Dear All,
After reading a huge and great amount of resources according the LME devices and the ON thermaltrak. Many thanks to Panson, Douglas Self, Charles Hansen enz. for the inspiration.
I would like to share my implementation.
Please feel free to comment and/or criticize. Any help is always more then welcome 😉
The idea behind this design is:
1: There should't be capacitors direct in series witth the signalflow. Hence the DC servo. The output of this servo should be fed to a protection system, that shuts's down the system by to much DC offset.
2: The amplifier should be a balanced bridged design. I love bridged designs.
3: Because it is an EF stage, the driver stage should be on a higher supply rails so it can fully saturate the end stage and distortion should be lower.
4: The common mode noise reduction on the input should be as good as possible, that is why the DC servo output function as a virtual ground as well
5: It should be able to supply good power at 4 ohm loads, hence the relative low power supply voltage for the end stage.
I look forward to your comments and/or suggestions 😉
With kind regards,
Bas
http://img215.imageshack.us/img215/417/thermaltrak1.gif
After reading a huge and great amount of resources according the LME devices and the ON thermaltrak. Many thanks to Panson, Douglas Self, Charles Hansen enz. for the inspiration.
I would like to share my implementation.
Please feel free to comment and/or criticize. Any help is always more then welcome 😉
The idea behind this design is:
1: There should't be capacitors direct in series witth the signalflow. Hence the DC servo. The output of this servo should be fed to a protection system, that shuts's down the system by to much DC offset.
2: The amplifier should be a balanced bridged design. I love bridged designs.
3: Because it is an EF stage, the driver stage should be on a higher supply rails so it can fully saturate the end stage and distortion should be lower.
4: The common mode noise reduction on the input should be as good as possible, that is why the DC servo output function as a virtual ground as well
5: It should be able to supply good power at 4 ohm loads, hence the relative low power supply voltage for the end stage.
I look forward to your comments and/or suggestions 😉
With kind regards,
Bas
http://img215.imageshack.us/img215/417/thermaltrak1.gif
Last edited:
Dear Stinius,
I changed the link. It should work now 😉 It works here in the Netherlands 😀
With kind regards,
Bas
Dear Bas
The link does not work here.
Maybe you could upload the picture here, not using an external link?
Cheers
The link does not work here.
Maybe you could upload the picture here, not using an external link?
Cheers
Dear bobodioulasso,
Thank you for the post. I know about the supersymmetry concept from Nelson Pass, but the topic you provide me is very interesting. I will read it carefully.
Thanks again.
With kind regards,
Bas
Dear bobodioulasso,
I have good experiences with the LME49810, and indeed this chip is more practical (more current, clipping indicator enz.) But according to one of the former designers from National on this Forum the LME49811 should sound superior to the LME49810.
The extra driver stage in the LME49810 is low biased. With the LME49811 the designer has some more freedom and choices of driver stage bias and configuration.
With kind regards,
Bas
Very nice.
One thing you can do is use the DC servo for DC offset protection. Since the opamp output never goes much over +/- 1v, use that fact to develop a signal that says your output has too much DC on it and use it to control something that "fixes" the problem. Say...shut down the amp. You don't show the PS for the opamp but I assume it is a bipolar supply. For current limiting I like Hall effect sensors. I also hate relays in series with the speaker. Use the Hall effect output to mute the audio input for a while to protect the amp.
One thing you can do is use the DC servo for DC offset protection. Since the opamp output never goes much over +/- 1v, use that fact to develop a signal that says your output has too much DC on it and use it to control something that "fixes" the problem. Say...shut down the amp. You don't show the PS for the opamp but I assume it is a bipolar supply. For current limiting I like Hall effect sensors. I also hate relays in series with the speaker. Use the Hall effect output to mute the audio input for a while to protect the amp.
or speaker relay opens and a higher resistance dummy load inserted into the output along with a BIG RED LED.
This allows you to measure what is happening without having to dismantle the circuitry to get it to start up.
Yes that was my intention. First I was thinking about a PIC, but I think I decide I don't want microcontrolers or other HF generators in this amp. So I might go the discrete way with logic IC's for protection.
An way to do it is with a relais between the PSU rails. I once saw this in a Peavey design. By DC faults, the PSU rails get shorted which in turns blows the circuit breaker. However there is a change this is not rapid enough to spare your expensive speakers.
I will study your suggestion 😉
I am agree, I don't want a relays in the output.
With kind regards,
Bas
I used to run a mixer board for some live performance events. One of my pet peeves were those pro-sound amps that would shutdown FOREVER when an a catastrophic event occurred, like acoustic feedback that overloaded it. The only way to get it back on-line was to cycle power. Not too convenient when the amps were backstage. At that point I decided that any amplifier I designed would NEVER shutdown unless it was absolutely necessary to do so. A large DC offset demands that the amp does shutdown and be repaired. Blowing the PSU rail fuses or popping a breaker is a good idea. I think it will be fast enough to protect the speakers. Fuses...fast. Breaker...not that fast. Overcurrent? Muting for a second or two should do it.
Dear,
I built a first prototype of this amplifier. Of course there is so much to tweak and to experiment with to tune and make the sound of an amplifier. However, I didn't like the sound as much as my previous LME49811/Sanken STD03 design. I found the (simple) design with the Sanken STD03 darlingtons far superior in 3D imaging, sustain and speed. (I don't talk about measurement speed in slewrate, frankly I don't care about that value, but I talk about speed of percussion, the ability to hear where the drummer hit on the snare drum, the ability to separate the kickdrum from the bass guitar).
I was figuring out, that it might be the 3rd stage in this 3 stage EF design that make the sound less. The Darlington's with the ME49810 are a 2 stage EF, and this is possible because of the high hFE of the STD03's. However I don't give up on the thermaltrak's. They must be great! and it must be me who didn't come up yet with a good enough design.
So I hope with your help I can get the best out of them, and I feel it is time for a different approach.
Obvious we need current, and with 2 stage EF we don't get the current we need with the hFE's of this transistor set.
However their might be another solution to stay with two stages and still get the required current.
If memory serves me well, The Spectral amplifiers and the (nightmare? 😀) Genesis Stealth B200 took this different approach. Namely parallel multiple EF stages.
Please take a look at my new schematics. I parallel 3 identical EF stages. Ecch power transistor has its own driver transistor.
Every pair should be biased separately, cause of the hFE differences.
Would this approach be possible?
And if this would be possible, would the LME49811 see to much capcitange due the many driver transistors?
Ever pair is bypassed with a 10uF/100nF cap set. This in attempt to improve switch off behavior.
I would like to get some input.
Thanks in advance.
If I design and order the PCB's any interest for cost-price?
See schematic here: http://img11.imageshack.us/img11/4223/thermaltrak2.gif
With kind regards,
Bas
I built a first prototype of this amplifier. Of course there is so much to tweak and to experiment with to tune and make the sound of an amplifier. However, I didn't like the sound as much as my previous LME49811/Sanken STD03 design. I found the (simple) design with the Sanken STD03 darlingtons far superior in 3D imaging, sustain and speed. (I don't talk about measurement speed in slewrate, frankly I don't care about that value, but I talk about speed of percussion, the ability to hear where the drummer hit on the snare drum, the ability to separate the kickdrum from the bass guitar).
I was figuring out, that it might be the 3rd stage in this 3 stage EF design that make the sound less. The Darlington's with the ME49810 are a 2 stage EF, and this is possible because of the high hFE of the STD03's. However I don't give up on the thermaltrak's. They must be great! and it must be me who didn't come up yet with a good enough design.
So I hope with your help I can get the best out of them, and I feel it is time for a different approach.
Obvious we need current, and with 2 stage EF we don't get the current we need with the hFE's of this transistor set.
However their might be another solution to stay with two stages and still get the required current.
If memory serves me well, The Spectral amplifiers and the (nightmare? 😀) Genesis Stealth B200 took this different approach. Namely parallel multiple EF stages.
Please take a look at my new schematics. I parallel 3 identical EF stages. Ecch power transistor has its own driver transistor.
Every pair should be biased separately, cause of the hFE differences.
Would this approach be possible?
And if this would be possible, would the LME49811 see to much capcitange due the many driver transistors?
Ever pair is bypassed with a 10uF/100nF cap set. This in attempt to improve switch off behavior.
I would like to get some input.
Thanks in advance.
If I design and order the PCB's any interest for cost-price?
See schematic here: http://img11.imageshack.us/img11/4223/thermaltrak2.gif
With kind regards,
Bas
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
correction
Dear,
One correction, I see I twisted the power supply indicators in the drawing.
The correct way must be of course,
The output transistors work on 27VDC. The driver transistors and the LME49810 on 37VDC.
With kind regards,
Bas
Dear,
One correction, I see I twisted the power supply indicators in the drawing.
The correct way must be of course,
The output transistors work on 27VDC. The driver transistors and the LME49810 on 37VDC.
With kind regards,
Bas
Dear Bobodioulasso,
No 😀 It is a drawing error, thanks for pointing me on it!
Please see the corrected schematic with also the voltages corrected.
http://img683.imageshack.us/img683/5425/thermaltrak22.gif
With kind regards,
Bas
Sebastiaan,
A provocative design; essentially a scalable "chip amp" that combines the LME49811 as a simple-to-implement, high-quality front-end and an inherently simple/stable ThermalTrak output-stage.
I haven't yet experimented with the ThermalTrak output-stage devices, so my observations may be way off. However, now that you've reduced the output device configuration to a two-stage Darlington EF, I believe that you need to alter your output-stage biasing scheme to account for the reduced number of cascaded devices in the EF output. Currently, you have three ThermalTrak bias diodes configured on each half of the push-pull EF output stage, which appears to me to leave the output stage seriously over-biased into Class-A operation (approximately 6-7A idle-current per output device). Or is my lack of experience with the ThermalTrak devices showing? 😉
Perhaps reduce the design to just two pairs of output devices per side of the balanced amplifier? Or would it be appropriate to reconfigure for four pairs of output devices per side, running the ThermalTrak diodes as two parallel strings of 4 diodes?
A provocative design; essentially a scalable "chip amp" that combines the LME49811 as a simple-to-implement, high-quality front-end and an inherently simple/stable ThermalTrak output-stage.
I haven't yet experimented with the ThermalTrak output-stage devices, so my observations may be way off. However, now that you've reduced the output device configuration to a two-stage Darlington EF, I believe that you need to alter your output-stage biasing scheme to account for the reduced number of cascaded devices in the EF output. Currently, you have three ThermalTrak bias diodes configured on each half of the push-pull EF output stage, which appears to me to leave the output stage seriously over-biased into Class-A operation (approximately 6-7A idle-current per output device). Or is my lack of experience with the ThermalTrak devices showing? 😉
Perhaps reduce the design to just two pairs of output devices per side of the balanced amplifier? Or would it be appropriate to reconfigure for four pairs of output devices per side, running the ThermalTrak diodes as two parallel strings of 4 diodes?
- Status
- Not open for further replies.