ECC88 with 12V heaters

ruffrecords · 2013-09-27 9:52 pm

kevinkr said:
Thanks, Ian for the comprehensive explanation, I can see the thought you have put into this and think your design choices are about as close to optimum as possible. Curious what if any effect does cranking up the quiescent current slightly (<= 10mA) have on linearity assuming adequate supply margin and dissipation in the 8416?

Dissipation becomes an issue. There is 150V across each triode which at 10mA is 1.5W dissipation for a total of 3W in the tube. There are various dissipation figures for the 6922 depending who's data sheet you read but none of them go as high as 3W. I have not acutally tried it at or near 10mA so I don't know how the linearity is affected. Simulations show that at 5.7mA and no NFB, the 6922 SRPP creates 2.1% distortion at 20V rms into 5K with a 300V supply. Changing the operating point to 11mA increases the distortion to 2.5% which seems counter intuitive, but if you reduce the load to 2K4 and repeat, the distortion drops to 2.3%. With a 10K load its close to 2.4% This is the old optimum load for the SRPP thing I think.

Cheers

Ian

ruffrecords · 2013-09-27 9:57 pm

kevinkr said:
Hi Ian,
This is the link I used to get to your page:"http://www.customtubeconsoles.com/"

The one in your sig line sends my IP (verizon.net) into conniptions.. lol

Oops! Thanks for pointing that out. It should now be correct. Took me ages to find out how to edit it. Fund it under the cryptic 'UserCP'

Cheers

Ian

kevinkr · 2013-09-27 10:06 pm

Great page by the way..

Eli Duttman · 2013-09-28 1:20 am

Ian,

If filament current is abundant, perhaps JJ's ECC99 as the buffer, instead of a 6922 "equivalent", would work out. You could crank I_B up, without fear of exceeding the plate dissipation limit.

ruffrecords · 2013-09-28 11:24 am

Eli Duttman said:
Ian,

If filament current is abundant, perhaps JJ's ECC99 as the buffer, instead of a 6922 "equivalent", would work out. You could crank I_B up, without fear of exceeding the plate dissipation limit.

The ECC99 is an interesting tube. I use it as a substitute for the 6922 to drive headphones. I run them at about 15mA. There's only one problem - the module these fit in is just wide enough to take a 6922 - the ECC99 is too tall.

I have ordered a few 12AV7 off eBay. The 12AT7, 12AV7 and 6922 all have a similar mu but gm increases as you go from the AT7 to AV7 and to 6922 so plate resistance drops in the same sequence. I think explains why the AT7 distorts so much. I am hoping the AV7 will be better and good enough for this application.

Cheers

Ian

ErikdeBest · 2013-09-28 1:36 pm

hi Ian,
an idea from my side. russians made the 6s4s and 6s3s, single triodes with quite a high gm, and still affordable. i thought about them because your requirements are quite 'hard' for a tube. two of these in series would draw 12,6v . just an idea

lordearl · 2013-09-28 1:52 pm

12FQ7 might be worth a shot. You won't get the middle shield like in the 6CG7/ECC88, however they're cheap, less gain but should work. Or 12DT8 (more gain).

SY · 2013-09-28 1:57 pm

lordearl said:
12FQ7 might be worth a shot.

Transconductance is low. ECC88 is not the equivalent of 6CG7, it's a 6DJ8.

lordearl · 2013-09-28 2:02 pm

Indeed - however in the absence of a reasonably priced exact equivalent, sometimes easier to modify a circuit to accommodate a different tube. Another benefit - the 12v options are quite good in that they stress filament transformers less than the 6v tubes.

ruffrecords · 2013-09-29 11:05 am

lordearl said:
12FQ7 might be worth a shot. You won't get the middle shield like in the 6CG7/ECC88, however they're cheap, less gain but should work. Or 12DT8 (more gain).

I use d 6CG7 tubes in my earlier designs because they are effectively a 6SN7 in a noval bottle. They have very low intrinsic distortion and in a mu follower they perform very well into a 10K load but they struggle into 5K. Despite that, they are a lot taller then the 6922 so there's not enough headroom to fit one in unfortunately.

Cheers

Ian

popilin · 2013-09-29 4:05 pm

Maybe a 5687, great linearity, reasonable gm, low Rp, same tall than ECC88...a bit current hungry perhaps...

Eli Duttman · 2013-09-29 4:53 pm

I have ordered a few 12AV7 off eBay. The 12AT7, 12AV7 and 6922 all have a similar mu but gm increases as you go from the AT7 to AV7 and to 6922 so plate resistance drops in the same sequence. I think explains why the AT7 distorts so much. I am hoping the AV7 will be better and good enough for this application.

As you are trying the 12AV7, consider the 5965, which is the industrial version and rather well thought of.

ruffrecords · 2013-09-30 11:30 am

Eli Duttman said:
As you are trying the 12AV7, consider the 5965, which is the industrial version and rather well thought of.

Yes, I am. Just got a couple in the post this morning.

Cheers

Ian

lordearl · 2013-10-09 10:55 am

Ian - any updates on the 5965 vs 12AV7? (a point you made about the 6CG7 a few years back put me on the right track with tube gear...so I hold your opinions in high esteem!)

ruffrecords · 2013-10-09 12:22 pm

lordearl said:
Ian - any updates on the 5965 vs 12AV7? (a point you made about the 6CG7 a few years back put me on the right track with tube gear...so I hold your opinions in high esteem!)

I have not got round to trying either of them yet. I have been looking for LTspice models for them but the only ones I have been able to find on the net give me syntax arrors or just don't work at all.

Cheers

Ian

jazbo8 · 2013-10-09 1:22 pm

Try this one:

*
* Generic triode model: 5965
* Copyright 2003--2008 by Ayumi Nakabayashi, All rights reserved.
* Version 3.10, Generated on Sat Mar 8 22:42:06 2008
* Plate
* | Grid
* | | Cathode
* | | |
.SUBCKT 5965_AN A G K
BGG GG 0 V=V(G,K)+0.61035688
BM1 M1 0 V=(0.024336203*(URAMP(V(A,K))+1e-10))**-2.1648233
BM2 M2 0 V=(0.40929668*(URAMP(V(GG)+URAMP(V(A,K))/24.272617)+1e-10))**3.6648233
BP P 0 V=0.0045273524*(URAMP(V(GG)+URAMP(V(A,K))/59.303235)+1e-10)**1.5
BIK IK 0 V=U(V(GG))*V(P)+(1-U(V(GG)))*0.010017704*V(M1)*V(M2)
BIG IG 0 V=0.0022636762*URAMP(V(G,K))**1.5*(URAMP(V(G,K))/(URAMP(V(A,K))+URAMP(V(G,K)))*1.2+0.4)
BIAK A K I=URAMP(V(IK,IG)-URAMP(V(IK,IG)-(0.0023786712*URAMP(V(A,K))**1.5)))+1e-10*V(A,K)
BIGK G K I=V(IG)
* CAPS
CGA G A 3p
CGK G K 4p
CAK A K 0.5p
.ENDS

ruffrecords · 2013-10-09 4:25 pm

@jazbo8,

Many thanks, that one actually works!! With 680R cathode resistors and NFB arranged for 10dB gain I get just over 0.6% distortion at +25dBV into 2400 ohms load which is pretty good.

Cheers

Ian

jazbo8 · 2013-10-09 5:23 pm

You are welcome, glad it worked out for you, all the credit goes to Ayumi who built the model. He has many more models at his site in case you are interested. To use those models in LTSpice, you need to manually edit the *.inc files by replacing all the "^" with "**" in the text. I think you will find those models superior to the ones by Koren, Munro, etc.

ruffrecords · 2013-10-09 11:58 pm

jazbo8 said:
You are welcome, glad it worked out for you, all the credit goes to Ayumi who built the model. He has many more models at his site in case you are interested. To use those models in LTSpice, you need to manually edit the *.inc files by replacing all the "^" with "**" in the text. I think you will find those models superior to the ones by Koren, Munro, etc.

Thanks for the link. That is a very impressive library of tube models. Should be more widely appreciated.

Cheers

Ian

jazbo8 · 2013-10-10 7:46 am

Yes, please help to spread the word - people should stop using the Koren and Munro models since the grid currents are poorly modeled (if at all). When you distribute the Ayumi's models, just be sure to include the top three lines of the SPICE code, as it's his copyrighted material.

ECC88 with 12V heaters

Member

Member

Administrator

Member

Attachments

Member

Member

Member

Ex-Moderator

Member

Member

R.I.P.

Member

Member

Member

Member

Ex-Moderator

Member

Ex-Moderator

Member

Ex-Moderator