I spent a few hours fine tuning this little amp.
What I found worked best for me was increasing the local feedback around the pentodes by increasing the 12k to 15K.
I then tried various levels of global feedback between 14dB and none and finally settled on exactly 6dB (150K with parallel 10pF from the 4 Ohm speaker tap with the triode input to 0V resistor of 5K6 - this resistor was 10K on Yves original circuit). This corresponded to a gain to the 4 Ohm Output of about 12.
More global feedback destroys the stereo image. Less and you don't have enough damping. 6dB was a good compromise - it halved the output impedance and didn't destroy the imaging, in fact I felt it made the image more focussed.
I've now declared this little amplifier job FINISHED. Well it needs a base plate and some rubber feet BUT its electronically finished.
Cheers,
Ian
What I found worked best for me was increasing the local feedback around the pentodes by increasing the 12k to 15K.
I then tried various levels of global feedback between 14dB and none and finally settled on exactly 6dB (150K with parallel 10pF from the 4 Ohm speaker tap with the triode input to 0V resistor of 5K6 - this resistor was 10K on Yves original circuit). This corresponded to a gain to the 4 Ohm Output of about 12.
More global feedback destroys the stereo image. Less and you don't have enough damping. 6dB was a good compromise - it halved the output impedance and didn't destroy the imaging, in fact I felt it made the image more focussed.
I've now declared this little amplifier job FINISHED. Well it needs a base plate and some rubber feet BUT its electronically finished.
Cheers,
Ian
I have the chassis and more concrete plans coming together for an EL84/6AU6 version of this amp....
http://www.diyaudio.com/forums/showthread.php?postid=647744#post647744
dave
http://www.diyaudio.com/forums/showthread.php?postid=647744#post647744
dave
Dave,
Some more experimental results of my version on this thread. So it was'nt actually "finished" after all, TYPICAL!!!
http://www.diyaudio.com/forums/showthread.php?s=&threadid=57411
As you seem to have anticipated the local feedback is severly attenuated by the rp of the diff amp triodes. Hence your version using 6AU6 pentodes would work MUCH better (higher rp from a pentode). Good luck with it.
As a hint from all my mucking about
You need enough local feedback to say halve the rp of the output tubes. This pushes the low frequency roll off (output tube rp and output tranny primary inductance) an octave lower and at the same time pushes the high frequency roll off (output tube rp and leakage inductance/effective primary capacitance) a bit higher - not necessarily a full octave because the high frequency roll off is 12dB/octave not 6dB/octave.
The more local feedback (the higher the high frequency roll off in the output tube/output tranny) the easier it is to apply global feedback and keep it stable BUT also the less global feedback you actually need for decent damping factor. A win/win situation. Less global feedback will do stunning things to your stereo image.
With the ECL86 version I also ran into problems with not enough gain to apply as much local feedback as I would have liked. Won't be a problem with 6AU6 pentodes - in fact you may end up with the opposite problem, too much gain.
All in all a great little circuit for screwing the best out of cheaper output trannies.
Cheers,
Ian
Some more experimental results of my version on this thread. So it was'nt actually "finished" after all, TYPICAL!!!
http://www.diyaudio.com/forums/showthread.php?s=&threadid=57411
As you seem to have anticipated the local feedback is severly attenuated by the rp of the diff amp triodes. Hence your version using 6AU6 pentodes would work MUCH better (higher rp from a pentode). Good luck with it.
As a hint from all my mucking about
You need enough local feedback to say halve the rp of the output tubes. This pushes the low frequency roll off (output tube rp and output tranny primary inductance) an octave lower and at the same time pushes the high frequency roll off (output tube rp and leakage inductance/effective primary capacitance) a bit higher - not necessarily a full octave because the high frequency roll off is 12dB/octave not 6dB/octave.
The more local feedback (the higher the high frequency roll off in the output tube/output tranny) the easier it is to apply global feedback and keep it stable BUT also the less global feedback you actually need for decent damping factor. A win/win situation. Less global feedback will do stunning things to your stereo image.
With the ECL86 version I also ran into problems with not enough gain to apply as much local feedback as I would have liked. Won't be a problem with 6AU6 pentodes - in fact you may end up with the opposite problem, too much gain.
All in all a great little circuit for screwing the best out of cheaper output trannies.
Cheers,
Ian
Damping factor of Yves
I found Yves's designe elegant, mainly because simmetry and simplicity.
I tried to do one with PCL82, that I found in my drawers. Usying Pentode output and the max feedback I can compatible with the needed gain to have a sensitivity of 1Vrms, the output resistance is too high.
There is some trick in the PCL86, apart the different mu of the first tube? How muc is it the dampoing factor of the amplifier proposed?
Thanks,
I found Yves's designe elegant, mainly because simmetry and simplicity.
I tried to do one with PCL82, that I found in my drawers. Usying Pentode output and the max feedback I can compatible with the needed gain to have a sensitivity of 1Vrms, the output resistance is too high.
There is some trick in the PCL86, apart the different mu of the first tube? How muc is it the dampoing factor of the amplifier proposed?
Thanks,
My opinion is that the Ultralinear connection is just about mandatory to get good damping with low output impedance. In pentode mode you need more feedback so you also need more open loop gain. 6GW8 (ECL86) will therefore be better than 6BM8 (ECL82).
Some thing you might try - cascode the diff amp front end. That is the triodes sitting on J FETs (matched) or even biploar transitors. This will increase the gain AND increase the effective rp of the diff amp stage which will help with the shunt feedback.
Cheers,
Ian
Some thing you might try - cascode the diff amp front end. That is the triodes sitting on J FETs (matched) or even biploar transitors. This will increase the gain AND increase the effective rp of the diff amp stage which will help with the shunt feedback.
Cheers,
Ian
Trying to get answers for you on using using a transistor for the bottom device and 6BM8 triodes in the top device of a cascode for each side of the diff amp.
I've simulated it and it should work BUT simulator clearly showed that Darlington connected BC547Bs for each lower device rather than just a single BC547B worked much better.
I've asked for guru advice here:
http://www.audioasylum.com/forums/tubediy/messages/87785.html
Let you know what responses I get or check it for yourself.
Cheers,
Ian
I've simulated it and it should work BUT simulator clearly showed that Darlington connected BC547Bs for each lower device rather than just a single BC547B worked much better.
I've asked for guru advice here:
http://www.audioasylum.com/forums/tubediy/messages/87785.html
Let you know what responses I get or check it for yourself.
Cheers,
Ian
- Status
- Not open for further replies.