|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
84.211.93.117
In Reply to: RE: Full music TJ 300B mesh Sofia traced plate curve posted by Balle Clorin on September 01, 2014 at 14:57:36
My 300BMr Kitics RH300B
At least in my amplifier 5000 turns out to be a good load.
It is not correct to only look at the 300B ,since the driver /gain distortion interacts heavily with the 300B. I can simulate my amp based on accurate spice models. And I find that 6SN7 does not have sufficient linear gain to drive a 300b fully, but that is what I am stuck with. My input clips before the 300B, so it is not optimal. But I think my amplifier sound wonderful with my speakers.I will live long and happily with this combo.The table below shows that in my amp 5k ohm load is still a good choice.
for Mr Kitics amp RH300b the 2500 load seem better than 5000. So that is that. By the way, my simulations tells me that different tubes will distort very different in each design, and evaluating tubes across design are may be very misleading.
Best regards
Balle
Edits: 09/06/14 09/06/14Follow Ups:
Great simulation - I guess your 300B model is good, but I am not sure about your ECC81 and 6SN7 models, which are also important for the end result.
Just as I wrote earlier, it is the circuitry of the amp that counts, not just the operating point or the load. And regardless of that technical issue, the remark about the sound into differing loads is a valid observation - although it may depend on the speakers sensitivity and tone as well.
I have recently improved all my spice models to a level of near perfection (all triode models I use)(by near perfection I mean that the simulated curves are virtually identical to those presented in the datasheets) and this has lead to further optimization of the RH300B (change of Rfb and Ra resistors on the ECC81). Output power has of course increased, so now I can count on approximately 13W at 5% distortion (which is the usual distortion level used). I should write a new blog entry documenting this optimization and several related topics, but just cannot find the time to do it.
I also get a different decay of harmonics than you get in your simulation, which is why I presume that it is probably caused by the imperfection of the ECC81 spice model (in particular).
http://rh-amps.blogspot.com/
I use Andrei Frolovs Curve captor. A fantastic tool that generates spice code for Norman Korens 8 parameter curves. It fits perfectly with any tube curve. The dots are points from datasheet curve , lines are the model. .I found program linked somewhere in this or DIY audio forumIn order to get accurate distortion calulation in LTSpice I need to specify a short time step and also specify " no data compression"
* ECC81 macro model
.subckt ECC81 P G K
koren8(0.03431957815,-0.02156357481,2.303890189,91.06322625,-77.3757358,61.37729547,-9.442653676,1.183113473)
.ends ECC81
* ECC81 LTSpice model
.subckt ECC81 P G K
Bp P K I=(0.03431957815m)*uramp(V(P,K)*ln(1.0+(-0.02156357481)+exp((2.303890189)+(2.303890189)*((91.06322625)+(-77.3757358m)*V(G,K))*V(G,K)/sqrt((61.37729547)**2+(V(P,K)-(-9.442653676))**2)))/(2.303890189))**(1.183113473)
.ends ECC81.6SN7 FITTED TO DATASHEET-GREEN CURve
* 6SN7WGTk8 LTSpice model
.subckt 6SN7WGTk8 P G K
Bp P K I=(0.02170661176m)*uramp(V(P,K)*ln(1.0+(-0.1418880099)+exp((5.336202785)+(5.336202785)*((22.20702463)+(-20.28905564m)*V(G,K))*V(G,K)/sqrt((-0.0001481184562)**2+(V(P,K)-(-11.62245799))**2)))/(5.336202785))**(1.347919701)
Cgk G K 2.4p ; 0.2p added
Cpk P K 1.2p ; 0.5p added
Cgp G P 4.2p ; 0.5p added
Rpk P K 1G ; to avoid floating nodes
d3 G K dx1
.model dx1 d(is=1n rs=2k cjo=1pf N=1.5 tt=1n)
.endsBalle
Edits: 09/06/14 09/06/14 09/06/14
Can you please explain the difference between the red and green curve for 6SN7?
For the ECC81 this is an illustration of the comparison between my model and the datasheet curves (as published on the blog).
I have superimposed the simulated curves with the graph from the datasheet.
How do you check your models?
http://rh-amps.blogspot.com/
1. The green curve is traced and fitted to the published datasheet buy Curve captor tool. The red curve is from a package of spice models from Duncan pages. It is clearly wrong/ too good and does not fit the published curves.
2. You can see the model fit in my first post. The dots are points from the datasheet curves. and,the lines ar the fitted model.
Best regards
Balle
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: