Can I trust SPICE? - diyAudio
Go Back   Home > Forums > Amplifiers > Tubes / Valves

Tubes / Valves All about our sweet vacuum tubes :) Threads about Musical Instrument Amps of all kinds should be in the Instruments & Amps forum

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 13th July 2007, 04:36 PM   #1
PeteN is offline PeteN  United Kingdom
diyAudio Member
 
Join Date: May 2007
Default Can I trust SPICE?

Hey all.

In the absence of any old equipment to salvage, a lottery win, and a big box of bits to 'experiment' with, I've been playing around with SPICE, anode curves and so on, and I've come to a circuit which in the simulation is, as far as I can tell, 'pretty good'.

It's a "simple" cathode coupled/common cathode/whatchamacallit using 12AX7 (ECC83). Design called for about 70Vp-p output from a 2V input. Simulation doesn't include volume pot so this is "full blast".

Schematic is here, an output trace here.

I searched the forums, and came across the Tube Library I'm using there, which is the Duncan Amps one (tube_im.olb), using the 12AX7 in place of a specific ECC83 model.

Outputs are about 74.4 (180) and 74.5 (360) Vp-p from each output (measured at the anode), with a 2V (peak) input sine wave at 1kHz. So gain is (74.4Vp-p/4p-p) = 18.6 and 18.625 respectively. NB in the trace it looks like the sines are slightly flat topped, but if you zoom in close enough or run the sim at a higher resolution, they aren't.

I was wondering what the odds are of it behaving this nicely in the real world? Obviously component values in the real world change as things warm up, age, and/or explode, but I'd like to think it was a pretty good representation?

Also I couldn't get a JFET CCS to "work" (aside from raising cathode voltage it had no discernible effects), anyone know if this is just a limitation of SPICE or the tube models?
__________________
-- Peter
  Reply With Quote
Old 13th July 2007, 04:48 PM   #2
kevinkr is offline kevinkr  United States
diyAudio Moderator
 
kevinkr's Avatar
 
Join Date: Sep 2004
Location: Boston, Massachusetts
Blog Entries: 6
It's far from an optimum implementation, the load resistors are far too low in value as is the tail resistor for use in a long tailed pair. (For low distortion load resistors should always be at least 2X - 3X larger than the rp of the tube they are loading. 12AX7A has an rp of about 66K - 88K depending on operating point.) The poor balance you are seeing is a reflection of that fact. This circuit is commonly known as an LTP.

Spice is only as good as the models you use with it, most tube models range from bad to good, few are great and most are not accurate for distortion sims. (fft) They are very good for gain, bandwidth, and reasonably good for operating point depending on the model.

Your jfet based current source didn't work because there probably isn't enough compliance voltage across it to approach the constant current region. (Pinch off voltage is probably way higher than what you have across that node.)

All manner of current sources work great in spice, that said jfet sources are generally cr*p. Use a couple of bipolars in a ring of two or similar.

Take a look at a couple of good treatises on line for tube circuit design.
The tubecad site by John Broskie is a good place to start.

Also I don't know what sim package you are using but if it is a student version I strongly recommend you download LTspice (Switchercad III) and add the tube models to it. Very powerful, no limitations on models or nodes, and pretty easy to use. Has its own support forum on yahoo as well.

Lots of stuff on tubes, current sources, etc. on wikipedia as well.
__________________
"Extraordinary claims require extraordinary evidence." - Carl Sagan
  Reply With Quote
Old 13th July 2007, 05:43 PM   #3
PeteN is offline PeteN  United Kingdom
diyAudio Member
 
Join Date: May 2007
Ah. The reason the anode loads were so small, was that with the larger ones, not only was there *far* too much anode swing, but with it came a *lot* of distortion and there was hardly any current through the tubes - <100uA. I was under the impression that having a tiny current would add problems (dis)charging up the various interelectrode capacitances, so I wangled it to have a reasonable current through it.

I'll have another go
__________________
-- Peter
  Reply With Quote
Old 13th July 2007, 08:14 PM   #4
Jeb-D. is offline Jeb-D.  United States
diyAudio Member
 
Jeb-D.'s Avatar
 
Join Date: Nov 2005
Location: SoCal
Since it seems like your new to Spice let me give you an intro

Generally, spice is pretty usefull, but gives you optamistic results. You just have to use it enough to know the fact from fiction.

Spice doesn't account for tolarances. So give yourself headroom, avoid designing things that require a pretty precise or fine tunning. Such as "this should give me just enough voltage to drive the output to full power".

Most tube models are much more linear in spice than real world. So THD is usually much higher in the real world. Especially in the non-linear region of the tube curves (low current). Spice will usually give you an exaggerated gain when in the non linear region as well.

Constant current plate loads don't swing as close to the rail in the real world. Spice will usually let the voltage swing to within a few volts of the supply rail (because most transistors are ideal in spice). In the real world all the CCS's ive made are lucky to get within 20V of the supply rail before clipping hard.

Spice doesn't show parts blowing up. So be sure to check that all the devices aren't theoretically exceeding any parameters before actually building it.
  Reply With Quote
Old 13th July 2007, 11:26 PM   #5
kevinkr is offline kevinkr  United States
diyAudio Moderator
 
kevinkr's Avatar
 
Join Date: Sep 2004
Location: Boston, Massachusetts
Blog Entries: 6
Quote:
Originally posted by PeteN
Ah. The reason the anode loads were so small, was that with the larger ones, not only was there *far* too much anode swing, but with it came a *lot* of distortion and there was hardly any current through the tubes - <100uA. I was under the impression that having a tiny current would add problems (dis)charging up the various interelectrode capacitances, so I wangled it to have a reasonable current through it.

I'll have another go
Good idea, and an indication that you have also probably selected the wrong tube for the job - try a 6DJ8..


__________________
"Extraordinary claims require extraordinary evidence." - Carl Sagan
  Reply With Quote
Old 14th July 2007, 06:09 AM   #6
Jeb-D. is offline Jeb-D.  United States
diyAudio Member
 
Jeb-D.'s Avatar
 
Join Date: Nov 2005
Location: SoCal
Quote:
try a 6DJ8.
Yeah try a 6DJ8 for less swing, but not at that low of current. I've heard they sound bad in current starvation.
  Reply With Quote
Old 14th July 2007, 06:47 PM   #7
PeteN is offline PeteN  United Kingdom
diyAudio Member
 
Join Date: May 2007
=) Well I had another go, and I've managed to get another "candidate solution" based on the advice here - so thanks chaps!

I also found a useful site which I'd overlooked before (guitar amp site, is probably why), and followed an example on there which (with a bit of tweaking) lead to this result. That site starts with a 12AT7 and redesigns for a 12AX7, but it was the method I was following, not the tube selection hehe.

Changes are that tubes are now ECC81/12AT7, seeing as the '83 had too much cow factor (moo... mu... OK so that was a terrible joke).
Loads are now about 3ra, at 36k (being closest E24 value to 35k, chosen for convenience at 350V/10ma).
Allowing 100V across the tail resistor for better CCS approximation, leaves 250V plate, which keeps this within max dissipation.
Tail now consists of two resistors in series, 420R cathode resistor and 18k tail. Quiescent current is 1.87mA per section.
Both tubes are now fairly balanced, 4.77% difference in amplitude, but I'm sure much less is desirable for Hi-Fi. I tried boosting up the tail resistance a bit, which helps, however I didn't want to go too high as Va max is 300V. Yet, as Va0 is 550V I increased the tail to 43k which (with RL increased to 36k to recover some gain) lowered the mismatching to 2.15% (70.1 and 68.6 Volts peak-to-peak on each triode section).
Is this insignificant enough for driving a push-pull output stage requiring on the order of 65Vp-p? (obviously volume control can+will be used stop over-driving the output stage)
Bearing in mind component tolerances times Spice inaccuracies, is it worth pursuing balance to less than 1% for example?

... again these readings come with the assumption that Spice is being reasonable!
__________________
-- Peter
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Capacitors--whom to trust? Quince Parts 92 17th April 2007 01:40 PM
Can I trust my Valve Tester gingertube Tubes / Valves 1 8th April 2005 09:14 AM
eBay: Can I trust this guy? Devil_H@ck Everything Else 10 31st July 2004 09:49 AM


New To Site? Need Help?

All times are GMT. The time now is 01:08 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2