#1 EL509 OTL Amplifier
Posted: Sat Jun 23, 2007 11:45 am
Thought I'd post the progress of the OTL amp on here. Don't really want to start going on on the WD forum about how wonderful a design sounds when it is derived from a commercial product.
Bruce Rozenblit does a similar thing to Nelson Pass and appears to let his discontinued and strangely some of his current production stuff enter the public domain after a period of time. Bruce's "Audio Reality" PDF download has loads of designs in it and I was particularly taken with his 80 watt OTL amp the Transcendent T8 which he has now replaced with the T16 Beast OTL commercially.
After hearing Andrews 6AS7 OTL at Owston I started investigating this type of amp. The attractions are obvious, not least the elimination of the need for expensive output iron. The problems are also challenging.
I have now reached the stage with this hobby where any further progress in my understanding is going to be accompanied by a great deal of expense if I continue along the single ended/ big output transformer route.
Much as I'd love to keep the SE faith I'm frankly bored with it. The sound is great but hobbyists need to be doing stuff and I don't want to build yet another SE amp. Seeing the lengths Paul has had to go to and to a lesser extent Nick to wring the ultimate performance out of the topology, I think a new direction would be in order, for me at least.
I'll put snippets on the WD forum but the bulk of the pictures and stuff will be in this thread.
I'm currently burning the tubes in using my existing single-ended amp. I've re biased the amp to give the PSU an easier time of it. The current these EL509s draw caused me a bit of concern with the PSU, which wasn't designed to handle 200mA per power tube. The chokes and mains transformer are only rated at 200mA and 250mA respectively so there was quite a lot of heat being generated. I've halved the current draw so the PSU is now only running flat out
instead of positively wilting at twice the rated draw.
This amp needs a lot of voltages from its PSU:
First we require a 500V 100mA pre/driver stage supply
Then a 120V negative bias supply for the output stage
Also a 240V 1.4 amp B+ supply for the output stage
Finally a 20V 8A supply for the series heater chain to the output stage
Sowter are winding me a power TX which will supply the 500V and the output stage B+ plus three heater supplies for the pre/splitter/driver stage.
I'll be using off the shelf transformers for the bias and the 20V 8A output stage series heaters.
The circuit itself uses half a 12AX7 to pick up the preamp signal and provide voltage gain. This is then DC coupled to a 12aU7 long tail differential phase splitter/driver where it is boosted a bit more. The signal is then once more DC coupled to a 12AU7 cathode follower buffer stage.
The upper buffer applies a corrective signal to compensate for its own degeneration by using three zener diodes in series across the anode/cathode circuits of the upper buffer, adding the zener network's output to the signal being fed to the upper bank of output tubes, which are themselves configured as cathode followers. Because the upper output bank has degeneration whilst the lower bank has gain this compensation is needed to avoid an unbalanced output. The way this compensation is applied is different to the Futterman scheme and is the basis of Rozenblit's patent
The lower bank of tubes is just fed a straight signal by the lower 12Au7 buffer and has no corrective boost applied, thus we end up with a symmetrical recombined signal at the output with only miniscule DC offset. This is important because the output stage is direct coupled to the speakers with no intervening capacitor.
In fact there is only one cap in the direct signal path per channel. This decouples the driver stage from the output stage to isolate the driver from the negative bias being fed to the output stage grids. Thus we have two DC coupled sections isolated from each other by the single cap.
It all sounds a bit complex in writing but the schematic looks simple if you don't panic.
I have four EL509s per channel as I'm building the 25 watt version. The 80 watt version uses eight EL509s per side. The amp gets its power by being class AB and it uses 22dB of negative feedback.
I'll post more as I get into the build. Should be interesting.
Steve
Bruce Rozenblit does a similar thing to Nelson Pass and appears to let his discontinued and strangely some of his current production stuff enter the public domain after a period of time. Bruce's "Audio Reality" PDF download has loads of designs in it and I was particularly taken with his 80 watt OTL amp the Transcendent T8 which he has now replaced with the T16 Beast OTL commercially.
After hearing Andrews 6AS7 OTL at Owston I started investigating this type of amp. The attractions are obvious, not least the elimination of the need for expensive output iron. The problems are also challenging.
I have now reached the stage with this hobby where any further progress in my understanding is going to be accompanied by a great deal of expense if I continue along the single ended/ big output transformer route.
Much as I'd love to keep the SE faith I'm frankly bored with it. The sound is great but hobbyists need to be doing stuff and I don't want to build yet another SE amp. Seeing the lengths Paul has had to go to and to a lesser extent Nick to wring the ultimate performance out of the topology, I think a new direction would be in order, for me at least.
I'll put snippets on the WD forum but the bulk of the pictures and stuff will be in this thread.
I'm currently burning the tubes in using my existing single-ended amp. I've re biased the amp to give the PSU an easier time of it. The current these EL509s draw caused me a bit of concern with the PSU, which wasn't designed to handle 200mA per power tube. The chokes and mains transformer are only rated at 200mA and 250mA respectively so there was quite a lot of heat being generated. I've halved the current draw so the PSU is now only running flat out
instead of positively wilting at twice the rated draw.This amp needs a lot of voltages from its PSU:
First we require a 500V 100mA pre/driver stage supply
Then a 120V negative bias supply for the output stage
Also a 240V 1.4 amp B+ supply for the output stage
Finally a 20V 8A supply for the series heater chain to the output stage
Sowter are winding me a power TX which will supply the 500V and the output stage B+ plus three heater supplies for the pre/splitter/driver stage.
I'll be using off the shelf transformers for the bias and the 20V 8A output stage series heaters.
The circuit itself uses half a 12AX7 to pick up the preamp signal and provide voltage gain. This is then DC coupled to a 12aU7 long tail differential phase splitter/driver where it is boosted a bit more. The signal is then once more DC coupled to a 12AU7 cathode follower buffer stage.
The upper buffer applies a corrective signal to compensate for its own degeneration by using three zener diodes in series across the anode/cathode circuits of the upper buffer, adding the zener network's output to the signal being fed to the upper bank of output tubes, which are themselves configured as cathode followers. Because the upper output bank has degeneration whilst the lower bank has gain this compensation is needed to avoid an unbalanced output. The way this compensation is applied is different to the Futterman scheme and is the basis of Rozenblit's patent
The lower bank of tubes is just fed a straight signal by the lower 12Au7 buffer and has no corrective boost applied, thus we end up with a symmetrical recombined signal at the output with only miniscule DC offset. This is important because the output stage is direct coupled to the speakers with no intervening capacitor.
In fact there is only one cap in the direct signal path per channel. This decouples the driver stage from the output stage to isolate the driver from the negative bias being fed to the output stage grids. Thus we have two DC coupled sections isolated from each other by the single cap.
It all sounds a bit complex in writing but the schematic looks simple if you don't panic.
I have four EL509s per channel as I'm building the 25 watt version. The 80 watt version uses eight EL509s per side. The amp gets its power by being class AB and it uses 22dB of negative feedback.
I'll post more as I get into the build. Should be interesting.
Steve
