Pass DIY Addict
Joined 2000
Paid Member
How's this for a dummy load? Just picked these up from MPJA.com a few weeks ago. The board that they are attached to measures approx 6x12 inches 
Four 16ohm 125w resistors in parallel. Two at a time (switch open) provides an 8ohm load capable of dissipating 250w. Close the switch and now there are four in parallel that provides a 4ohm load capable of dissipating almost 500w.
For an input signal, I'll use a CD that I have with test tones as specific frequencies and just put that track on repeat while I work. I presume the input signal should be nearly "full line-level" rather than something that comes out of preamp-style volume control that is a low signal (like -30dB or so)...
I'm ready
Four 16ohm 125w resistors in parallel. Two at a time (switch open) provides an 8ohm load capable of dissipating 250w. Close the switch and now there are four in parallel that provides a 4ohm load capable of dissipating almost 500w.
For an input signal, I'll use a CD that I have with test tones as specific frequencies and just put that track on repeat while I work. I presume the input signal should be nearly "full line-level" rather than something that comes out of preamp-style volume control that is a low signal (like -30dB or so)...
I'm ready
Last edited:
Pass DIY Addict
Joined 2000
Paid Member
I think I have successfully measured and adjusted the AC current gain for the amp. I used my CD player's line level output on the back and just ran this straight into the amp with a 60Hz sine wave that I have on a test disk from when I built my subwoofer. I connected my dummy load and set it to 4 ohms and started my measurements
Using WuffWaff's choice of 1k3 for R13 (I largely just copied his design), I measured AC voltage across the source resistor on one of the paralleled mosfets and the AC voltage across the output resistor. Using the calculations I found in the Wiki and other threads here, I calculated AC current gain at just shy of 50%. Since I am looking to optimize the amp for a 4-ohm load, my target (using the AXE1-2 spreadsheet) is 60%.
I first tried replacing the 1k3 resistor with 1k2, but this took me in the wrong direction and reduced AC current gain to 46.5%. I then put the 1k3 in series with a 221ohm resistor which yielded 57% AC current gain. The last stop for the night was putting the 1k3 in series with a 332ohm resistor - the result was 60% AC current gain, right on target.
The strange part is that my dummy load resistors never really got hot. They were clearly warm to the touch, but I could comfortably grasp them without any trouble. Based on what others here have said, I expected them to get hot enough to burn my fingers...
So, it looks like my final configuration looks like:
- 21.4v rails
- 7.3A bias - I measure 0.407v drop across each of my source resistors, so I have .407/.333 = 1.2A bias per fet * 3 fets per bank * 2 banks = 7.33A bias per side.
- 94w into 8 ohms - according to calculation in the AXE1-2 spreadsheet
- 167w into 4 ohms - again, from the spreadsheet
Relative DC offset tracks VERY consistently below 0.08 at all times (warm or cold)
Absolute DC offset is harder to nail down. I can adjust it to 0 with the lid off, but then it climbs when I put the lid on and the inside temperature rises. I think I want to look around and see if I can find a decent ventilated lid rather than just drilling holes in the top plate. This might end up looking nicer...
The final task is to bring it into the EE lab and see if I can get some actual measurements of output capability. I am not sure how to do this, though. I know I need a signal generator, my dummy load, and a scope to watch the output waveform. I presume I just run a pure sine wave in, slowly increase the input volume, and put the test leads of the scope across the dummy load and watch the resulting wave form until I see clipping. Is the scope able to record/display output power while displaying the output waveform?
Using WuffWaff's choice of 1k3 for R13 (I largely just copied his design), I measured AC voltage across the source resistor on one of the paralleled mosfets and the AC voltage across the output resistor. Using the calculations I found in the Wiki and other threads here, I calculated AC current gain at just shy of 50%. Since I am looking to optimize the amp for a 4-ohm load, my target (using the AXE1-2 spreadsheet) is 60%.
I first tried replacing the 1k3 resistor with 1k2, but this took me in the wrong direction and reduced AC current gain to 46.5%. I then put the 1k3 in series with a 221ohm resistor which yielded 57% AC current gain. The last stop for the night was putting the 1k3 in series with a 332ohm resistor - the result was 60% AC current gain, right on target.
The strange part is that my dummy load resistors never really got hot. They were clearly warm to the touch, but I could comfortably grasp them without any trouble. Based on what others here have said, I expected them to get hot enough to burn my fingers...
So, it looks like my final configuration looks like:
- 21.4v rails
- 7.3A bias - I measure 0.407v drop across each of my source resistors, so I have .407/.333 = 1.2A bias per fet * 3 fets per bank * 2 banks = 7.33A bias per side.
- 94w into 8 ohms - according to calculation in the AXE1-2 spreadsheet
- 167w into 4 ohms - again, from the spreadsheet
Relative DC offset tracks VERY consistently below 0.08 at all times (warm or cold)
Absolute DC offset is harder to nail down. I can adjust it to 0 with the lid off, but then it climbs when I put the lid on and the inside temperature rises. I think I want to look around and see if I can find a decent ventilated lid rather than just drilling holes in the top plate. This might end up looking nicer...
The final task is to bring it into the EE lab and see if I can get some actual measurements of output capability. I am not sure how to do this, though. I know I need a signal generator, my dummy load, and a scope to watch the output waveform. I presume I just run a pure sine wave in, slowly increase the input volume, and put the test leads of the scope across the dummy load and watch the resulting wave form until I see clipping. Is the scope able to record/display output power while displaying the output waveform?
Last edited:
Pass DIY Addict
Joined 2000
Paid Member
I think I understand the output power calculation... Is the example below accurate?
I=V/R, so presume I measure output voltage swing of 38v (+19 for a high and -19v for a low) into an 8 ohm load before I see clipping on the scope, so I = 38/8 = 4.75A. I then multiply 4.75A by the one-way measured voltage sing of 19v (yielding 90w into an 8 ohm load).
Repeating the above example with a 4 ohm load, I might still see 38v before clipping sets in so then I get I = 38/4 = 9.5A, then multiply 9.5A by the observed 19v and get 180w into 4 ohms.
Or am I approaching this from entirely the wrong direction?
What is a typical input level (in volts) before the amp goes into clipping? Or does this vary by pre-amp specifications?
Thanks!
I=V/R, so presume I measure output voltage swing of 38v (+19 for a high and -19v for a low) into an 8 ohm load before I see clipping on the scope, so I = 38/8 = 4.75A. I then multiply 4.75A by the one-way measured voltage sing of 19v (yielding 90w into an 8 ohm load).
Repeating the above example with a 4 ohm load, I might still see 38v before clipping sets in so then I get I = 38/4 = 9.5A, then multiply 9.5A by the observed 19v and get 180w into 4 ohms.
Or am I approaching this from entirely the wrong direction?
What is a typical input level (in volts) before the amp goes into clipping? Or does this vary by pre-amp specifications?
Thanks!
Pass DIY Addict
Joined 2000
Paid Member
Hello Eric,
nice to hear the amp is working. If it is ready to play some music you should try the 50% AC-Current-Gain setting. The gain is normally bigger than the 3dB in loudness you get into 4 Ohms.
Try to set the absolute dc offset with the lid on the amp. This will work fine. I connected the current source fet with the case to keep the temperature down and better coupled to the rest of the amp. This reduced the starting up voltage (abs DC) a lot.
William
nice to hear the amp is working. If it is ready to play some music you should try the 50% AC-Current-Gain setting. The gain is normally bigger than the 3dB in loudness you get into 4 Ohms.
Try to set the absolute dc offset with the lid on the amp. This will work fine. I connected the current source fet with the case to keep the temperature down and better coupled to the rest of the amp. This reduced the starting up voltage (abs DC) a lot.
William
Pass DIY Addict
Joined 2000
Paid Member
Thanks for the suggestions, William! With the current source fet (Q6) did you just attach it to the chassis, or did you mount it on the heat sink?
Your comment about the AC current Gain setting is a little puzzling to me (I've been fiddling around with the AXE1-2 spreadsheet that you created). Will a setting of 50% work better for 4-ohm loads than setting it to 60%? Its easy enough to change...
One more detail - did you use a solid or a perforated top panel?
Thanks!
Your comment about the AC current Gain setting is a little puzzling to me (I've been fiddling around with the AXE1-2 spreadsheet that you created). Will a setting of 50% work better for 4-ohm loads than setting it to 60%? Its easy enough to change...
One more detail - did you use a solid or a perforated top panel?
Thanks!
Hi,
well chassis and heat sink is more or less the same in my case. I´ve attached the fet to the top of the case wich is directly bolted to the heat sinks. I can also (just) adjust the pot to set the absolute dc offset when the lid is on. Must be some pics somewhere. What is you startup abs dc offset value?
60% will give you more power into 4 Ohms, 50% will sound better........ The difference sounds a lot in power but won´t be that much in loudness.
My top panel has 3 holes in it (around 40mm in diameter), I don´t use a bottom panel.
William
well chassis and heat sink is more or less the same in my case. I´ve attached the fet to the top of the case wich is directly bolted to the heat sinks. I can also (just) adjust the pot to set the absolute dc offset when the lid is on. Must be some pics somewhere. What is you startup abs dc offset value?
60% will give you more power into 4 Ohms, 50% will sound better........ The difference sounds a lot in power but won´t be that much in loudness.
My top panel has 3 holes in it (around 40mm in diameter), I don´t use a bottom panel.
William
Pass DIY Addict
Joined 2000
Paid Member
Absolute DC offset runs about 3v at startup and slowly decreases, probably takes an 45-60 mins to stabilize. Relative offset starts off about 0.002v and slowly climbs to about 0.06 or so at the same time.
With the fet, is the point for it to thermally track with the chassis, or just to give it additional thermal stability to keep it from changing temperature too quickly?
Eric
With the fet, is the point for it to thermally track with the chassis, or just to give it additional thermal stability to keep it from changing temperature too quickly?
Eric
Hi,
3V is not very much. I start at 8V. What value did you use for the McMillan resistors? You could double them for 6V at startup, and do a lot for sound quality.
Did you couple the input diff pair to eachother? My relative offset would stay the same between warm and cold.
The point with the fet was to keep it a bit cooler so the difference between cold and warm would become smaller.
William
3V is not very much. I start at 8V. What value did you use for the McMillan resistors? You could double them for 6V at startup, and do a lot for sound quality.
Did you couple the input diff pair to eachother? My relative offset would stay the same between warm and cold.
The point with the fet was to keep it a bit cooler so the difference between cold and warm would become smaller.
William
Pass DIY Addict
Joined 2000
Paid Member
Hi William,
I switched R12 back to 1k3 for 50% AC current gain and pulled Q6 from the PCB and attached it with wire leads for placement flexibility. I need to let it warm up and explore bias settings again.
I am using 3k3 for the McMillan resistors. By doubling them, do you mean moving to 6k resistors? I'm all for improvements to sound quality!
The input differential pair was matched to within 0.001v and are mounted side by side on one of the main heatsinks centered above three power mosfets so they track thermally with one another. They seem to behave very, very well. There is an image of them here http://www.diyaudio.com/forums/pass-labs/28336-aleph-x-builders-thread-37.html#post2069790
It sounds like the best option is to try and keep the current source fet at either as steady a temperature as possible, or for it to track with the main heatsink so that it stabilizes quickly. Perhaps I'll experiment with heat sink mounting and back of the chassis mounting to see which performs better.
These are exactly the reasons that I have not yet done final assembly on the other two amps yet. Thanks guys!
Eric
I switched R12 back to 1k3 for 50% AC current gain and pulled Q6 from the PCB and attached it with wire leads for placement flexibility. I need to let it warm up and explore bias settings again.
I am using 3k3 for the McMillan resistors. By doubling them, do you mean moving to 6k resistors? I'm all for improvements to sound quality!
The input differential pair was matched to within 0.001v and are mounted side by side on one of the main heatsinks centered above three power mosfets so they track thermally with one another. They seem to behave very, very well. There is an image of them here http://www.diyaudio.com/forums/pass-labs/28336-aleph-x-builders-thread-37.html#post2069790
It sounds like the best option is to try and keep the current source fet at either as steady a temperature as possible, or for it to track with the main heatsink so that it stabilizes quickly. Perhaps I'll experiment with heat sink mounting and back of the chassis mounting to see which performs better.
These are exactly the reasons that I have not yet done final assembly on the other two amps yet. Thanks guys!
Eric
Hi Eric,
3k3 is very low. I startet with 4k7 and noticed improvements until I reached 10k. The startup absolute DC value grows proportionally but goes down very fast. With 10k it was around 8v. With the Jfet input I had to raise the value to 22k. If I remember well the bass suffered with the lower values.
I did experiment with a can of ice spray and the only part that really showed a big influence on absolute dc offset was Q6. Keeping it at a constant temp from startup would be very good but nearly impossible. There are a few other things that influence abs dc offset like the mains voltage. That´s why it will never be 0V all the time.
William
3k3 is very low. I startet with 4k7 and noticed improvements until I reached 10k. The startup absolute DC value grows proportionally but goes down very fast. With 10k it was around 8v. With the Jfet input I had to raise the value to 22k. If I remember well the bass suffered with the lower values.
I did experiment with a can of ice spray and the only part that really showed a big influence on absolute dc offset was Q6. Keeping it at a constant temp from startup would be very good but nearly impossible. There are a few other things that influence abs dc offset like the mains voltage. That´s why it will never be 0V all the time.
William
Pass DIY Addict
Joined 2000
Paid Member
I can change the McMillan resistors to 8k to 10k without any problem - I appreciate the suggestion!
Hmmm... I'm trying to envision a place in/on the chassis that will remain at constant temperature while the amp is up and running. Only one potential solution comes to mind:
Drill a hole in the back panel of the chassis to run the wires for Q6 outside of the amp. Mount Q6 to a relatively small heat sink that is mounted with spacers or nylon such that its base is parallel to, but raised up from the back panel. Thus, we have stability from the mass of the sink and from ambient air temperature, but it is de-coupled from the immediate heat of the chassis.
What do you think of this approach?
Hmmm... I'm trying to envision a place in/on the chassis that will remain at constant temperature while the amp is up and running. Only one potential solution comes to mind:
Drill a hole in the back panel of the chassis to run the wires for Q6 outside of the amp. Mount Q6 to a relatively small heat sink that is mounted with spacers or nylon such that its base is parallel to, but raised up from the back panel. Thus, we have stability from the mass of the sink and from ambient air temperature, but it is de-coupled from the immediate heat of the chassis.
What do you think of this approach?
Hi,
it could work but I don´t think it is really neccessary. The 8V I see at startup are down to 3V within a few minutes and are near the target value after 20 minutes. It is not critical to have some abs dc voltage at the output.
The only thing that is not so good is that it causes the transformers to hum because they are loaded assymetrically.
William
it could work but I don´t think it is really neccessary. The 8V I see at startup are down to 3V within a few minutes and are near the target value after 20 minutes. It is not critical to have some abs dc voltage at the output.
The only thing that is not so good is that it causes the transformers to hum because they are loaded assymetrically.
William
Pass DIY Addict
Joined 2000
Paid Member
Uhhmm, I've lost you here... What causes the transformer to hum due to asymmetrical loading?