janneman said:So it's the *principle* of synmmetrical diff input stgae that cuases the issue not the implementation itself per se.
Jan Didden
I think it is mostly the other way round. You can use emitter degeneration on the VAS transistors too. With BJT’s it is only difficult to get good 'electrical’ symmetry if you attempt to get excessive gain from the input stage.
Cheers,
Glen
1.695mA through R209 and R210 (8.2k and 85mV)
This is a fifteen dollar amp. . .
Working on value of the little blue caps.
Pretty sure about cap across be of Q207/Q208. . .collector has voltage and the
cap straddles the trace.
I'll take a closeup of C103/C203 silver caps. . .
This is a fifteen dollar amp. . .
Working on value of the little blue caps.
Pretty sure about cap across be of Q207/Q208. . .collector has voltage and the
cap straddles the trace.
I'll take a closeup of C103/C203 silver caps. . .
I'm glad that we are reminded that this is a $15 amp. That is what makes other amps so expensive, comparatively. Better parts.
I've come to the conclusion on this amp.
First, I'm going to use two amps (mono).
Second, each amp is going to serve as a preamp and amp
by using one channel as preamp (15dB gain max, followed
by a pot, then into the second amp (15 dB gain). So far it
works well.
Third, I'm going to further raise the bias on the one channel
acting as preamp much higher to ensure class A operation
on the preamp half.
One question. . . should I remove the cap on the second
input since the output had less than 5mV of DC offset?
One observation. . . when using higher resistance as the
load the amp seems to produce less heat.
Example:
8 ohms load with 8V produced 8W and 1A of current
while
100 ohm load with 28.3V produced 8W and 283mA or current
both situations created 8W at the load. . . the 100 ohm load
created the same energy using 1/4 the current. . . the amp
stays really cool. . . efficient delivery of energy to the load!
Why not create amps with really big voltage swings and
speakers with higher impedances. This would allow for the
high wattage amp but produce far less heat at the amp. . .
seems more efficient. A 1000 watt amp would only need a
+/- 316V DC supply and 3.16 amps.
First, I'm going to use two amps (mono).
Second, each amp is going to serve as a preamp and amp
by using one channel as preamp (15dB gain max, followed
by a pot, then into the second amp (15 dB gain). So far it
works well.
Third, I'm going to further raise the bias on the one channel
acting as preamp much higher to ensure class A operation
on the preamp half.
One question. . . should I remove the cap on the second
input since the output had less than 5mV of DC offset?
One observation. . . when using higher resistance as the
load the amp seems to produce less heat.
Example:
8 ohms load with 8V produced 8W and 1A of current
while
100 ohm load with 28.3V produced 8W and 283mA or current
both situations created 8W at the load. . . the 100 ohm load
created the same energy using 1/4 the current. . . the amp
stays really cool. . . efficient delivery of energy to the load!
Why not create amps with really big voltage swings and
speakers with higher impedances. This would allow for the
high wattage amp but produce far less heat at the amp. . .
seems more efficient. A 1000 watt amp would only need a
+/- 316V DC supply and 3.16 amps.
gni said:One observation. . . when using higher resistance as the
load the amp seems to produce less heat.
Example:
8 ohms load with 8V produced 8W and 1A of current
while
100 ohm load with 28.3V produced 8W and 283mA or current
both situations created 8W at the load. . . the 100 ohm load
created the same energy using 1/4 the current. . . the amp
stays really cool. . . efficient delivery of energy to the load!
The differnece here is that with 100 ohm load the amp was running full signal - where the efficiency can be around 70%. The catch is you can't get much more with that 100 ohm load - with 8 ohms you can push more watts. When you get to full signal with 8 ohms, the efficiency will approach 70% also.
It seems that with a high efficiency driver and
the same driver having a high impedance would
make a better system. The amp would be operating
more efficient and the driver would still create enough
output.
In the end the whole process leads me to believe that
there is a better way to transfer energy to the driver
while increasing the efficiency of the whole system.
Why go digital if you operate your Class AB at 70%
more of the time than you could operate a Class D
amplifier?
Is it possible to have custom drivers made with
100 ohm voice coils instead of 16 or 8 or 4?
Would such driver have a reduced impedance peaks
thus making it a better load?
the same driver having a high impedance would
make a better system. The amp would be operating
more efficient and the driver would still create enough
output.
In the end the whole process leads me to believe that
there is a better way to transfer energy to the driver
while increasing the efficiency of the whole system.
Why go digital if you operate your Class AB at 70%
more of the time than you could operate a Class D
amplifier?
Is it possible to have custom drivers made with
100 ohm voice coils instead of 16 or 8 or 4?
Would such driver have a reduced impedance peaks
thus making it a better load?
gni said:In the end the whole process leads me to believe that
there is a better way to transfer energy to the driver
while increasing the efficiency of the whole system.
Why go digital if you operate your Class AB at 70%
more of the time than you could operate a Class D
amplifier?
But you wouldn't be operating at 70% efficiency more of the time. You would end up with even higher rails (kilovolts, maybe) in order to operate backed off 10-20dB for normal listening. For a given output power and operational headroom, efficiency wouldn't be much better than it is now.
It is true that higher Z is a tad more efficient (a few points) compared to a lower Z because of unavoidable I^2R losses which increase at high currents. But when you jack up the overall impedance you start getting into high-voltage specific design problems - second breakdown, surface leakage currents, stray capacitance, shock hazards at the speaker terminal...
Hi,
in days when we were all a lot younger, most speakers were 15ohm or 16ohm full range drivers that were very efficient.
Then we moved to 8ohm as the norm.
Now many like to use 1ohm to 4ohm.
I recommend you stay with 8ohm and just ignore the lower impedances.
In general the higher impedance speakers tend to be more efficient. Look at some of the professional (PA) drivers that are available in both 4ohm and 8ohm. Often the 8ohm are slightly more efficient.
Now look at the amplifier that has to drive these loads.
As the impedance is halved, the current sent to the load is doubled.
Losses due to resistance are I^2 * R. Double the current with the same size cable/resistor/semiconductor and you have 4times the losses.
Quartering the load impedance leads to 16times the losses.
Work out what happens when you couple up 2ohm and 1ohm loads.
in days when we were all a lot younger, most speakers were 15ohm or 16ohm full range drivers that were very efficient.
Then we moved to 8ohm as the norm.
Now many like to use 1ohm to 4ohm.
I recommend you stay with 8ohm and just ignore the lower impedances.
In general the higher impedance speakers tend to be more efficient. Look at some of the professional (PA) drivers that are available in both 4ohm and 8ohm. Often the 8ohm are slightly more efficient.
Now look at the amplifier that has to drive these loads.
As the impedance is halved, the current sent to the load is doubled.
Losses due to resistance are I^2 * R. Double the current with the same size cable/resistor/semiconductor and you have 4times the losses.
Quartering the load impedance leads to 16times the losses.
Work out what happens when you couple up 2ohm and 1ohm loads.
Once, I did some tests on complementary differentials[CD] and single ended differential[SD] both having CCS and loaded with current mirror at collectors. Same Iq=2mA per leg was set.
The differences in key areas were as follows
: Slew rate was symmetric with CD, whereas slewrate was non-symmetric with SD
: Overdrive capability was very poor in SD, the amp's output was clipping asymmetrically, whereas CD forced equal clipping during low impedance load drive.
: Power Bandwidth extention was more in case of CD
: The THD at HF region[5khz upwards] was very low as compared to what we obtained with SD.
: PSRR was 80dB in CD and 69dB in SD
🙂
The differences in key areas were as follows
: Slew rate was symmetric with CD, whereas slewrate was non-symmetric with SD
: Overdrive capability was very poor in SD, the amp's output was clipping asymmetrically, whereas CD forced equal clipping during low impedance load drive.
: Power Bandwidth extention was more in case of CD
: The THD at HF region[5khz upwards] was very low as compared to what we obtained with SD.
: PSRR was 80dB in CD and 69dB in SD
🙂
AndrewT said:Hi,
in days when we were all a lot younger, most speakers were 15ohm or 16ohm full range drivers that were very efficient.
Because back then people were willing to trade a little space in the room for a speaker you could actually hear when driven with a five watt amp. Not anymore.
Losses due to resistance are I^2 * R. Double the current with the same size cable/resistor/semiconductor and you have 4times the losses.
Unfortunately, semiconductor power handling is worse than halved at double the voltage. This places a practical limit on rail voltage (power). The optimum impedance may be 8 ohms (or even 10 or 20) for a 50 watt amp. Start adding zeros and it's not anymore.
Workhorse said:Once, I did some tests on complementary differentials[CD] and single ended differential[SD] both having CCS and loaded with current mirror at collectors. Same Iq=2mA per leg was set.
How did you load the CD with a current mirror at the collectors???? That doesn't work.....except in Spice.
wg_ski said:
How did you load the CD with a current mirror at the collectors???? That doesn't work.....except in Spice.
Yes, current mirror on collectors of complimentary differentials and it works in reality and spice ofcourse.
could you please tell why did you said it will not work.......?
🙂
wg_ski said:
How did you load the CD with a current mirror at the collectors???? That doesn't work.....except in Spice.
I think, you must be thinking about how to define the predictable VAS bias current in complementary symmetry design using current mirrors on collectors of differentials, well there exist a way by which you can do this.....................Think....!!!!!!😉
Performance Data
One channel driven at 60Hz sine wave into resistive load. This little
amp did quite well. The protection circuit only activated into the 4
and 2 ohm load. I ran the 100, 32, and 24 ohm loads for several
hours each with the heatsink getting slightly warm. The 16 and 8
ohm loads I ran shortly since the power resistors were beyond their
ratings. The 4 and 2 ohm were run for 10 seconds at a voltage just
below the trip of the protection. The heatsinks were very hot after
the 4 and 2 ohm loads tests. The amp at idle has +/- 34.5 VDC rails.
My meter doesn't say it is a true rms meter (high speed sampling)
100 ohm load
34.1V [344mA] (11.83W)
32 ohm load
33.1V [1.03A] (34.2W)
24 ohm load
32.7V [1.363A] (44.5W)
16 ohm load
31.81V [1.99A] (63.2W)
8 ohm load
29.33V [3.67A] (107.5W)
4 ohm load
11.93V [2.98A] (35.58W)
Protection circuit activated
2 ohm load
5.85V [2.91A] (17.1W)
Protection circuit activated
One channel driven at 60Hz sine wave into resistive load. This little
amp did quite well. The protection circuit only activated into the 4
and 2 ohm load. I ran the 100, 32, and 24 ohm loads for several
hours each with the heatsink getting slightly warm. The 16 and 8
ohm loads I ran shortly since the power resistors were beyond their
ratings. The 4 and 2 ohm were run for 10 seconds at a voltage just
below the trip of the protection. The heatsinks were very hot after
the 4 and 2 ohm loads tests. The amp at idle has +/- 34.5 VDC rails.
My meter doesn't say it is a true rms meter (high speed sampling)
100 ohm load
34.1V [344mA] (11.83W)
32 ohm load
33.1V [1.03A] (34.2W)
24 ohm load
32.7V [1.363A] (44.5W)
16 ohm load
31.81V [1.99A] (63.2W)
8 ohm load
29.33V [3.67A] (107.5W)
4 ohm load
11.93V [2.98A] (35.58W)
Protection circuit activated
2 ohm load
5.85V [2.91A] (17.1W)
Protection circuit activated
Hi,
power drops by 0.7dBV changing from 16r0 to 8r0 loading.
This is a good sign.
Shame the protection prevents testing into 4r0.
A 4r0 resistive load is less stressful than an 8ohm reactive load.
If we had seen that the amp was happy driving a 4r0 load, even for just 1 to 2seconds with a cold heatsink , we would have a better indication of what the amplifier is capable of driving.
power drops by 0.7dBV changing from 16r0 to 8r0 loading.
This is a good sign.
Shame the protection prevents testing into 4r0.
A 4r0 resistive load is less stressful than an 8ohm reactive load.
If we had seen that the amp was happy driving a 4r0 load, even for just 1 to 2seconds with a cold heatsink , we would have a better indication of what the amplifier is capable of driving.
I've never done a test at all those values. My meter
isn't a true rms meter. It seems more like it is reading
the peak value since the 100 ohm value is so close to
the rail voltage. If that were true then I need to lower
all the 'peak' values to rms values.
Yet my source AC is 123.2 VAC 60 Hz. . . that should be rms!
Our AC has been almost 125 VAC in the past -- we
always run hot. We had power company check and
there was a problem with the capacitor bank and the
adjusted it. At the time my volt meter was 0.3 V higher
than the power company meter. So. . . my meter acts
like a rms meter at 120 VAC but it can't be correct at
60 Hz and 34.1 V. . . that is almost the supply rails of
the amp. . .
100 ohm load
34.1V peak
24.1V vrms
32 ohm load
33.1V peak
23.4V rms
24 ohm load
32.7V peak
23.1V rms
16 ohm load
31.81V peak
22.5V rms
8 ohm load
29.33V peak
20.74V rms
4 ohm load
11.93V peak
8.43V rms
2 ohm load
5.85V peak
4.13V rms
isn't a true rms meter. It seems more like it is reading
the peak value since the 100 ohm value is so close to
the rail voltage. If that were true then I need to lower
all the 'peak' values to rms values.
Yet my source AC is 123.2 VAC 60 Hz. . . that should be rms!
Our AC has been almost 125 VAC in the past -- we
always run hot. We had power company check and
there was a problem with the capacitor bank and the
adjusted it. At the time my volt meter was 0.3 V higher
than the power company meter. So. . . my meter acts
like a rms meter at 120 VAC but it can't be correct at
60 Hz and 34.1 V. . . that is almost the supply rails of
the amp. . .
100 ohm load
34.1V peak
24.1V vrms
32 ohm load
33.1V peak
23.4V rms
24 ohm load
32.7V peak
23.1V rms
16 ohm load
31.81V peak
22.5V rms
8 ohm load
29.33V peak
20.74V rms
4 ohm load
11.93V peak
8.43V rms
2 ohm load
5.85V peak
4.13V rms
Originally posted by AndrewT A 4r0 resistive load is less stressful than an 8ohm reactive load.
I think I understand. . . the 8 ohm reactive load will draw more current
depending on the phase than a purely 4 ohm load.
I am not referring to peak short term transient currents into highly reactive speaker loads.gni said:I think I understand. . . the 8 ohm reactive load will draw more current
depending on the phase than a purely 4 ohm load.
Simply a 45degrees to 60degrees phase angle load. This load dissipates very high power in the output and driver devices that exceeds the power dissipation in the same devices when driving a resistive load of half the impedance.
It's very stressful as far as SOAR is concerned.
I am beginning to think that a more representative test load should be 2r7 for 8ohm capable amplifiers and 1r3 for 4ohm amplifiers. But this is continuous maximum power testing with cold heatsinks and must be time limited to just a few seconds to prevent device failures due to a falling SOAR with increasing temperature.
If your amp is protected to prevent operation into a 4r0 resistive load then I wonder if it can reliably drive an 8ohm reactive load.
New Test Results
TIP41/TIP42 outputs
6A max current
100V Max voltage (around 34V on this system)
65W Total Power Dissipation
Junction Temperature +150 °C Maximum
When testing the higher loads the protection would
engage when rapidly reducing the signal from max
to min. Strange behavior.
100 ohm load
34.3V @ 60Hz
34.1V @ 120Hz
34.3V @ 1001Hz
16 ohm load
31.81V @ 60Hz
31.46V @ 120Hz
31.97V @ 1001Hz
8 ohm load
29.33V @ 60Hz
28.77V @ 120Hz
29.06V @ 1001Hz
4 ohms load
11.93V @ 60Hz (trip protection) 2.98A and 35.58W
15.54V @ 120Hz (trip protection) 3.89A and 60.37W
24.25V @ 1001Hz (trip protection) 6A and 147W
TIP41/TIP42 outputs
6A max current
100V Max voltage (around 34V on this system)
65W Total Power Dissipation
Junction Temperature +150 °C Maximum
When testing the higher loads the protection would
engage when rapidly reducing the signal from max
to min. Strange behavior.
100 ohm load
34.3V @ 60Hz
34.1V @ 120Hz
34.3V @ 1001Hz
16 ohm load
31.81V @ 60Hz
31.46V @ 120Hz
31.97V @ 1001Hz
8 ohm load
29.33V @ 60Hz
28.77V @ 120Hz
29.06V @ 1001Hz
4 ohms load
11.93V @ 60Hz (trip protection) 2.98A and 35.58W
15.54V @ 120Hz (trip protection) 3.89A and 60.37W
24.25V @ 1001Hz (trip protection) 6A and 147W
- Status
- Not open for further replies.
- Home
- Amplifiers
- Solid State
- What's this Amplifier Design