Hi,
I have made 5 years back the Crescendo ME amplifiers from Elektor which are used in bridge mode to drive my peerless XLS12 sub with passive radiator.
Sofar, I haven't made an amplifier from my own hand, this will hopefully change in the future.
I've made a circuit in LTspice;
In the attachment you can find the schematic, FFT analysis 220W@4 and a phase margin plot.
The amp itself is not that complex. It's a complementary design with a differential vas which are driving directly a 2SK1530/2SJ201 OPS.
The vbe multiplier used in the schematic will be replaced by a 2SK2733 Fet multiplier (have no spice model for that one).
I used miller compensation around the Vas stage and lag at the feedback network.
May I ask to take a look at the attachments and give me some feedback. The goal is to use two of these amps in bridge mode to drive a subwoofer with an impedane of 8Ohm / 250Wrms.
Thanks
EDIT 23/02/2013
The project is finished, see: http://www.diyaudio.com/forums/soli...n-subwoofer-wideband-duty-13.html#post3382545
I have made 5 years back the Crescendo ME amplifiers from Elektor which are used in bridge mode to drive my peerless XLS12 sub with passive radiator.
Sofar, I haven't made an amplifier from my own hand, this will hopefully change in the future.
I've made a circuit in LTspice;
In the attachment you can find the schematic, FFT analysis 220W@4 and a phase margin plot.
The amp itself is not that complex. It's a complementary design with a differential vas which are driving directly a 2SK1530/2SJ201 OPS.
The vbe multiplier used in the schematic will be replaced by a 2SK2733 Fet multiplier (have no spice model for that one).
I used miller compensation around the Vas stage and lag at the feedback network.
May I ask to take a look at the attachments and give me some feedback. The goal is to use two of these amps in bridge mode to drive a subwoofer with an impedane of 8Ohm / 250Wrms.
Thanks
EDIT 23/02/2013
The project is finished, see: http://www.diyaudio.com/forums/soli...n-subwoofer-wideband-duty-13.html#post3382545
Attachments
Last edited:
I ve tested such a complementary differential with a differential VAS.
Make sure that the pairs are VBE matched , otherwise the results
are a nightmare.
The two VAS pairs must have the same VBE for the two sides,
while for the input differentials only the same polarity pairs must
have very close matching.
For a subwoofer amp , i would stay with the crescendo topology..
Make sure that the pairs are VBE matched , otherwise the results
are a nightmare.
The two VAS pairs must have the same VBE for the two sides,
while for the input differentials only the same polarity pairs must
have very close matching.
For a subwoofer amp , i would stay with the crescendo topology..
Last edited:
Using drivers will reduce distorsion at high frequencies but will
also reduce stability, wich can be solved by increasing the compensations caps.
For a woofer amp , drivers are not strictly necessary.
Also , the power mos you re using dont suit without temperature compensation,
since they have a positive temperature coefficient , contrary to hitachi/renesas
lateral fets 2SK1058/2SJ162 who allow for a simple resistor to set the bias.
also reduce stability, wich can be solved by increasing the compensations caps.
For a woofer amp , drivers are not strictly necessary.
Also , the power mos you re using dont suit without temperature compensation,
since they have a positive temperature coefficient , contrary to hitachi/renesas
lateral fets 2SK1058/2SJ162 who allow for a simple resistor to set the bias.
You've made me doubting the differential vas. At this moment I have running the Old Crescendo's (2x 2SK135/2X 2SJ50) and the Crescendo ME's.
If I'm building a new amp, I really wan't to use another topology. This to learn and to hear what the sonic difference could be. Even if it is not really necessary for a subwoofer amp.
I've the same amp as in post nr1 redrawn with a darlington vas. Simulation wise I get identical results with identical compensation cap values. But the one with the darlington VAS shows 0.8V DC offset, I don't understand were that comes from? If I short circuit the high pass input filter, the DC offset decreases to 50mV.
The reason wy I've choosen to use the vertical's instead of the Laterals is:
1: Output wattage is bigger with the verticals Vds = 2.5V vs 12V
2: The vertical's are easy to obtain
3: The vertical's have proven me to be verry rugged (if the gate stoppers are mounted as close as possible to the gates) (have burned about 20 of them until the leight in my head has turned on)
If either the differential or darlington VAS is better suit to be used in bridge configuration, I really would like to hear the reason for this. When I see the time in a couple of days I start building this on a vero board, just to test a bit.
If I'm building a new amp, I really wan't to use another topology. This to learn and to hear what the sonic difference could be. Even if it is not really necessary for a subwoofer amp.
I've the same amp as in post nr1 redrawn with a darlington vas. Simulation wise I get identical results with identical compensation cap values. But the one with the darlington VAS shows 0.8V DC offset, I don't understand were that comes from? If I short circuit the high pass input filter, the DC offset decreases to 50mV.
The reason wy I've choosen to use the vertical's instead of the Laterals is:
1: Output wattage is bigger with the verticals Vds = 2.5V vs 12V
2: The vertical's are easy to obtain
3: The vertical's have proven me to be verry rugged (if the gate stoppers are mounted as close as possible to the gates) (have burned about 20 of them until the leight in my head has turned on)
If either the differential or darlington VAS is better suit to be used in bridge configuration, I really would like to hear the reason for this. When I see the time in a couple of days I start building this on a vero board, just to test a bit.
Attachments
I would say that there are 4 versions of output devices that can be used in Power amplifiers.
BJT - silicon
BJT - germanium
FET - lateral
FET - vertical.
I believe that all 4 need to be approached differently at the design stage.
I further believe that one cannot simply swap between any of these types when an amplifier is designed to perform properly with only one of those types.
BJT - silicon
BJT - germanium
FET - lateral
FET - vertical.
I believe that all 4 need to be approached differently at the design stage.
I further believe that one cannot simply swap between any of these types when an amplifier is designed to perform properly with only one of those types.
A few details might have to change. Sounds like a challenge.I've the same amp as in post nr1 redrawn with a darlington vas. Simulation wise I get identical results with identical compensation cap values. But the one with the darlington VAS shows 0.8V DC offset, I don't understand were that comes from? If I short circuit the high pass input filter, the DC offset decreases to 50mV.
Can you please post the LTspice file? It would save us the trouble of drawing the circuit again.
Just change the extension from '.asc' to '.txt' otherwise the forum software will not allow it.
I've the same amp as in post nr1 redrawn with a darlington vas. Simulation wise I get identical results with identical compensation cap values. But the one with the darlington VAS shows 0.8V DC offset, I don't understand were that comes from? If I short circuit the high pass input filter, the DC offset decreases to 50mV.
Since the differential s collector resistor were not increased , the darlington
VAS cant work properly as the voltage drop in this resistors is not
enough to reach an adequate Vbe...
Increase R5/6/7/8 to about 1.5 to 1.8K...
That said , using a darlington VAS will not bring a lot of perfs while
reducing stability , so the miller caps should be increased in respect
of the original value.
Hi Andrew,
I will go for the Vertical's, I have no doubt about that. What are your thoughts for taking care of during a design with VFet's? Idle current of the VAS is important when driving directly the OPS, Vas idle current is +-20mA.
Ingenieus,
I zipped the .asc file and put the spice models for the BJT's in text on the circuit.
Wahab,
You're right, the voltage across the collector resistors from the diff stage was only 680mV's. I increased the resistor values to 1K8 ==> DC offset decreased to 25mV's.
Concerning the compensation cap values, when using the original values (22pF) I'm still having a phase margin of 88°.
Greetz
I will go for the Vertical's, I have no doubt about that. What are your thoughts for taking care of during a design with VFet's? Idle current of the VAS is important when driving directly the OPS, Vas idle current is +-20mA.
Ingenieus,
I zipped the .asc file and put the spice models for the BJT's in text on the circuit.
Wahab,
You're right, the voltage across the collector resistors from the diff stage was only 680mV's. I increased the resistor values to 1K8 ==> DC offset decreased to 25mV's.
Concerning the compensation cap values, when using the original values (22pF) I'm still having a phase margin of 88°.
Greetz
Attachments
it's alive
Thanks Ron,
I know the circuits of Rod Elliot. But it's more fun when it comes entirely out of my own hand this time.
I got the diff vas circuit running on vero board without the OPS. I used a headphone with a 2k2 resistor in series, connected at the output of the VAS.
I've played some music and heard nothing strange. I hooked it on to my signal generator with a sinus of 1KHz,10Khz and a square wave of 1kHz and 10kHz. The 10kHz square looked not that good, could it be that the load of 2K2 of my headphone is to big?
What else should be interesting to test to evaluate this circuit?
Thanks Ron,
I know the circuits of Rod Elliot. But it's more fun when it comes entirely out of my own hand this time.
I got the diff vas circuit running on vero board without the OPS. I used a headphone with a 2k2 resistor in series, connected at the output of the VAS.
I've played some music and heard nothing strange. I hooked it on to my signal generator with a sinus of 1KHz,10Khz and a square wave of 1kHz and 10kHz. The 10kHz square looked not that good, could it be that the load of 2K2 of my headphone is to big?
What else should be interesting to test to evaluate this circuit?
Attachments
Last edited:
The solution to the DC offset problem is very simple. A 100uF capacitor between the 430 ohm resistor in the feedback network (R25) and ground brings the DC offset right down to 230 microvolt.
The capacitor uncouples the resistor from the DC path. If you want a more detailed explanation, I will have to look a few things up to make sure that I've got it right.
The capacitor uncouples the resistor from the DC path. If you want a more detailed explanation, I will have to look a few things up to make sure that I've got it right.
Those differential pairs were just crying out for current mirror loads, so I added them. The THD at 20kHz is about a quarter of what it was before. The frequency response is fine for a sub amp, but I opened it up a bit for general purpose use. With some tweaking and optimization of values there might be some more improvements forthcoming.
Another thing - it might be better to use a MOSFET for thermal tracking. I'm not sure that a BJT Vbe multiplier goes well with MOSFETs. Does anybody have an opinion on this?
Another thing - it might be better to use a MOSFET for thermal tracking. I'm not sure that a BJT Vbe multiplier goes well with MOSFETs. Does anybody have an opinion on this?
Attachments
Last edited:
Interesting Ingenieus,
How did you conclude that the diff pair needed the current mirrors? Do you compare the collector current in both legs, or what do you use as trigger to go for a current mirror?
I've never used current mirrors befors.
With the optimised current mirror (as Bob Cordell calls these in his book) the symmetrical topology should work.
Concerning the vbe multiplier, as I explained in post #1 ,I will be using a 2SK2733 mosfet, but due to the lack of the spice model for this,...
Thanks for the input!
How did you conclude that the diff pair needed the current mirrors? Do you compare the collector current in both legs, or what do you use as trigger to go for a current mirror?
I've never used current mirrors befors.
With the optimised current mirror (as Bob Cordell calls these in his book) the symmetrical topology should work.
Concerning the vbe multiplier, as I explained in post #1 ,I will be using a 2SK2733 mosfet, but due to the lack of the spice model for this,...
Thanks for the input!
Andrew,
I don't fully understand your explanation? Are you saying that a current mirror is effective when, for example a differential VAS is used ( symmetrically loaded).
Is your experience that an amp with resistor loaded LTP sounds better than one with current mirrors?
Greetz
I don't fully understand your explanation? Are you saying that a current mirror is effective when, for example a differential VAS is used ( symmetrically loaded).
Is your experience that an amp with resistor loaded LTP sounds better than one with current mirrors?
Greetz
I have one family of amps that I particularly like the sound of. It is resistor loaded.
The Leach is also resistor loaded but that didn't seem quite as nice.
On this Forum there seems to be a significant minority that believe the resistor loading is better than the mirror loading.
Recently there have been a few mirror loaded designs with symmetrical loading that are reported to sound good.
I think you will have to assess that for yourself.
The Leach is also resistor loaded but that didn't seem quite as nice.
On this Forum there seems to be a significant minority that believe the resistor loading is better than the mirror loading.
Recently there have been a few mirror loaded designs with symmetrical loading that are reported to sound good.
I think you will have to assess that for yourself.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
- Home
- Amplifiers
- Solid State
- Amp design for subwoofer and wideband duty