Hi!
I would like some comments of my implementation on a N-channel amplifier. I hope no rules have been broken by borrowing pieces from different amplifiers to assemble this one.
The output is very much similiar to the one used by Nelson Pass in his MOSFET Citation 12 and the input stage is inspired by some of Anthony Holton's schematics. I've also added a DC-Servo to remove the need of a big electrolytic cap in the feedback.
The BD136 on the schematic will be replaced by MJE350, but I did not have the spice implementation of it. A zobel network is also present at the output but not shown in the schematic. My goal was an amplifier that used only parts that I already had = cheap.
I'm exhibiting some crossconduction at high frequencies but much less than in both the Citation 12 and Anthony's N-Channel amp, so I guess this is something that might need to be tweaked once the amplifiers is build?
Feel free to make any comments as i would appreciate that very much!
-Pelle
I would like some comments of my implementation on a N-channel amplifier. I hope no rules have been broken by borrowing pieces from different amplifiers to assemble this one.
The output is very much similiar to the one used by Nelson Pass in his MOSFET Citation 12 and the input stage is inspired by some of Anthony Holton's schematics. I've also added a DC-Servo to remove the need of a big electrolytic cap in the feedback.
The BD136 on the schematic will be replaced by MJE350, but I did not have the spice implementation of it. A zobel network is also present at the output but not shown in the schematic. My goal was an amplifier that used only parts that I already had = cheap.
I'm exhibiting some crossconduction at high frequencies but much less than in both the Citation 12 and Anthony's N-Channel amp, so I guess this is something that might need to be tweaked once the amplifiers is build?
Feel free to make any comments as i would appreciate that very much!
-Pelle
An externally hosted image should be here but it was not working when we last tested it.
Thanks! I'll try to change the value of the collector resistor and/or Q13. I'll probably end up having to change them once the amplifier is done and tested.
-Pelle
The pictures if btw a gif-image that I've put on my server. I did not manage to attach any pictures directly when I posted so I used the 'Insert hyperlink' instead.
-Pelle
The pictures if btw a gif-image that I've put on my server. I did not manage to attach any pictures directly when I posted so I used the 'Insert hyperlink' instead.
To prevent cross-conduction I would try replacing R36 by a lower value resistor or maybe by a higher value resistor and a 10mA current source like the one that charges the gate of M18 in paralell. This would ensure that M19 turns off fast enough
To get big schematic pictures in small manageable files they should be captured as 16 color GIFs or PNG files, never as JPEG
To place images directly in the forum they should be uploaded to your own web-server and referred from the post as images, not urls or attachments
Images placed as attachments are not directly visible from the threads and increase information traffic demands on the web-server of the forum
To get big schematic pictures in small manageable files they should be captured as 16 color GIFs or PNG files, never as JPEG
To place images directly in the forum they should be uploaded to your own web-server and referred from the post as images, not urls or attachments
Images placed as attachments are not directly visible from the threads and increase information traffic demands on the web-server of the forum
Pelle: Nice and straightforward design! However the big problem is (like the Citation 12 clone) that there is really high rail loss. At each rail this design starts clipping at some 10 Volts under the rail.
In practical life this means that to build a 100W amp you will have to use not 44V rail voltage but 54V, to satisfy the amplifier's rail loss. It wastes power like a 180 Watt amplifier, so you power supply and heat sinks need to be size: 180 Watt. Still the amp put out only 100 Watt.
Not a very efficient design.
R36 why not make it 50 Ohms, and then place Q13 on the heat sink as well. R17 could be 10 Ohms. Then it will work without cross conduction.
All the best from
Lars
In practical life this means that to build a 100W amp you will have to use not 44V rail voltage but 54V, to satisfy the amplifier's rail loss. It wastes power like a 180 Watt amplifier, so you power supply and heat sinks need to be size: 180 Watt. Still the amp put out only 100 Watt.
Not a very efficient design.
R36 why not make it 50 Ohms, and then place Q13 on the heat sink as well. R17 could be 10 Ohms. Then it will work without cross conduction.
All the best from
Lars
Thanks for your replies! It's great that there are no bigger faults to the design. All the things you have commented on are easy to adjust even when the PCB is made.
Lars Clausen: Why is this special to this design? M18 is driven as a source follower and would have the same minimum voltagedrop as many other mosfet-amps? According to simulation this design clips at 30V positive and 29V negative, so that's more like 5V at each rail. Will there be any difference when I build the amp?
MikeB: As I said in the begining "The BD136 on the schematic will be replaced by MJE350, but I did not have the spice implementation of it."
-Pelle
Lars Clausen: Why is this special to this design? M18 is driven as a source follower and would have the same minimum voltagedrop as many other mosfet-amps? According to simulation this design clips at 30V positive and 29V negative, so that's more like 5V at each rail. Will there be any difference when I build the amp?
MikeB: As I said in the begining "The BD136 on the schematic will be replaced by MJE350, but I did not have the spice implementation of it."
-Pelle
Pelle said:
Sorry, next time i will read...
Edit: you can download spicemodel here:
http://www.onsemi.com/site/products/summary/0,4450,MJE350,00.html
pro: I don't no what the problem might be. 80dB, that's the openloop gain of this amp, so it sounds like all feedback is gone in your simulation.
MikeB: I've also changed the value of R14 to 22k.
I now have a complete amplifier on my desk and I will hopefully try it tomorrow. An initial test with +-12V showed that no dc was present at the output and that it did amplify the noise made by my finger touching the input. This was without DC-servo and with a blockingcap at the feedback.
I'll get back to you when I know some more.
-Pelle
MikeB: I've also changed the value of R14 to 22k.
I now have a complete amplifier on my desk and I will hopefully try it tomorrow. An initial test with +-12V showed that no dc was present at the output and that it did amplify the noise made by my finger touching the input.
This was without DC-servo and with a blockingcap at the feedback.I'll get back to you when I know some more.
-Pelle
Commentable Thoughts
I agree with the Lars Statement regarding the High Rail Loss.
Perhaps in our design we used push pull drivers and the VAS stage has voltage of about +-65VDC and Driver+Mosfets are operated at 58VDC to get maximum Efficiency.
Secondly Gate resistors must be of values ranging from 15 To 25 ohms to eliminate the cross-conduction at high Frequencies of order@20KHZ.
Thirdly, increase the driver stage current also.
With Regards,
WorkHorse Audio Group
Lars Clausen said:
I agree with the Lars Statement regarding the High Rail Loss.
Perhaps in our design we used push pull drivers and the VAS stage has voltage of about +-65VDC and Driver+Mosfets are operated at 58VDC to get maximum Efficiency.
Secondly Gate resistors must be of values ranging from 15 To 25 ohms to eliminate the cross-conduction at high Frequencies of order@20KHZ.
Thirdly, increase the driver stage current also.
With Regards,
WorkHorse Audio Group
Hi!
I have increased the current through the phase-inverting driver to the same current as in the VAS (approx. 8mA). In simulation I get no crossconduction, even at 500kHz. But I have to measure the actually amplifier to confirm this.
To raise the voltage at the vas is not an option for me as it will increase the cost of the amp, so I have to live with the somewhat bad efficiency.
-Pelle
I have increased the current through the phase-inverting driver to the same current as in the VAS (approx. 8mA). In simulation I get no crossconduction, even at 500kHz. But I have to measure the actually amplifier to confirm this.
To raise the voltage at the vas is not an option for me as it will increase the cost of the amp, so I have to live with the somewhat bad efficiency.
-Pelle
Now I have done the first real test with +-30V. But the result was not what I expected. I got oscillation at 10MHz, about 100mV peak-peak. Didn't have time to test any modifications today, so it has to wait untill next week.
Here is a picture of the schematic I used for testing:
And here is the amplifier:
-Pelle
Here is a picture of the schematic I used for testing:
An externally hosted image should be here but it was not working when we last tested it.
And here is the amplifier:
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
-Pelle
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.