Keypunch: Finally i have opportunity to answer your questions.
I copy them here:
1..I would say yes.
2..For 100 Watts in 8 Ohms use a 33-0-33 V AC for 200 Watts
use a 44-0-44 V, for 50 Watts use a 24-0-24 V. For 15 W use a 12-0-12 V transformer.
3..Change C2 to 1 nF for 75 kHz.
4..I would guess there is no significant sound difference between
different IR devices, so it only depends on loadability. Of course you want to keep your devices not too overdimensioned, for low
gate capacitance, and thus lower distorsion in the treble region.
I would say IRFP250N goes a long way. Changing to another brand (cell technology) may also change the sound of the amplifier.
5..Definitely go for 200% loadability. Just to be safe.
6..No a black anodization is (if intact) in fact insulating enough
for low voltages as found here. (< 1 kV).
7..I would say use the IRF9630's as drivers, and forget about
the IRF630's. Unless you want to build amplifiers in the 50W
region.
I hope these answers are helpful.
Best regards
Lars
I copy them here:
1..I would say yes.
2..For 100 Watts in 8 Ohms use a 33-0-33 V AC for 200 Watts
use a 44-0-44 V, for 50 Watts use a 24-0-24 V. For 15 W use a 12-0-12 V transformer.
3..Change C2 to 1 nF for 75 kHz.
4..I would guess there is no significant sound difference between
different IR devices, so it only depends on loadability. Of course you want to keep your devices not too overdimensioned, for low
gate capacitance, and thus lower distorsion in the treble region.
I would say IRFP250N goes a long way. Changing to another brand (cell technology) may also change the sound of the amplifier.
5..Definitely go for 200% loadability. Just to be safe.
6..No a black anodization is (if intact) in fact insulating enough
for low voltages as found here. (< 1 kV).
7..I would say use the IRF9630's as drivers, and forget about
the IRF630's. Unless you want to build amplifiers in the 50W
region.
I hope these answers are helpful.
Best regards
Lars
Keypunch: For your quick reference here is a spreadsheet that
shows output powers for various transformer voltages. Rectifier losses and approximate rail losses are also taken into account.
Obviously the very high powers require the number and breakdown voltage of the MOSFET's to be increased. And the smaller powers allow for a decrease in those same factors.
All the best from
Lars
shows output powers for various transformer voltages. Rectifier losses and approximate rail losses are also taken into account.
An externally hosted image should be here but it was not working when we last tested it.
Obviously the very high powers require the number and breakdown voltage of the MOSFET's to be increased. And the smaller powers allow for a decrease in those same factors.
All the best from
Lars
Thanks for your Answers and Follow Up Chart
Hi Lars,
Thanks for your reply of yesterday (Post #62):
<http://www.diyaudio.com/forums/showthread.php?postid=472824#post472824>
To my initial questions (Post #49). After I saw you reply yesterday I started to do more research, mostly concerning how output power as function of power supply is calculated. I created an quick and dirty spreadsheet based on one set of formulas I found:
<http://www.signaltransfer.freeuk.com/powerout.htm>
I have not had much time to find out if there are as complete and/or possibily decenting points of view on the power supply calculations. If you do not object, I was thinking of opening up a seperate thread or finding one that discusses such things in greater details rather than using this thread for the power supply related calculations. I provided the above link so you have a frame of reference what I found and been doing calculations on using my best novice guess for the various losses and such.
I really think you for your follow up post (#63) today:
<http://www.diyaudio.com/forums/showthread.php?postid=473663#post473663>
I decided I needed to calculate power supply amp requirements for the various impedances as means to determine against Id what limits exist. I was doing some more research from my calculations by looking at the various IRFP Datasheets. Very kind, and thoughtful you did this spreadsheet calculations and identified the overall losses. Makes me wonder if you are reading my mind all the way from Denmark
I do have some other questions I actually had as of the last set I posted, but I wanted to just break things up a bit. A few are less so questions and more a case of some small items missing on the schematic. I need a few days before I can reply with questions and points of likely omission on the schematic.
Thanks again for your answers and follow up spreadsheet power/impedance calculations of today (Post #63). They are very much welcome as I still trying to guess at some of the losses the formulas I am trying to use with the values you have provided in excellent detail in the spreadsheet calculations posted today.
Regards,
John L. Males
Willowdale, Ontario
Canada
13 September 2004 17:46
Hi Lars,
Thanks for your reply of yesterday (Post #62):
<http://www.diyaudio.com/forums/showthread.php?postid=472824#post472824>
To my initial questions (Post #49). After I saw you reply yesterday I started to do more research, mostly concerning how output power as function of power supply is calculated. I created an quick and dirty spreadsheet based on one set of formulas I found:
<http://www.signaltransfer.freeuk.com/powerout.htm>
I have not had much time to find out if there are as complete and/or possibily decenting points of view on the power supply calculations. If you do not object, I was thinking of opening up a seperate thread or finding one that discusses such things in greater details rather than using this thread for the power supply related calculations. I provided the above link so you have a frame of reference what I found and been doing calculations on using my best novice guess for the various losses and such.
I really think you for your follow up post (#63) today:
<http://www.diyaudio.com/forums/showthread.php?postid=473663#post473663>
I decided I needed to calculate power supply amp requirements for the various impedances as means to determine against Id what limits exist. I was doing some more research from my calculations by looking at the various IRFP Datasheets. Very kind, and thoughtful you did this spreadsheet calculations and identified the overall losses. Makes me wonder if you are reading my mind all the way from Denmark
I do have some other questions I actually had as of the last set I posted, but I wanted to just break things up a bit. A few are less so questions and more a case of some small items missing on the schematic. I need a few days before I can reply with questions and points of likely omission on the schematic.
Thanks again for your answers and follow up spreadsheet power/impedance calculations of today (Post #63). They are very much welcome as I still trying to guess at some of the losses the formulas I am trying to use with the values you have provided in excellent detail in the spreadsheet calculations posted today.
Regards,
John L. Males
Willowdale, Ontario
Canada
13 September 2004 17:46
Samuel Jayaraj: The BD135/136 drivers don't get warm at all. In my prototype i have even used SMD versions in SOT23 with no problem (only tested at up to 200 Watts RMS).
You can use also BD139/140.
BC517 does not have to be strictly mounted on the heatsink UNLESS it's a fan cooled amplifier. Then the airflow could lead the feedback heating away from the BC517, and so you would need to mount it on the heatsink. The output setup is quite stable, and will unlikely run away.
pro: The output power and transformer voltages are RMS. Losses may vary a bit, so don't count too much on the last two decimals of the calculation. It is done from this formula:
(VAC - (Rectifier Loss + Rail Loss)^2) / RL. Plain Ohms Law.
VAC = Transformer Secondary Voltage in AC
RL = Load Resistance. I.e. 8 Ohms.
Rajeev: With the 15 Ohms pull-down resistor, with an average of about 6 Volts over it, in the negative phase, you get a Id of 400 mA, and at 98 - 6 V (92V) You will dissipate maximum 37 Watts in it. Since it only conducts half of the time, and then with less than the 92V VDS, you will in practical life only dissipate something like 12-15 Watts in it, as an average. So you should mount it on the heat sink.
Over 700 people have already downloaded the schematic since Thursday
All the best from
Lars
You can use also BD139/140.
BC517 does not have to be strictly mounted on the heatsink UNLESS it's a fan cooled amplifier. Then the airflow could lead the feedback heating away from the BC517, and so you would need to mount it on the heatsink. The output setup is quite stable, and will unlikely run away.
pro: The output power and transformer voltages are RMS. Losses may vary a bit, so don't count too much on the last two decimals of the calculation. It is done from this formula:
(VAC - (Rectifier Loss + Rail Loss)^2) / RL. Plain Ohms Law.
VAC = Transformer Secondary Voltage in AC
RL = Load Resistance. I.e. 8 Ohms.
Rajeev: With the 15 Ohms pull-down resistor, with an average of about 6 Volts over it, in the negative phase, you get a Id of 400 mA, and at 98 - 6 V (92V) You will dissipate maximum 37 Watts in it. Since it only conducts half of the time, and then with less than the 92V VDS, you will in practical life only dissipate something like 12-15 Watts in it, as an average. So you should mount it on the heat sink.
Over 700 people have already downloaded the schematic since Thursday
All the best from
Lars
Mhm, i think the output RMS power with 115v rails is more like something around 760W RMS to 8 ohm load..
(115V - 5V) / sqr. root 2 = ~78 V RMS to load
(78 x 78) / 8 = ~760W RMS
Otherwise the schematics is very interesting (great job Lars ), and maybe ill replace my 2x 1.2kW (@ 4ohm load) amp with bipolars with that one when ill have time
regards
(115V - 5V) / sqr. root 2 = ~78 V RMS to load
(78 x 78) / 8 = ~760W RMS
Otherwise the schematics is very interesting (great job Lars
), and maybe ill replace my 2x 1.2kW (@ 4ohm load) amp with bipolars with that one when ill have timeregards
skaara said:Mhm, i think the output RMS power with 115v rails is more like something around 760W RMS to 8 ohm load..
(115V - 5V) / sqr. root 2 = ~78 V RMS to load
(78 x 78) / 8 = ~760W RMS
Otherwise the schematics is very interesting (great job Lars
regards
And the fact that the voltages listed are already RMS voltages would not perhaps mean that the power is indeed (115-(4+5))*(115-(4+5))/8= some 1400 W?
Hi,
If you get hold of the Hitachi 1977 design you can build a
mirror image of it and use N mosfets for both of the upper
and lower sides of the pushpull. I tried it with the original
design using P mosfets. Moreover, you can get rid of the
usual Vbias pot!
Randy Slone's book has the original design too.
By mirror image I mean you use NPN differentials as input
followed by a VAS which is a NPN differentials.
I don't have my own PC yet so I can't send schematics!
Peace.
If you get hold of the Hitachi 1977 design you can build a
mirror image of it and use N mosfets for both of the upper
and lower sides of the pushpull. I tried it with the original
design using P mosfets. Moreover, you can get rid of the
usual Vbias pot!
Randy Slone's book has the original design too.
By mirror image I mean you use NPN differentials as input
followed by a VAS which is a NPN differentials.
I don't have my own PC yet so I can't send schematics!
Peace.
Thanks Lars
A few questions ;---
1, Did you make all those tests with a supply of +/- 90v DC,
2, What is the max MFD of cap (charging current ) that a 35A bridge rect can handle ,
3, Will this N-channel Mosfet Amp sound better than an amp of the same capacity with bipolar complementry devices in the output ?
A few questions ;---
1, Did you make all those tests with a supply of +/- 90v DC,
2, What is the max MFD of cap (charging current ) that a 35A bridge rect can handle ,
3, Will this N-channel Mosfet Amp sound better than an amp of the same capacity with bipolar complementry devices in the output ?
Hi Samuel,
It's true- I got rid of the bias pot! And it doesn't matter if it's
BJTs, lateral mosfets, or V-mosfets. I'm learning now how to
post a schematic on the web so I can send you guys the
circuit. I want to know if others have found out that the 1977
Hitachi design could be modified this way.
Peace.
It's true- I got rid of the bias pot! And it doesn't matter if it's
BJTs, lateral mosfets, or V-mosfets. I'm learning now how to
post a schematic on the web so I can send you guys the
circuit. I want to know if others have found out that the 1977
Hitachi design could be modified this way.
Peace.
Follow Up Questions
Hi Lars,
Ok, I have some followup questions to some of your answers here adn some new ones. I actually have a few new ones and possible omissions in the schematic so I will break things up for ease to different postings in this thread.
Ok, concerning the above posting:
1) Great, just waht I need in Amp.
2) I will defer general PSU questions for this Amp to a new posting in this thread.
3) Just changing C2 affects both HP and LP -3db points? I may need to experiment with both the HP and LP -3db points. Any chance of providing the formula or advising what/where the formula would for the input design you have can be found?
4) "it only depends on loadability. Of course you want to keep your devices not too overdimensioned, for low gate capacitance, and thus lower distorsion in the treble region."
a) Sorry Lars, I am not sure what is meant by "devices not too overdimensioned". Can you explain? No need for deep technical details. Just enough for someone that knows how to plug into formuals and what goes where in the formulas type answer. I am not assuming it is a formula based answer, just best way I know at moment to guide you on the level of detail I will or not need.
b) Which of Ciss, Coss, Crss (if any) is considered the gate capacitance? Is it a directly listed item on the datasheet? If so, I not been able to figure which item on the datasheet it is. Could you advise me which one is if listed directly, otherwise how I ffind or calculate out?
c) I will assume gate capacitance distortion impact is a function of frequency as well. That being said at what frequency or function of gate capacitance value does it start to impact. I am asking as you will see below in other points and questions I may want to consider alternate IRF devices.
d) I have heard paralleling output devices is considered to have a negative impace on trebel and highs. Would this be related to device gate capacitance? If so how and is it a simple case of know the gate capacitance for the device and then applying basic parallel formula for capacitance to determine the capacitance the load sees?
e) "I would say IRFP250N goes a long way." means provides best overall balance between power and treble/high distortion? What would be the best device for the lowest and cleanest treble/high? Would it matter if the output load was a direct driver connection with no passive crossover or not?
f) For safety margin at higher output I was thinking I would use a single plus and minus set of IRFP260Ns and the (yet to be refined PSU loss calculations) datasheet would suggest for a single set of IRFP260N's (none paralleled) I could have the following safe supply voltage:
Rload Pout Iavg Ipk Vrail Vac Vpk
1.00 1302.05 14.9316 46.9091 52.60 42.00 46.9091
1.50 923.13 10.2655 32.2500 52.60 42.00 48.3750
2.00 714.50 7.8213 24.5714 52.60 42.00 49.1429
4.00 374.89 4.0060 12.5854 52.60 42.00 50.3415
6.00 254.04 2.6926 8.4590 52.60 42.00 50.7541
8.00 192.10 2.0278 6.3704 52.60 42.00 50.9630
For a woofer application I belive I could have 2 parallel sets of either IRFP250N or IRFP260N and be well within safe margins of the above? Am I deducting the information from the device datasheets correctly?
g) For a single set of IRFP250N's or IRFP260N's (none paralleled) the following would be well within the device range:
Rload Pout Iavg Ipk Vrail Vac Vpk
1.00 728.62 11.1698 35.0909 39.60 32.00 35.0909
1.50 516.58 7.6792 24.1250 39.60 32.00 36.1875
2.00 399.83 5.8508 18.3810 39.60 32.00 36.7619
4.00 209.79 2.9968 9.4146 39.60 32.00 37.6585
6.00 142.16 2.0142 6.3279 39.60 32.00 37.9672
8.00 107.50 1.5169 4.7654 39.60 32.00 38.1235
Would you agree?
5) Just to clarify we are both speaking same. To me the way I meant the question to mean and based on your answer I would deduce the 200% loadability means to effect that use device at 0.5 * (Rated) Pd (and I will assume that takes into account Id ratings) in a constant RMS sense and with a PSU design that can deliver enough amps to allow for 0.5 * (Rated) Idm?
6) Why if the black anodization is sufficient insulation that guidance is to use mica for insulation of output power device to heatsink? Does this black anodization not break down over time? One has to dill holes in the heatsink to enable a mechanical connection to heatsink, does this not pose a shorting concern with screw shorting the inner drilled area to the output device tab? Any suggestions beside Mica or a technique that can be used to provide the thermal conduction while insulating the output device tab from the heatsink?
7) I assume the suggestion for using the IRF9630 as drivers is when using the IRF630 as output driver? Using the power supply formulas link I noted in a prior posting and their suggested loss calculation values (for now, with RMS Constant of 1.3) I have the following (ballpark) results (pending better loss calculations) I would believe are acceptable to use the IRF630 instead of the IRFP240N, and using the IRF9630 instead of the IRFP9240:
Rload Pout Iavg Ipk Vrail Vac Vpk
1.00 388.89 8.1603 25.6364 29.20 24.00 25.6364
1.50 275.72 5.6102 17.6250 29.20 24.00 26.4375
2.00 213.40 4.2744 13.4286 29.20 24.00 26.8571
4.00 111.97 2.1894 6.8780 29.20 24.00 27.5122
6.00 75.88 1.4715 4.6230 29.20 24.00 27.7377
8.00 57.38 1.1082 3.4815 29.20 24.00 27.8519
Would you agree in general for loads >= 4 ohms?
Lars, that completes this posting. It is very late and this took me longer than I expected to create the posting, even though I has completed the various calculations with a spreadsheet I created in past few days. Again no hurry for your reply. I suspect it will be a day or two before I can post the two or three otehr postings if questions and/or schematic oversights (cosmetic, not technical, I am not that well versed to know when there is a technical faw in a design)
Regards,
John L. Males
Willowdale, Ontario
Canada
17 September 2004 02:07
Hi Lars,
Ok, I have some followup questions to some of your answers here adn some new ones. I actually have a few new ones and possible omissions in the schematic so I will break things up for ease to different postings in this thread.
Lars Clausen said:
Ok, concerning the above posting:
1) Great, just waht I need in Amp.
2) I will defer general PSU questions for this Amp to a new posting in this thread.
3) Just changing C2 affects both HP and LP -3db points? I may need to experiment with both the HP and LP -3db points. Any chance of providing the formula or advising what/where the formula would for the input design you have can be found?
4) "it only depends on loadability. Of course you want to keep your devices not too overdimensioned, for low gate capacitance, and thus lower distorsion in the treble region."
a) Sorry Lars, I am not sure what is meant by "devices not too overdimensioned". Can you explain? No need for deep technical details. Just enough for someone that knows how to plug into formuals and what goes where in the formulas type answer. I am not assuming it is a formula based answer, just best way I know at moment to guide you on the level of detail I will or not need.
b) Which of Ciss, Coss, Crss (if any) is considered the gate capacitance? Is it a directly listed item on the datasheet? If so, I not been able to figure which item on the datasheet it is. Could you advise me which one is if listed directly, otherwise how I ffind or calculate out?
c) I will assume gate capacitance distortion impact is a function of frequency as well. That being said at what frequency or function of gate capacitance value does it start to impact. I am asking as you will see below in other points and questions I may want to consider alternate IRF devices.
d) I have heard paralleling output devices is considered to have a negative impace on trebel and highs. Would this be related to device gate capacitance? If so how and is it a simple case of know the gate capacitance for the device and then applying basic parallel formula for capacitance to determine the capacitance the load sees?
e) "I would say IRFP250N goes a long way." means provides best overall balance between power and treble/high distortion? What would be the best device for the lowest and cleanest treble/high? Would it matter if the output load was a direct driver connection with no passive crossover or not?
f) For safety margin at higher output I was thinking I would use a single plus and minus set of IRFP260Ns and the (yet to be refined PSU loss calculations) datasheet would suggest for a single set of IRFP260N's (none paralleled) I could have the following safe supply voltage:
Rload Pout Iavg Ipk Vrail Vac Vpk
1.00 1302.05 14.9316 46.9091 52.60 42.00 46.9091
1.50 923.13 10.2655 32.2500 52.60 42.00 48.3750
2.00 714.50 7.8213 24.5714 52.60 42.00 49.1429
4.00 374.89 4.0060 12.5854 52.60 42.00 50.3415
6.00 254.04 2.6926 8.4590 52.60 42.00 50.7541
8.00 192.10 2.0278 6.3704 52.60 42.00 50.9630
For a woofer application I belive I could have 2 parallel sets of either IRFP250N or IRFP260N and be well within safe margins of the above? Am I deducting the information from the device datasheets correctly?
g) For a single set of IRFP250N's or IRFP260N's (none paralleled) the following would be well within the device range:
Rload Pout Iavg Ipk Vrail Vac Vpk
1.00 728.62 11.1698 35.0909 39.60 32.00 35.0909
1.50 516.58 7.6792 24.1250 39.60 32.00 36.1875
2.00 399.83 5.8508 18.3810 39.60 32.00 36.7619
4.00 209.79 2.9968 9.4146 39.60 32.00 37.6585
6.00 142.16 2.0142 6.3279 39.60 32.00 37.9672
8.00 107.50 1.5169 4.7654 39.60 32.00 38.1235
Would you agree?
5) Just to clarify we are both speaking same. To me the way I meant the question to mean and based on your answer I would deduce the 200% loadability means to effect that use device at 0.5 * (Rated) Pd (and I will assume that takes into account Id ratings) in a constant RMS sense and with a PSU design that can deliver enough amps to allow for 0.5 * (Rated) Idm?
6) Why if the black anodization is sufficient insulation that guidance is to use mica for insulation of output power device to heatsink? Does this black anodization not break down over time? One has to dill holes in the heatsink to enable a mechanical connection to heatsink, does this not pose a shorting concern with screw shorting the inner drilled area to the output device tab? Any suggestions beside Mica or a technique that can be used to provide the thermal conduction while insulating the output device tab from the heatsink?
7) I assume the suggestion for using the IRF9630 as drivers is when using the IRF630 as output driver? Using the power supply formulas link I noted in a prior posting and their suggested loss calculation values (for now, with RMS Constant of 1.3) I have the following (ballpark) results (pending better loss calculations) I would believe are acceptable to use the IRF630 instead of the IRFP240N, and using the IRF9630 instead of the IRFP9240:
Rload Pout Iavg Ipk Vrail Vac Vpk
1.00 388.89 8.1603 25.6364 29.20 24.00 25.6364
1.50 275.72 5.6102 17.6250 29.20 24.00 26.4375
2.00 213.40 4.2744 13.4286 29.20 24.00 26.8571
4.00 111.97 2.1894 6.8780 29.20 24.00 27.5122
6.00 75.88 1.4715 4.6230 29.20 24.00 27.7377
8.00 57.38 1.1082 3.4815 29.20 24.00 27.8519
Would you agree in general for loads >= 4 ohms?
Lars, that completes this posting. It is very late and this took me longer than I expected to create the posting, even though I has completed the various calculations with a spreadsheet I created in past few days. Again no hurry for your reply. I suspect it will be a day or two before I can post the two or three otehr postings if questions and/or schematic oversights (cosmetic, not technical, I am not that well versed to know when there is a technical faw in a design)
Regards,
John L. Males
Willowdale, Ontario
Canada
17 September 2004 02:07
Lars Clausen said:Thijs: As a matter fact i did:
With a normalized bias current of 33 mA i measured the following:
1 kHz: 33 mA
10 kHz: 37 mA
100 kHz: 56 mA
200 kHz: 60 mA
300 kHz: 71 mA
400 kHz: 94 mA
500 kHz: 145 mA
At these measurements i adjusted the input signal up, to maintain
full amplitude on the output, even when the gain starts to fall off above 330 kHz.
If i maintained a constant input signal, the bias + cross conduction actually falls off above 250 kHz, because the amplitude starts to slowly decrease at that frequency. So the max. bias + cross conduction is 65 mA. Nothing to worry about
IRF640N devices used in the above test.
A similar test performed with IRFP250N showed no significant difference in the result. (Surprisingly enough).
I would say that lars's design does not suffer from cross conduction in this test, which is very good.
32mA rise at 250KHz is not bad at all considering driving 12 mosfet. which each have drain-source capacitance of ~300pF.
table of power values
Hi keypunch, your tables of values for outputs into various loads has gone awry.
200watts into 8ohms needs a Vpk of 56.6volts and this requires a Vrail of about 56.6 + 4 to 10 =61 to 67 volts when loaded probably 70volts (50vac) on light loading.
The other values for I & P may be OK. but all the Vrail & Vpk values are far too low
regards Andrew T.
Hi keypunch, your tables of values for outputs into various loads has gone awry.
200watts into 8ohms needs a Vpk of 56.6volts and this requires a Vrail of about 56.6 + 4 to 10 =61 to 67 volts when loaded probably 70volts (50vac) on light loading.
The other values for I & P may be OK. but all the Vrail & Vpk values are far too low
regards Andrew T.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.