As I create my first amplifier diagram, I eagerly await your feedback

Amplifiers with that many power transistors are not easy to make PCBs for. Making everything align properly on the heat sink and fit into the chassis will make you go through multiple “generations” before everything FITS TOGETHER properly enough to actually be able to build it. Unless you’re going to get all those parts custom manufactured and machined from CAD drawings. And pay thousands of dollars to get it done.

Then it has to work electrically, as you intended it to. Following Selfs 11 cardinal rules concerning EM compatibility get exponentially harder with each pair you add.

It has to be designed together, electrically, mechanically and thermally as a system. Physical construction needs to be worked out, long before a schematic is finalized. I don’t see anything “wrong” with the design, and other than some minor optimization, will WORK, and work safely/reliably if the physical construction is up to snuff.
 
  • Like
Reactions: 1 users
R52 and R53 I would move to 1 ohm
These are driver / outputs you may overload them before the others kick in .
The design here lends itself to 4 1n4148's as the VAS bias circuit with 1 ohm driver resistors.
Your preamp won't pass any signal the first 5532 has no resistor in its feedback 2 *27k would work. Your volume pot is in wrong spot. Reverse position with 10k feedback resistor.
220 ohm amp feedback to 1k.
 
Last edited:
  • Like
Reactions: 1 user
R25, R26, D11, and D12 are in the wrong place. See Leach's schematic.

The abundance of 100 ohm resistors tells me that the designer did not know how to calculate the values, or whether the resistor was necessary. The amplifier probably works, but not very well.

This board has quite a few well-designed amplifiers that you can build.
Ed
 
  • Like
Reactions: 1 user
Transistors 2SD669, 2SC5171, 2SA1930, 2SC4793, 2SA1837, 2SC2705, 2SA1145 are all obsolete and hard to obtain. I strongly advise not to use any of those you can buy on eBay or Aliexpress. Those are all fake. There may be some that are genuine, but the chance of getting them is slim. In the place of 2SC5171/2SA1930, you can use TTC011B/TTA006B pair. Others may have suggestions for the substitute for 2SC2705/2SA1145 pair.

MJL21193G and MJL21194G have limited fT at 4MHz. Why not use 2SC5200/2SA1941 (or current models, TTC5200/TTA1941) instead since you are already using them?

By using lateral MOSFET (like ECW20P20 / ECW20N20 pair from Exicon, 200V, 16A, 250W parts), you can worry less about the thermal management of the output stage. Since this is your first project, you may want to reduce some issues that are dependent on execution of the design, like where and how to mount the transistors for the thermal compenstation of the output stage.

I would make the input stage transistors cascoded at this voltage.

As others have mentioned, you may want to consider the value and neccesities of some 100 ohm resistors. Some are not required and some should be in different values.

Good luck and enjoy your first build!

Satoru
 
  • Like
Reactions: 1 user
Amplifiers with that many power transistors are not easy to make PCBs for. Making everything align properly on the heat sink and fit into the chassis will make you go through multiple “generations” before everything FITS TOGETHER properly enough to actually be able to build it. Unless you’re going to get all those parts custom manufactured and machined from CAD drawings. And pay thousands of dollars to get it done.

Then it has to work electrically, as you intended it to. Following Selfs 11 cardinal rules concerning EM compatibility get exponentially harder with each pair you add.

It has to be designed together, electrically, mechanically and thermally as a system. Physical construction needs to be worked out, long before a schematic is finalized. I don’t see anything “wrong” with the design, and other than some minor optimization, will WORK, and work safely/reliably if the physical construction is up to snuff.
I much appreciate your warm comments.
Localy made an amplifire chassis (50 cm x 20 cm x 8 cm) will cost you between $12 and $15. I've included a picture of the chassis.
They will build the chassis to my specifications if I pay extra. Additionally, heat sinks are accessible, and I was able to get one that will enable me to create the PCB the correct size.
 

Attachments

  • cabinets.jpg
    cabinets.jpg
    10.4 KB · Views: 39
R52 and R53 I would move to 1 ohm
These are driver / outputs you may overload them before the others kick in .
The design here lends itself to 4 1n4148's as the VAS bias circuit with 1 ohm driver resistors.
Your preamp won't pass any signal the first 5532 has no resistor in its feedback 2 *27k would work. Your volume pot is in wrong spot. Reverse position with 10k feedback resistor.
220 ohm amp feedback to 1k.
point acknowledged; I'll share it once again after revisions. I much appreciate your warm comments.
 
R25, R26, D11, and D12 are in the wrong place. See Leach's schematic.

The abundance of 100 ohm resistors tells me that the designer did not know how to calculate the values, or whether the resistor was necessary. The amplifier probably works, but not very well.

This board has quite a few well-designed amplifiers that you can build.
Ed
point acknowledged; I'll share it once again after revisions. I much appreciate your warm comments.
 
@satour, I appreciate your feedback. You are 100% correct on transistors. I'll take your recommendations into account. The MJW3281A/MJW1302A power transistor (copy-paste issue) will be used. I've made a list of several transistors, and I'll include your suggested transistor as well.

Driver Transistors

NPNPNPhFE minPdVcb/VceIcHzpF
LTP/200MHz
2N55512N5401800.31W180V0.6A100 MHz
Pre-Driver/150MHz
2N55512N5401800.31W180V0.6A100 MHz6
C2240A9702000.3W120V0.1A50 MHz3
A92A42700.055W18V0.005A20 MHz
VAS/Bias/100MHz
C2705A1145800.8W150V0.05A200 MHz1.8
2SC34232SC1360705W150V0.05A200 MHz2.5
D669B649601W120V1.5A140MHz14
MJE340MJE3503020W300V0.5A10 MHz
BD139BD1404012W80V1A50 MHz
Driver 1st/30MHz
TIP 31CTIP 32C2040W140V3A3 MHz
TIP 41CTIP 42C2065W140V6A3 MHz
C4793A183710050W230V1A100 MHz20
2SC51712SA193010020W180V2A200 MHz26
MJE15032

MJE15034

MJE15028

MJE15030
MJE15033

MJE15035

MJE15029

MJE15031
5050W250V

350V

120V

120V
8A30 MHz
C2073A9404025W150V1.5A1MHz70
Driver 2nd/30MHz
C5198194155100W160V10A30 MHz220

Power Transistor


NPNPNPhFEhFEPdVcb/VceIcHzpFBrand
TTC5200TTL1943 80-160150W230V15A30 MHz145Toshiba
2SC52002SA1943 55-160150W250V15A30 MHz200Toshiba
2SC59492SA2121 55-160220W200V15A30 MHz270Toshiba
2SC51982SA1941 55-160100W160V10A30 MHz320Toshiba
MJW3281AMJW1302A 50-125200W230V15A30 MHz600ON Semi
NJW0281GNJW0302G86/7975-150150W250V15A30 MHz400ON Semi
MJL21194MJL2119343/7825-75200W400V16A4 MHz500ON Semi
NJL3281DNJL1302D 75-150200W260V15A30 MHz600ON Semi
MJL4281AMJL4302A 80-250230W350V15A35 MHz600ON Semi
 
There is nothing wrong with 4 MHz output transistors if you are using them in a triple EF with sustained beta high fT parts in both of the other positions. It will likely give you less trouble with stability with that many spread out all over the place. Fast parts don’t like long distributed buses with lots of inductance that you cannot remove. Keep the base stoppers on the outputs, but drop them to 1 or 2 Ohms. You need something, but not much. Bump the Rbe to between 5 and 20 ohms to increase the current in the C5200 (or C5242, Fairchild’s easier to find version of C5198) drivers. That will also improve switch off time in the 4 MHz outputs.
 
  • Like
Reactions: 1 user
Hi there, The schematic could not be completed without your help. Many thanks.
I attempted to include your suggestions into the diagram (Attachment).
I think the Q13/15 & Q14/16 are a darlington pair. Should I keep these two or swap them out for 2SC517 & 2SA1930?

Thanks.
opu
 

Attachments

  • new_amp_01.png
    new_amp_01.png
    355.4 KB · Views: 59
Last edited:
With due respect, that is not design but a mishmash of bits and pieces copypasted from all over the world.

I can not kludge together a Ferrari transmission, a Trabant chassis, 4 various sized wheels, a carburetor, a Diesel engine, and say: "hey, look at the car I designed", even if all those are car parts and work fine in their original designs.

Errors here and there, some gross, such as the zero gain input buffer (meaning NO signal will get through), the potentially infinite gain (or at least open loop) following next stage, the 100 ohm input impedance, etc.

As suggested above, start studying the basics, then go step by step.

As in there are some 15-20 steps to reach this level of complexity.
 
I use discrete darlingtons in VAS. With the first collector tied to ground and with Rbe set to force about 2 mA in the first transistor. If you do that you can use beefy second transistors and not worry about a 25 or 30 pF Cob ruining things.
It sounds like Sziklai Darlington, huh? I don't know a lot about electronics. I bought some Chinese PCBs with a watts range of 100–300, which is how I found out about amplifiers. Drawing the PCB diagram is my first task before testing those. Then I gathered a few online schematics.
and I observed that more or less all are the same, so I thought, why not boost the wattage by adding a power transistor? I had no idea that this required a lot of Math! A few publications, such as "Douglas Power amp design handbook" and "Designing Audio Power Amplifier" by Bob Cordell, I also download."Many things I don't understand, and few things I do!!
I greatly appreciate you sharing your knowledge with me.
 
With due respect, that is not design but a mishmash of bits and pieces copypasted from all over the world.

I can not kludge together a Ferrari transmission, a Trabant chassis, 4 various sized wheels, a carburetor, a Diesel engine, and say: "hey, look at the car I designed", even if all those are car parts and work fine in their original designs.

Errors here and there, some gross, such as the zero gain input buffer (meaning NO signal will get through), the potentially infinite gain (or at least open loop) following next stage, the 100 ohm input impedance, etc.

As suggested above, start studying the basics, then go step by step.

As in there are some 15-20 steps to reach this level of complexity.
Hello, you may not be familiar with the guy from Bangladesh known as "Leepu." His full name is Leepu Nizamuddin Awlia. You can search for him on Google by typing in "leepu and pitbull" or "chop shop london garage."
You will be captivated by his tale.
I am attempting to learn the process of building an amplifier.
Thanks
 
  • Like
Reactions: 1 user
Hello, you may not be familiar with the guy from Bangladesh known as "Leepu." His full name is Leepu Nizamuddin Awlia. You can search for him on Google by typing in "leepu and pitbull" or "chop shop london garage."
You will be captivated by his tale.
I am attempting to learn the process of building an amplifier.
Is this post (and the thread in general) some sort of AI learning process step/result?
Post a huge amount of unrelated words and see "who bites the hook"?
 
  • Like
Reactions: 1 user
Is this post (and the thread in general) some sort of AI learning process step/result?
Post a huge amount of unrelated words and see "who bites the hook"?
Oh, sorry, I am in part responsible , at least for the car related bit.

I compared copypasting random circuit bits to assembling wild cars out of completely unrelated parts.

Lo and behold, there is such a guy, Leepu, successfully assembling wild cars using that method.

what-car-is-this-Leepu.jpg


He hosts a successful "reality" TV show in none less than History Channel.

Even funnier, he is Bangladeshi like Mr Opusadat :eek:

Proof of "small world", "reality beats fiction" and so on :D
Clearly a gifted Mechanic.

To see other cars he made:
 
  • Like
Reactions: 1 user
Every single amplifier I’ve ever built has been a hodge-podge of “things I found somewhere”. The only complete fails over the last decade or two have been “the damn thing won’t fit together”, forcing me to start over. No, having uncontrollable oscillations using a pair of TIP142/7 outputs per channel doesn’t count - I ended up putting in Sankens and it fired up the first time after that - and I’ve been using it for months now. No other design changes, only using $4 outputs instead of $2 ones.
 

Attachments

  • 87922EA8-2164-44D0-8ED5-7677998B2D77.jpeg
    87922EA8-2164-44D0-8ED5-7677998B2D77.jpeg
    70.4 KB · Views: 24