I prefer the EL84 with 20% UL tap.
Mullard and Marconi (N709 = EL84 ) have designs with a 20% tap and Dynaco ST35 initially used a 31.6% tap (IIRC) and later used 20% tap.
If you route the cathode through the 8 Ohm speaker winding then you do not need NFB, distortion stays below 2%.
Mullard and Marconi (N709 = EL84 ) have designs with a 20% tap and Dynaco ST35 initially used a 31.6% tap (IIRC) and later used 20% tap.
If you route the cathode through the 8 Ohm speaker winding then you do not need NFB, distortion stays below 2%.
Last edited:
To drop the voltage you can use a "swinging choke", this has the advantage of reducing the load on your mains transformer and stabilizing the DC voltage with varying load. A swinging choke is NOT the same as an ordinary choke due to the very high voltage swings in them. They are also expensive. (Sowter in the UK has them)
Measure the wieght and measure the resistance across the HT winding. Then you can look at the Hammond transformers and "guess" the approximate current capabilities.
Never ever hook it up directly to the mains.
An option may be to use a standard 230 to 230V insulating transformer, you should be able to find one of those. Another option is to find two identical lets say 230 to 12 transformers and then you couple the two 12 volt windings. One transformer is to the mains stepping down to 12V and then the other transformers' 12 volt winding brings it back up to 230V but it is isolated from the mains.
Measure the wieght and measure the resistance across the HT winding. Then you can look at the Hammond transformers and "guess" the approximate current capabilities.
Never ever hook it up directly to the mains.
An option may be to use a standard 230 to 230V insulating transformer, you should be able to find one of those. Another option is to find two identical lets say 230 to 12 transformers and then you couple the two 12 volt windings. One transformer is to the mains stepping down to 12V and then the other transformers' 12 volt winding brings it back up to 230V but it is isolated from the mains.
Last edited:
Schade feedback can be applied to triodes as well so there is no reason why it can't benefit UL. Its something of a misconception to say that you are aiming for pentode power with triode behaviour. All Schade actually does is lower the Output impedance to whatever it is applied to and also linearilzes the response like any feedback arrangement.
Shoog
Shoog
Well, the reason why I ask is that an EL84 with the right load (usually lower value) and 20-25% UL connection can have almost the same power as a pentode with half the THD and the internal resistance of a triode. Applying shunt feedback you reduce even more the internal resistance while keeping high gain and power.
...so why not? Same applies to small amount of cathode feedback, that would have been my next question.
...so why not? Same applies to small amount of cathode feedback, that would have been my next question.
The only thing with shunt feedback is if the driver is not a stiff load - it tends to collapse well before you can achieve the potential power of the output stage pentode. The reality is that it can only deliver a relative small proportion of that power cleanly.
In simulations I performed cathode feedback was better able to deliver clean higher power than the RH designs - because it is less demanding on the driver stage.
Shoog
In simulations I performed cathode feedback was better able to deliver clean higher power than the RH designs - because it is less demanding on the driver stage.
Shoog
Last edited:
If you are using a pentode as the driver then it overcomes the limitations of the driver stage. Even a very low power pentode is able to resist the current coming back across the feedback resistor and perform its duty.
There are two good ways top drive a Schaded output tube - with a transformer ( or autotransformer) or with a pentode.
Shoog
There are two good ways top drive a Schaded output tube - with a transformer ( or autotransformer) or with a pentode.
Shoog
I have not gone back to using a 12A*7 since I started to use a 5670 / 2C51 / Russian 6N3P-EV as driver. The main reason is that it is more linear, can work with about 7 ~ 8dB NFB and the result is a far gentler overload when overdriven.
I have also gone to "correctly sizing" the OPT - for an EL84 SE I use a 7 Watt Transcendar OPT. Too much iron seems to loose some of the finest details. I have now a pair of 10 Watt Transcendar OPT's on order for a PSE EL84 but honestly I have my doubt that it wil work out better than the single EL84. Won't know till I try.
( this was after I have read a (IIRC) German story of someonewho replaced his 300B with three EL84 in parallel on the same OPT and prefers that to the 300B.)
Results of 20% UL + about 8dB GNB were about the same for 20% UL + cathode feedback on a 8 Ohm tap and no GNFB.
I did model Schade feedback and in the past build a 6L6GC with Schade feedback (and James mains, OPT's and choke) and was not impressed. It will give better results with cr@p OPT's but the moment you put in a decent OPT the whole scenario changes.
You have to ask yourself why Schade feedback was never implemented in higher quality amplifiers - Schade FB is a lot easier to implement, no frequency correction necessary, only a resistor. The cheap stuff (broadcast radios etc) just did not have any NFB. The slightly better stuff had PP without NFB and up the ladder it goes. Matter of economics.
UL+cathode FB differed in the very low levels of output, around the 50 ~ 100 mW range when the speaker impedance was changed from a nominal 8 Ohm load. In the end I moved away from the cathode feedback. Did try it briefly but Dianna Krall and Alison Krauss were not having the same ambience. Perhaps I did not give it enough time, perhaps it might work better with different speakers but the Proac Tablette 10's, the Wharfedale Denton's anniversary models and the AVI Neutron III all had the same results. YMMV.
I have also gone to "correctly sizing" the OPT - for an EL84 SE I use a 7 Watt Transcendar OPT. Too much iron seems to loose some of the finest details. I have now a pair of 10 Watt Transcendar OPT's on order for a PSE EL84 but honestly I have my doubt that it wil work out better than the single EL84. Won't know till I try.
( this was after I have read a (IIRC) German story of someonewho replaced his 300B with three EL84 in parallel on the same OPT and prefers that to the 300B.)
Results of 20% UL + about 8dB GNB were about the same for 20% UL + cathode feedback on a 8 Ohm tap and no GNFB.
I did model Schade feedback and in the past build a 6L6GC with Schade feedback (and James mains, OPT's and choke) and was not impressed. It will give better results with cr@p OPT's but the moment you put in a decent OPT the whole scenario changes.
You have to ask yourself why Schade feedback was never implemented in higher quality amplifiers - Schade FB is a lot easier to implement, no frequency correction necessary, only a resistor. The cheap stuff (broadcast radios etc) just did not have any NFB. The slightly better stuff had PP without NFB and up the ladder it goes. Matter of economics.
UL+cathode FB differed in the very low levels of output, around the 50 ~ 100 mW range when the speaker impedance was changed from a nominal 8 Ohm load. In the end I moved away from the cathode feedback. Did try it briefly but Dianna Krall and Alison Krauss were not having the same ambience. Perhaps I did not give it enough time, perhaps it might work better with different speakers but the Proac Tablette 10's, the Wharfedale Denton's anniversary models and the AVI Neutron III all had the same results. YMMV.
Last edited:
What I've seen in PP amps is that the shunt feedback "balances" the driver and output stage, and it's best is achieved when both "collapse" together: too less shunt feedback and the output tube won't work as linear as it could, too much shunt feedback and the driver won't be able to drive the power tube. It's just a matter of finding the right values.
There's an implementation on Radio Designers Handbook, then Wolcott amps always used shunt feedback, and I'm sure many more. I also feel that once a brand stick on a method, it often keeps optimizing that one vs changing the whole system. It is true that is easy to implement, but it's not that easy to find the right values to optimize the combination of the driver and the output stage. Considering that, a standard gnfb is way easier.
No because what you are depending on is a high fixed resistance which is able to resist the feedback current from overwhelming the driver current. Only pentodes provide this. Transformers behave differently but also work.
Schade is no magic bullet solution, it does what it does well but brings its own demands on the design.
Shoog
I would take the typical operating characteristics as my guide
http://www.r-type.org/pdfs/6au6-1.pdf
Depending on what B+ you have available I would go with something like:
250V Plate Voltage given a B+ of 350V and a load resistor of approx 20K and a feedback resistor of 100K
125V screen derived by a simple voltage divider with a smoothing cap
Plate current 7.6mA and Screen current 3mA
Cathode resistor should be about 520R
I would take this as my starting point and use Eagle to refine these values.
The main quality that is important is that the plate resistance of the driver is constant at whatever plate voltage (ie signal amplitude) it is operating at. This rules out triodes which have a variable and low plate resistance dependent on their gain.
But here's the kicker - cathode feedback from the output valve plate to driver cathode can match this whilst using a simpler triode as the driver.
Shoog
http://www.r-type.org/pdfs/6au6-1.pdf
Depending on what B+ you have available I would go with something like:
250V Plate Voltage given a B+ of 350V and a load resistor of approx 20K and a feedback resistor of 100K
125V screen derived by a simple voltage divider with a smoothing cap
Plate current 7.6mA and Screen current 3mA
Cathode resistor should be about 520R
I would take this as my starting point and use Eagle to refine these values.
The main quality that is important is that the plate resistance of the driver is constant at whatever plate voltage (ie signal amplitude) it is operating at. This rules out triodes which have a variable and low plate resistance dependent on their gain.
But here's the kicker - cathode feedback from the output valve plate to driver cathode can match this whilst using a simpler triode as the driver.
Shoog
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.