Mr. Self likes 22u , I use a 22u presently. I have seen anywhere between 47u and .1u on commercial designs. As we are just decoupling a simple shunt regulator, "just adequate" quality will do ... no blackgates on my Vbe , please.but that never happens in this industry
does C86 need to be 22uF - a smaller value allows use of a higher quality cap ?
OS
Last edited:
No problem Carlo's , this is good. I never wanted for this , one thing led to another. We still must build up to the "brother" level , this takes time.
OS
Mr. Self likes 22u , I use a 22u presently. I have seen anywhere between 47u and .1u on commercial designs. As we are just decoupling a simple shunt regulator, "just adequate" quality will do ... no blackgates on my Vbe , please.
OS
Ahh, more guessing
the quality of the cap does matter , more tommorrow..
I read that they are no longer in production.
Where I do see them is in esoteric "audiohaus's".. where they want your soul in return 
. They are made with carbon "doped" electrolyte , right ?? Have you researched an alternative ??OS
No, not at this stage. But I bought quite a few about five years ago, and still have some left, which I treasure. They are very useful for the ultimate low noise application.
It is never nice to be reminded that some of these audiophool components are actually very, very good. I have found that electros specced to 100KHz for switch mode use are excellent for audio, and well priced. UCC make such a range.
Hugh
It is never nice to be reminded that some of these audiophool components are actually very, very good. I have found that electros specced to 100KHz for switch mode use are excellent for audio, and well priced. UCC make such a range.
Hugh
Funny you mention 100k , I was sourcing the signal path caps and came across these from panasonic (the EE or FM series) http://www-images.panasonic.com/industrial/components/pdf/ABA0000CE108.pdf
for modern switching apps. and cheap , Mouser has them (33u@35vfor $.11) very low impedance at HF.
OS
I don't know if having a model for the darlington will matter, a 2N5089 driving the usual transistor used here would probably work.
But if I were to predict, I would say that the results at AC won't be much different than those of the C-B capacitor mod. In both cases a hurdle overcome is the Ib of the transistor. The Darlington uses a whole other device to fix this, with little impact on other factors. For instance, the resistor network reduces OLG of the stage by its ratio. The cap mod fixes this as well as the Ib problem by adding only 1 passive component.
Gm is still the same, which is the second hurdle. Compared with the cap mod, using a darlington is still worse, and with the cap mod added to the darlington, performance is still about the same.
A better option is to use CFP instead. This reduces Gm as well as Ib, and sims much better. Add the cap mod and it works wonders.
Sims to follow, let's see if I'm right.
- keantoken
Lower impedance is better, as far as I can tell.
At the frequencies you describe, adding any more active components will only degrade performance in my experience. This is why the single-transistor version works so well there.
The darlington doe have better DC impedance, but CFP is most "bang for buck". Although instability might tip the scales a bit.
I'll add a bypass cap, 22uF .02ohms 22uH and see how that goes.
- keantoken
At the frequencies you describe, adding any more active components will only degrade performance in my experience. This is why the single-transistor version works so well there.
The darlington doe have better DC impedance, but CFP is most "bang for buck". Although instability might tip the scales a bit.
I'll add a bypass cap, 22uF .02ohms 22uH and see how that goes.
- keantoken
Wait a sec, I neglected cap parasitics the first time around. The first plot are the results. This assumes .075ohms ESR, 22nH ESL, 22uF. ESL is worst-case.
Second plot adds the same spec'd caps as bypass. I thought the spike would be worse, since using the same caps adds the same resonances, and the caps will resonate supposedly, but it turns out the resonance comes from the 22nH resonating with the BD139's Cob.
BTW, OS, how do I upload to the site? I made a digital instrument that sounds really nice with Bach, which I'd like to share (It's somewhere between a harp, harpsichord, guitar, and piano). Think "Switched-On Bach" except with only one instrument. I thought I'd upload a few and if there was interest I could upload more.
- keantoken
Second plot adds the same spec'd caps as bypass. I thought the spike would be worse, since using the same caps adds the same resonances, and the caps will resonate supposedly, but it turns out the resonance comes from the 22nH resonating with the BD139's Cob.
BTW, OS, how do I upload to the site? I made a digital instrument that sounds really nice with Bach, which I'd like to share (It's somewhere between a harp, harpsichord, guitar, and piano). Think "Switched-On Bach" except with only one instrument. I thought I'd upload a few and if there was interest I could upload more.
- keantoken
Attachments
So.. the single transistor vbe with the cap mod is the winner.
You can upload here : HFS / This is the "root" of the server. Click the upload folder , a upload button appears on the top right .. then , just like DIYA "manage attachments" , 10 "choose file" buttons appear. Digital instruments are cool , check out my signal generator/instrument (below) Could not blow the "supersym" even with all those waves. OS
Attachments
A 33 ohm resistor in series with C1 or the "cap mod" works. That is assuming that by the time the amp is on the PCB, we will have any control over resonances at 200MHz!
The file is uploaded! Look in the Uploads folder at OS's website for BWV973.mp3, and please tell me what you think! I may be able to improve the sound by rendering differently.
- keantoken
The file is uploaded! Look in the Uploads folder at OS's website for BWV973.mp3, and please tell me what you think! I may be able to improve the sound by rendering differently.
- keantoken
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.