Valve Shunt Regulator Design Walk-Through

We all start somewhere
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#166 Ripple

Post by IslandPink »

My result with your circuit, shown from 16 sec to 20 sec after startup .
Attachments
Mike_TRex_ripple16s.JPG
"Once you find out ... the Circumstances ; then you can go out"
User avatar
Mike H
Amstrad Tower of Power
Posts: 20157
Joined: Sat Oct 04, 2008 5:38 pm
Location: The Fens
Contact:

#167

Post by Mike H »

No that's that 3.5 Hz oscillation again. Look at the time interval between each peak. I had to make C8 larger to stop it. Also run it for 20 - 60 seconds then look at the end of it, 4 sec.s isn't really long enough for it all to properly settle down yet.
 
"No matter how fast light travels it finds that the darkness has always got there first, and is waiting for it."
User avatar
Mike H
Amstrad Tower of Power
Posts: 20157
Joined: Sat Oct 04, 2008 5:38 pm
Location: The Fens
Contact:

#168

Post by Mike H »

IslandPink wrote:Ah. I had a feeling Stephie's models would be more sophisticated.
That's not a guarantee that's what it means, looks unnecessarily elaborate to me. But then I don't understand the math, but, fewer elements tend to run faster. Usually.

I was going to look into making a Koren 6N1P model, but then found it would be identical to the already existing Koren ECC88.

All this could be academic anyway as the models were originally created to make a really basic audio amplifier stage, this kind of application is 'pushing the envelope'. As it was I had to tweak R4 to get U2 - U3 biased so they could work sensibly.
 
"No matter how fast light travels it finds that the darkness has always got there first, and is waiting for it."
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#169

Post by IslandPink »

I'll check the ripple again this evening with a longer settling time as you suggest .

I think you've twisted my arm on the ECC88 in any case, as I realised the lower Gm of the 6N1P puts the bias of the lower triode at only about -1.2V for 10mA current - a little too close to grid-current territory for my liking , especially if you go down to the 2A3 case for 320V-out on the supply . ECC88/6N23P/6N24P should be around -2.2V or a bit more. So, I've been on the case getting some of those sorted.
I also checked the gain of the cascode and it's about x440 with ECC88's , so there we go - every little helps to reduce the Z-out .

Should have the bits to do the supply in a week or so, then all I need is some dull weather and a couple of days free .
"Once you find out ... the Circumstances ; then you can go out"
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#170

Post by IslandPink »

IslandPink wrote:I'll check the ripple again this evening with a longer settling time as you suggest .
Ok, with your triode-5 model, this is interesting . With 10u sub for the C6 cap it's dead smooth as you suggest - down in the uV range.
If you use anything from 1u to 4.7u , the 3.5Hz ripple is there , at ~10-20mV . At 6.8uF , it goes super-smooth, as with 10u . I wonder if this is a real effect or an artifact of the models ?
I think I'll make sure I can fit different sized caps on the board I use for this, so some real tests can be made on the hardware .
"Once you find out ... the Circumstances ; then you can go out"
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#171

Post by IslandPink »

Also interesting - and you may have discovered something v.useful here, Mike : if I do analysis of a pulse current load ( 0.21A baseline, 0.2sec pulse to 0.33A , with sharp sides, then down again ) , the case of C6 = 1u shows a big voltage dip of 40-50V on the B+ , the case of C6 = 10u shows only a few volts dip . Seems like the dip & recovery time is dependant on C6 and presumable on the time constant of C6 with the 300k to ground .
"Once you find out ... the Circumstances ; then you can go out"
User avatar
Mike H
Amstrad Tower of Power
Posts: 20157
Joined: Sat Oct 04, 2008 5:38 pm
Location: The Fens
Contact:

#172

Post by Mike H »

Yes I think so, the oscillation is caused because it's 'hunting', i.e. the low value of C6 is making it over-compensate each way so it goes too far then has to recover. Which is too far the other way again, hence the oscillation. It may well be true in the real circuit as well.

Only thing is, in the real circuit, would you in fact notice a 3.5 Hz oscillation of only a few millivolts?

Would you even look for it, if you didn't know you needed to look for it? :D
 
"No matter how fast light travels it finds that the darkness has always got there first, and is waiting for it."
User avatar
Mike H
Amstrad Tower of Power
Posts: 20157
Joined: Sat Oct 04, 2008 5:38 pm
Location: The Fens
Contact:

#173

Post by Mike H »

IslandPink wrote:I think you've twisted my arm on the ECC88 in any case, as I realised the lower Gm of the 6N1P puts the bias of the lower triode at only about -1.2V for 10mA current - a little too close to grid-current territory for my liking , especially if you go down to the 2A3 case for 320V-out on the supply . ECC88/6N23P/6N24P should be around -2.2V or a bit more. So, I've been on the case getting some of those sorted.
As I keep trying to explain, the spice models are not real valves, but a bunch of equations that "work like" :D

The equations can only follow (and only as best they can) Ia versus Vg @ Va, as like on the datasheet graph. Any other subtleties are outside their scope.

In this respect, a model for ECC88 is practically identical to a model for 6N1P, as it turns out.
 
"No matter how fast light travels it finds that the darkness has always got there first, and is waiting for it."
User avatar
Nick
Site Admin
Posts: 15711
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#174

Post by Nick »

As I keep trying to explain, the spice models are not real valves, but a bunch of equations that "work like"
Yes, but the models are based on the physics of valves which have been well defined since the 40's, its not just a curve fitting exercise. I agree its not real, but Spice is far from a toy,

The equations are not just pulled out of the air,
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
User avatar
Mike H
Amstrad Tower of Power
Posts: 20157
Joined: Sat Oct 04, 2008 5:38 pm
Location: The Fens
Contact:

#175

Post by Mike H »

AAARRRRRGGHHH!!!!!!!


Misconstrued again
 
"No matter how fast light travels it finds that the darkness has always got there first, and is waiting for it."
User avatar
Nick
Site Admin
Posts: 15711
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#176

Post by Nick »

Well, maybe I did, but "but a bunch of equations that "work like"" To me work like infers that its just luck that the same results happens, I was just trying to say that the connection between the models and the physics is more than works like.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#177

Post by IslandPink »

'A bunch of equations that work like' is the basis of any form of computer modelling - even the most sophisticated stuff. The better the models , the more accurately you can simulate . We wouldn't be turning over £100M a year if the bunch of equations that work like a lens system was just a rough approximation that you couldn't trust .

I agree you can't exactly say how a circuit will sound, based on Spice, but you can predict quite a bit about how it will sound .

Anyway, I'm poised to re-build the Power Supply this weekend . I have almost every piece gathered together, the 6N24P's arrived a couple of days ago ( 3 days transit from Lithuania ! ) , and the weather is cooperating by turning sh]te . All systems go :!:
"Once you find out ... the Circumstances ; then you can go out"
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#178

Post by IslandPink »

See this thread :

http://www.audio-talk.co.uk/phpBB2/view ... c&start=60

for the actual hardware !
"Once you find out ... the Circumstances ; then you can go out"
Post Reply