Readers Respond: Another Mailbag

This month we take a look at some recent mail sent to this column. While a variety has come in, much of it has been concerned with computer-related issues. "Ouch!" you say? If you're worried that that means another round of computer hardware and tech support horror stories, that's not quite true. This time, we'll touch on some considerations about the computer operating system (OS), as opposed to the underlying hardware. But, we'll also focus at greater length on a range of Spice issues.

Windows 95-Related Items: A letter on Windows 95 weaknesses came from an old friend and fellow audiophile, Dave White, of Baraka-Intracom.

Hi Walt.
I found it amusing to read your April 6 Computer Tech Support comments. I wanted to forward a few of my own related experiences.

My office system crashed last week after installing Norton Utilities 3.0, and then would only fire up in safe mode. First, I went through the device manager trying to correct the problem, with no success. Then, I tried reinstalling Windows, but no luck—I kept getting a password error. I was forced to reformat the hard drive and reinstall everything, after trying to backup data files to a floppy in DOS. To say I was unhappy was an understatement.

I don't consider Windows 95 a robust OS. While testing software at my present company (as well as a previous one), we do fresh Windows installs before every software test, as the only way to ensure a consistent environment. To safely run Windows 95 on a system which must keep running, I now follow these rules:

1. Never install software you don't absolutely have to have forever.
2. During install, shut down every single background program (those little lower right-hand corner icons).
3. Never install trial version (timeout) SW. The released version from many SW vendors cannot be installed over the trial version. Plus, you also can end up with constant error messages.
4. Be cautious installing software without an accompanying uninstall program. If you have problems,you may be stuck with it. Always backup your registry files (user.dat and system.dat) before installing.
5. Don't use disk cleanup or move programs, and don't let your "C:" drive get down to less than 100 Mbytes of free space. These cleanup programs are like playing "Russian Roulette" with a fully loaded gun. Any program you install will have a substantial amount of data installed to C:, even though you tell it to install to D:, and it will show as installed on D: in Windows Explorer.
6. Don't install software with automated online upgrade features. I get sick of the windows popping up telling me to upgrade at a $5 per month fee. Why wasn't the software properly done in the first place?

As you are aware, Windows 95 uses the registry as the "brains" of the system, and it doesn't take much for it to get corrupted. If you have a lot of software installed and are dependent on your machine, you should take the time to get a book explaining the registry and how to maintain and/or modify it.

Hi Dave! Many thanks for the list of Windows 95 cautions. I can relate to some of your experiences, as several months back I had installed Norton Utilities 3.0, expressly for the touted crash-management recovery features. I didn't have problems at first, but the constant update reminders drove me crazy. It did catch program crashes and prevented lock-ups. Then, more recently, I experienced a hard lock-up from which the Norton Utilities couldn't recover, and I lost some data. I then decided that such limited and costly protection wasn't necessary after all. Bye-bye, Norton Utilities!

Crash management and/or prevention utilities, such as those by Norton, are really symptomatic of basic limitations in the underlying OS. It makes much more sense to me for that system itself to be rock-solid—which Windows 95 simply isn't. No user of an OS should ever need to buy books about how the OS stores basic program data, just to be able to recover more quickly from crashes!

Spice Correspondence: Lots of Spice mail was received on the Feb. 9 and May 1 columns, which I consider a healthy improvement over OS and other PC-related glitches!

A reader wrote in regarding a problem accessing the "free Spice" URL in the May 1 column. He may have made a typo. Or, a subtle change of case within a typed-in URL can cause things to fail (UNIX systems don't tolerate this). A full and correct URL is repeated here—type it in exactly:

http://www.repairfaq.org/~filipg/HTML/FAQ/BODY/F_Free_Spice1.html#FREESPICE_001

Vacuum Tube Spice Models: Earles L. McCaul of Raytheon Missile Systems wrote about his experiences with vacuum tube and transistor guitar amplifiers, and the desirability for more tube Spice models. It's likely he had no idea what an info avalanche this might unleash...

I read with interest your May 1 commentary on "Spice Programs," and would like to speak up for the "Antiquity-Boys," those of us who still design, work with, and use vacuum-tube equipment, e.g. musicians! There are literally millions of Fender, Mesa-Boogie, Kendricks, etc. guitar amplifiers still in use. So long as the musicians using them perceive these amplifiers as a "part of the musical instrument" process, there will be vacuum tubes around. And, somebody will be needed to design the amps!

As a musician and EE, I've studied and used both tube and transistor amplifiers. At first, tubes were something to be "replaced" by the smaller, lighter, more efficient transistors. Then reality set in, as side-by-side playing showed tubes to be more "warm" sounding and easier to "play into distortion" than were transistor amps. Sure, transistors had more power and less weight, but they didn't sound "familiar." So, they were immediately branded "bad" by any and every musician worth his guitar strings. Except for bass amplification, guitarists considered tubes "IN" and transistors definitely "OUT."

What I would like to see is a Spice library suite for the most common vacuum tubes, such as triodes: 12AT7, 12AU7, and 12AX7/7025/ECC83; power pentodes: 6V6GT, 6L6GC, 6CA7/EL34, 6550, and 7027; and rectifiers 5AR4, 5U4, and GZ34. Vacuum tubes aren't dead, they're just DAMN expensive!

In my first response to Earles, I referred him to Marshall Leach's work, among the better known papers on Spice models for vacuum tubes.¹ See Leach's web site at http://www.ee.gatech.edu/users/207/ for more information, as well as references to using Spice for other tasks.

I also alluded to several other available sources, which I have now tracked down. Bear with me here, as there is lots of Spice info on tubes. Others published are the models by Intusoft,^2,3 and those of Broydé and Clavelier,⁴ and Rydel.⁵ There also have been some letters on tube Spice modeling, which provide an overview of several approaches.^6,7

The publication Glass Audio has had numerous articles on Spice-based preamp design and related modeling techniques.^8,9 A recent AES paper also discusses an improved model for triodes,¹⁰ and more Glass Audio articles and letters address modeling improvements and preamp circuits. ^11-13

Some time after my first correspondence with Earles McCaul, Tom Bruhns of Hewlett-Packard wrote in after reading the second Spice column, and offered some timely modeling advice.

Hi Walt,
Just read your May 1 column in Electronic Design, and I didn't notice a pointer to: duncanamps.simplenet.com/spice.html.

Duncan has models there that can be hard to find; for example, some vacuum-tube models. He also has lots of pointers to simulators and sources of other models.

I didn't catch your earlier articles, but thought you'd enjoy a really simple circuit example that can cause convergence problems:

It consists of a series circuit of a step generator (say, 0 V before t = 1 ms and 20 V after), a diode (e.g., 1N4001), an inductor, and a capacitor.

For convenient time scaling, make the LC resonate around 1 kHz. Run a default .tran simulation, then try again with integration=gear. It is just a resonant-charging circuit, but it is also the simplest circuit I've found, which causes Spice to remind me that I still have a head, and it's worthwhile using it!

In standard Berkeley Spice3f5, and in other Spices I've tried, default integration results in oscillation of the voltage between the diode and inductor. Adding resistance in various places doesn't seem to help it converge, at least if the resistance is high enough (or low enough, if in series) to not ding the Q too much. Gear integration makes it behave.

I find Spice to be a very good tool to ensure that I put appropriate thought into my circuits. Sometimes running a simulation points out shortcomings of Spice, sometimes shortcomings of the macromodels (when used), and sometimes shortcomings of the circuit itself. But all that's OK, as my head can sort it out just fine. I value all inputs early in the design. And, sometimes it's been valuable late in the design, too. I've used Spice with very abstracted models of my circuit to analyze performance of nonlinear feedback systems, and it has saved me huge amounts of bench time.

Thanks for the input, Tom. I found Duncan Munro's web site to be quite a treat, as not only does he have online many of the models that Earles McCaul requested, but lots of other info as well. See the Figure for Duncan Munro's 6SN7 plots. As you most likely know, you can sign up for e-mail notification on model updates as they appear. Also, thanks for your insights using Spice as one element of an overall design.

I noted that some of Duncan Munro's models exist in PSpice, as well as 3f4 formats. For those wondering about what seems like duplication, this arises from the existence of some very useful functions within PSpice, which aren't available in more generic Spice versions. Duncan explained it thusly:

These (functions) are ones that I specifically use with my models, and are not available in either 2g6 or 3f4; they are purely proprietary to PSpice:

".PARAM"—allows parameters to be passed, "LIMIT\{x,min,max\}"—to constrain a value.

There is no way round the lack of .PARAM in 2g6 or 3f4. If you refer to the PSpice file, you will see that two models have been written, and all of the tube types come from parameters being passed. Nice and neat.

In the 3f4 version, you will see that each model is described individually. In actual fact, they are all exactly the same model; it's just that some of the parameter values in them are different.

TIP: The list of various vacuum tube Spice model references below is by no means a complete list. While reasonably extensive, it also can be expanded considerably from citations within. You also can supplement your info on Spice via the web, by doing an AltaVista search: http://www.altavista.digital.com/. Use a search entry of +"vacuum tube" +spice, and you'll get enough hits to keep you busy for some time.

I Meet Electronic Design Readers: In the May 1 column, I had a brief note about hoping to meet some of the column readership at upcoming seminars. Now that the tour is over, I'm happy to say that more than a few folks did stop by to say hello, identifying themselves as column readers.

My thanks, first for reading the column, and second, for stopping by and saying hello. It was a real treat to meet some of the readership firsthand.

As always, reader comments are invited on Spice (or other) topics. Enjoy not just the reading, but also those warm tube sounds !

References:

1. W. M. Leach, Jr., "Spice Models for Vacuum-Tube Amplifiers," Journal of Audio Engineering Society, Vol. 43, No. 3, March 1995.
2. L. Meares, "A Spice Model for a Vacuum Tube," Intusoft Newsletter, Feb. 1989.
3. "Modeling Vacuum Tubes, Parts 1 and 2," Intusoft Newsletter, Feb., Apr. 1994.
4. F. Broydé and E. Clavelier, "Modélisation et Simulation des Circuits à tubes avec ISSpice 3," Electronique Radio Plans, No. 553, Dec. 1993.
5. C. Rydel, Simulation of Electron Tubes With Spice," Preprint No. 3965, Presented at 98^th AES Convention, May 1995.
6. E. Pritchard; W. M. Leach Jr., "Comments on 'Spice Models for Vacuum-Tube Amplifiers'; author's reply," Journal of Audio Engineering Society, Vol. 45, No. 6, June 1997.
7. F. Broydé, E. Clavelier, C. Hymowitz, C. Rydel; W. M. Leach Jr., "Comments on 'Spice Models for Vacuum-Tube Amplifiers'; author's reply," Journal of Audio Engineering Society, Vol. 45, No. 6, June 1997.
8. N. Koren, "Spice and the Art of Preamplifier Design, Part 1," Glass Audio, Vol. 9, No. 2, 1997.
9. Norman Koren, "Spice and the Art of Preamplifier Design, Part 2," Glass Audio, Vol. 9, No. 4, 1997.
10. Walter Sjursen, "Improved Spice Model for Triode Vacuum Tubes," Journal of Audio Engineering Society, Vol. 45, No. 12, Dec. 1997.
11. John Harper; Norman Koren, "Spice It Up; author's reply," Glass Audio (Letters), Vol. 10, No. 1, 1998.
12. Stefano Perugini, "A PSpiced Preamp With THD Cancellation," Glass Audio, Vol. 10, No. 2, 1998.
13. Jean-Charles Mailett, "Algebraic Technique for Modeling Triodes," Glass Audio, Vol. 10, No. 2, 1998. (see web site with triode and rectifier models at http://www.lynx.bc.ca/~jc/tubes.html)