[sneared]

When Apheared posted his AD-823 / EL-2001 amp I was really excited. I thought that this would be a portable amp that could compete with the likes of OBH-11/SEs and MG-Heads.

Man was I disappointed. The "CMoy Killer" amp sounded really "rough" to me. I thought maybe it was a circuit topology or layout problem. So I built it with buffers inside and outside of the feed back loop, and finally settled on the multi-loop layout. The multi-loop sounded slightly better, but the roughness was still there.

Jude said it "Gradoized" his Senns, and PPL complained of "grain."
The CMoy I built on a Hansen board sounded better to me than this amp. And I really couldn't stand to listen to it with Grados.

On to other projects (actually I was ready to quit DIY). But I had those AS8620s sitting around so I pulled out my old board and plugged 'em in.

The difference is night and day. The sound is wonderful. And now it does compete with those non-portables. And the sound is great with SR60s and HD600s, classical, jazz and whatever.

A few specifics: the AD-8620 with one EL-2001 per channel draws about 8 ma at idle compared to 14 ma with the AD-823. Slightly worse DC offset of about 2-3 mV compared to <1mv with the AD-823. My amp is set up with a gain of 5 and runs off of two 9v batteries, and a BUF-634 virtual ground driver.

Save yourself the effort, skip the AD823.

Next, I started looking at the power supply. The ripple on the supply rails is pretty horrendous. I substituted a Didden-Jung regulated power supply for the batteries and the sound was perceptively better.

Any ideas on how to get the ripple to more tolerable levels? Sheer rail capacitance doesn't seem to help. Could be that 9v batteries just suck?! Could also be one reason that the amp seems to take on the character of the virtual ground driver..?

 

[zbuddah]

cool so that is "the" opamp to get eh?

gonna have to get myself a few after i figure out what i'm doing.

 

[ppl]

In no way should an EL-2001 / AD-8610 sound Rough at all. Evean the AD-823 will sound Better than Eather the cmoy or the OBH-11 i cant say about the SE version nor the MG Head as i havent heared these. an AD823/EL-2001 is in the same class as the Headroomm Cosmic. the 823 dose have Grain and this was why i could not wait intill the 8620 Dual came out. I think you are doing somthing seriousl wrong. I cantt make any recomendations unless i see your schematic. But i would like to make a few points. There must be at least a 100-470 ohm resistor between the Opamp output and the Buffers input. This is not optional it is required. Next for The SR-60 one EL-2001 is not Enough you need at least (4) 2001's in parralell of better yet an EL-2008 if you can get some. What is your idea of sheer rail capacitence? to me 30,000uf + Per rail is normal and lessis a compromise otherwise you will have Powersupply Ripple. Next what Kind of 9 volt batteries are you using if thay are Alkaline then yes thay Suck use a Good NI-MH that when new and fully charged can put at least 6 amps into a short circuit for a half second. I test this with my DMM on the 10 amp DC current scale just put the leeds directly on the batt termanals. The new NI-MH from Radio shack and the Gold Peak meet this. most other brands do not the ones in a plastic case are junk do not use them Only about 2 amps. the same as a fresh alkiline. While you are not going to use this mutch current it dose indicate the batteries internal resistence. I have found batteries sound better than evean a good AC Supply. 9 Volt batteries in gen do suck and if possible use AAA or AA cells.

USE A CURRENT SOURCE ON THE OPAMP Biasing the opamp into class A with a current source is manditory. now some folks can't tell the Difference but I can without evean trying. look up the schematic to my portable amp on hasnam's site and use these resistor values around the Multi-loop feedback. in addition use the cap multiplier and Current source isolating the opamp from the output supply, otherwise the ripple on the rails will go right into the opamp just like it was going on the input.

 

[sneared]

ppl-

Thanks for the suggestions.

I don't think I was doing anything wrong with the AD823. It ended up almost exactly a Jung design. To my ears, the AD823 just sounded "rough". I really believe this is what you refer to as "grain" It was NOT oscillation. I built the amp several different ways to make sure that it wasn't a design/layout issue. The resistor between the AD823 and the single EL2001 was adjusted with a scope and function generator to give no under/overshoot. I ended up using 100 ohms, but I also experimented with values above and below this to the point of losing frequency response, or driving the amp into oscillation. As I said, this was completely eliminated by using the AD8620, even without using a current source. I'm still experimenting with that. Still, the improvement of the 8620 over the 823 was so dramatic that I just can't see spending any effort on it. Do you think the 823 is superior?

Haven't tested the 9volts like that because I figured they'd probably be toast. But your point may be correct that the batteries couldn't supply the current. I was using Rayovac NiMHs. I have tried AA packs with some improvement. I was almost considering carrying around SLAs.

Haven't tried the cap multiplier 'cause I'm always at least six months behind your suggestions.

Is there any way to get around having so much rail capacitance? When I get up to those levels, the amp sounds slow to me even using cerrafines and blackgates.

And I'm still trying to figure out how you isolated the output buffer rails from the AD8620 supply rails.

Really appreciate all the infromation you've been putting out!

BTW, Have you actually A/B'd the Cosmic with an AD823/EL2001?

 

[ppl]

Hell No i do not think the 823 is better than an 8620 the Sound of the 8620 is lightyears ahead of the 823 Like i sed the 823 sounded Grany to me and thats why i was waiting for the 8620 to come out to replace the 823 in my pocket amp that was starting to iritate me. But my ears are very sensitive to this as Systems that will Drive me out of earshot to there sound seem to be ok for most others so i must have some sort of over-sensitivity to this type of Distortion. I love the AD-8610 & 8620 these are wonderfull sounding very transparrent and Clean. The only opamp that evean comes close is the 8065, But this requires carfull implimentation to sound good and is too fast for an EL-2001 and maby 2002. I use this with a set of (4) OPA-633's, Nice. Back to the 8620 it also cosumes less power than the 823 as you noticed. I just cant recomend this Opamp enough.

It sounds like you did your homework I am impressed as i thought only myself went through as mutch testing as you did. So Do use the Cascode current source but use change the resistors value from 100 ohms to 1K. the 100 ohms was the value that simulation looked good with the resultent transient measurements and frequency response on the Prototype also looked good but the sound was smoother with 1K. Moving along to the 100 ohm resistor between the opamp and buffer was your tests with the actual phones connected this matters, at what valur did you notice instability and what value produced a rolloff and under what conditions. I ask because i will add this to my data base on this part. Good luck and great to year from you again.


You must have mised this since hasnam has posted it sometime ago but hear it is again go down to portable Amp Doc. that is in MS word format. my latest design of that amp are now on the second Generation and take that concept to the next level.

http://homepage.mac.com/WebObjects/F...eSharing1.html when i get around to it ill try to post the Schematic.

Yes i borrowed a frends cosmic and did some AB tests and the cosmic sounded Flat with no guts and I heared noise like i do from my PC so i gess the cosmics switching supply caused that.

 

[tangent]

I think you're hearing the aggressive tendency of the AD823, sneared. When driving the headphones alone, this tendency isn't so prevalent, probably because the op-amp is struggling with the load. When you buffer it, the op-amp is doing voltage gain in about the ideal environment, so the chip's true character comes through. In a META42, an AD823 does sound a touch edgy with high-impedance headphones, and with Grados it's basically unlistenable. Yet, AD823s in a Hansen is quite listenable with Grados -- a touch weak on the bass perhaps, certainly not aggressive.

I have no idea yet how much a current source will tame this in a META42-like configuration (EL2001, multiloop). I intend to try it fairly soon, though. The AD823 is the best low-voltage op-amp I have, so I'm on the lookout for ways to rescue it.

 

[sneared]

PPL-

Thanks for more suggestions. As usual you have so many ideas that I'm constanly behind the curve. And my biggest problem is fitting all that stuff inside the "portable" enclosure! I have not tried to stack the buffers or try the EL2008. See, I was so unhappy with the amp with the AD823 I was ready to give up DIY. Now there's hope!

I can't remember the values for oscillation and frequency roll off. Mainly because I reconfigured the amp so many times. I have a feeling that the frequency roll off may have been stray capacitance since after I rebuilt the amp several times, it certainly was not optimally laid out. Likewise with oscillation. I recall the oscillation occurred at very low resistor values like 0-10 ohms and even then was kind of a random thing. I did have the actual phones in the circuit (my SR60s, not my HD600s!). I haven't looked closely at the META42 board, but I would guess it would take care of some of the layout issues.

Tangent-

Could be about the Hansen board. I have not driven the cans with the AD823 alone since Apheard's design used a buffer, and I would be violating one of PPL's most adament design requirements . I have always used it with a buffer. All I did was substitute the AD8620 for the AD823 in my amp. Maybe I'm overly sensitive to the edginess as you describe it. Like I said, I couldn't stand to listen to music through the amp with Grados. And HD600s sounded too "raw." Just substituting the AD8620 took all that away. Extremely smooth, but with all the punch and dynamics needed. I'll post some frequency response tests like I performed with the OBH-11/SE as soon as my beer supply is replenished.

 

[erix]

ppl, I'm wondering if you could explain some portions of the schematic from the link you posted. I've studied this thing for hours over the past 6 months or so but I have a couple of questions.

I have a feeling most of them have to do with stability as it pertains to high-speed design but I wonder if you can explain it in lay-terms?

R1, the 10 Ohm resistor from the power rails to the supply pins on the opamp. What do they do?

R5, the 470 Ohm resistor before the volume pot - What does it do? Compensate for mistracking Volume pot? It's the same value as R8 - is it there to eliminate DC offset when using non-FET input opamps?

R12, the 4.7 Ohm from inverting input to ground via R8 - What does it do?

Q1, J1, & C1 - Is this the Capacitance Multiplier? What is the Transistor and FET?

The Cascode Current sources seem to be coming, one from the positive rail and the other from the negative - I thought you only needed one from the negative rail?

Last one!

U1, D1, and J7 in the power supply section. Is D1 a zener? What type? What is U1 and what does it do? J7 - is this some sort of polarity protection?

Thanx kindly for your help!!

ok,
erix

p.s. OCF in 9 days and counting!!

 

[ppl]

R1 is to isolate the Audio signal ground including the Feedback as well as Vol cont and Resistor in Par with opamps + Input. The 10 ohm value is low enough to be transparrent to Audio frequencies but highly Lossy at Rf Frequencies. this prevents any indictive coulpeling fron eather the Output or any Supply By pass ground return. This amp i think dose not use a current source as it was my First gen. the Fet transistors on the Supply rails are Current sources floating the opamp's voltage supply upon the Main rails but providing about 20 dB of reduction in any nioise on the mail rails for any reason be it load or powersupply Ripple. the capacitent multiplier is Q1 and Q2 this amplifies the 1,000uF capacitor c1 & C11 by about 250 times. The resistors on the opamp supply rails are used in alot of High end Audio circuits These resistors along with the Small caps on the Opamp rails C3,C19 and C9 & C10 provide high frequency decoupling. this is most often the only isolation most designs provide to isolate the opamp from the noise on the powersupply rails. it is generly recomended in High Speed opamps to be stable. I have however in this design added the cap Multiplier as well as the current source to dramiticaly improve the isolation from the lower than Audio range up to where the capacitence of the Transistors become dominent. this is at an ureal frequency and is alot higher than the Audio video and Vhf range. the 470 ohm resistor prior to the pot Buffers the Pot from ultralow source impedances. pots should not be driven with less that 1/100 th its resistence to avoid an overdampened condition. u1,D1 & J7 form a simple but accurate dead battery indicator when the battery voltage gets to one half the LED gose out. thus indicating a battery needing evean tghough the amp may still be playing. U1 is a 3.3 Volt Bule Ultrabright LED D1 is a Zener connected Transistor for low noise. it is set at about 8 volts. the transistor J7 is a Current source, i dont just use these on Opamps. the current source asures a constsnt curent to the LED untill compleat dropout thus the LED stays Bright untill it gose out with out getting gradualy dimmer as a resistor would do.

Hope i havent forgot anytthing>

 

[erix]

Just what I needed!!

ok,
erix