.com/mcprod/...at=Power+Amplifiers&prodid=1115&product=MC2KW
--HK
- Status
- Not open for further replies.