Charles,
These are the values i am currently using.. I will refer to the schematic for reference as to which parts i am talking about.
The intergrator Cap is a 681 ceramic
R28 ( feedback resistor) 5k6. ( this is probably too small) ??
R21,R27 - 4k7
R34 - 330k
R20 - 6k8
C38 - 121 Ceramic
Anyway any thought / suggestions as to better values would be great.
Regards
Peter.
These are the values i am currently using.. I will refer to the schematic for reference as to which parts i am talking about.
The intergrator Cap is a 681 ceramic
R28 ( feedback resistor) 5k6. ( this is probably too small) ??
R21,R27 - 4k7
R34 - 330k
R20 - 6k8
C38 - 121 Ceramic
Anyway any thought / suggestions as to better values would be great.
Regards
Peter.
If your designations are still the same as on the schematic of your 2nd post then R28 (parallel to integrator C) is much too low since you have almost no NFB that way. A value around 100k would be a good value to start with.
For the integrator cap I'd use 330pF. I would leave the rest as is if you really want that huge gain of 70.
The above calculations are valid for carrier-based operation using a 2Vpp triangle of 250 kHz and feedback takeoff before the output filter. I'd try it that way first and then change the NFB network for post-filter takeoff.
Regards
Charles
Pierre, how is your one working BTW ?
For the integrator cap I'd use 330pF. I would leave the rest as is if you really want that huge gain of 70.
The above calculations are valid for carrier-based operation using a 2Vpp triangle of 250 kHz and feedback takeoff before the output filter. I'd try it that way first and then change the NFB network for post-filter takeoff.
Regards
Charles
Pierre, how is your one working BTW ?
Charles,
I have made those changes. It is now running sooooo well... I havn't yet run it though RMAA. But the sound quality is fantastic. Bass, Mid and trebble resonse are REALLY good.
Even when the amplifier is clipping it still sounds really good. I will fix up my dummy load and try and get some measurements with RMAA at near full power into 4 and 2 ohms.
Thank you for helping me, It has made such a difference.
Regards
Peter
I have made those changes. It is now running sooooo well... I havn't yet run it though RMAA. But the sound quality is fantastic. Bass, Mid and trebble resonse are REALLY good.
Even when the amplifier is clipping it still sounds really good. I will fix up my dummy load and try and get some measurements with RMAA at near full power into 4 and 2 ohms.
Thank you for helping me, It has made such a difference.
Regards
Peter
Congratulations if you are happy with the results. We hope to see some RMAA results. Is the feedback taken now BEFORE filter?
If you have read the thread I started, I was very happy also with this mode, but wanted to have a load-independent freq. response, so I started to think about POST-filter feedback.
I am going to post some more info in this thread today.
Best regards
If you have read the thread I started, I was very happy also with this mode, but wanted to have a load-independent freq. response, so I started to think about POST-filter feedback.
I am going to post some more info in this thread today.
Best regards
Thank you. I am still trying to determine why i am getting such A high thd & imd reading in RMAA, with only just a loop back cable from speaker out to line in.. Until i sort that out my readings will all be thrown off...
Yes I am now taking my feedback from before the output filter. I may attempt to move it later by setting up a PID as charles has suggested.
However I did my first test run at 2 ohms (5 minutes) and all is stable. Although my Fets are starting to get a bit toasty. I think i need lower values of gate resistors. ( currently 8 ohms).
I measured the power output while running just at clipping into 2 ohms on that run, I managed to 1650Watts RMS. I am really happy with this, Considering how small the circuit is and how little heatsinking there really is.
I will post RMAA results soon..
Regards
Peter.
Yes I am now taking my feedback from before the output filter. I may attempt to move it later by setting up a PID as charles has suggested.
However I did my first test run at 2 ohms (5 minutes) and all is stable. Although my Fets are starting to get a bit toasty. I think i need lower values of gate resistors. ( currently 8 ohms).
I measured the power output while running just at clipping into 2 ohms on that run, I managed to 1650Watts RMS. I am really happy with this, Considering how small the circuit is and how little heatsinking there really is.
I will post RMAA results soon..
Regards
Peter.
In order to get valid results with RMAA, you must ensure you get the maximum from your soundcard in loopback mode. Use a cable that you know is good.
RMAA is very tricky with the levels, if you don't get the right one, you will get too high distortion or the program will warn you telling that it has too low signal. Use the "Adjust I/O levels" button, and select "loopback test". In the version I have (v5.3), it then shows a spectrum analyzer and level bars. Open the Windows volume controls and adjust both line in and wave out / master out so you get the correct level.
Have in mind that RMAA allows you to change the test frequency under the "Test Options -> test signals" menu. Start with 1 KHz and sort things right with your soundcard alone.
When you insert the amplifier, be sure to have a voltage divider / potentiometer so you can get the best signal. You will have to play with the levels also to get the optimum noise level.
I will do some measurements next week with my amp, also, so if I find something useful I will post it here or the other thread.
Have luck and keep us posted, please!
Best regards.
RMAA is very tricky with the levels, if you don't get the right one, you will get too high distortion or the program will warn you telling that it has too low signal. Use the "Adjust I/O levels" button, and select "loopback test". In the version I have (v5.3), it then shows a spectrum analyzer and level bars. Open the Windows volume controls and adjust both line in and wave out / master out so you get the correct level.
Have in mind that RMAA allows you to change the test frequency under the "Test Options -> test signals" menu. Start with 1 KHz and sort things right with your soundcard alone.
When you insert the amplifier, be sure to have a voltage divider / potentiometer so you can get the best signal. You will have to play with the levels also to get the optimum noise level.
I will do some measurements next week with my amp, also, so if I find something useful I will post it here or the other thread.
Have luck and keep us posted, please!
Best regards.
Hello Pierre / phase_accurate,
I have just run some more tests with RMAA.
The IMD and THD have improved tremendously.
Have a look at
http://tesla.reidconsulting.com.au/results/Comparison.htm
At 230Watts RMS the THD is about 0.4% and IMD is about 0.3%. I think these reading should actually be less as the THD of my sound card is already 0.191 %
Pierre, The fets i used were IRFP260N's 50A, 200V HexFets.
Regards
Peter
I have just run some more tests with RMAA.
The IMD and THD have improved tremendously.
Have a look at
http://tesla.reidconsulting.com.au/results/Comparison.htm
At 230Watts RMS the THD is about 0.4% and IMD is about 0.3%. I think these reading should actually be less as the THD of my sound card is already 0.191 %
Pierre, The fets i used were IRFP260N's 50A, 200V HexFets.
Regards
Peter
One thing i just noticed from the plots is that My high frequency -3db point is quiet low. To shift this Up is it just a matter of reducing the inductance of my output filter / using a smaller bypass capacitor on the output, currently it is 1.3uF. Output Inductor is a 103 , 10A.
Regards
Peter
Regards
Peter
Hi,
your results look nice, much better than before. Congratulations!
I did not think over the mathematical description, but the output capacitance depends on the load, you want to connect. A higher impedance of the speaker needs a smaller output cap, IMHO.
BTW, in contrast to a prior post anywhere in this classD-forum I had better experiences using stranded wire (120x0,1) on the output inductor (RM10, N48, Al 250nH) than solid one. The coil remained cool. But I reached only rail voltages up to 25V indeed, before the stage was killed by shoot through. Maybe, the things change a lot with growing rail voltages.
Regards, Timo
your results look nice, much better than before. Congratulations!
I did not think over the mathematical description, but the output capacitance depends on the load, you want to connect. A higher impedance of the speaker needs a smaller output cap, IMHO.
BTW, in contrast to a prior post anywhere in this classD-forum I had better experiences using stranded wire (120x0,1) on the output inductor (RM10, N48, Al 250nH) than solid one. The coil remained cool. But I reached only rail voltages up to 25V indeed, before the stage was killed by shoot through. Maybe, the things change a lot with growing rail voltages.
Regards, Timo
Congratulations, the results are much better now.
About the measurements, I think that either your sound card is really crappy or that you can adjust levels so you can get much lower loopback distortion. With a normal SB-64 I get about 0.007% distortion, if I remember ok.
About the cutoff frequency, if you have taken feeback before filter, it is mainly determined by it (provided that your previous stages are not limiting it already). What exact coil part number are you using (103 is the value, 10uH, I assume?). With 10uH and 1.3uF at 3.3ohms you should have about 60KHz -3dB bandwidth with some peaking at 35KHz. (use 680nF instead). You will have a great load dependency though.
If the coil is designed for 10A, it is too low (at 125Vpp on 3.3ohms you have almost 19A peak), so it can be greatly contributing to distortion due to saturation at higher power levels. Although you can correct this to SOME extent with post-filter feedback, it is better to dimension the coil correctly.
The mosfets are quite good in terms of Rds(on), but they have a high gate charge and capacitance, what rise/fall times are you getting at the output of the mosfets? You are not using any dead-time adjustment, apart from the 47ohms||diode in the gates. Dead-time has a great impact on THD, perhaps you can get better results by lowering that 47 ohms (but be sure that the amp works reliable, test the amp. during hours, because you can kill your mosfets if dead-time is too low)
Your project looks promising, try to optimize feedback a bit more (first tune up your soundcard) and I would recommend to change the coil and go for post-filter NFB.
Could you post a photo of your board, if you don't mind?
Best regards,
Pierre
About the measurements, I think that either your sound card is really crappy or that you can adjust levels so you can get much lower loopback distortion. With a normal SB-64 I get about 0.007% distortion, if I remember ok.
About the cutoff frequency, if you have taken feeback before filter, it is mainly determined by it (provided that your previous stages are not limiting it already). What exact coil part number are you using (103 is the value, 10uH, I assume?). With 10uH and 1.3uF at 3.3ohms you should have about 60KHz -3dB bandwidth with some peaking at 35KHz. (use 680nF instead). You will have a great load dependency though.
If the coil is designed for 10A, it is too low (at 125Vpp on 3.3ohms you have almost 19A peak), so it can be greatly contributing to distortion due to saturation at higher power levels. Although you can correct this to SOME extent with post-filter feedback, it is better to dimension the coil correctly.
The mosfets are quite good in terms of Rds(on), but they have a high gate charge and capacitance, what rise/fall times are you getting at the output of the mosfets? You are not using any dead-time adjustment, apart from the 47ohms||diode in the gates. Dead-time has a great impact on THD, perhaps you can get better results by lowering that 47 ohms (but be sure that the amp works reliable, test the amp. during hours, because you can kill your mosfets if dead-time is too low)
Your project looks promising, try to optimize feedback a bit more (first tune up your soundcard) and I would recommend to change the coil and go for post-filter NFB.
Could you post a photo of your board, if you don't mind?
Best regards,
Pierre
Hi Pierre,
Yes, I will take photos soon and post them. The circuit boards are just home made pcb's. I am putting in a PCB panel to be manufactured sometime in the next few weeks, so once i have sorted my design i will route up a board that will be all SMD, it should be pretty small too..
I the rise / fall time of my mosfet output stage is currently 67/65 nS. Swinging from -80v to +80v. So its not too slow. btw, my gate resistors are now 8R2. No I am not using any more dead time control other than the diode || resistor.
With regards to moving the feedback post filter, I want to do that but i have to do a bit more research into the PID side of things, I might see if charles can help me out a bit with that.
I might have to make up a suitable output inductor out of a little powdered torroid i guess. Btw that output inductor is a 104, 100uH , I might have it wrong in my schematic.
Ok, I will try using a 680nF cap
Regards
Peter Reid.
Yes, I will take photos soon and post them. The circuit boards are just home made pcb's. I am putting in a PCB panel to be manufactured sometime in the next few weeks, so once i have sorted my design i will route up a board that will be all SMD, it should be pretty small too..
I the rise / fall time of my mosfet output stage is currently 67/65 nS. Swinging from -80v to +80v. So its not too slow. btw, my gate resistors are now 8R2. No I am not using any more dead time control other than the diode || resistor.
With regards to moving the feedback post filter, I want to do that but i have to do a bit more research into the PID side of things, I might see if charles can help me out a bit with that.
I might have to make up a suitable output inductor out of a little powdered torroid i guess. Btw that output inductor is a 104, 100uH , I might have it wrong in my schematic.
Ok, I will try using a 680nF cap
Regards
Peter Reid.
Ok. Then your filter cutoff freq. is 6.2 KHz. You must reduce the coil to something in the 15-40uH range and re-calculate the capacitor depending on the load range you plan to use it. If you need help on this, tell me.
Have you tried to deliver a lot of power (<300w) during a long time? How do your mosfets feel? Are you using a large heatsink?
Best regards
Have you tried to deliver a lot of power (<300w) during a long time? How do your mosfets feel? Are you using a large heatsink?
Best regards
Okay I thought my output inductor was a little big... It would be good to get some help with working out how many turns of wire i'll need on the powered iron core.. The Core ID = 25mm OD = 40mm
I have run the amplifier continously at 1100W RMS output. heatsink is only a little warm after an hour, fets are same temp as heatsink... Yep am using quite a large heatsink.. 0.55'c/W
Regards
Peter
I have run the amplifier continously at 1100W RMS output. heatsink is only a little warm after an hour, fets are same temp as heatsink... Yep am using quite a large heatsink.. 0.55'c/W
Regards
Peter
Well, the number of turns depends on the core material. You need to know the AL value in order to calculate it. Have a look at micrometals site. They have good iron power toroids and even a software to calculate the parameters for all their cores.
I am currently using a Wilco drum core, you can test with that also, for example a 27uH one rated at 20A. However, I will move to a toroidal for EMI reasons.
Mmmm, 0.55 K/W is quite a large heatsink! Have you measured the overall efficiency of the amplifier?
Best regards,
Pierre
I am currently using a Wilco drum core, you can test with that also, for example a 27uH one rated at 20A. However, I will move to a toroidal for EMI reasons.
Mmmm, 0.55 K/W is quite a large heatsink! Have you measured the overall efficiency of the amplifier?
Best regards,
Pierre
Yep, I understand that. I beleive mine permiability of this torroid core is about 75mu not 100% sure though. So I guess i'll need around 25 turns I should be safe.
Yes, I realise the heatsink is a bit big, but is useful in the design / testing process. Once i have made my new pcbs i will use a smaller heatsink so it will fit into a 1 Rack unit case.
I am yet to measure the effeciency of the amplifier.
Regards
Peter.
Yes, I realise the heatsink is a bit big, but is useful in the design / testing process. Once i have made my new pcbs i will use a smaller heatsink so it will fit into a 1 Rack unit case.
I am yet to measure the effeciency of the amplifier.
Regards
Peter.
Can anyone suggest a size of wire that would be appropriate for a 30uH Inductor that is going to be running from +/- 80V. And is going to be run at 2 ohms ...
I was thinking something around 1.8mm
I am going to wind 25-30 turns on a powdered iron toroid core i have that have a relative permability of 75
Thanks.
Peter.
I was thinking something around 1.8mm
I am going to wind 25-30 turns on a powdered iron toroid core i have that have a relative permability of 75
Thanks.
Peter.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.