Hello,
I create and build this power amplifier.
Ideas for building were to check mr. Lavardin theory about thermal distortions.
I used low feedback, open loop gain about 50db. Traditional distortion is low till half power. Currents for first stage 1mA, for VAS 2,5 mA, for EF 4,6 / 12 / 200 mA.
Sound is clear, voices and instruments have clear separation.
For listening was used Klipsh RF-3 speakers, sensitivity 98db.
I am looking for ideas how to improve this amplifier.
I create and build this power amplifier.
Ideas for building were to check mr. Lavardin theory about thermal distortions.
I used low feedback, open loop gain about 50db. Traditional distortion is low till half power. Currents for first stage 1mA, for VAS 2,5 mA, for EF 4,6 / 12 / 200 mA.
Sound is clear, voices and instruments have clear separation.
For listening was used Klipsh RF-3 speakers, sensitivity 98db.
I am looking for ideas how to improve this amplifier.
Member
Joined 2009
Paid Member
I have looked at these claims before. In the experiments I saw posted on the web, simple topologies are compared with more complex topologies. The use of CFP and of Cascodes. The results seem to be positive - less distortion and clearer sound. The theory makes sense, but it is claimed that memory distortion is an serious issue and that it's removal makes a big difference to the sound.
BUT the issue I have is that the experiments posted on the web do not prove that 'memory distortion' is a real practical issue that affects sound quality.
The reason is that these more complex topologies are inherently superior in technical performance even if you ignore ALL aspects of memory distortion from the analysis. The CFP has very low distortion compared with a single device. A Cascode reduces the Early Effect and greatly reduces Miller capacitance. You can't separate out the huge performance improvements from these more complex topologies due to better linearity and higher OLG from their impact on so-called memory distortion.
These are all very good design approaches and make for good amplifiers. But I don't see any evidence that 'memory distortion' is an issue for sound quality, or that it has been vanquished as an issue with these improvements. But it sure does make for some fun marketing.
I would like to add that the website you linked is very nicely presented and some good quality work is presented there. I admire the effort taken to do the work and share it for us to see.
BUT the issue I have is that the experiments posted on the web do not prove that 'memory distortion' is a real practical issue that affects sound quality.
The reason is that these more complex topologies are inherently superior in technical performance even if you ignore ALL aspects of memory distortion from the analysis. The CFP has very low distortion compared with a single device. A Cascode reduces the Early Effect and greatly reduces Miller capacitance. You can't separate out the huge performance improvements from these more complex topologies due to better linearity and higher OLG from their impact on so-called memory distortion.
These are all very good design approaches and make for good amplifiers. But I don't see any evidence that 'memory distortion' is an issue for sound quality, or that it has been vanquished as an issue with these improvements. But it sure does make for some fun marketing.
I would like to add that the website you linked is very nicely presented and some good quality work is presented there. I admire the effort taken to do the work and share it for us to see.
Last edited:
But the greatest thermal cycling is in the power output devices. So that is where I would expect thermal distortion, not in the pre stages.
jan
Hit the point, it is the output stage that needs a reduction in thermal distortion.
Have been searching for a topology that reduces thermal distortion in output stage for the last few days, but it seems that there is no published articles about such an output stage...
Using ThermakTrak output transistors, plus exceptional Vbe temperature compensation has a larger thermal memory effect than some of the other specialized cascodes etc. in the Lavaradin Amp paper.
Andrew C Russell
"Ideas on Temperature Compensation for Emitter Follower Triples in Class AB Audio Amplifiers"
Andrew uses opamps with precise multi-slope gain to maintain the lowest distortion temperature independent output bias points.
yes, 50 dB more loop gain for a total 100 dB over the audio range is much more powerful than discrete device limited matching and lousey tracking
low error feedback network, high loop gain, two-pole compensation
and then maybe input Q bootstrap cascode if you want to think you've done something about a non-problem
low error feedback network, high loop gain, two-pole compensation
and then maybe input Q bootstrap cascode if you want to think you've done something about a non-problem
Dan D'Agostino amplifiers reviews are very positive. These amplifiers have AB emitter followers in output stage. Emitter resistor is 1 ohm. Such resistor ensure that Vbe (measured between base and emitter resistor connection to output) is stable over temperature range. This idea can be adapted for low power stages.
For instance Simulations shows that MJE15030 having 5 ohm emitter resistor and 70ma bias have stable Vbe over temperature range 25-150C.
BD139 have same behavior with 45ma bias and 10ohm emitter degeneration resistor. This working point used in Lehmann headphone amplifier having great reviews as well.
I found that exist patents claiming what traditional amplifiers having heat element gives better sound. There talking that semiconductor crystals should kept in at least 75C. From this temperature increasing of sound quality begins. For instance ouptut transistors in NAIM amplifiers have gasket of around 4mm between heatsink. Maybe this gasket ensure high working temperature of output transistor.
Maybe these ideas can be used in output stage design.
For instance Simulations shows that MJE15030 having 5 ohm emitter resistor and 70ma bias have stable Vbe over temperature range 25-150C.
BD139 have same behavior with 45ma bias and 10ohm emitter degeneration resistor. This working point used in Lehmann headphone amplifier having great reviews as well.
I found that exist patents claiming what traditional amplifiers having heat element gives better sound. There talking that semiconductor crystals should kept in at least 75C. From this temperature increasing of sound quality begins. For instance ouptut transistors in NAIM amplifiers have gasket of around 4mm between heatsink. Maybe this gasket ensure high working temperature of output transistor.
Maybe these ideas can be used in output stage design.
Heat propagates by diffusion so does carrier recombination in base zone of BJT.The only efficient thermal compensation in output is the Quad triple cascade and Elvee Circlophone.
A very simple idea is to keep the voltage amp predrivers at constant temp . Not much power is required for that but PCB must be designed accordingly.
A very simple idea is to keep the voltage amp predrivers at constant temp . Not much power is required for that but PCB must be designed accordingly.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.