Hmm... 0.309% is what I'm getting using Ayumi's 6AU6 model and my 12AU7WA/6189 model. The quality of your model(s) has a great deal of influence on the accuracy of the THD results in LTspice or any other simulation software.
BTW, here's a Laplace with a correction factor for gain. I.E. 5mVin=5mVout.
Code:
LAPLACE=(1+3.18e-3*s)*(1+7.5e-5*s)/((1+3.18e-4*s)*9.8759)
Hi cogsncogs, I'm using Ayumi's models for both tubes, so the different results are likely be caused by the 12AU7 model
Hey cogsncogs, thanks for your input on this.
Following kevinkr's findings that real life THD will be 3 or 4 times worse than the spice prediction, that 0.3% THD you're getting with 5mV input would translate to 0.9% in real life. That would be about what I'd expect from this pair of tubes, especially with the 6AU6 set to such high gain.
Is the 12AU7 Ayumi model that far off? (It could be.) Would you mind pointing us to your 12AU7WA/6189 model? I'd like to try it out and compare it to the Ayumi model.
Using your 12AX7 model (from the Mazda datasheet), I've been able to get an RCA-style RIAA (with Ayumi-model 12AT7 cathode follower) to output 1V RMS with <0.05% THD (all 2nd harmonic) in LTspice. Assuming that that result is 4X more optimistic than real life, I could expect something around 0.2% THD at 1V RMS out into 100k ohms in real life. That's my figure to beat.
As I've mentioned elsewhere, I built an RCA RIAA preamp with cathode follower for my brother, something like 20 years ago. It's power supply is primitive; just your basic, unregulated CRCRC, with a basic DC heater supply (not regulated). I had it in for a checkup and got to hear it again. It's really pretty good; quiet enough for its intended use with MM carts. It may not be the highest resolution, most detailed phono stage you've ever heard, but it's really easy to listen through. I can see why it's a classic. Were I to do it over again, I'd improve the EQ and use a ZVN0545 source follower to buffer each 12AX7, and use regulated PSU's for both plate and heater supplies.
--
Following kevinkr's findings that real life THD will be 3 or 4 times worse than the spice prediction, that 0.3% THD you're getting with 5mV input would translate to 0.9% in real life. That would be about what I'd expect from this pair of tubes, especially with the 6AU6 set to such high gain.
Is the 12AU7 Ayumi model that far off? (It could be.) Would you mind pointing us to your 12AU7WA/6189 model? I'd like to try it out and compare it to the Ayumi model.
Using your 12AX7 model (from the Mazda datasheet), I've been able to get an RCA-style RIAA (with Ayumi-model 12AT7 cathode follower) to output 1V RMS with <0.05% THD (all 2nd harmonic) in LTspice. Assuming that that result is 4X more optimistic than real life, I could expect something around 0.2% THD at 1V RMS out into 100k ohms in real life. That's my figure to beat.
As I've mentioned elsewhere, I built an RCA RIAA preamp with cathode follower for my brother, something like 20 years ago. It's power supply is primitive; just your basic, unregulated CRCRC, with a basic DC heater supply (not regulated). I had it in for a checkup and got to hear it again. It's really pretty good; quiet enough for its intended use with MM carts. It may not be the highest resolution, most detailed phono stage you've ever heard, but it's really easy to listen through. I can see why it's a classic. Were I to do it over again, I'd improve the EQ and use a ZVN0545 source follower to buffer each 12AX7, and use regulated PSU's for both plate and heater supplies.
--
Last edited:
Thanks for this!
To use this, do I place an input signal AC voltage in my schematic and place this LAPLACE= string as a spice directive in my schematic? Or does this string have to be written into the .ac directive?
--
It's not perfect, but untill we get a better way to model triodes, it will have to do.
Code:
*==============================================================
* 6189/12AU7WA LTSpice model
* Modified Koren model (6 parameters): mean fit error 0.238074mA
* Traced by Wayne Clay using PlotDigitizer and Curve Captor v0.9.1
* on 3/22/2014 from Tung-Sol datasheet.
* ==============================================================
.subckt 12AU7WA P G K
Bp P K I=
+ (0.02753937477m)*uramp(V(P,K)*ln(1.0+(-0.1580044043)+exp
+ ((1.904441567)+(1.904441567)*((26.38270501)+(23.18833909m)
+ *V(G,K))*V(G,K)/V(P,K)))/(1.904441567))**(1.309056183)
Cgp G P 2.2p ; 0.7p added
Cgk G K 2.3p ; 0.7p added
Cpk P K 0.6p ; 0.2p 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)
.ends 12AU7WA
Code:
* 6189/12AU7WA macro model
* Modified Koren model (6 parameters): mean fit error 0.238074mA
* Traced by Wayne Clay using PlotDigitizer and Curve Captor v0.9.1
* on 3/22/2014 from Tung-Sol datasheet.
.subckt 6189 P G K
koren6(0.02753937477,-0.1580044043,1.904441567,26.38270501,23.18833909,1.309056183)
.ends 6189
* 6189/12AU7WA Spice 3F4 model
* Modified Koren model (6 parameters): mean fit error 0.238074mA
* Traced by Wayne Clay using PlotDigitizer and Curve Captor v0.9.1
* on 3/22/2014 from Tung-Sol datasheet.
.subckt 6189 P G K
Bp P K I=(0.02753937477m)*uramp(V(P,K)*ln(1.0+(-0.1580044043)+exp((1.904441567)+(1.904441567)*((26.38270501)+(23.18833909m)*V(G,K))*V(G,K)/V(P,K)))/(1.904441567))^(1.309056183)
Cgp G P 2.2p ; 0.7p added
Cgk G K 2.3p ; 0.7p added
Cpk P K 0.6p ; 0.2p 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)
.ends 6189
* 6189/12AU7WA PSpice model
* Modified Koren model (6 parameters): mean fit error 0.238074mA
* Traced by Wayne Clay using PlotDigitizer and Curve Captor v0.9.1
* on 3/22/2014 from Tung-Sol datasheet.
.subckt 6189 P G K
Gp P K VALUE={(0.02753937477m)*limit(V(P,K)*log(1.0+(-0.1580044043)+exp((1.904441567)+(1.904441567)*((26.38270501)+(23.18833909m)*V(G,K))*V(G,K)/V(P,K)))/(1.904441567),0.0,1.0e16)**(1.309056183)}
Cgp G P 2.2p ; 0.7p added
Cgk G K 2.3p ; 0.7p added
Cpk P K 0.6p ; 0.2p 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)
.ends 6189
* 6189/12AU7WA PSpice model for Micro-Cap
* Modified Koren model (6 parameters): mean fit error 0.238074mA
* Traced by Wayne Clay using PlotDigitizer and Curve Captor v0.9.1
* on 3/22/2014 from Tung-Sol datasheet.
.subckt 6189 P G K
Gp P K VALUE={(0.02753937477m)*limit(V(P,K)*ln(1.0+(-0.1580044043)+exp((1.904441567)+(1.904441567)*((26.38270501)+(23.18833909m)*V(G,K))*V(G,K)/V(P,K)))/(1.904441567),0.0,1.0e16)**(1.309056183)}
Cgp G P 2.2p ; 0.7p added
Cgk G K 2.3p ; 0.7p added
Cpk P K 0.6p ; 0.2p 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)
.ends 6189
* 6189/12AU7WA LTSpice model
* Modified Koren model (6 parameters): mean fit error 0.238074mA
* Traced by Wayne Clay using PlotDigitizer and Curve Captor v0.9.1
* on 3/22/2014 from Tung-Sol datasheet.
.subckt 6189 P G K
Bp P K I=(0.02753937477m)*uramp(V(P,K)*ln(1.0+(-0.1580044043)+exp((1.904441567)+(1.904441567)*((26.38270501)+(23.18833909m)*V(G,K))*V(G,K)/V(P,K)))/(1.904441567))**(1.309056183)
Cgp G P 2.2p ; 0.7p added
Cgk G K 2.3p ; 0.7p added
Cpk P K 0.6p ; 0.2p 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)
.ends 6189
* 6189/12AU7WA CircuitMaker model
* Modified Koren model (6 parameters): mean fit error 0.238074mA
* Traced by Wayne Clay using PlotDigitizer and Curve Captor v0.9.1
* on 3/22/2014 from Tung-Sol datasheet.
.subckt 6189 P G K
Bp P K I=(0.02753937477/1.0e3)*uramp(V(P,K)*ln(1.0+(-0.1580044043)+exp((1.904441567)+(1.904441567)*((26.38270501)+(23.18833909/1.0e3)*V(G,K))*V(G,K)/V(P,K)))/(1.904441567))^(1.309056183)
Cgp G P 2.2p ; 0.7p added
Cgk G K 2.3p ; 0.7p added
Cpk P K 0.6p ; 0.2p 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)
.ends 6189Try this 12AT7 Mazda model:
Code:
* --------------------------------------------------------------
* 12AT7_mz3 LTspice model
* Modified Koren model (8 parameters): mean fit error 0.149064mA
* Traced by Wayne Clay on 3/3/2014 using Curve Captor v0.9.1
* from Mazda Belvu datasheet
* --------------------------------------------------------------
.subckt 12AT7-mz3 P G K
Bp P K I=
+ (0.02790974004m)*uramp(V(P,K)*ln(1.0+(-0.01128450125)+exp((2.13112283)+
+ (2.13112283)*((98.95747475)+(-218.6570898m)*V(G,K))*V(G,K)/sqrt((24.5022294)**2+
+ (V(P,K)-(-33.23286842))**2)))/(2.13112283))**(1.220154214)
Cgk G K 2.9p ; 0.7p added
Cgp G P 2.2p ; 0.7p added
Cpk P K 1.7p ; 0.2p added
Rpk P K 1.0G ; to avoid floating nodes in mu-follower
d3 G K dx1
.model dx1 d(is=1n rs=2k cjo=1pf N=1.5 tt=1n)
.ends 12AT7-mz3
Well, I corrected my simulation settings as suggested by Hans Polack (timestep error)
Then compared Ayumi 12AU7 model against cogsncogs 12AU7WA
Huge difference in THD but, 3d harmonic with both models is held to -120dB under the fundamental. What changes is 2nd h, at -80 dB with one model, -60dB the other..
(we are talking in the Tokarev circuit, in the exact same conditions, w/o any changes)
Then compared Ayumi 12AU7 model against cogsncogs 12AU7WA
Huge difference in THD but, 3d harmonic with both models is held to -120dB under the fundamental. What changes is 2nd h, at -80 dB with one model, -60dB the other..
(we are talking in the Tokarev circuit, in the exact same conditions, w/o any changes)
Most of what i know, I learnt from Allen Wright's pre-amp cook book. So if you want valves - here are valves. According to his web site - this was the pre-amp that Allen used.
Attachments
It's not my schematic, I would guess its actually 1Meg. There is another resistor at output of the phono stage (after coupling capacitor) that is labelled 1m. This would logically be 1Meg as well.
do you have any pics to share? i exchanged emails with him, very nice fellow...
hope he lives longer despite his health concerns....
Do I take a certain kind of voltage source from my pre-existing symbol library, and apply the LAPLACE string to that?
Here's a schematic you've uploaded with a LAPLACE statement in it:
Is the LAPLACE statement applied to the E1 voltage source? If yes, then what is that E1 device called in the LTspice symbols library? I haven't used that one yet...
Thanks for any help you can send my way, and my apologies for drifting off-topic here.
--
No photos, sorry. Patrick was down here in Melbourne looking after health concerns so hopefully things will improve. Send him another email I am sure he would be pleased with some correspondence.
On a brighter note - we had a nice Japenese dinner together and I showed him the schematic of my diy mosfet amps. 2 days later I get an email back with a new detailed schematic with lots of improvements. He is a very clever designer and very generous with his time and advice. I just need the time now to implement these improvements.
Cheers, Hazard
Sent from my SM-G900I using Tapatalk
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.