Lars, how did you measure the cross conduction (or the lack of)?
Did you actually monitor the currents from each half and subtract them from each other?
Did you actually monitor the currents from each half and subtract them from each other?
Re: table of power values
Hi Andrew,
Thanks for your comments. I do not have enough knowlege or experience to say the tables are correct. I created a spreadsheet based on a link
<http://www.signaltransfer.freeuk.com/powerout.htm>
I mentioned in a prior post, #64:
<http://www.diyaudio.com/forums/showthread.php?postid=473700#post473700>
If you concur the formulas of the link above I based my values on as being correct, then I will go and recheck that I implemented the math correctly in the spreadsheet. I am quite ok with spreadsheets and have created far more complex ones. It is possible in the cut and past of the rows I messed soemthing up and need to fix any such error in cut and paste of translation of formulas to the spreadsheet. That said, either I implemented the math wrong or I simply need to have better understadn and accounting of PSU losses. As I mentioned in post #64, I will likely open a tread on PSU calculations/losses for different PSU topologies. I did do some spot checking of my PSU results with Lars table and I seemed close. I assumed my differences were result of different, and likely incorrect, loss factor and RMS Constant values. If that is case, once I feel I have no more questions or discussion about Lar's amp design, then I will decide on next new thread I need to open should I not find what I need to know in existing threads on PSU's. All I am trying to do is not let things get too off topic in this thread.
I do appreciate your points Andrew, and I am most welcome to hear them. I believe I eluded to fact in my #64 post and smaller sense my last post, #78, I may not have right formulas, or as research shown me depends on amp class which formulas and constants are used.
I just do not have the time with the learning and research curve for this level of Audio to try to start new thread on PSU, let alone I already think I may need to start a few threads for what I want to find out if I cannot find out in other treads. The research is very time consuming for me at least as I have much more to read, cross reference and sort out the technical knowledge and any implied assumptions the information makes on application use.
I hope you can understand that despite what appears to be intelligent and informed questions or understanding I am am very new to this level of theory. The intelligent and informed postings I have made have beed with much research on my part so I do in fact ask intelligently and be able to understand the kind and experienced answers of others. I am sure I have much to learn, and I am doing my best to do so and pace the amount of information I have to digest. It is hard as there are many good points and matters of theory and practice raised in diyAudio, as well as many devoted diy personal pages. You have no idea how I wish it did not take so much research and learning. Fact is it is only way and I am not lazy at all. I want to be through in what I do as I want to make sure I di not make teh common mistakes/pitfalls, while also being able to perpahs pool experience and knowledge from various sources to create a superset design/implementation. For example the PSU can have one or two bridges, or one can have a PSU for gain/input stage and one for the output driver stage. One can have a choke or not in the PSU. I do not want to discuss any of those in this thread. My point is I have long list of things to learn, and bit by bit I get through it through a combination of research, questions to forums and/or opening threads as time permits for me.
All I ask is patience, and if one feels there is a misguided or incorrect piece of information to say so, that way I know I need to do more research and learning.
Regards,
John L. Males
Willowdale, Ontario
Canada
17 September 2004 06:15
Hi Andrew,
Thanks for your comments. I do not have enough knowlege or experience to say the tables are correct. I created a spreadsheet based on a link
<http://www.signaltransfer.freeuk.com/powerout.htm>
I mentioned in a prior post, #64:
<http://www.diyaudio.com/forums/showthread.php?postid=473700#post473700>
If you concur the formulas of the link above I based my values on as being correct, then I will go and recheck that I implemented the math correctly in the spreadsheet. I am quite ok with spreadsheets and have created far more complex ones. It is possible in the cut and past of the rows I messed soemthing up and need to fix any such error in cut and paste of translation of formulas to the spreadsheet. That said, either I implemented the math wrong or I simply need to have better understadn and accounting of PSU losses. As I mentioned in post #64, I will likely open a tread on PSU calculations/losses for different PSU topologies. I did do some spot checking of my PSU results with Lars table and I seemed close. I assumed my differences were result of different, and likely incorrect, loss factor and RMS Constant values. If that is case, once I feel I have no more questions or discussion about Lar's amp design, then I will decide on next new thread I need to open should I not find what I need to know in existing threads on PSU's. All I am trying to do is not let things get too off topic in this thread.
I do appreciate your points Andrew, and I am most welcome to hear them. I believe I eluded to fact in my #64 post and smaller sense my last post, #78, I may not have right formulas, or as research shown me depends on amp class which formulas and constants are used.
I just do not have the time with the learning and research curve for this level of Audio to try to start new thread on PSU, let alone I already think I may need to start a few threads for what I want to find out if I cannot find out in other treads. The research is very time consuming for me at least as I have much more to read, cross reference and sort out the technical knowledge and any implied assumptions the information makes on application use.
I hope you can understand that despite what appears to be intelligent and informed questions or understanding I am am very new to this level of theory. The intelligent and informed postings I have made have beed with much research on my part so I do in fact ask intelligently and be able to understand the kind and experienced answers of others. I am sure I have much to learn, and I am doing my best to do so and pace the amount of information I have to digest. It is hard as there are many good points and matters of theory and practice raised in diyAudio, as well as many devoted diy personal pages. You have no idea how I wish it did not take so much research and learning. Fact is it is only way and I am not lazy at all. I want to be through in what I do as I want to make sure I di not make teh common mistakes/pitfalls, while also being able to perpahs pool experience and knowledge from various sources to create a superset design/implementation. For example the PSU can have one or two bridges, or one can have a PSU for gain/input stage and one for the output driver stage. One can have a choke or not in the PSU. I do not want to discuss any of those in this thread. My point is I have long list of things to learn, and bit by bit I get through it through a combination of research, questions to forums and/or opening threads as time permits for me.
All I ask is patience, and if one feels there is a misguided or incorrect piece of information to say so, that way I know I need to do more research and learning.
Regards,
John L. Males
Willowdale, Ontario
Canada
17 September 2004 06:15
SonnyA: The cross conduction test was first performed with a single pair of IRF640N and later with a single pair of IRFP250N. (200W / 400W RMS). So the currents would obviously have to be multiplied by the number of output pairs. Still an increase to double the value of BIAS current is not a problem.
PerAnders: I removed the load and Zobel network (no signs of instability).
Then i applied signal to run the amplifier at clipping level, and finally tuned the frequency to the points listed, while measuring the current flowing from + to - rails. (Since no loading was in contact with the output terminals or GND).
Rajeev Luthra:
1..No i performed the tests at +/- 60V rails. But since capacitances drop
with higher Vds, i would guess it's not getting any worse even at higher voltages.
2..I would say it's >200 Ampere. You will typically see a 1/5
charging cycle, so your peak current is Idc * 5 * 1.41.
3..Hmm good question. The sound qualities of this design is open and dynamic, but not ultra detailled. It has perfect properties for use as a PA amplifier, it has a happy and engaging sound to it.
But can not compare with a fast non-feedback design such as The End Millennium, when it comes to High End properties like detailling, space and placement in sound stage.
Just the use of paralleled output devices, will destroy most detailling in any amplifier, MOSFET or BJT. So for hifi: use single MOSFET output devices, you get an improvement to a decent level of performance, but i still wouldn't call it 'high end'. Though well on level with many of the conventional audio kit's found (normal speed < 500 kHz/ with feedback loops).
The shown design is in the category: 'PA high power amplifier', and as shown does not have high-end audio capabilities. The design can be modified to meet audio specs, but my best bet is you can't meet both uses in any one design. If you were to try, the same thing would happen as happens every time you try to make a design have too many uses:
It would end up doing everything equally bad.
If you want to want to move 10 tons of goods, use a truck, if you want to haul on the freeway at 120 MPH, use a Porsche. But the truck won't run 120 MPH any more than the 10 tons will fit into the back seat of the Porsche 😉
I am working on an 'audio adaption' of the N-Channel amplifier and may make it into some kind of product in the future, if the sound can be on high enough level.
All the best from
Lars
PerAnders: I removed the load and Zobel network (no signs of instability).
Then i applied signal to run the amplifier at clipping level, and finally tuned the frequency to the points listed, while measuring the current flowing from + to - rails. (Since no loading was in contact with the output terminals or GND).
Rajeev Luthra:
1..No i performed the tests at +/- 60V rails. But since capacitances drop
with higher Vds, i would guess it's not getting any worse even at higher voltages.
2..I would say it's >200 Ampere. You will typically see a 1/5
charging cycle, so your peak current is Idc * 5 * 1.41.
3..Hmm good question. The sound qualities of this design is open and dynamic, but not ultra detailled. It has perfect properties for use as a PA amplifier, it has a happy and engaging sound to it.
But can not compare with a fast non-feedback design such as The End Millennium, when it comes to High End properties like detailling, space and placement in sound stage.
Just the use of paralleled output devices, will destroy most detailling in any amplifier, MOSFET or BJT. So for hifi: use single MOSFET output devices, you get an improvement to a decent level of performance, but i still wouldn't call it 'high end'. Though well on level with many of the conventional audio kit's found (normal speed < 500 kHz/ with feedback loops).
The shown design is in the category: 'PA high power amplifier', and as shown does not have high-end audio capabilities. The design can be modified to meet audio specs, but my best bet is you can't meet both uses in any one design. If you were to try, the same thing would happen as happens every time you try to make a design have too many uses:
It would end up doing everything equally bad.
If you want to want to move 10 tons of goods, use a truck, if you want to haul on the freeway at 120 MPH, use a Porsche. But the truck won't run 120 MPH any more than the 10 tons will fit into the back seat of the Porsche 😉
I am working on an 'audio adaption' of the N-Channel amplifier and may make it into some kind of product in the future, if the sound can be on high enough level.
All the best from
Lars
Ok, Lars but this was maybe a simplification of the measurement. The best way to determine this is to attach an oscilloscope with either a current probe (for those lucky people which have one 🙂 ) or via a small current shunt. Much of your currents were probably caused by parasite capacitances.
Per Anders: I did use a small shunt, but not to connect a scope. I connected a true RMS multimeter, in DC mode, which i think is fully adequate to determine whether there is any sign of cross conduction. (Which was the point of my investigation).
I agree that the rise in Iq may well be from output capacitances of the MOSFET's. (Some 5 nF each).
If you are right in this assumption, it shows there is no cross conduction at all 😉
All the best from
Lars
I agree that the rise in Iq may well be from output capacitances of the MOSFET's. (Some 5 nF each).
If you are right in this assumption, it shows there is no cross conduction at all 😉
All the best from
Lars
I imagine that you will kid yourself when you don't see the "off" level which you can't pick with anything else than an oscilliscope.
Suppose that you measure in the positive rail and have perfect conduction for positive half period and the absolutely zero for the negative. This will create DC which is perfectly normal. If you connect an oscilloscope you will be able to see how much the "off" level changes.
Suppose that you measure in the positive rail and have perfect conduction for positive half period and the absolutely zero for the negative. This will create DC which is perfectly normal. If you connect an oscilloscope you will be able to see how much the "off" level changes.
Peranders: I agree if you are trying to spot some parts that have a turn-off delay, when under heavy load. Which i don't have in this setup. But just to make sure i will conduct the test you propose, and post here later.
🙂
🙂
I repeated the test now using an oscilloscope to determine if any cross conduction was to be found under load conditions, as suggested by PA. Current was measured in the negative rail
using a small non inductive shunt resistor, and a 100 MHz scope.
At full power (200 Watts) @ 10 kHz, less than 10 mA of cross conductive current was found in the deadzone.
At 100 kHz, the figure increased to 80 mA.
This setup uses a single pair of the slow and highly capacitive
IRFP260 (old version). It's reverse and output capacitances match up with 4 pairs of IRFP250N (!!!)
using a small non inductive shunt resistor, and a 100 MHz scope.
At full power (200 Watts) @ 10 kHz, less than 10 mA of cross conductive current was found in the deadzone.
At 100 kHz, the figure increased to 80 mA.
This setup uses a single pair of the slow and highly capacitive
IRFP260 (old version). It's reverse and output capacitances match up with 4 pairs of IRFP250N (!!!)
Lars Clausen said:
Would you be willing to send me one also? I would pay for shipping and other charges if needed. Even if I could just get a PCB that would be great. It doesn't even have to be assembled, thats the best part 🙂
Let me know if you could.
Thank you very much!
I would also like to get pcb drawing when u will make it, so i can etch my pcbs..😀
regards,
Jure
regards,
Jure
Lars,
Thanks , I am a small PA operator and also repair audio gear , actually I was looking for a design with Irfp250s as they are easily avilible here in India and due to their bulk use in inverters they are cheaper and more easily avilible than bipolar devices . Now as you have said " It has perfect properties for use as a PA amplifier, it has a happy and engaging sound to it . " I have decided to to make your amp , I hope it will sound good in sub bass too ? I was not very confident regarding the stability of mosfet output stages , in simple designs that can be attempted by DIY guys , but you have proved me wrong ,
As the Irfp250s are 200v devices I think the maximum voltage
transformer I can safely use with these devices is 63v-0-63v , how much MFD filter caps , and VA transformer would you recomend for a stereo version of the Zeta if I were to drive 780w or more per ch at 4ohms ( as per your chart ) .
If you give me the GO signal I will complete it within a month , I am not an Amp designer but I can make good hand painted pcbs , and rest of the layouts of a well designed amp project .
Thanks , I am a small PA operator and also repair audio gear , actually I was looking for a design with Irfp250s as they are easily avilible here in India and due to their bulk use in inverters they are cheaper and more easily avilible than bipolar devices . Now as you have said " It has perfect properties for use as a PA amplifier, it has a happy and engaging sound to it . " I have decided to to make your amp , I hope it will sound good in sub bass too ? I was not very confident regarding the stability of mosfet output stages , in simple designs that can be attempted by DIY guys , but you have proved me wrong ,
As the Irfp250s are 200v devices I think the maximum voltage
transformer I can safely use with these devices is 63v-0-63v , how much MFD filter caps , and VA transformer would you recomend for a stereo version of the Zeta if I were to drive 780w or more per ch at 4ohms ( as per your chart ) .
If you give me the GO signal I will complete it within a month , I am not an Amp designer but I can make good hand painted pcbs , and rest of the layouts of a well designed amp project .
Rajeev Luthra: I think you will like this amplifier! for this high pwer, use a 2 kVA transformer, and 47.000 uF per rail. (Totally 2 pcs.).
Others: The PCB i'm going to make will be slightly different than the proposed circuit. It is using a single MOSFET output stage. As you might know i am not in the PA field (even though i actually started out building PA amplifiers about 20 years ago).
So the PCB will be with single output devices, and power from 50 to 500 Watts in 8 Ohms. The output devices can range from IRFP150N to i.e. APT50M65LLL, which is actually good for up 1000 Watts RMS, with the right cooling. Even bigger single devices are available.
But the PCB does NOT have 6 pairs of IRFP250N in the output stage.
Others: The PCB i'm going to make will be slightly different than the proposed circuit. It is using a single MOSFET output stage. As you might know i am not in the PA field (even though i actually started out building PA amplifiers about 20 years ago).
So the PCB will be with single output devices, and power from 50 to 500 Watts in 8 Ohms. The output devices can range from IRFP150N to i.e. APT50M65LLL, which is actually good for up 1000 Watts RMS, with the right cooling. Even bigger single devices are available.
But the PCB does NOT have 6 pairs of IRFP250N in the output stage.
PerAnders: I finally got a DC hooked up, so here are some photos of the currents flowing in the negative rail.
The first one is taken with full output, over a 8 Ohms resistor load. In series with the negative rail is a 0R27 resistor to measure current flow. You can see the amplitude is 1.3 - 1.4 Vp so this gives a peak current of 5 Ampere. I suspect my load resistor climbs a little bit above 8 Ohms at the 10 kHz.
And here is the bottom line blown up to 5 mV (19 mA) per div. The small arrow on the left side of the screen is 0 mA.
At 100 kHz you can see the 8 Ohms load resistor starting to climb seriously. It is a large 300 Watt wirewound resistor, and it's not exactly non inductive 😉 But still no sign of any serious cross conduction.
The scope was halted when shooting, to avoid the load resistor from leaving burnmarks on the workbench 😀
The first one is taken with full output, over a 8 Ohms resistor load. In series with the negative rail is a 0R27 resistor to measure current flow. You can see the amplitude is 1.3 - 1.4 Vp so this gives a peak current of 5 Ampere. I suspect my load resistor climbs a little bit above 8 Ohms at the 10 kHz.
An externally hosted image should be here but it was not working when we last tested it.
And here is the bottom line blown up to 5 mV (19 mA) per div. The small arrow on the left side of the screen is 0 mA.
An externally hosted image should be here but it was not working when we last tested it.
At 100 kHz you can see the 8 Ohms load resistor starting to climb seriously. It is a large 300 Watt wirewound resistor, and it's not exactly non inductive 😉 But still no sign of any serious cross conduction.
An externally hosted image should be here but it was not working when we last tested it.
The scope was halted when shooting, to avoid the load resistor from leaving burnmarks on the workbench 😀
And here some pics of the setup:
You can see the big load resistor to the left. Other than that, transformer, rectifier, caps, and the amplifier PCB with added N channel output. The PCB is 35 by 45 mm ( 1.4by 1.8 inches).
A closeup of the output stage (with the IRFP260's) The P-channel
driver used here is a IRF9640, it's glued on the heat sink with a piece of kapton insulator. There are no insulators between the IRFP's and the anodized heat sink. The Al_2O_3 surface of a heat sink has a dielectric strength of 16.7 kV per mm. Since the surfaca layer is 10 microns it can withstand up to 167 Volts not counting the thermal grease layer that will provide a few extra microns of distance to the aluminium.
And here the back of the board with the folded cascode symmetrical voltage amplifier, drivers etc. The source resistor (white) is only mounted to enable current measurements, and not nessescary in the working amplifier.
An externally hosted image should be here but it was not working when we last tested it.
You can see the big load resistor to the left. Other than that, transformer, rectifier, caps, and the amplifier PCB with added N channel output. The PCB is 35 by 45 mm ( 1.4by 1.8 inches).
An externally hosted image should be here but it was not working when we last tested it.
A closeup of the output stage (with the IRFP260's) The P-channel
driver used here is a IRF9640, it's glued on the heat sink with a piece of kapton insulator. There are no insulators between the IRFP's and the anodized heat sink. The Al_2O_3 surface of a heat sink has a dielectric strength of 16.7 kV per mm. Since the surfaca layer is 10 microns it can withstand up to 167 Volts not counting the thermal grease layer that will provide a few extra microns of distance to the aluminium.
An externally hosted image should be here but it was not working when we last tested it.
And here the back of the board with the folded cascode symmetrical voltage amplifier, drivers etc. The source resistor (white) is only mounted to enable current measurements, and not nessescary in the working amplifier.
Looks good! The IRF9640 is not too big for the phase inverter? I guess that IRF9610/20 is enough. In this position the current, and the dissipation is not too high, and lower power devices means lower capacitance.
Your load resistor looks nice😀
I'm not brave enough to forshake the insulators. 10micron is too thin in my point of view. Easy to destroy it with some scratch...
sajti
Your load resistor looks nice😀
I'm not brave enough to forshake the insulators. 10micron is too thin in my point of view. Easy to destroy it with some scratch...
sajti
Sajti
I think to use IRF9620 the value of10ohm resistor will have to be increased to decrease the dissipation in this device , which may alter the performance of the amp
I think to use IRF9620 the value of10ohm resistor will have to be increased to decrease the dissipation in this device , which may alter the performance of the amp
rajeev luthra said:
10 ohm resistor means about 300mA, which is 15W if the rails are 50V. IRF9620 can handle this power, so not necessary to increase the value of the mentioned resistor.
sajti
N channel amp!!
Hi Lars!!,
I was looking for a design with Irfp250s, This is one of the best topology I have seen so far. And I have decided to to make your amp , I hope it will sound good.
I have some questions about your amp...
1. I think the C10 +ve is connected wrongly Bet'n R54&C11.
2. Are all the resistors 1/4 watts except 0.33 ohms (5 watts).
3.can I use higher power & ft devises Like 2SB649,2SD669 for T12, T13.
4. Do I need to place T11, T12& T13 towards the heatsink mounted with IRFP250 OR do I need to Put a small heat sink for T12,T13 & T14.
5. I think you have already said this but I would like to ask again if I can use IRFP640 for T14, Instead of IRFP9240.
6. This Might be silly,But In your schematic There is no Drain,Source indicaters on the devices, if you can clarify this.
7. How much Bias current need to be provided for this amp.
8. Finally the most important if you can propose a short circuit protection.
Thanks for the simply great design.
Arasuk
Hi Lars!!,
I was looking for a design with Irfp250s, This is one of the best topology I have seen so far. And I have decided to to make your amp , I hope it will sound good.
I have some questions about your amp...
1. I think the C10 +ve is connected wrongly Bet'n R54&C11.
2. Are all the resistors 1/4 watts except 0.33 ohms (5 watts).
3.can I use higher power & ft devises Like 2SB649,2SD669 for T12, T13.
4. Do I need to place T11, T12& T13 towards the heatsink mounted with IRFP250 OR do I need to Put a small heat sink for T12,T13 & T14.
5. I think you have already said this but I would like to ask again if I can use IRFP640 for T14, Instead of IRFP9240.
6. This Might be silly,But In your schematic There is no Drain,Source indicaters on the devices, if you can clarify this.
7. How much Bias current need to be provided for this amp.
8. Finally the most important if you can propose a short circuit protection.
Thanks for the simply great design.
Arasuk
In post 92 Lars has written the following ;-
(( the PCB will be with single output devices, and power from 50 to 500 Watts in 8 Ohms. The output devices can range from IRFP150N to i.e. APT50M65LLL, which is actually good for up 1000 Watts RMS, with the right cooling. Even bigger single devices are available. ))
Lars
You really are a GENIOUS , 1000 watts from single pair of output
devices , here you have taken the ACTUAL advantage of N-channel , as only in N-channel such high power devices are avilible .
For amps up to 100w I would prefer complementry output stage , as it would not make much differance in the cost .
(( the PCB will be with single output devices, and power from 50 to 500 Watts in 8 Ohms. The output devices can range from IRFP150N to i.e. APT50M65LLL, which is actually good for up 1000 Watts RMS, with the right cooling. Even bigger single devices are available. ))
Lars
You really are a GENIOUS , 1000 watts from single pair of output
devices , here you have taken the ACTUAL advantage of N-channel , as only in N-channel such high power devices are avilible .
For amps up to 100w I would prefer complementry output stage , as it would not make much differance in the cost .
I was just wondering if the n-channel design by Lars (exceptional otherwise, of-course) is a bit lower in efficiency compared to other Class A/AB designs?