CA3280 spice model?

From circuitbending to homebrew stompboxes & synths, keep the DIY spirit alive!

Moderators: Kent, luketeaford, Joe.

Post Reply
User avatar
discomicke
Common Wiggler
Posts: 144
Joined: Wed Sep 22, 2010 7:16 pm
Location: Stockholm

CA3280 spice model?

Post by discomicke » Fri Nov 06, 2020 4:35 pm

Does anyone have a CA3280 spice model to share ? I'm not smart enough to make one from the datasheet.

User avatar
devinw1
Super Deluxe Wiggler
Posts: 2112
Joined: Tue Aug 07, 2018 11:20 am
Location: Portland, OR
Contact:

Re: CA3280 spice model?

Post by devinw1 » Fri Nov 06, 2020 5:36 pm

You could try the LTSPice Yahoo group? I remember doing so a while back and never finding one though :(.

User avatar
discomicke
Common Wiggler
Posts: 144
Joined: Wed Sep 22, 2010 7:16 pm
Location: Stockholm

Re: CA3280 spice model?

Post by discomicke » Fri Nov 06, 2020 5:51 pm

Thanks! Someone actually did post a very simplified version in the group almost exactly 2 years ago. At first glance pins for input current and bias current is missing from the model.

Posting it here in-case someone ends up at this thread via searches:
*_smn_descriptioninfo
*Dual, 9MHz, Operational Transconductance Amplifier (OTA)
*_smn_enddescriptioninfo
*_smn_developerinfo
*IIT
*_smn_enddeveloperinfo
*_smn_packageinfo
*PDIP-16
*_smn_endpackageinfo
*_smn_nodesinfo
*shape triangle
*1 l +
*2 l -
*3 r out
*_smn_endnodesinfo
*_smn_componenttype
*Opamp
*_smn_endcomponenttype

.SUBCKT CA3280E 1 2 3
* A= 100000 RI= 1.5e+012 RO= 6.3e+007 VSW+= 13.7 VSW-= -14.3
* Vos= 0.00025 Ibs= 3e-006 Ios= 3e-007 SR+= 1.25e+008
* fu= 9e+006 fp2= 1e+032 CC= 3e-011
Vos 4 1 DC 0.00025V
Ib1 4 0 3.15e-006A
Ib2 2 0 2.85e-006A
G1 0 5 4 2 0.0464159
G2 0 6 5 0 7.3676e-007
G3 0 3 6 0 7.3676e-007
Ri 4 2 1.5e+012ohm
R1 5 0 1000ohm
R2 6 0 6.3e+007ohm
R3 3 0 6.3e+007ohm
C1 5 0 1.76839e-006
C2 6 0 2.52627e-041
Cc 5 0 3e-011
.ENDS

Don T
Super Deluxe Wiggler
Posts: 1366
Joined: Sun Feb 02, 2014 3:11 pm

Re: CA3280 spice model?

Post by Don T » Sat Nov 07, 2020 7:03 pm

I thought there was one for the CA3080? You could just use two of those, correct?

Aether Machine
Learning to Wiggle
Posts: 31
Joined: Mon Jan 27, 2020 2:32 pm
Location: elsewhere
Contact:

Re: CA3280 spice model?

Post by Aether Machine » Sat Nov 07, 2020 7:11 pm

You could probably just use the LM13700 model, if you drive Id with a current source (the one major difference between the 2 parts) it will be almost identical

User avatar
neil.johnson
Super Deluxe Wiggler
Posts: 1693
Joined: Thu Apr 08, 2010 3:51 pm
Location: Cambridge, UK
Contact:

Re: CA3280 spice model?

Post by neil.johnson » Mon Nov 09, 2020 2:51 am

Have you checked out the website of Don Sauer, the designer of the LM13700?
http://www.idea2ic.com

Neil
Random ramblings :: http://www.njohnson.co.uk
SSI parts and more :: https://www.cesyg.co.uk

Joel W. Walker
Learning to Wiggle
Posts: 5
Joined: Sun Mar 28, 2021 5:42 pm

Re: CA3280 spice model?

Post by Joel W. Walker » Sun Apr 04, 2021 11:59 am

Hi All -- First post.

I also needed a model and a complete schematic for the CA3280 (or now the AS3280). Here's my best guess, based on a compilation of resources as listed below, most importantly the OTA paper by Achim Gratz.

At risk of reviving a slightly dormant thread, I'll post it here in case it is useful to others. Comments appreciated.

Complete Schematic:

CA3280.png

Spice Subcircuit, tested in LTSpice
Copy from between (but not including) the #### lines following and save as a plain text file called CA3280.sub
Locate the directory path where LTSpice keeps its library files on your system
For example, on my Mac it is: "/Users/joel/Library/Application\ Support/LTspice/lib/"
Place the CA3280.sub file in the "sub/" folder under the main library path

######################################
* CA3280 spice model 04/21 by Joel W. Walker <joelwwalker@gmail.com>
* Referencing: data sheets distributed by Intersil and Texas Instruments,
* the LM13700 spice model shared by Don Sauer (idea2ic.com),
* the CA3080 transistor models shared by John P. Doty <jpd@noqsi.com>,
* "Operational Transconductance Amplifiers" by Achim Gratz (Synth.Stromeko.net), and
* "Comments On The New 13600 And 3280 OTA Devices" by Serge Tcherepnin
* in Electronotes #113 by Bernie Hutchins (electronotes.netfirms.com)
.SUBCKT CA3280 LIN EMT IAB VEE OUT VCC INN INP
Q01 IAB N01 VEE QNX
Q02 N02 N01 VEE QNX
Q03 N03 IAB N02 QNX
Q04 N04 N01 VEE QNX
Q05 EMT IAB N04 QNX
Q06 N05 N01 VEE QNX
Q07 EMT IAB N05 QNX
Q08 N06 N01 VEE QNX
Q09 N07 IAB N06 QNX
Q10 N08 IAB N01 QNX
Q11 N03 INN EMT QNX
Q12 N07 INP EMT QNX
Q13 LIN N09 VEE QNX
Q14 N10 N09 VEE QNX
Q15 N11 LIN N10 QNX
Q16 N12 N09 VEE QNX
Q17 N13 LIN N12 QNX
Q18 N14 N09 VEE QNX
Q19 N15 LIN N14 QNX
Q20 N16 N09 VEE QNX
Q21 N11 LIN N16 QNX
Q22 N17 LIN N09 QNX
Q23 INN INN N11 QNX
Q24 INP INP N11 QNX
Q25 N18 N18 VEE QNX
Q26 N19 N18 VEE QNX
Q27 OUT N19 N18 QNX
Q28 N20 N20 VCC QPX
Q29 N13 N20 VCC QPX
Q30 INN N13 N20 QPX
Q31 N21 N21 VCC QPX
Q32 N15 N21 VCC QPX
Q33 INP N15 N21 QPX
Q34 N22 N22 VCC QPX
Q35 N03 N22 VCC QPX
Q36 N19 N03 N22 QPX
Q37 N23 N23 VCC QPX
Q38 N07 N23 VCC QPX
Q39 OUT N07 N23 QPX
R01 N08 VCC 2K
R02 N17 VCC 2K
.MODEL QNX NPN(IS=21.48F XTI=3 EG=1.11 VAF=80 BF=550 ISE=50F NE=1.5
+ IKF=10M XTB=1.5 BR=.1 ISC=10F NC=2 IKR=3M RC=10 CJC=800F
+ MJC=.3333 VJC=.75 FC=.5 CJE=1.3P MJE=.3333 VJE=.75 TR=30N
+ TF=400P ITF=30M XTF=1 VTF=10 CJS=5.8P MJS=.3333 VJS=.75)
.MODEL QPX PNP(IS=50F XTI=3 EG=1.11 VAF=80 BF=100 ISE=130F NE=1.5 IKF=1M
+ XTB=1.5 BR=1 ISC=0 NC=2 IKR=0 RC=0 CJC=4P MJC=.3333
+ VJC=.75 FC=.5 CJE=1.4P MJE=.3333 VJE=.75 TR=500N TF=23N ITF=.1
+ XTF=1 VTF=10 CJS=5.5P MJS=.3333 VJS=.75)
.ENDS CA3280
######################################

Spice Symbol, tested in LTSpice
Copy from between (but not including) the #### lines following and save as a plain text file called CA3280.asy
Symbol files go into the "sym/" subdirectory of the LTSpice library path located above
To keep your parts organized create a new folder under "sym/" with whatever name you like, e.g. "MyParts/"
Place the CA3280.asy file in that folder and restart LTSpice
The Draft -> Component tool should now show your new subfolder with your new part inside

######################################
Version 4
SymbolType CELL
RECTANGLE Normal -64 -72 64 72
WINDOW 0 0 -72 Bottom 2
WINDOW 3 0 72 Top 2
SYMATTR Value CA3280
SYMATTR Prefix X
SYMATTR Description Half CA3280 OTA by Intersil
SYMATTR ModelFile CA3280.sub
PIN -64 -48 LEFT 8
PINATTR PinName LIN
PINATTR SpiceOrder 1
PIN -64 -16 LEFT 8
PINATTR PinName EMT
PINATTR SpiceOrder 2
PIN -64 16 LEFT 8
PINATTR PinName IAB
PINATTR SpiceOrder 3
PIN -64 48 LEFT 8
PINATTR PinName VEE
PINATTR SpiceOrder 4
PIN 64 48 RIGHT 8
PINATTR PinName OUT
PINATTR SpiceOrder 5
PIN 64 16 RIGHT 8
PINATTR PinName VCC
PINATTR SpiceOrder 6
PIN 64 -16 RIGHT 8
PINATTR PinName INN
PINATTR SpiceOrder 7
PIN 64 -48 RIGHT 8
PINATTR PinName INP
PINATTR SpiceOrder 8
######################################

Cheers - Joel 8-)
Last edited by Joel W. Walker on Sun Apr 04, 2021 2:34 pm, edited 1 time in total.

User avatar
oldenjon
Ultra Wiggler
Posts: 960
Joined: Sun Feb 26, 2012 1:28 am
Location: Sacramento, CA

Re: CA3280 spice model?

Post by oldenjon » Sun Apr 04, 2021 12:05 pm

Very timely, I was just looking for this yesterday!
"We all dance by ourselves, that's why we only play electronic music"

http://synthscalledbeasts.tumblr.com/

Joel W. Walker
Learning to Wiggle
Posts: 5
Joined: Sun Mar 28, 2021 5:42 pm

Re: CA3280 spice model?

Post by Joel W. Walker » Sun Apr 04, 2021 12:11 pm

oldenjon wrote:
Sun Apr 04, 2021 12:05 pm
Very timely, I was just looking for this yesterday!
Awesome! Please let me know if it works for you ...

User avatar
KSS
Super Deluxe Wiggler
Posts: 5311
Joined: Mon Jan 25, 2016 7:28 am

Re: CA3280 spice model?

Post by KSS » Sun Apr 04, 2021 6:34 pm

Back in the day there was much back and forth on the Electronotes newsletter and Analog Heaven mailing list WRT the 3280 and its actual 'distance' from the datasheet. You might recognize a few of the names in those discussions, Serge, Rossum, Patchell. <-- Patchell's site used to have about 9 pages covering details of the 3280.

Using a 136 or 13700 model will miss most of the reasons you'd need to use a 3280 instead of it.
As always, but here especially, Details matter. If they don't then you can use something besides a 3280.

Joel W. Walker
Learning to Wiggle
Posts: 5
Joined: Sun Mar 28, 2021 5:42 pm

Re: CA3280 spice model?

Post by Joel W. Walker » Sun Apr 04, 2021 9:04 pm

Hi KSS --

Thanks for the info -- I have indeed had a terrible time finding good info on the 3280, and have been frustrated by the datasheet "distance" you referenced. I'm quite interested in the old information / conversations you referenced, which are mostly previously unknown to me, and have mostly not popped up in my searches.

I may be misunderstanding your comment, but note that the model I posted is not a 13700 -- it is an honest attempt to build a 3280 "from scratch" using what clues I could find. The LM13700 was a reference just for the layout of the Wilson mirrors. Also, the low-level transistor models are not tuned to the 3280, but rather borrowed from an old tuning of the 3080. However, I suspect that the topology and layout may be very close to reality.

Do you suspect that the schematic I drew and posted is wrong in any specific ways? Or, do you suspect that the low-level transistor details are substantially off base?I would be quite interested to know either, if so.

Thanks -
Joel

User avatar
KSS
Super Deluxe Wiggler
Posts: 5311
Joined: Mon Jan 25, 2016 7:28 am

Re: CA3280 spice model?

Post by KSS » Sun Apr 04, 2021 10:02 pm

@Joel W. Walker

First, Welcome to MuffWiggler! I should have put that in the first reply!

I'm sorry. I did not mean to reflect or imply negatively on your work in any way. I think it's WONDERFUL that you are putting forth the effort to do this!
I was responding more to ensure that anybody reading did not equate the 3280 with its lesser cousins. And to note that the datasheet for it is known to contain errors.

Jim Patchells site is still up and hosted by old crow here.
http://www.oldcrows.net/~patchell/
I haven't seen any post from Jim since many years ago he began going blind. If he is still responding anywhere, I expect it would be AH.

Edit: http://www.oldcrows.net/~patchell/archi ... hives.html
http://www.oldcrows.net/~patchell/archi ... 280_1.html /edit

The public available Electronotes are at: http://electronotes.netfirms.com The paper copies of non public back issues are not presently for sale.

The AH aka Analog Heaven mailing list is still active at: http://analogue-heaven.1065350.n5.nabble.com/

Would Alfa be responsive to helping you create a better SPICE model for their part? Certainly seems worth asking if you haven't already. They may not want to share everything for competitive reasons -though they must know anybody with skill in that area can 'pop the top' and have a look for themselves. But they might be inspired and enthused to have a verified reliable model for *their* specific silicon.

Joel W. Walker
Learning to Wiggle
Posts: 5
Joined: Sun Mar 28, 2021 5:42 pm

Re: CA3280 spice model?

Post by Joel W. Walker » Mon Apr 05, 2021 8:31 am

Hi KSS --

Thanks very kindly for the warm welcome!

I appreciate the links as well - it's an amazing body of work this community has compiled. Nice idea regarding contacting Alfa. I know people are happy that the part is being manufactured again. I agree, it seems that they would have an interest in facilitating reliable simulation to support sales and adoption of the component. I'll consider sending them a note.

By the way, I also did several calculations to understand the function and operation of the 3280 circuit. I could post that too (though it might take some effort to make things more presentable) if there is an interest.

Cheers - Joel

User avatar
oldenjon
Ultra Wiggler
Posts: 960
Joined: Sun Feb 26, 2012 1:28 am
Location: Sacramento, CA

Re: CA3280 spice model?

Post by oldenjon » Mon Apr 05, 2021 10:37 am

Another place you might look for info is SDIY's archive. There looks to be several pages of discussion about the CA3280 there, and I know that I've seen contributions from Patchell in various threads.
https://synth-diy.org/pipermail/synth-diy/

*edit* is this the same as analogue heaven?
"We all dance by ourselves, that's why we only play electronic music"

http://synthscalledbeasts.tumblr.com/

User avatar
devinw1
Super Deluxe Wiggler
Posts: 2112
Joined: Tue Aug 07, 2018 11:20 am
Location: Portland, OR
Contact:

Re: CA3280 spice model?

Post by devinw1 » Mon Apr 05, 2021 11:36 am

One of the larger differences between the 3280 and the LM13700 is the linearizing diode scheme. Make sure that is correct on the model if you are looking at the 3280 seriously.

Bernie Hutchins
Learning to Wiggle
Posts: 38
Joined: Sat Jun 27, 2020 3:19 pm

Re: CA3280 spice model?

Post by Bernie Hutchins » Mon Apr 05, 2021 11:43 am

The cited Electronotes 113 has been a posted free on my EN site for many years. Bernie

User avatar
neil.johnson
Super Deluxe Wiggler
Posts: 1693
Joined: Thu Apr 08, 2010 3:51 pm
Location: Cambridge, UK
Contact:

Re: CA3280 spice model?

Post by neil.johnson » Mon Apr 05, 2021 5:38 pm

oldenjon wrote:
Mon Apr 05, 2021 10:37 am
Another place you might look for info is SDIY's archive. There looks to be several pages of discussion about the CA3280 there, and I know that I've seen contributions from Patchell in various threads.
https://synth-diy.org/pipermail/synth-diy/

*edit* is this the same as analogue heaven?
Jim Patchell is still very much alive and kicking, often posting on Facebook. I'll point him at this thread.

Synth-dIy is not analogue heaven. I'm surprised more folks in this part of muffs are not subscribed.

Neil
Random ramblings :: http://www.njohnson.co.uk
SSI parts and more :: https://www.cesyg.co.uk

Joel W. Walker
Learning to Wiggle
Posts: 5
Joined: Sun Mar 28, 2021 5:42 pm

Re: CA3280 spice model?

Post by Joel W. Walker » Tue Apr 06, 2021 2:23 pm

oldenjon wrote:
Mon Apr 05, 2021 10:37 am
Another place you might look for info is SDIY's archive. There looks to be several pages of discussion about the CA3280 there, and I know that I've seen contributions from Patchell in various threads.
https://synth-diy.org/pipermail/synth-diy/
Hi oldenjon -- thanks so much for the links.
devinw1 wrote:
Mon Apr 05, 2021 11:36 am
One of the larger differences between the 3280 and the LM13700 is the linearizing diode scheme. Make sure that is correct on the model if you are looking at the 3280 seriously.
Hi devinw1 -- Yes, that is indeed quite important. I believe these features are correct in the schematic I drew and the associated spice model. The linearizing scheme employed here has the difference of diode voltage drops "paralleled" with the difference of base-emitter drops on the main long-tailed transistor pair. This is consistent with the drawing in the old crows / Patchell notes linked by KSS, as well as the AS3280 / CA3280 Datasheet and the illustration by Achim Gratz / Stromeko (figure 11). This is in distinction to the older scheme in the LM13700 (figure 16), where the series sum of diode and diff-amp base-emitter drops is equivalent across the two sides. I would certainly appreciate another set of eyes to look things over in detail ...
Bernie Hutchins wrote:
Mon Apr 05, 2021 11:43 am
The cited Electronotes 113 has been a posted free on my EN site for many years. Bernie
Hi Bernie -- it's an honor for you to chime in, and much appreciated. Yes, the excerpts from Electronotes 113 were very interesting to me. Unfortunately, I was not able to access the original discussion of the 3280 from EN107, which Serge references in EN113.

The most relevant bit from EN113 was the line "In the case of the 3280, the cause is the fact that twice the I_d is reflected, and thus each of the diodes sees 1/2 of 2 times I_d. Where is this found in the data sheet?" This helped me to reconcile the drawing from Achim Gratz / Stromeko, which shows clusters of four current mirrors for both I_d and I_ABC, whereas the AS3280 / CA3280 Datasheet shows only two. However, I believe the details on the transistor-level structure of those current mirrors are correct in the datasheet (with two legs omitted to save space in each case), whereas the mirrors in the Gratz diagram are just indicated symbolically. Other shortcomings in the data sheet (also clarified by the Gratz diagram) are that *separate* lines from the I_d and I_ABC current mirror clusters source / sink the diodes / differential pair on either side of the OTA (related to the "missing legs" issue, which is central to the comment by Serge), whereas the data sheet shows single lines that are connected to both sides. (Also, the arrow directions are unconventional in some cases, but this is minor.) Finally, both the datasheet and the Gratz diagram are unspecific with regards to the construction of the 5 remaining single current mirrors. Four of these are PNP type, connected to the positive rail; the fifth is NPN type, and connected to the output (together with one from the prior group of 4). Gratz seems to hint in section 4.5 that the PNP mirrors are probably of standard Wilson type, as in the LM13700 (and without the extra Darlington pair from the CA3080). I've drawn all 5 as 3-transistor Wilson mirrors. Of course, it could be that there are still "hidden details", though I suspect the most important bits of the topology are correct. It could also be that the transistor models (drawn from an older analysis of the 3080 as stated above) are inaccurate in a relevant way, though I suspect the dependence there to be weak (e.g. in the standard way that specific values of Beta can be somewhat unimportant, as long as they are much larger than 1).

Neil -- thanks also for your input! That would be super if Jim Patchell wished to add some thoughts.

Cheers -
Joel 8-)

PS -- I may still post some original mathematical analysis if there is interest. The 3280 is a remarkable bit of engineering, and there are a couple of fascinating higher-order cancellations that I've not seen described elsewhere.

User avatar
neil.johnson
Super Deluxe Wiggler
Posts: 1693
Joined: Thu Apr 08, 2010 3:51 pm
Location: Cambridge, UK
Contact:

Re: CA3280 spice model?

Post by neil.johnson » Tue Apr 06, 2021 3:29 pm

Why don't you join synth-diy and discuss this topic there? Achim is still active on the list. Jim is probably still on list as well.

Neil
Random ramblings :: http://www.njohnson.co.uk
SSI parts and more :: https://www.cesyg.co.uk

Post Reply

Return to “Music Tech DIY”