Calcs for op amp loading vs THD?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Markey US US 8,233,643
Also forgot about Chalupa who published an AES paper. So that makes at least 4 of us.
Both Chalupa and Markey burn off the DC CM voltage whereas I AC-coupled it with films after the floating stage.

Here discrete "British" Mic Pre modified to be "flying" and for much lower HD.

DC coupled using cascoding level shifter. Birt differential follows, with both common mode servo and differential DC servo.

My design also included fully electronic gain switching with 6dB coarse steps in the "flying input" (using now discontinued J108 SMT versions that give ~ 6 Ohm fully saturated) and fine adjustment in the Birt stage.

It was intended to be more or less functionally parallel to PGA-2500 with the need for AC coupling removed and some other PGA-2500 limitations removed.

I would now use discretes instead of Birt and a level shifted (use MOSFET's) classic feedback loop to input emitters.

Thor
 
it's kind of a solution searching for a hypothetical problem.

I disagree.

Capacitors are the least "ideal" parts around, electrolytic capacitors more so.

All "Audio Grade" designed for low distortion Capacitors (and yes, they have on average 40dB+ less hd than generic electronic cap's) seem to be EOL'ed.

So if you want a high quality "transparent" Mic Preamp getting rid of electrolytic capacitors is essential.

Sure, you can use 47uF/100V DC MPK from Wima at 45mm x 65mm x 57mm and at the bargain price of only 34USD if you buy 40pcs.

Doing it electronically costs a fraction and uses what is basically existing circuitry that needs to paid for anyway and is just somewhat extended using a few extra parts that are cheap.

Thor
 
Capacitors are the least "ideal" parts around, electrolytic capacitors more so.
Aside from distortion and bandwidth concerns, I avoid 'lytics because I probably won't be able to afford to have a service department, and the last thing I need in a decade or two is units coming back. It makes for some interesting design challenges. There's a lot you can do with a 1µ/63V WIMA PPS. I'll be interested to see whether Murata expands their high-value C0G line; they're at 470n now, albeit only 25V.
 
Last edited:
Aside from distortion and bandwidth concerns, I avoid 'lytics because I probably won't be able to afford to have a service department, and the last thing I need in a decade or two is units coming back. It makes for some interesting design challenges. There's a lot you can do with a 1µ/63V WIMA PPS.

Agreed, but you need to keep current noise low.
At that point it's desirable to use J-Fet's where current noise is low, can always parallel a few for low Voltage noise.

How about 4 x 2SK2145 as four separate parallel differentials? Suitable VAS * buffer.

With a bootstrapped input resistance you can even go 100nF coupling cap with acceptable noise levels.

I'll be interested to see whether Murata expands their high-value C0G line; they're at 470n now, albeit only 25V.

They are SMD, you can parallel a few. Easy. Panasonic and others also have excellent SMD film types.

Thor
 
I disagree.
yup
Capacitors are the least "ideal" parts around, electrolytic capacitors more so.
All passive components have their performance limits
All "Audio Grade" designed for low distortion Capacitors (and yes, they have on average 40dB+ less hd than generic electronic cap's) seem to be EOL'ed.
I wrote about capacitors back in the 80s also... There was a lot of drama in the audiophile press back then about capacitors (and everything else).
So if you want a high quality "transparent" Mic Preamp getting rid of electrolytic capacitors is essential.
Show me the data... I am repeating myself but heavy lifting for electrolytic caps in a mic preamp is generally in series with the gain pot of the more common topology. At high gain the cap can be in series with single digit resistance (some manufacturers cheat by specifying their preamps at low gain). The DC blocking capacitors in series with the mics, isolating the phantom voltage should be good quality but are not working into very low impedance (nominally 2k ohm driven from 150-200 ohm mic source impedance).
Sure, you can use 47uF/100V DC MPK from Wima at 45mm x 65mm x 57mm and at the bargain price of only 34USD if you buy 40pcs.
When I was over all mixers at Peavey (not considered audiophile) I used truckloads of 22uF caps in that circuit node. I recall one problem from DC current leakage in a new capacitor series that introduced noise. My suspicion in retrospect is that the new series of caps was not properly formed in, but I didn't waste my time back then solving the vendor's problem. I just black balled that capacitor series and used a quiet one.
Doing it electronically costs a fraction and uses what is basically existing circuitry that needs to paid for anyway and is just somewhat extended using a few extra parts that are cheap.
Cheap is subjective. I discussed this technology with a respected console designer (I respected him, now RIP). Incorporated into a large console there would be economies from being able to share the higher voltage rails, etc. His consoles were considered premium audio quality and liberally used DC servos to keep audio path capacitors small and linear. He had no interest in flying a mic preamp up to phantom voltage. Alternately you could drive pin one of the mic input down to a negative voltage but that can cause new problems in real world applications.
Another idea I bounced off my friend back then (this was years ago), was to use the phantom power switch to short across the input dc blocking capacitors when phantom voltage was switched off. This would completely remove those evil capacitors from the audio path when not being used. At the time it was still popular to make DIY passive mixers that routinely needed make up gain stages. There was no reason to need phantom power for a make up gain stage.

Of course do what floats your boat... Capacitors have been well studied over the decades and modern capacitors are superior to the old poop we had to work with.

JR

PS: Do they make phantom powered mics without caps in their audio path?
 
Yours JR is/was more elegant. In Chalupa, Markey, Kirkwood et al and possibly thor we cheated by bracketing supplies around phantom Vcm.

Not quite. My circuit uses a somewhat higher rail than +48V and uses CCS/Cascoding with a classic discrete mic pre frontend, in effect forming a flying, ultra low noise and distortion voltage to current converter.

There is no additional supply except this and the +/- 12...24V Op-Amp Supply.

The Birt circuit operated as "virtual ground" and audio is injected as current into the virtual ground node via cascoding transistors acting as level shifters.

My design was for +/-18V rails with the virtual grounds at -9V. So it can operate with inputs all across 0V to +48V common mode.

As all the audio is current mode internally there is no signal voltage to speak off internally.

For DC, using matching and zero Tempo current sources etc. to minimise offset in the input.

Arranging a differential servo running on +/-18V rails but allowing to monitor and servo the input transistor emitter was one of the more fun and challenging design issues.

By minimizing all issues in the first place meant it was not THAT challenging.

Result, < 1mV output offset, Quad OPA for Servo duties (OPA1679) and 1/2 Quad OPA for signal (OPA1679 again).

Gain control switching J-FETS needed optocouplers, but are still somewhat preferred to relays in my books.

In my books the whole circuit falls under "prior art" and "obviousness". Really mostly tres primitive. The "how to make it actually work properly" I book under experience and "trade secret".

Thor
 
Last edited:
Show me the data...

Cyril Bateman. Capacitor distortion article series in Wireless World.


Cheap is subjective.

Ok, low cost, which is objective.

Of course do what floats your boat... Capacitors have been well studied over the decades and modern capacitors are superior to the old poop we had to work with.

Again, I must disagree. The cheapest vendor from China SMD badge jobs we see everywhere ( including with big brand names on them - I know the factories personaly) are worse than the crappy old poop.

PS: Do they make phantom powered mics without caps in their audio path?

My own designs are based on Shoeps and use P-MOS followers, so have very high input impedance. C0g SMD ceramics (essentially distortion free) or SMD film cap's in very manageable values can be used (10nF) and are free from anything measureable.

But it varies.

Thor
 
Not quite.


In my books the whole circuit falls under "prior art" and "obviousness". Really mostly tres primitive. The "how to make it actually work properly" I book under experience and "trade secret".

Thor
I won't argue about "elegance" that is clearly subjective, but I still haven't seen any 3 transistor differential stages in the art the did what mine did. Of course everything is obvious after we see it.
===

Yes, COG/NPO caps are very linear, but don't tell the audiophiles that you are using ceramic caps. ;)

JR
 
I still haven't seen any 3 transistor differential stages

I have seen 3 transistor differentials in WW in the 90's. More recently as "class I" output stage. All of which may not be the same as yours.

Yes, COG/NPO caps are very linear, but don't tell the audiophiles that you are using ceramic caps. ;)

Nope, just call them "TDK C0G Capacitors" and don't mention "ceramic" and you are good.

BTW, when it comes to passives, the peak for fairly affordable high audio quality was in the mid 90's, ever since it went downwards for anything counting as commodity.

I blame the beastly Chinese, starting with the late 90's bad electrolytic capacitor plague.

I also blame the woke ejits following Saint Greta, who forced dihydrogen monoxide based electrolytes with sub 10 years shelf life on the industry, banning lead etc. et al, blah, blah, blah.

Thor
 
I have seen 3 transistor differentials in WW in the 90's. More recently as "class I" output stage. All of which may not be the same as yours.
No definitely not the same thing but interesting ... maybe after beer o'clock I'll look at it closer. I respect Cordell's work.

media-1174242-class-i-output-stage-fig6t.jpg

Taking feedback from both emitter degeneration resistors reminds me of another one of my abandoned circuit ideas. Back in the 90s I crafted a way to manage the Class A current for a cheap guitar power amp using plastic darlington output devices. They were difficult to bias reliably using typical Vbe multipliers and thermal coupling to the heat sink. I made one work serviceably using a cheap transistor array sensing voltage across emitter resistors, but could not get any of the amp engineers interested. I was going to use this approach in a small 35W Karaoke amp I needed, but my boss wouldn't risk it... Nothing refines a circuit design like mass production but my boss at the time was a ball-less a_hole.
Nope, just call them "TDK C0G Capacitors" and don't mention "ceramic" and you are good.
I'm already good, but I never had much success marketing to audiophiles (too much truth).
BTW, when it comes to passives, the peak for fairly affordable high audio quality was in the mid 90's, ever since it went downwards for anything counting as commodity.

I blame the beastly Chinese, starting with the late 90's bad electrolytic capacitor plague.
I blame the customers and low expectations.
I also blame the woke ejits following Saint Greta, who forced dihydrogen monoxide based electrolytes with sub 10 years shelf life on the industry, banning lead etc. et al, blah, blah, blah.

Thor
Oops stop press, we agree... :unsure:

JR
 
dihydrogen monoxide based electrolytes
I've heard that stuff is deadly but I drink it every day. I get mine from Chuck Norris' ranch. A few added drops of food-grade H2O2 provides a nice O2 rush.

but my boss at the time was a ball-less a_hole.
Couldn't stop laughing.

My goal for input electrolytic capacitor elimination - besides eliminating the stored charge - was reduced 1/f noise from the rising source impedance from input cap reactance as frequency decreases. That and leakage/noise effects/long phantom turn-on-to-quiet-operation time.

My Protoboard wasn't quiet enough to experiment but my flat MC preamp, which is direct-coupled, has low enough noise to run experiments to see what 1/f does with electrolytic-coupled and lower-value film-coupled inputs.
 
Couldn't stop laughing.
it still isn't funny (to me) now decades later... He literally spanked me at the kiss ass corporate politics game..... I survived to see him ultimately walk the plank but not soon enough for my taste.
My goal for input electrolytic capacitor elimination - besides eliminating the stored charge - was reduced 1/f noise from the rising source impedance from input cap reactance as frequency decreases. That and leakage/noise effects/long phantom turn-on-to-quiet-operation time.
yup imperfect along with many other weak links in the chain.
My Protoboard wasn't quiet enough to experiment but my flat MC preamp, which is direct-coupled, has low enough noise to run experiments to see what 1/f does with electrolytic-coupled and lower-value film-coupled inputs.
I remember phono preamps . :cool: I hear vinyl is coming back.;)

JR
 
No definitely not the same thing but interesting ... maybe after beer o'clock I'll look at it closer. I respect Cordell's work.

media-1174242-class-i-output-stage-fig6t.jpg

Taking feedback from both emitter degeneration resistors reminds me of another one of my abandoned circuit ideas. Back in the 90s I crafted a way to manage the Class A current for a cheap guitar power amp using plastic darlington output devices. They were difficult to bias reliably using typical Vbe multipliers and thermal coupling to the heat sink. I made one work serviceably using a cheap transistor array sensing voltage across emitter resistors, but could not get any of the amp engineers interested. I was going to use this approach in a small 35W Karaoke amp I needed, but my boss wouldn't risk it... Nothing refines a circuit design like mass production but my boss at the time was a ball-less a_hole.

I'm already good, but I never had much success marketing to audiophiles (too much truth).

I blame the customers and low expectations.

Oops stop press, we agree... :unsure:

JR

OK I went back and chewed on that class I? amplifier schemo... It looks more like a science fair project than practical design. Reading the text the output stage is running at something like 130mA quiescent current... that is high, most bipolar maps run around 25 mA quiescent current. The dual opposite polarity LTP in parallel, with extra devices bootstrapping the LTP current appears to be a thermal stability circuit.
===
My 3 transistor differential was a much different animal, but just another piece of unfinished art, that shall remain unfinished.

JR
 
OK I went back and chewed on that class I? amplifier schemo... It looks more like a science fair project than practical design. Reading the text the output stage is running at something like 130mA quiescent current... that is high, most bipolar maps run around 25 mA quiescent current.

25mA is WAY too low. You need around 26mV across the emitter resistor of each transistor to get optimum "Class B" according to Mr Self (most of us call that Class AB). If you use 0.22R emitter resistors then:

26mV/220mOhm = 118mA

The dual opposite polarity LTP in parallel, with extra devices bootstrapping the LTP current appears to be a thermal stability circuit.

Kind of.

It is actually a circuit that prevents the "non-active" transistor from cutting off and provides a stable, thermally independent bias.

As a result essentially all of the common "Class AB" distortion mechanisms EXCEPT GM doubling/halving (doubling if you from a Class B view, halving if you view from a Class A view) which is generally amenable to using NFB to correct.

It is a clever little circuit.

I have something simpler that does the same job, but that's beside the point.

Thor
 
At that point it's desirable to use J-Fet's where current noise is low, can always parallel a few for low Voltage noise.
Yes. Two pair of of sub-nV/rtHz FETs at reasonable Ib for Graeme style mic pre purposes is a good direction, yes? Assuming Q1a||Q2a and Q1b||Q2b per side?
How about 4 x 2SK2145 as four separate parallel differentials? Suitable VAS * buffer.
That I will look up. Line input, you’re saying? Would love to see an example schemo, doesn’t have to be particularly crafty.
With a bootstrapped input resistance you can even go 100nF coupling cap with acceptable noise levels.
There’s an SSL 9K line input like that, monitor section IIRC. But you’re more referring to meat-and potatoes unbalanced work within a given module, I’m guessing.
They are SMD, you can parallel a few. Easy. Panasonic and others also have excellent SMD film types.
‘Tis a pity ECHU-X stops at 220n and 50V (too close to 48V!), they’re 2%. Actually I think that value is only in 16V. WIMA PPS 63V starts at 5% but are bulk only in that tolerance. I have a few things I’ve drawn where my best plan is to do proto runs by batching 10% 1u/63V, which is sub-optimal. The Murata NP0 are 50V up to 220n and 25V in 330n and 470n. 5% all around for those but I suppose that’s a given for an NP0 rating. I’m mixing metaphors here by talking about EQ and mic pre and line in applications in the same landscape. Strange how availability of PPS and NP0 parts substantially directs a whole lot of other decisions…
 
Last edited:
Back
Top