The Frugalamp by OS

Status
Not open for further replies.
Jump to Latest
#481
ostripper said:
Share ..man , Share. I have both running simutaniously..
Click to expand...


Here's what I did:

Add a AC signal voltage to the feedback loop (the red arrow in pic), right click this and st AC Analysis Magnitude to .5V, phase 0.
Ground the input.

Open the AC analysis sim and form the expression like I have in the pic (click "add expression" to do this), but using the nets in your schematic. Run it.
 

Attachments

  • 114.png
    114.png
    50.2 KB · Views: 416
#482
Hmm, I may not have been as far off as I thought. I've been testing full power THD and mistaking the high readings as a simulator problem. I've now found that this is almost never done!

I'll have to get my hands on Multisim and have a good play.

ostripper and MJL21193, you two are seriously spoiling me for choice of designs here 😀
 
#485
MJL21193 said:
I went back through this thread to where Andy first demonstrated the method to you. I have LT4 on this computer (never used) and i looked through the tutorial. Took a while to figure out how to do the same in MS, but I got there.
Just another weapon in my arsenal. 🙂
Click to expand...

Hi John,

Just a minor nitpick. It looks like you are plotting -AB rather than AB. When the amp simulation is correctly configured for negative feedback and it shows the amp as working, the phase of the loop gain AB should be 0 deg at very low frequencies. Then as frequency increases, it should become negative, eventually hovering at -90 deg over a broad frequency band, then finally falling more negative than that in the MHz region. So you need to put a minus sign in your loop gain computation. Not a big deal at all, but just keeps things consistent with what OS is doing.
 
#489
MJL21193 said:
Thanks again Andy,
I have it straightened out now. I didn't even notice the phase was reversed 😱

Better?
Click to expand...

Looks really good. I would not try to lead compensate this bad boy, as the magnitude looks picture perfect. Lead comp would flatten out the high-frequency loop gain and likely degrade the stability.

Now that you guys are no longer using LTspice, I think I will go back to my own project, which is being neglected at the moment 🙂.
 
#491
jaycee said:
I still get the "noise" with BD139/140, but I don't get the nasty, spiky behaviour. However, the KSA1381/KSC3503 models have worked in the past. Likewise, thsi is the first time I've seen the "noise". Something's definitely changed in LTSpiceIV to cause this
Click to expand...

Maybe too little too late, but try edit model and delete (or comment) QCO=0.05 and RCO=50.1187. For me that helps much (in LTSpice). More details here: http://www.diyaudio.com/forums/showthread.php?postid=1686931#post1686931
 
#492
andy_c said:


Looks really good. I would not try to lead compensate this bad boy, as the magnitude looks picture perfect. Lead comp would flatten out the high-frequency loop gain and likely degrade the stability.

I think I will go back to my own project, which is being neglected at the moment 🙂.
Click to expand...


Thanks Andy,
Very good performance in real life too - very stable.

Don't go too far, we will funk up somewhere else and the same advice and fixes apply to both.
🙂
 
#494
ostripper said:

LT's FFT has more resolution...
Click to expand...


That's only the way it appears, as MS just shows the magnitude of the harmonics and not the hash between. I believe it is every bit as accurate and resolved as LT's.
With that said, I don't think you can put too much faith in this. The overall THD yes, but the distribution of harmonics should be used as a rough guide only and not gospel.
 
#495
I changed FA3 (again,updated every instance of it)..
Went back to the good old "blameless", tacked on some
cascodes and widlars and finally have a amp that
will take any model.. ksa1381,mje350,BD-XX, or even the
new one, KSA1221(what a squirrely device)...
FA3.jpg


With 4 different devices it clips the same (symmetrically and
without sticking/saturation)
has the same UGP, and almost non-existant H2/3/5 (sub .001%)
To be sure I did this one on both LT4 and NI circuit design suite
10 (ahem)..

Sometimes simple IS better..
OS
 
#499
Split up R2 into 2 x 4k7. (or 2 x 5k6 for ~5 mA into diodes)
Put the C6 cap from between the two 4k7 to V+ supply.

(4k7//4k7) x 22uF, will give a lowpass RC filter at about 3 Hz

And of course do same thing for R6, C9 on the negative side.
 
Status
Not open for further replies.
Top Bottom