Hi Mike,
I took it up to about 37 VDC, being realistic about the test. It's now hammering away. I'm running it just below clipping. Just looked, okay it's into clipping. Disco is fun for this stuff ("Knock on Wood" right now), the entire place is thumping.
I can say with certainty that the overload characteristics are pretty good. I'm impressed. So I'm going to let it cook. I must say it's hard to concentrate at the moment.
Mike, MORE POWER!!!!!!!!
Looks like I've gone to the dark side. -"Car Wash" is starting.
-Chris
I took it up to about 37 VDC, being realistic about the test. It's now hammering away. I'm running it just below clipping. Just looked, okay it's into clipping. Disco is fun for this stuff ("Knock on Wood" right now), the entire place is thumping.
I can say with certainty that the overload characteristics are pretty good. I'm impressed. So I'm going to let it cook. I must say it's hard to concentrate at the moment.
Mike, MORE POWER!!!!!!!!
Looks like I've gone to the dark side. -"Car Wash" is starting.
-Chris
Hi Mike,
2° C across the thermal pad to the heatsink. Heatsink is approx 40.4 °C. The heatsink is my standard test unit, 10" X 7" X 1". Both channels are mounted to it. I'm running it hard. Now, should I connect my dummy loads in parallel? I am looking at bias stability.
Now, if this performance can be at least maintained at a higher power level you will have a design to be really reckoned with. This one is already hitting the mark in this power class.
-Chris
2° C across the thermal pad to the heatsink. Heatsink is approx 40.4 °C. The heatsink is my standard test unit, 10" X 7" X 1". Both channels are mounted to it. I'm running it hard. Now, should I connect my dummy loads in parallel? I am looking at bias stability.
Now, if this performance can be at least maintained at a higher power level you will have a design to be really reckoned with. This one is already hitting the mark in this power class.
-Chris
Hi Mike,
I'm just leting it cool down now. The bias went up to 26 mA warm. Don't worry, at some point I'll get them hot.
Yes, the speakers are a simple 2 way design with zobels across the woofers. Those poor 8" Vifa's, just for the test bench. I will pound it some more with the 8 ohm dummy loads across as well.
My choice so far is going to be the MJW devices, unless On Semi comes out with better ones. The driver transistors need improvement. I'm talking quality, not power here. I only got the MJL21196G's, the MJL21195G's got lost in the system. I'm trying to chase those down now. I'll try those when I get them.
Looking for more signal transistors. My Japanese supplier is basically gone now. Those normally work better in the 2 watt down catagory. So I have to set up and test the Linear Systems parts. I want to compare those to some other readily available parts. I do have a couple 2SK146 and 2SK240 pairs to try as well. Reference only at this point. I may even have uPA68H hiding somewhere.
Al, I forget what you were using for outputs. Could you remind me please? Also, did you match then (best you can)?
If you can do THD measurements, set the bias for the lowest THD and lowest current draw. Measure with a 4 ohm load as well, low output power with a 10 ~ 20 KHz sine signal. Your minimum power level will be set by the sensitivity of your THD meter. This should be a worst case for the amplifier as crossover distortion will reduce as the output goes up. I like to look at the residual to see where the noise or harmonics are. If your oscillator has high THD, consider RMAA and a sound card. Be careful you don't kill the sound card!
-Chris
Edit: Al, just saw your post. I'm not surprised really. I have found that matching the NPN to PNP gains make a positive difference on sound quality. All the way back, that's why I don't like the drivers I'm using right now.
I'm just leting it cool down now. The bias went up to 26 mA warm. Don't worry, at some point I'll get them hot.
Yes, the speakers are a simple 2 way design with zobels across the woofers. Those poor 8" Vifa's, just for the test bench. I will pound it some more with the 8 ohm dummy loads across as well.
My choice so far is going to be the MJW devices, unless On Semi comes out with better ones. The driver transistors need improvement. I'm talking quality, not power here. I only got the MJL21196G's, the MJL21195G's got lost in the system. I'm trying to chase those down now. I'll try those when I get them.
Looking for more signal transistors. My Japanese supplier is basically gone now. Those normally work better in the 2 watt down catagory. So I have to set up and test the Linear Systems parts. I want to compare those to some other readily available parts. I do have a couple 2SK146 and 2SK240 pairs to try as well. Reference only at this point. I may even have uPA68H hiding somewhere.
Al, I forget what you were using for outputs. Could you remind me please? Also, did you match then (best you can)?
If you can do THD measurements, set the bias for the lowest THD and lowest current draw. Measure with a 4 ohm load as well, low output power with a 10 ~ 20 KHz sine signal. Your minimum power level will be set by the sensitivity of your THD meter. This should be a worst case for the amplifier as crossover distortion will reduce as the output goes up. I like to look at the residual to see where the noise or harmonics are. If your oscillator has high THD, consider RMAA and a sound card. Be careful you don't kill the sound card!
-Chris
Edit: Al, just saw your post. I'm not surprised really. I have found that matching the NPN to PNP gains make a positive difference on sound quality. All the way back, that's why I don't like the drivers I'm using right now.
Hi Al,
Sounds like a plan. Post your results please.
You could easily make a jig for grading output devices. I have one for mosfets and one for BJT's. I'm interested to find out what your betas were. The reason I like the MJW0281 and MJW0302 is that the compliments are close. Also, the spread is very low between devices.
I have high hopes to find more On Semi products with these characteristics. For new designs, I would spec these or the MJL equivalents. I hope it's the same die on a bigger heat spreader.
Now we need a 3 or 4 A driver transistor with these same characteristics.
-Chris
Sounds like a plan. Post your results please.
You could easily make a jig for grading output devices. I have one for mosfets and one for BJT's. I'm interested to find out what your betas were. The reason I like the MJW0281 and MJW0302 is that the compliments are close. Also, the spread is very low between devices.
I have high hopes to find more On Semi products with these characteristics. For new designs, I would spec these or the MJL equivalents. I hope it's the same die on a bigger heat spreader.
Now we need a 3 or 4 A driver transistor with these same characteristics.
-Chris
Hi Mike,
No, I'm getting a parts order together right now. I'll try that a bit later. I may need to wait until tomorrow when everyone clears the building.
The amplifier is very, very well behaved. I have some problems to work out involving my computers. I need a '98 machine to run a database program and my scanner. I want to run XP for testing. So I'll be moving the sound card and RMAA around.
Once I have that all sorted out, testing can begin. Then I'll do some mods and evaluate each step.
Updates are available again. I ..... am ..... very ..... afraid.
-Chris
No, I'm getting a parts order together right now. I'll try that a bit later. I may need to wait until tomorrow when everyone clears the building.
The amplifier is very, very well behaved. I have some problems to work out involving my computers. I need a '98 machine to run a database program and my scanner. I want to run XP for testing. So I'll be moving the sound card and RMAA around.
Once I have that all sorted out, testing can begin. Then I'll do some mods and evaluate each step.
Updates are available again. I ..... am ..... very ..... afraid.
-Chris
I try to match the drivers, but that seems impossible...
While the mje15030 is a fine device (hfe = 110 "always" 1..35ma 5..36v) the mje15031 seems to be a catastrophy, in no way a complementary...
It changes its hfe dramatically with Vce, while it can be matched at vce = 5v, it will rise its hfe up to 500 at 36v.
This behaviour makes it useless for symasym, having a low current vas.
I used an mje15031 from an older batch, this gave 7 to 4mv across the basestoppers, reducing thd ~ the same amount.
I guess the mje15034/35 (much more modern device, matching to 5% confirmed by roender) will be much better choice or some japanese.
The mje15030/1 seems to be stoneage.
I might reactivate my quasi complementary, complementary transistors are... a fairy tale.
(kidding)
Mike
While the mje15030 is a fine device (hfe = 110 "always" 1..35ma 5..36v) the mje15031 seems to be a catastrophy, in no way a complementary...
It changes its hfe dramatically with Vce, while it can be matched at vce = 5v, it will rise its hfe up to 500 at 36v.
This behaviour makes it useless for symasym, having a low current vas.
I used an mje15031 from an older batch, this gave 7 to 4mv across the basestoppers, reducing thd ~ the same amount.
I guess the mje15034/35 (much more modern device, matching to 5% confirmed by roender) will be much better choice or some japanese.
The mje15030/1 seems to be stoneage.
I might reactivate my quasi complementary, complementary transistors are... a fairy tale.
(kidding)Mike
Now what might cause this strange phenomenon? Hmmm... You can generate around 50kV with a Tesla Coil...
I suppose my work here is done... I have succeeded in restarting this thread and have even noted an issue, though a very small one.
One more thing - I have run across a CCS/CM page on the 4QDTec website. It shows a better CM than the Wilson, though there are thermal issues. It is diagram 6. I was wondering if it would be a thermal problem if this was implemented in SymAsym. I know that this type of imbalance isn't crucial, but I was just wondering what the group opinion on it was. Here is the link:
http://www.4qdtec.com/csm.html
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.