Hi,
you set the bias current in the output stage with the load disconnected.
When the load is disconnected, the bias current in the upper and lower half MUST EXACTLY balance.
The difference you have measured is either due to including the currents flowing to other parts of the circuit or measurement error.
BTW, most amplifiers draw different supply currents from the +ve & -ve supply rails and the difference flows back through the ground (usually the signal ground-oops).
FET output stages do not work too well with a low bias current. Many designers recommend very much higher output bias currents. The most extreme recommendation I have read is Borbely, >=100mA per output pair and ~=500mA total bias. This requires a very big heatsink.
Have a read/search of Cordell's FET vs BJT thread.
you set the bias current in the output stage with the load disconnected.
When the load is disconnected, the bias current in the upper and lower half MUST EXACTLY balance.
The difference you have measured is either due to including the currents flowing to other parts of the circuit or measurement error.
BTW, most amplifiers draw different supply currents from the +ve & -ve supply rails and the difference flows back through the ground (usually the signal ground-oops).
FET output stages do not work too well with a low bias current. Many designers recommend very much higher output bias currents. The most extreme recommendation I have read is Borbely, >=100mA per output pair and ~=500mA total bias. This requires a very big heatsink.
Have a read/search of Cordell's FET vs BJT thread.
marus said:My finished quasi amp...
I have 2mV DC offset (out-to-gnd) with input conected to GND.
(too bad that when i conect it to computer sound card i get some noise.... but that's my computer fault)
Bias curent 27mV on positive rail and 29mAon negative rail ... i couldn't set them both equal (why?)
Anyway has a beautiful and clear sound...
I wander ... if I increase the bias curent to 100mA ... would it be possible ? ... would it increase the performance ?
Hi Marus,
Congratulations on a successful build. I haven't even tested the prototype yet, so thank you for proving everything is ok. 2mV on the output is a great result.
With regard to the unequal bias current, how are you measuring it? You could try turning it up to 40mA and see if it evens out a bit. I would not go above 50mA though it won't make much difference in performance.
Finally what voltage rails are you using?
Cheers and well done again.
Quasi
Oh Nice!
Nice to hear, your To247 Version is working!
Still I go for the t0220, i personally hope its heaving a better SOA behaviour than the single-pair version.
I'll fit that AMP into a Bi-amped system for the low frequencies.
Can't wait to make my boards.....still I have to wait till the 9th of feb to start the project.
Greetings
Black
Nice to hear, your To247 Version is working!
Still I go for the t0220, i personally hope its heaving a better SOA behaviour than the single-pair version.
I'll fit that AMP into a Bi-amped system for the low frequencies.
Can't wait to make my boards.....still I have to wait till the 9th of feb to start the project.
Greetings
Black
Hi,
2pair of IRFP460 can do 200W into 8ohm @ 60degree phase angle and Tc<=65degC.
4ohm goes to about 300W @ 50degree phase angle but for short term SOA only (about 100mSec).
The rail voltage is modelled at +-61Vdc at the bottom of the ripple so you need to add half ripple to get the reading that your multimeter gives when on full load.
I would expect the rail voltage at the output devices to drop by 2V to 5V compared to the regulated quiescent voltage. It would appear that a regulated supply of between +-63Vdc to +-67Vdc should hit your target. Does this amp need a higher voltage on the source follower side drivers & voltage amp?
Remember that the regulated supply should be designed for the maximum continuous DC current required. All of the peak current into the amp will be met by the post regulator smoothing/decoupling capacitors. This peak current for an Re=6r (8ohm speaker) would be 9Apk per channel. Smoothing caps of about 2mF/Apk to 3mF/Apk usually do for an unregulated supply and may be suitable for your regulated supply. This would be +-36mF to +-54mF shared between the channels.
2pair of IRFP460 can do 200W into 8ohm @ 60degree phase angle and Tc<=65degC.
4ohm goes to about 300W @ 50degree phase angle but for short term SOA only (about 100mSec).
The rail voltage is modelled at +-61Vdc at the bottom of the ripple so you need to add half ripple to get the reading that your multimeter gives when on full load.
I would expect the rail voltage at the output devices to drop by 2V to 5V compared to the regulated quiescent voltage. It would appear that a regulated supply of between +-63Vdc to +-67Vdc should hit your target. Does this amp need a higher voltage on the source follower side drivers & voltage amp?
Remember that the regulated supply should be designed for the maximum continuous DC current required. All of the peak current into the amp will be met by the post regulator smoothing/decoupling capacitors. This peak current for an Re=6r (8ohm speaker) would be 9Apk per channel. Smoothing caps of about 2mF/Apk to 3mF/Apk usually do for an unregulated supply and may be suitable for your regulated supply. This would be +-36mF to +-54mF shared between the channels.
marus said:I used +/- 40V for testing...
I think 2 pairs of IRFP460 will be better for me
Can I reach 200W / 8 ohms with 2 pairs ? What supply I will use ? +/- 65 V ?
Also i think at a swiching power supply K6 of 700W... will be enought for 2 x 200 W class AB ?
With a conventional well built power supply, 65v rails will provide about 170 watts RMS into 8 ohms. You will need about 70 volt rails to get 200 watts. This allows for some power supply sag as well as losses in the amp module of around 6 volts.
As AndrewT suggests a regulated supply of 64 volts will give you 200 watts.
A switching power supply of 700 watts wll be adequate for a stereo pair of 200 watt modules.
Another thing to consider when going for higher powers with the smaller PCB is thermal coupling. On the smaller PCB the output FETs mount on a bracket rather than directly on the heatsink and this means that they will run hotter. In a domestic situation the module will be fine but avoid PA type applications and limit high power bench tests to only a few seconds.
Cheers
Q
undoubtedly. Doesn't need any theoretical proof.BlacK_Chicken said:
Even two 125W devices are more robust than a dual device in a 250W package.
This time one needs to look at the Rth j-c and Rth c-s.
The twin devices are allways cooler running than the dual package. Cooler running = higher SOA. end of proof.
anatech said:
Chris,
Some seem to think the carbon resistors have some sonic value when used for the source resistors whilst using metal film or oxide for the rest of the resistors. Is this more because the carbon are not inductive or is there some other reason why some suggest carbon resistors be used for the source resistors?
Regards,
John L. Males
Willowdale, Ontario
Canada
07 February 2007 09:53
P.S. I am still working on the specs links gents. I have not forgotton. A major schedule change in last week did not afford me the time I thought I would have. jlm
Hi John,
Many people think a lot of things. I can't begin to figure out why. The sort that posts that "I've tried this and it sounds better" is often in error. We are guarantied not to be objective on something we've worked on. Finding friends willing to risk hurting your feelings is also rare.
Carbon composition types are proven to be noisier, proven to drift in value and proven to have a voltage coefficient. The is no way an engineer would use one unless he/she were after this type of distortion. They do well in RF circuits only because they are not inductive by nature, therefore they will be used as gate / grid stoppers and some snubbers in audio.
-Chris
Many people think a lot of things. I can't begin to figure out why. The sort that posts that "I've tried this and it sounds better" is often in error. We are guarantied not to be objective on something we've worked on. Finding friends willing to risk hurting your feelings is also rare.
Carbon composition types are proven to be noisier, proven to drift in value and proven to have a voltage coefficient. The is no way an engineer would use one unless he/she were after this type of distortion. They do well in RF circuits only because they are not inductive by nature, therefore they will be used as gate / grid stoppers and some snubbers in audio.
-Chris
Hi Chris,
I know all to well about the myth vs fact. One reason I want to build my own amps is to do as much as I can to stick to facts and with a DIY build the things that can have the most factual positive result can be done as there is no accountant dreaming $$ or trying to cut $$ on the cost to increase marketing $$ margins.
I asked in case the carbon resistors characteristics might not be as important as a source resistor. I am leaning towards using metal or oxide source resistors to avoid inductance. One idea I had that seems to keep looking like best option is to use three 2W metal film resistors in parallel for the source resistors. Either 1R0 or 1R5 to net out to 0R33 or 0R50 as the source resistance. This is a far less expensive approach than finding and purchasing the 5W non-inductive resistors. This may mean one more modification of the PCB of the many I have made. Oh well. I am starting to get good at making graphics based edits to the board, even if some edits I do can take hours.
Regards,
John L. Males
Willowdale, Ontario
Canada
07 February 2007 11:35
I know all to well about the myth vs fact. One reason I want to build my own amps is to do as much as I can to stick to facts and with a DIY build the things that can have the most factual positive result can be done as there is no accountant dreaming $$ or trying to cut $$ on the cost to increase marketing $$ margins.
I asked in case the carbon resistors characteristics might not be as important as a source resistor. I am leaning towards using metal or oxide source resistors to avoid inductance. One idea I had that seems to keep looking like best option is to use three 2W metal film resistors in parallel for the source resistors. Either 1R0 or 1R5 to net out to 0R33 or 0R50 as the source resistance. This is a far less expensive approach than finding and purchasing the 5W non-inductive resistors. This may mean one more modification of the PCB of the many I have made. Oh well. I am starting to get good at making graphics based edits to the board, even if some edits I do can take hours.
Regards,
John L. Males
Willowdale, Ontario
Canada
07 February 2007 11:35
John,
The problem is that if you have enough space and pad area to accomodate 2 or 3 of them in parallel. Sometimes you just don't or you have to anticipate that when you first design your board. I've tried putting resistors in series and parallel to get what is required in my recent super leach and its no fun doing that. Furthermore, not too many guys like to do the board in the first place.
The problem is that if you have enough space and pad area to accomodate 2 or 3 of them in parallel. Sometimes you just don't or you have to anticipate that when you first design your board. I've tried putting resistors in series and parallel to get what is required in my recent super leach and its no fun doing that. Furthermore, not too many guys like to do the board in the first place.
anatech said:
Chris,
Might you be suggesting that the 5W wirewound resistor perhaps has a better surge rating compared to three carbon film/oxide 2W resistors in parallel?
bigpanda said:
bigpanda,
Ths "stock" quasi PCB likely does not have room on it for 3 X 2W source resistors. With all the editing I have been doing to the PCB, in part to accomodate some part size differences that are larger than what the PCB is able to handle, it is not a big deal for me to edit the PCB to accomodate the extra space required for 3-4 2W resistors in parallel as alternate to the 5W wirewound resistor. The layout of the quasi "classic" PCB lends itself to making such changes with a graphics editing program with reasonable success. The process is time consuming, but the quasi PCB layout happens to enable such changes to be at least practical with a graphics editing program.
Regards,
John L. Males
Willowdale, Ontario
Canada
08 February 2007 15:52
Hi John,
Wirewound do tend to have much better overload ability than any other type. Do you need more overload capacity?
Since there is an easy sub to make, why not take that route? Did you go and have a look? If you find them, you can lookup each datasheet and compare yourself. Don't forget to derate the resistors you intend to use as they will not be running at ambient temperature with what you are suggesting.
-Chris
I haven't a clue.
Wirewound do tend to have much better overload ability than any other type. Do you need more overload capacity?
Since there is an easy sub to make, why not take that route? Did you go and have a look? If you find them, you can lookup each datasheet and compare yourself. Don't forget to derate the resistors you intend to use as they will not be running at ambient temperature with what you are suggesting.
-Chris