SPICE Amplifier Simulation

Hello All,

I'm getting sucked deeper and deeper into the DIY hole. As a physics major, I've been imprinted with the desire to know every facet of a system, the why's as well as how's, so to speak. So when it comes to sound hardware (especially amps), the lure of schematic philosophies is just as attractive to me as the ultimate goal of a proper sound system. This leads to endless theoretical wanderings and lots and lots of research. I've seen a lot of the accepted DIY designs in theory, specifically the Pass brethren, but I've not been blessed with a wide personal hands-on experience.

So, what I'm wondering is, is anyone else out there spending time in the theoretical bent? Specifically, who here has done any SPICE simulation for all or part of any amplifier design? The achieve a precise, reliable result, are there any pitfalls that should be observed? How hard is it to get distortion characteristics from a sim that match the real-world circuit output? Are there any reasonably complete compenda of various stage designs, and if so where can I find such a source? Which sub-circuit schemes are accepted as refined and perfected, and which are considered open to opinion? And above all, where can I find a wide assortment of SPICE netlists to demonstrate real-world amplifier results? I would like these in order to get a feel for distortion characteristics, frequency response, and all the other characteristics of an amp design, and also to experiment with various combinations. If anyone has any personally-developed SPICE netlists they would like to share, I would whole-heartedly welcome them and the author's experience. I would prefer to stick to solid-state (I'm a new breed student that doesn't trust components that incandesce), but as I'm also interested in electrostatic speakers, I'm not completely close-minded to high voltage-output tube designs. Thanks everyone.
- Jonathan
 
Computer Aided Design Work

Jonathan,

All of my work on new designs is done with computer modeling. This can be quite time consuming since it may take months of simulations before a soldering iron is ever turned on. This results in minimum wasted parts and inhalation of solder fumes.

Distortion results done with computer simulation can be quite close to what is measured after the equipment is actually built. The largest difference will be when the simulation states that distortion levels will be around -147 DB or better. These figures are rarely reachable with real live parts even at low levels. Keep in mind that Low distortion figures are only a small part of the game. There are many other things that are actually more important as long as the distortion levels are reasonable.

Computer modeling will not compensate for lack of design ability or knowledge of what will work and what will not. It is thus just one of many tools that are available to use. It is also easy to become satisfied with simulated results and never actually build any equipment. This is a big drawback since it reduces actual hands on experiance.


John Fassotte
Alaskan Audio

[Edited by alaskanaudio on 11-24-2001 at 04:41 AM]
 
You might find <a href="http://www.geocities.com/ResearchTriangle/Node/2356/">GEORGE KRILOV AUDIO SCHEMATICS</a>
of interest, has lots of simulations of various Amp designs e.g. with/without feedback, current sources, symetrical without diff first stage. Not sure if he has built much but he has simmulated alot.

Its a Russian/English site, some of the pages haven't been translated yet.

Regards
James
 
I use CircuitMaker 2000 (a scaled-down version of Protel) with great results, either in digital or analog circuits. For example, i've been using it with the JLH'96 to measure bandwidth, noise, feedback, power dissipation and stuff like that. It makes things easier for me before i even touch the solder iron.
 
In my previous post I forgot to mention the simulator package that I use here in Fairbanks, Alaska.

I have been using Multisim 2001 Professional since it came out and find that it suits the requirements of doing audio work very well. It includes a distortion analyzer and also a bode plotter along with thermal other simulations. Pretty much everything that is needed and quite easy to use.

Multisim 2001 was a upgrade from Electronics Workbench in its various forms and versions. Multisim 2001 Professional has progressed into a fantastic package that is faily inexpensive,(depending on your point of view of course). It is a serious piece of software with great capabilites. If anyone is thinking about buying a software package to do audio simimulation work I highly recommend buying it.

The only draw backs that I can complain about is the some what slow update rate on the screen when scrolling around or moving parts around and perhaps how it lays down its simulated wires. They could improve on these portions of the software. These problems could be assoaciated using the Windows operating system since many other oftware packages that do simulation work appear to suffer from the same problems.

I do not use any of the schematic capture or printed circuit board layout addon features that are available for Multisim 2001 Professional. For this purpose I use Hywire II. It is extremely efficient and easy to use and does not suffer from slow screen updates as things are moved around.

Some may shy away from HYWire II because it is a DOS program. Quite frankly it would be a mistake to do so. I have been using it Since the late 1980's. And have kept it current through the years with the upgrades that Wintek has offered. The technical support has also been very good.

As a comparison of files sizes generated between a DOS based program and a Windows based program. I have drawn the same schematic in Protel 99SE and HYWire II. Hywire II used 134Kb of disk space for the drawing while Protell used a gastly 10MB+ with all the overhead the program has. I did this was a number of years back when I was thinking about changing software. I have also tested software packages by other vendors and have also said no thanks to those. I will stick with what I have.

John Fassotte
Alaskan Audio

[Edited by alaskanaudio on 11-25-2001 at 02:30 AM]
 
I would also like to point out a software for schematic drawing and SPICE simulation. It is named Micro-Cap. They just recently introduced the 7th version of the program. As the program is fairly expensive I use the demo version. But as they have a 50 component (not cicuit node) limit for simulation, it is still well possible to simulate most of my DIY ideas in one piece.
http://www.spectrum-soft.com/index.shtm

The two best sides of the program in my mind is the ease to define the SPICE models for any given circuit (just copy-paste in text mode) and the way the circuit components are placed and the wireing is done.

To: Alaskanaudio. If you get the chance, take a look at the software. I have also tried Circuit Maker, Electronics Workbench, Multisim etc, but I can say I totally fell in love with the ease to draw circuits when I started to use Micro-Cap....

**

Hope it doesn't bother you that the above information seems lot like a marketing hype. I really am just a happy user of the program and not their marketing person :)
 

Nelson Pass

The one and only
Paid Member
2001-03-29 12:38 am
I've used a number of these packages, and they
are very nice, but I have to warn you that the
models aren't perfect, particularly with regard
to Mosfets. I often get behaviour which only
approximates reality.

Distortion predictions? Forget it.
 
ergo,

I do have the evaluation version of MC7 and will play with it sometime in the future. The layout looks very nice.

Reply to Mr. Pass's comment.

Mr. Pass has mentioned that we can forget about doing distortion measurements with simulations. I believe that he is in error since I depend on these measurements during my design work.

As a example I have posted a file on my web site that illustrates a descrete Opamp I have designed for use within my next power amplfier and other projects. The link for downloading this 2 page file is:

http://www.audioamps.com/diyprojects/miscinfo/0202har.pdf

This file shows the expected 2nd and 3rd harmonic distortion figures at a output level of 9 Vrms when working into a 2.21K load. The 0Db referance point is the 9 Volt RMS output level. As you will see the 2nd Harmonic is expected to be -101 DB while the third harmonic should be at around -112 DB down from the 9 Vrms output level.

This level is about ten times higher than the level I expect to use. The simulated and actual measurements of this particular circuit are very close. There is just a very slight rise in distortion at 100Khz when measured with test equipment.

The expected phase relationship of the harmonics are also shown. The proper phase relationships of the harmonics is in my opinion very important to excellent perceived sound quality.

My problem with simulated distortion levels is when they are expected to be very low. When this occurs I have to increase the output levels of the device being simulated so that the distortion figures become measureable with my test equipment. When I can do such verification my confidence in the simulated distortion measurements increase each time I use them.

Many simulation programs allow various spice models to be used for the same device. Thus we can get expected performance values when a parts may be at one extreme or the other of its expected performance.

There are also various ways to model MOSFETS and some papers un the subject indicate that some are extremely accurate. Spice models and simulations can also draw the curves you find in the device data sheets sheets to see how close the models are to the real curves provided by various manufacturers.

If anyone has a problem downloading the file I mentiond please let me know.

My email address is: [email protected]



John Fassotte
Alaskan Audio

[Edited by alaskanaudio on 11-26-2001 at 09:11 PM]
 

argo

Member
2001-02-12 3:35 pm
Estonia
Ergo

Thanks for a great software suggestion (or should I say - tänan hea soovituse eest). First impressions of a program are really positive. Do you also know some similar package for PCB layout? I currently have Eagle Layout Editor evolution version but trying to work with it just drives me nuts.

Argo
 
Hi .....

I'm also very interested in the circuit analysis stuff... so I also simulate a lot in SPICE... and indeed it is very handy .. but I also get some results that just cannot be true... like the push-pull MOSFET simulations .... the complementairy MOSFET devices are just to simple simulated and thus you get results you cannot depend on .... but heee that's a fun game too... to improve your SPICE simulation...:)

greetings,
Thijs
 
to:tschrama

Try the 'text' mode (lower left corner). Micro-Cap accepts the standard SPICE models you can get from manufacturers web pages etc. In that way you will get much more precise models. It still ain't same as in real world but should get you much closer.... It helps to have an idea of the theory behind text mode SPICE models and how to configure them in text mode.

I too believe that the simulator is good for evaluation of dc modes and operating point confirmation, but the measurement and most importanly for audio, listening, is still better done in real world.

Ergo
 
spice models for lateral mosfets

Hello all, this is my first post on this forum. I also wanted to say that this is quite possibly one of the coolest places I have found yet for audio. :D

My question related directly to this thread. alaskanaudio and ergo, you seem to indicate that you have MOSFET models for simulation. Are they lateral MOSFETs? I have looked everywhere, and nobody seems to have any models pre-made for lateral MOSFETs (e.g. 2sj162/2sk1058, buz900p/buz905p, or ecx10p16/ecx10n16).

Worse yet, the MOSFET template I have for my simulator, PSPICE, seems to be geared toward vertical MOSFETs. The template curves aren't the same as the lateral FETs, and I can't seem to change certain params to get the graphs to come out right. So even when I try and make my own models, that doesn't work :mad: .

I appreciate that fact that everyone likes their simulators, but I really don't want to change from the one I use now. Can anybody give me some helpful info? Even some documentation would be welcome. I believe it is also possible to code the model in text mode using custom equations, but I really want to avoid that route if at all possible.

FYI, I want this for a specific amp design I am building. This design uses lateral MOSFETs, which, given that this is my first project, I would really rather not deviate from.

Thanks in advance, and I hope everyone is gearing up for an enjoyable holiday season!
 
Spice Models- Where?

I'm attempting a Spice simulation of a Leach Amp. However, I am missing models for the following devices. I have been searching the manufacturers' sites and came up empty. Does anyone know where to get them? I am using Simetrix 4.0, which uses Spice3 models or HSPICE .MOD files.

1N4004
1N4148
1N5250B
2N3439
243940
2N5415
2N5416
2N6474
2N6476
MJ15001
MJ15002
MJ15003
MJ15004
MJE15030
MJE15031
MPS8098
MPS8099
MPS8598
MPS8599
 
BUZ Lateral Mosfet availability

Users of BUZ900 series MOSFETS

Martinez & Associates,Inc. is pleased to announce that we have been
designated
the sole US distributors for Magnatec's Lateral MOSFET product Line.

We can be reached at:

Martinez & Associates, Inc.
234 Boston Post Road
Wayland, MA 01778
Tel: (508) 358-0131
Fax: (508) 358-2361
email: [email protected]

We accept payment by check or VISA, Mastercard, Discover and American
Express credit cards.

We appreciate your interest in Magnatec.

John E. Martinez


I have succesfully purchased BUZ mosfets from them, and obtained spice models.
dshortt9
 
Mundi

Spice models for the MJ15001 and MJ15002 are available from the ONSemi website. For the other MJ/MJE devices you will probably need to use a model from the ONSemi site for a generally similar device. There are several sites on the web with a model for the 1N400x series diodes and the 1N4148. Can't help with the other transistors as I have not had a need to look for these.

Geoff
 
I've used Spice a few times and found it helpful to get a rough idea of what is going on in a circuit and specifically to predict what the "ideal" circuit ought to perform like. I use this to compare with what it actually does in order to help track down the causes of the differences. This is very useful.

Like Nelson I would be shy to use it to make really accurate predictions. It's one of those garbage in garbage out situations. The simulation is ignorant of a lot of real-world stuff like lead inductance, stray capacitances, ground line resistances, parasitics and real life psus. You could attempt to include models for all of this but your life would be over before you completed them! I believe some simulators allow you to run Monte Carlo sims on the component parameters.