Be sure the series L of the supply to the circuit are as low as possible. Short, low L/Z distribution is best. One way is to make a wide, closely spaced traces for low L, high C pcb path. For DIY'ers this also helps prevent ringing and possible oscillations in wideband circuits. At low freqs small series L may not matter. However, at low freqs the DIY'er needs to keep signal and power returns from one another.
The more advanced forum members might want to look into the books by Don White (ICT) on EMI control et al as many of the digital products, if sold, would have to pass UL EMI/RFI or other countries requirements and it starts at the pcb to contain. Not for this Headphone amp subject but think if you do such work, it just becomes habit when designing pcb for anyting else - audio, even..... just becomes habit.
Back on your heads !
The more advanced forum members might want to look into the books by Don White (ICT) on EMI control et al as many of the digital products, if sold, would have to pass UL EMI/RFI or other countries requirements and it starts at the pcb to contain. Not for this Headphone amp subject but think if you do such work, it just becomes habit when designing pcb for anyting else - audio, even..... just becomes habit.
Back on your heads !
Last edited:
??? if you cut a ground plane up into areas, you increase the inductance (Z) of that ground plane. This is often seen when there is one ground plane on the same level as the signal traces. Better to have ground plane on seperate pcb level or opposite side from singal traces. Not the same thing here with seperate ground planes.
Last edited:
No complainst here about what you have said..... remember though - a circuit capable of 1MHz and beyond can become unstable with uncontrolled stray L/c in the layout.
[My background is in RF/microwave precision measurments (secondary standards lab), Power systems and their grounds from 230Kv to 13.8KV to 480 v/3000A buss and their grounds and for pulsed power experiments of 10 megaWatt levels. Grounding of complex systems with computer facility system grounds, control system grounds, ac/dc/pulsed power grounds all together in large experimental facilities. Instrumentation for underground nuclear 'events' back in the day. Ground planes for experimental plasma excellerator using Maxwell cap banks discharging 100Kv into gas/plasma/ emi/rfi/emp etc etc etc. I understand the grounding issues from power to microwaves and how they interact in a system. --- just in case anyone is interested. ]
Ah, thanks for that background! I'll remember and now know where to ask my questions!
I have played with 60Kv into laser capacitor banks in my career. But what specifically is a Maxwell capacitor bank? Google comes up with a brand name, and I presume that is not it.
There actually is such a thing as a flux capacitor! But it uses magnetic flux and behaves as a capacitor not an inductor! The phase angle is what you would expect from a capacitor because of the way the field is loaded!
Maxwell Technolgy out of san diego have the nitch on HV discharge caps... yes, were also used in flash lamps for pumping lasers to higher energy levels.
No. Thats fine. can also put a bypass cap across the 10 ohms so the ground is low at Hf and only affective at 'hum' frequencies.
Also can try using shunt caps, as shown on pcb, withOUT lowest esr for damping, if needed.
Ohh, it's coming!
I spent some time on vacation taking a closer look at the article and some of your comments intrigued me and I'll definitely have some questions. Some things you describe jibe with what one forum member (ex. Nat Semi) discussed with me in a few private emails.
quote on page 32 of LA vol. 3
@RNMarsh
On page 32, I read the following:
"I also preferred the arrangement because the input and the feedback occurred in the same transistor - making the most direct correction."
Forum member audioman54 made a similar comment re. the Nat Semi LME chips (that would be Mark Brasfield who once worked on the LME chips). I had never given it any thought until I read his postings and the same comments in his personal correspondence with me. Many moons ago, I bought Walt Jung's book on audio op amps and read other articles where you see feedback and signal input are separate and so I always thought that's the best way. So your comment caught my eye because it is similar to what he told me.
Are the benefits measurable? Or one of those things where you can hear improvement? Can you expand a bit on this comment and your experience?
@RNMarsh
On page 32, I read the following:
"I also preferred the arrangement because the input and the feedback occurred in the same transistor - making the most direct correction."
Forum member audioman54 made a similar comment re. the Nat Semi LME chips (that would be Mark Brasfield who once worked on the LME chips). I had never given it any thought until I read his postings and the same comments in his personal correspondence with me. Many moons ago, I bought Walt Jung's book on audio op amps and read other articles where you see feedback and signal input are separate and so I always thought that's the best way. So your comment caught my eye because it is similar to what he told me.
Are the benefits measurable? Or one of those things where you can hear improvement? Can you expand a bit on this comment and your experience?
Last edited:
My own work in this area dates back to the 1970's. I was developing toplogy similar to the now current-mode feedback amps at the time. I now call it a diamond amp configuration. It is very common now. I'll get some old schematics and old articles put up here, soon. But if you want the short story history - go to the Headphone amp article in Linear Audio. Too much to duplicate here. Anyway, it was WAY before IC makers started doing it this way and now have many variants.
Boy, I never thought I would be this old that i would be writing about audio history with myself in it. I'll tell you all that if you spend every day and evening and weekends for twenty-five years studying, have no children (1) until late in life and are willing to keep doing this while your spouses (4) walk off.... and have access to the worlds largest technical library and great minds involved in research around you ---- you can far out distance mainstream. But it will cost you. So, I am willing to give what I know and can to help others take a shorter, easier path.
Boy, I never thought I would be this old that i would be writing about audio history with myself in it.
I'll tell you all that if you spend every day and evening and weekends for twenty-five years studying, have no children (1) until late in life and are willing to keep doing this while your spouses (4) walk off.... and have access to the worlds largest technical library and great minds involved in research around you ---- you can far out distance mainstream. But it will cost you. So, I am willing to give what I know and can to help others take a shorter, easier path.
Last edited:
kinda tough when Harris was selling CFA in early '70s, unless "WAY before" was months, weeks?
Walt Jung's History chapter in his (free, online) Op Amp Applications tome is worth reading
Google turns up a 1972 ref to HA2500 second source
many cirucit topologies were implemented in potted modules, hybrid circuits before semi processes allowed fully monolithic impelementaitons
Ha! 1 wife and 4 kids for me! That is why I am so farrrrrr behind. I won't catch up.
BTW, I have tried the approach recommended by audioman54 and was pleased that it worked. Your comment caught my eye, along with others in the article. So hopefully I'm not too pesky and receive a big bill from you in the mail.
More questions to come. And thanks for the upcoming material.
I missread Walt's history re HA-2500
I've poked thru a little more of the history - the HA-2500 is a conventional diff pair input amp
wikipedia credits CFA op amps as dating to 1980's patents - so Marsh could easily have anticipated those patents
but later in Walt's history he shows a 1937 tube amp "CFA" circuit
such series feedback amps became ubiquitous with single polarity input Q bjt amp circuits
I find it hard to believe no discrete bjt amp with complementary/offset cancelling front end was ever made in the prior 2+ decdes of discrete complementary Q availbilty before the fast complementary monolithic "CFA" op amps of the '80s
I've poked thru a little more of the history - the HA-2500 is a conventional diff pair input amp
wikipedia credits CFA op amps as dating to 1980's patents - so Marsh could easily have anticipated those patents
but later in Walt's history he shows a 1937 tube amp "CFA" circuit
such series feedback amps became ubiquitous with single polarity input Q bjt amp circuits
I find it hard to believe no discrete bjt amp with complementary/offset cancelling front end was ever made in the prior 2+ decdes of discrete complementary Q availbilty before the fast complementary monolithic "CFA" op amps of the '80s
Last edited: