Cascode with ECC81(12AT7)=>Morgan Jones

[SY]

Just for the record:

ECC88/10mA CCS/1kHz/14VRMS out (average of 20 samples)

2nd -43dB
3rd -69dB
4th -84dB
5th -88dB

ECC81/2mA CCS/1kHz/14VRMS out (average of 8 samples, JJ)

2nd -63dB
3rd -89dB
4th -96dB
5th <-100dB

ECC81s that were not JJ type measured, on average, about 6dB worse in distortion, still a better performance at moderately large voltage swings than ECC88.

I think you're looking at those curves the wrong way. Brian Beck has argued that the figure of merit for linearity ought to be the change in mu with voltage. I think he's dead right about that. It is instructive to look at those graphs on the respective datasheets.

 

[AudioAngel]

Ok I've discarded the idea of building a cascode diff amp with this tubes.
I'll go back to a solid state preamp. No, just a joke.
Following reinhard's hint I had a look on BAT products. Therefore I'll try a design using a single stage design like BAT with a balanced diff amp and plate loaded triodes. But I think its better to open a new thread with my questions concerning this topic.
However I still want to know wehter my design is correct or not and what can be improved. Now its more an exercise and lerning task.

Raphael

 

[SY]

For a line stage, you don't need much gain unless you're using really odd sources and/or power amps. Low mu tubes are your friend. I hate wasting a 12B4 in preamp service, but it really does make a good line amp.

 

[AudioAngel]

Hello,

just for information: I found a schematic where Rainer von der Linde uses a ECC81 cascode for a line stage. It seems that this schematic is taken from his book. Can you say something to that?

Raphael

 

[moamps]

Since it was I who referred you to that particular schematic, I wish to comment. Rainer zur Linde uses the same topology for the line stage and phono stage (grounded cathode amplifier, cascode, cathode follower), which is a most unusual design choice. The open loop gain of the line amplifier is 56dB and NFB is 37dB, which gives a closed loop gain of 19dB. I'm pretty sure that none of the more experienced valve designers among us would choose that path in designing a line stage, which is what SY has been trying to tell you. Simply put, the cascode may be good for the phono stage but not necessarily for the line stage.

Regards,
Milan

 

[kevinkr]

Since it was I who referred you to that particular schematic, I wish to comment. Rainer zur Linde uses the same topology for the line stage and phono stage (grounded cathode amplifier, cascode, cathode follower), which is a most unusual design choice. The open loop gain of the line amplifier is 56dB and NFB is 37dB, which gives a closed loop gain of 19dB. I'm pretty sure that none of the more experienced valve designers among us would choose that path in designing a line stage, which is what SY has been trying to tell you. Simply put, the cascode may be good for the phono stage but not necessarily for the line stage.

Regards,
Milan

There are certain other oddities in his line stage designs as well, things like volume and balance controls inside the feedback loop.. See p.80 -81, and 98 in "Build Your Own Audio Valve Amplifiers" for examples of this. He uses multiple mu-followers in the first cited example and a cascode in the second.

How much gain do you need in a line stage, and to get it down to reasonable levels lots of FB must be used as previously pointed out.

Note in the phono stage he uses a cascode in the second stage where its low miller capacitance is large irrelevant, but worse he places the theoretically higher gain stage in the second position whereas normally you'd do the reverse for better overall noise performance. (see p.96)

Simple would be better, a single 5842 driving a really good plate to line transformer would run circles around these designs. I prefer dht, but not for the newbie...

 

[reinhard]

It is the bad PSRR why he put the cascode in the second stage,
that is Audio Research Style, see SP 10

Reinhard

 

[kevinkr]

It is the bad PSRR why he put the cascode in the second stage,
that is Audio Research Style, see SP 10

Reinhard

The ARC SP10 has extremely heavily regulated supplies so poor PSRR is not a consideration in that design. See schematic here:
http://www.arcdb.ws/SP10/ARC_SP10_schematic2.JPG

Not to mention the fact that the first stage actually has pretty mediocre PSRR. (More or less just the ratio of RL/Rp)

IMO from a noise performance standpoint this is still a mistake, admittedly mitigated by the parallel 6DJ8 in the first stage (SP10) and large amounts of global feedback around the whole thing.

 

[the_manta]

Thank you SY !
Nice info. I think i have to make new experiments with my ECC81's ;-))

The JJ's are frame grid tubes like the ECC88 or other, not like the orginal design i think. (I don't have those, I just heard it and saw on pics)
Maybe I should buy some of those. At the moment i only have old german ones (Valvo, Lorenz, tele..)
We will see
Manta