I don't see no obvious problem with this version. Except maybe a Wilson Mirror
might have be worth an extra transistor? Screen current subtracts out the tail,
never seen by OPT matching equation. All good so far, driver will make or break it.
might have be worth an extra transistor? Screen current subtracts out the tail,
never seen by OPT matching equation. All good so far, driver will make or break it.
Does this mean the ECL86 is no longer usable in this design or is there a variation of the circuit that will work with ECL86?
There is active auto-biasing in this circuit.
Many pentodes could probably drop right in.
Adjust R5 if bias is grossly incorrect.
Many pentodes could probably drop right in.
Adjust R5 if bias is grossly incorrect.
ECL86 pentode section should work well. Power is a little lower, but the quality should be good, and the build cost very low.
The models for it are in the same Koren library as the JJ valves, and I would say these are good.
The triode section may be a bit short on gm to keep it to a Spud, unless some of the damping factor and distortion is traded off by increasing the impedance of the Schade network.
As Ken has said, the driver will make or break the design as a whole. I have got something in mind to post shortly, which I hope will do it justice!
The models for it are in the same Koren library as the JJ valves, and I would say these are good.
The triode section may be a bit short on gm to keep it to a Spud, unless some of the damping factor and distortion is traded off by increasing the impedance of the Schade network.
As Ken has said, the driver will make or break the design as a whole. I have got something in mind to post shortly, which I hope will do it justice!
Today I took the RS 120+120:12V Toroidal 100VA trafo and unwound some turns. Reconnecting the mains, I measured 2.7V reduction for every 10 turns stripped off. I took it down to about 7.7V (ie about 8000:8 Ohms) and tested the amp again.
For EL84 at 210V/52mA dc balanced (but without screen current compensation and without Schade network) I got a little over 7W clean output.
The simulation promises over 8W clean, so I better get to work on the driver.
A PL82 gave just about the same output, but needed some action on the trimmer to balance the dc current.
For EL84 at 210V/52mA dc balanced (but without screen current compensation and without Schade network) I got a little over 7W clean output.
The simulation promises over 8W clean, so I better get to work on the driver.
A PL82 gave just about the same output, but needed some action on the trimmer to balance the dc current.
Hey jkeny,
There are no problems using the triode section as driver as long as you add the MOSFET follower. To use half a ECC83 as Schade driver alone is a total dead end as it can not supply enough current.
If we go for a fully grown "nonspud" with EL84 a pentode-driver is ideal, this one is with 6AU6 but better ones as E280F or D3a could be used:
There are no problems using the triode section as driver as long as you add the MOSFET follower. To use half a ECC83 as Schade driver alone is a total dead end as it can not supply enough current.
If we go for a fully grown "nonspud" with EL84 a pentode-driver is ideal, this one is with 6AU6 but better ones as E280F or D3a could be used:
An externally hosted image should be here but it was not working when we last tested it.
Rod,
It might be the JJ models that are wrong. Not that the models themselves should be at fault, but the curves published by JJ are not to be trusted.
It might be the JJ models that are wrong. Not that the models themselves should be at fault, but the curves published by JJ are not to be trusted.
Lars,
I think the models are good because they are are very good match for the waveforms on the circuit I built.
I think they were made by measuring real parts, rather than the data sheets...
Rod,
In a previous post you said they didn´t match as you didn´t get the indicated power😉.
JJs that I first used in our guitaramps proved to be both inconsistent and unreliable. This has shown specially with their 7591S. EL84 might be another story.
About your screen-compensation it seems like you must choose a lot lower screenvoltage than anodevoltage. How big must the difference be?
Lars, I didn't get the indicated power because I was sweating the poor EL84 under the 2,6K:8 Ohm transformer! The toroidals measured 13,3V open circuit.
I have heard a lot about poor quality in JJ octal valves. But the EL84 is a winner, best sounding and reliable, the best way to get those old AC30s singing again!
To set the screen voltage, find out the maximum (peak) screen current you expect. Then set R15 so that the screen current drops about 20V with this max current. Then set screen voltage, say 30V lower to give a margin.
R16,17,18 are set to be the same as R15. You can set R15 to drop only 10V peak if you like, but that will degrade the accuracy a little.
So with EL84, I expect no more than 20mA peak at 210V in a good circuit, so R15=1K for 20V dropped. So VG2 can be 25V at least below B+. But I like VG2 to be as low as possible, because that will reduce the screen current during music playing.
With 8K ohm load, there is no difference in power output for VG2=270V or 210V.
I posted my driver circuit idea:
http://www.diyaudio.com/forums/tubes-valves/153318-shunt-cascode-power-valve-driver.html#post1949200
http://www.diyaudio.com/forums/tubes-valves/153318-shunt-cascode-power-valve-driver.html#post1949200
Member
Joined 2009
Paid Member
I'm starting to understand more of this now (tubes that is) and this thread was something I remember looking at last year.
Did any of these ideas get built - any updates ???
Did any of these ideas get built - any updates ???
Hey Bigun,
Didn´t you read the whole thread? Rod built the outputstage IRL. The driver wasn´t so interesting as it was quite straightforward.
Didn´t you read the whole thread? Rod built the outputstage IRL. The driver wasn´t so interesting as it was quite straightforward.
Member
Joined 2009
Paid Member
Sorry, I didn't mean to be lazy, but it seemed as if there was only a prototype and it was built to compare with Spice models rather than as an amplifier to listen to. I didn't get an sense as to how it sounds ??
Hey Bigun,
What would anyone elses opinion matter? It would tell you nothing as taste is individual. Also noone would admit that their own design sounds crappy😉.
Though we have built other PP-amps on this theme that we like.
Build to find out, in a SE version first as the PP is rather complicated. The SE is a fairly std Schade. As I see it this is an amp for those who already have a stash of ECL86 lying around.
Personally I would build a Schade SE EL84 like the one I designed for Klimon as a first. It has the same topology but without the MOSFET and with another more suitable driver tube.
What would anyone elses opinion matter? It would tell you nothing as taste is individual. Also noone would admit that their own design sounds crappy😉.
Though we have built other PP-amps on this theme that we like.
Build to find out, in a SE version first as the PP is rather complicated. The SE is a fairly std Schade. As I see it this is an amp for those who already have a stash of ECL86 lying around.
Personally I would build a Schade SE EL84 like the one I designed for Klimon as a first. It has the same topology but without the MOSFET and with another more suitable driver tube.
revintage: all of your pics in this thread are gone. is there a final schem you favoured that you have handy? Else I'll copy Rod's last.
- Status
- Not open for further replies.