Hello Folks,
I have an Adcom GFA 5500 power amp for many years. There is nothing wrong with this beautiful amp. However, after reading many articles written by hi-fi mentors I decide to upgrade this amp with a new circuit.
This circuit is a combination of various legendary ones designed by different gurus.
The input stage utilized JFETs in accordance with Mr. John Curl doctrine: JFETs sound more sophisticate. As JFET has low BVdss (about 30V) cascode is unavoidable. The gain (VAS) stage resembles that of Mr. Randy Slone's Optimos. Although in his original circuit a cascode stage was used, it is substitued by MosFET for simplicity. The output characteristics were similar except the former had a current input while the latter used voltage input. The output stage is Mr. Nelson Pass favaorite. It is comprised of 6 complementary MosFETs pairs connected in parallel.
I have the intention of connecting a resistor across source terminals of the input JFETs but this would violate the super symmetry copyright/patent of Mr. Nelson Pass. Only with his kind permission would this modification be implemented.
This circuit is by no means complete. There are components without values. As I do not own any SPICE software those values will be figured out by trial and error. I need your help to fill in the blanks.
Please feel free to make any suggestions, alterations, modifications or corrections to the circuit.
I anticipate your valuable help.
with kind regards,
circuitscc
I have an Adcom GFA 5500 power amp for many years. There is nothing wrong with this beautiful amp. However, after reading many articles written by hi-fi mentors I decide to upgrade this amp with a new circuit.
This circuit is a combination of various legendary ones designed by different gurus.
The input stage utilized JFETs in accordance with Mr. John Curl doctrine: JFETs sound more sophisticate. As JFET has low BVdss (about 30V) cascode is unavoidable. The gain (VAS) stage resembles that of Mr. Randy Slone's Optimos. Although in his original circuit a cascode stage was used, it is substitued by MosFET for simplicity. The output characteristics were similar except the former had a current input while the latter used voltage input. The output stage is Mr. Nelson Pass favaorite. It is comprised of 6 complementary MosFETs pairs connected in parallel.
I have the intention of connecting a resistor across source terminals of the input JFETs but this would violate the super symmetry copyright/patent of Mr. Nelson Pass. Only with his kind permission would this modification be implemented.
This circuit is by no means complete. There are components without values. As I do not own any SPICE software those values will be figured out by trial and error. I need your help to fill in the blanks.
Please feel free to make any suggestions, alterations, modifications or corrections to the circuit.
I anticipate your valuable help.
with kind regards,
circuitscc
An externally hosted image should be here but it was not working when we last tested it.
You need to move R17 & R22 to between the output stage and the driver stage, not in the middle of the driver stage....
And you don't really need R8 and R13....
And maybe a temperature compensated bias circuit....
Otherwise it looks simple and good.
Soren
And you don't really need R8 and R13....
And maybe a temperature compensated bias circuit....
Otherwise it looks simple and good.
Soren
Soren,
Good catch! R17 and R22 should be put between the driver and the output stages. When the Adcom GFA 5500 was reviewed I found that there were 2 sets of +/- supplies. The lower voltage high capacity rails fed the output stage while a slightly higher +/- supply catered the input and driver stages. That is to say R17, R22, R8 and R13 will be removed from the circuit.
Meanwhile, I'm going over R. Slone's Hi-Power Audio Amp Construction Manual. An update to the component values was made in accordance with his designs.
Thanks again for your valuable input.
with kind regards,
circuitscc
Good catch! R17 and R22 should be put between the driver and the output stages. When the Adcom GFA 5500 was reviewed I found that there were 2 sets of +/- supplies. The lower voltage high capacity rails fed the output stage while a slightly higher +/- supply catered the input and driver stages. That is to say R17, R22, R8 and R13 will be removed from the circuit.
Meanwhile, I'm going over R. Slone's Hi-Power Audio Amp Construction Manual. An update to the component values was made in accordance with his designs.
Thanks again for your valuable input.
with kind regards,
circuitscc
Since my proposed circuit was posted I received many inputs from Diyers. The majority of them asked me to replicate famous designs by Mr. Nelson Pass, OptiMos by Mr. R. Slone and to some extent, the JC-xx ot ML-xx by Mr. John Curl. Also submitted were original circuits of some commercial designs. While I admired their one of a kind, outstanding, musical and natural sounding amplifiers I am not going to copy and make them available. As this would offend them, infringe their copyrights and violate their patents. Only if you can obtain the designer's consent shall I put the circuit on to a PCB.
When any power amp circuit is analyzed, it eventually boils down to some fundamental circuits and theories. These building blocks and theories were the ones that we learned from books, high school and college. Engineers and designes would select useful ones and skillfully combined them together to suit their goal. Then the amps were presented to us as commercial products, books and amp circuits along with calculated values. The great site Passdiy.com is a representative of the latter through which Nelson (Papa) brings to us numerous beautiful diy projects. Worth to mentioned are the books 'Audio Power Amp Design Handbook' by Mr. Douglas Self, 'Hi-power Audio Amp Construction Manual' and 'The Audiophiles's Project Sourcebook' by Mr. Randy Slone. Also unforgettable is the Leach Amps by Mr. W. Marshall Leach of Georgia IT. Their designs have become today's mainstream.
I'm not here to promote their publications. Instead, to share some of my previous encounters on building their amp circuits. I started to build them when I was in college. Then my first job provided with me the funds needed to continue my diy journey on more powerful amps. Quest for better amps goes on and on while hurdles show up one after the other.
My first project was built on a proto (or universal) circuit board with pads and holes only. It worked well for simple circuits but as complexity grew soldering had become time consuming and prone to errors. Some of these errors had cost me expensive parts. Many projects were half built and shelved forever as a result. Until recently I made a semi-proto board for my self to reduce the mentioned burdens. These boards are custom made for similar circuitry(s) which share a common goal, such as a power amp. That is to say the board will be designed to accommodate a complex amplifier circuit while a builder can decide which component location to be used or omitted. This idea can be illustrated with a circuit which was drafted over the weekend.
The circuit is a combination of various building blocks such as constant current source, differential (long tail) pair, current mirror, VAS, current limiter and an output stage consisted of complementary MosFets. Please submit your own building block if it is not included in this draft. The circuit may look complicate at first but the builder has the option to select the required block and simply replaced the unused ones with a jumper or leave it blank (such as a capacitor). It also provides you with the flexibility to experiment a simple amp in the beginning and to extend the SAME assembly to more complicated designs down the road.
You may notice that in the input stage are complememtary pairs made up of NPN and PNP transistors. Leave those locations blank and complete the connections with resistors or jumpers as appropriate.
In the negative feedback path are excessive resistors/caps not needed for an actual circuit. They are included for testing purposes. This amp can be tested without an output stage. Once the input and driver stages are test OK the redundant parts can be removed. That may save you the risk of destroying your expensive power MosFets if the assembly is first tested with them connected.
Please feel free to add circuits, blocks that are not displayed in my first draft.
My idea is to make a board that can test and accommodate different amp circuits with similar topologies. With the SAME circuit board you can build simple or complicate amps as you wish.
I'm not here to promote their publications. Instead, to share some of my previous encounters on building their amp circuits. I started to build them when I was in college. Then my first job provided with me the funds needed to continue my diy journey on more powerful amps. Quest for better amps goes on and on while hurdles show up one after the other.
My first project was built on a proto (or universal) circuit board with pads and holes only. It worked well for simple circuits but as complexity grew soldering had become time consuming and prone to errors. Some of these errors had cost me expensive parts. Many projects were half built and shelved forever as a result. Until recently I made a semi-proto board for my self to reduce the mentioned burdens. These boards are custom made for similar circuitry(s) which share a common goal, such as a power amp. That is to say the board will be designed to accommodate a complex amplifier circuit while a builder can decide which component location to be used or omitted. This idea can be illustrated with a circuit which was drafted over the weekend.
The circuit is a combination of various building blocks such as constant current source, differential (long tail) pair, current mirror, VAS, current limiter and an output stage consisted of complementary MosFets. Please submit your own building block if it is not included in this draft. The circuit may look complicate at first but the builder has the option to select the required block and simply replaced the unused ones with a jumper or leave it blank (such as a capacitor). It also provides you with the flexibility to experiment a simple amp in the beginning and to extend the SAME assembly to more complicated designs down the road.
You may notice that in the input stage are complememtary pairs made up of NPN and PNP transistors. Leave those locations blank and complete the connections with resistors or jumpers as appropriate.
In the negative feedback path are excessive resistors/caps not needed for an actual circuit. They are included for testing purposes. This amp can be tested without an output stage. Once the input and driver stages are test OK the redundant parts can be removed. That may save you the risk of destroying your expensive power MosFets if the assembly is first tested with them connected.
Please feel free to add circuits, blocks that are not displayed in my first draft.
My idea is to make a board that can test and accommodate different amp circuits with similar topologies. With the SAME circuit board you can build simple or complicate amps as you wish.
circuitscc,
Very nice idea to use different gurus concept in a single amp.
Good for us who can not decide which layout to use. 😎 Maybe to add some Hiraga or JLH flavour too? (kidding)
I can not quite understand why "connecting a resistor across source terminals of the input JFETs" is violating Mr. Pass' su-sy patent? I understood a su-sy concept applies only to a balanced design (two amps in a bridge, as I understand it). Am I wrong with that? Or maybe you already assumed that your circuit will be balanced too? 😕
Very nice idea to use different gurus concept in a single amp.
Good for us who can not decide which layout to use. 😎 Maybe to add some Hiraga or JLH flavour too? (kidding)
I can not quite understand why "connecting a resistor across source terminals of the input JFETs" is violating Mr. Pass' su-sy patent? I understood a su-sy concept applies only to a balanced design (two amps in a bridge, as I understand it). Am I wrong with that? Or maybe you already assumed that your circuit will be balanced too? 😕
Referring to schematics in Post #1:
The J175 is rather noisy. I would just use 2SK170 / 2SJ74 as everyone else, and they are still available in tens of thousands.
For cascoding, you can of course use BJT as shown, but you may also consider using J111/J174 a la Borbely. I have used that very effectively and it makes the circuit much simpler. Here the somewhat higher noise of the J174 does not affect the final output so directly.
For the output stage, you really want to use truely complementary devices, so I would use 2SK1530/2SJ201 there any time. Again they are still available in thousands. There are other devices I might consider, such as UHC Mosfets, but they require thermal compensation, so a bit more complicated.
And since I have suggested Toshiba devices for 2 stages out of three, we might of course also replace the IRF610/9610 with 2SK2013/2SJ313, to complete the (Japanese) landscape.
Have not attempted to change your topology. There is no need to.
😉
BR,
Patrick
The J175 is rather noisy. I would just use 2SK170 / 2SJ74 as everyone else, and they are still available in tens of thousands.
For cascoding, you can of course use BJT as shown, but you may also consider using J111/J174 a la Borbely. I have used that very effectively and it makes the circuit much simpler. Here the somewhat higher noise of the J174 does not affect the final output so directly.
For the output stage, you really want to use truely complementary devices, so I would use 2SK1530/2SJ201 there any time. Again they are still available in thousands. There are other devices I might consider, such as UHC Mosfets, but they require thermal compensation, so a bit more complicated.
And since I have suggested Toshiba devices for 2 stages out of three, we might of course also replace the IRF610/9610 with 2SK2013/2SJ313, to complete the (Japanese) landscape.
Have not attempted to change your topology. There is no need to.
😉
BR,
Patrick
EUVL,
I agree with you that the J1xx Fets can be noisy as they are designed for switching circuits. I am going to use 2SJ102/2Sk246 instead.
I'm also concerned at the gm of the input Fet at high frequency, when the inter-electrode cap of the following MosFet became significant. Any suggestion?
circuitscc
I agree with you that the J1xx Fets can be noisy as they are designed for switching circuits. I am going to use 2SJ102/2Sk246 instead.
I'm also concerned at the gm of the input Fet at high frequency, when the inter-electrode cap of the following MosFet became significant. Any suggestion?
circuitscc
Miljacc,
Your are right. A resistor across the emitters of a differential pair is not a patent of anyone. When I review the manual of a SRS760 spectrum analyzer, the front end had resistors, not just one, across the emitters. Thanks very much for your tip.
circuitscc
Your are right. A resistor across the emitters of a differential pair is not a patent of anyone. When I review the manual of a SRS760 spectrum analyzer, the front end had resistors, not just one, across the emitters. Thanks very much for your tip.
circuitscc
The 2SJ103/2sk246 has way too low gm.
I would stick to cascoded 2SK170/2SJ74.
Also the source degeneration at the JFETs are too high.
See
Distortion in JFET input stage circuits
Patrick
I would stick to cascoded 2SK170/2SJ74.
Also the source degeneration at the JFETs are too high.
See
Distortion in JFET input stage circuits
Patrick
- Status
- Not open for further replies.