Current Mirrors and first post here

[vynuhl.addict]

Hi everyone, my name is Colin and I am new to this forum, well as a poster atleast but have been reading the postings for months whiile twiddling away with my own projects..My first amplifier started as a modified P3A and being happy and unhappy with aspects of the sound it changed quite alot. I just recently emailed Hugh Dean from aspen audio concerning a couple technical questions i had on different topologoes and to be honest, didnt really expect an answer, let alone a detailed response like I was given regarding feedforward and current mirrors. Hugh is unbelievable and isanely gracious like the rumours suggest and absolutely elegant with his words and help and I assure him that one day when I am in a better financial position I would strongly consider buying an Aksa kit but right now my heart is in trying to create something that is great sounding myself, with maybe 10% originality but it looks like that may just come from my transistor chhoice thus far (o: judging by the different topologies out there

I had been inspired by the p3a and aksa to build an amplifier based originally on the P3A schematic, with a change in the output arrngement using 4 paralelled output transistors in emitted follower cofiguration and same with the drivers mje15032/15033 to use at +/- 47 volts. vas uses 2sa1145, my local par
ts store were backordered on the 2sa1360(from what i gather hugh uses the npn compliment 2cs3423 in his lifeforce vas) and 2sa1145 specs matched to the nth, it is just a lower power dissipation part with the desirable 2.5pf output capacitance. input differential uses 2n5551..


I have tried the citation 12 version of balancing in the differential with the resistor biasing and left it in for a month. from a sound perspective it is rich and fullsounding but for some reason seemed a little lacking in the dynamics department with really round bass.


I found the same kind of story with the transistor current source biased to 3ma, highs were a little better, but like above had that round sound, i have never heard tubes aside from guitar amps but imagine this round sound is what tube buffs like which must be even order distortion characteristics???.Dont get me wrong , they both sound great, but being a drummer of 22 years things like cymabls and leading and trailing edges of percussion seemed to get slowed down and for the life of me i couldnt tell the difference between a dark sounding zildjian k and a brights cymbal like a sabian aax.

Now i know current mirrors in differentials seem to be taboo butafter borrowing the current mirror idea from mr self with mirror degeneration resistors, 100ohm emitter degenertion resistors on the differential with the transistor,led,resistor current source biased to 4.4 ma detail,air and unbelieveable imaging and details jumped out of the woodwork, so far that is with out being edgy or brittle, i can hear the tones from cymballs again ), reflections off of recording location walls..


This is my part time hobby, and 90% of my decisions are based subjective listening but aurally I dont see what the issued is soundwise from mirrors in the differential?..

If someone can guide me to making a superb sounding resistor blanced differential it would be much apreciated..


Thanks
Colin

 

[ilimzn]

This is my part time hobby, and 90% of my decisions are based subjective listening but aurally I dont see what the issued is soundwise from mirrors in the differential?..
If someone can guide me to making a superb sounding resistor blanced differential it would be much apreciated..
Thanks
Colin

Hehe I always wondered about that myself so far I had pretty much zero reasons to avoid a current mirror - perhaps only if I was in some sort of a contest to use a minimum of transistors in a design
Still, while you are experimenting, you might want to try a 'middle way' - self balanced current source termination of the differential stage. This is essentially a current mirror, but the common base node is AC bypassed to the supply rail. Within the mirror, this makes the transistor that is not diode-connected, act as a current source nearly exactly one half that of the tail current. This gives you a balanced LTP. Keep in mind that the maximum current swing available to drive the VAS is now halved to +-1/2 of the tail current (with regular current mirror, it is theoretically +- the full tail current), so the total gm of the stage is halved.

 

[lineup]

Emitter resistors
in a differential pair
can be used, but many times are not used.

If used, a good target value is to use 2 resistor
with Ohm value that gives something like
0.050 - 0.200 Volt drop across each resistor at the selected current.
I guess the most optimal value,
would depend on what transistors are used in the differential pair.
.......................................................................

I use most often around 0.100V drop.

If for example, I run current source at 2 mA
and 1 mA in each transistor + resistor
this would render emitter resistors have value 100 Ohm.

0.001 ampere x 100 ohm = 0.100 volt
.......................................................................


Hugh Dean is an australian gentleman.

Another Amplifier man,
Douglas Self has got some ideas and hints - about differential emitter resistors:

See:
Distortion in Power Amplifiers.
( a classical, legendary investigation in details of power amplifier design )
Link:
http://www.dself.dsl.pipex.com/ampins/ampdept.htm


lineup
Lineup Audio Lab

 

[kASD]

Hugh is unbelievable and isanely gracious like the rumours suggest and absolutely elegant with his words and help

Well said,

He is Remarkable Gentleman

 

[lineup]

Re: Re: Current Mirrors and first post here

I have tried the citation 12 version of balancing in the differential with the resistor biasing and left it in for a month.
from a sound perspective it is rich and full sounding
but for some reason seemed a little lacking in the dynamics department with really round bass.
-----
Now i know current mirrors in differentials seem to be taboo but after borrowing the current mirror idea from mr self with mirror degeneration resistors, 100ohm emitter degenertion resistors on the differential with the transistor,led,resistor current source biased to 4.4 ma detail,air and unbelieveable imaging and details jumped out of the woodwork, so far that is with out being edgy or brittle, i can hear the tones from cymballs again ), reflections off of recording location walls..
-----------

If someone can guide me to making a superb sounding resistor balanced differential it would be much apreciated.

.

Hehe I always wondered about that myself so far I had pretty much zero reasons to avoid a current mirror - perhaps only if I was in some sort of a contest to use a minimum of transistors in a design
.

I know some Aspen Hugh Dean amplifiers use resistor biasing in input pair.
And his amplifiers are said to SOUND very good.
Those who built them, ARE IN LOVE.

From theoretical viewpoint CCS source, no doubt, should be the better.
But audio sounds does not always follow technical and logical expected performance.


I am no 'subjectivist' at all!!!!!!
Yet, the best audio sound I ever heard is my own:

Single Ended RESISTOR biased input pair voltage amplifier, with local feedback.
Feeding Single End Follower Metal Film RESISTIVE loaded output.
There is No Global Feedback, from output.
All stages in full Class A.

Don't ask me why this gives such a subjectively fantastic sound performance.
It would be very easy to build a similar circuit using current sources, mirrors and whatever you like
and so improve most all test data to another higher level.

-------------------------------------------------------------------------
I am talking, of course, of my Discrete 12 Volt supply,
SE Class A 4 Bipolar Transistors Headphone amplifier.
You can get the full schematic of latest updated version
and Amplifier Data with my description and instructions here:

Discrete Single End TRUE Class A HeadPhone Amplifier


Regards
lineup

 

[vynuhl.addict]

lllmzn, the capacitor you speak of, is this similar to what is done in the blameless amplifier design for the differential CCS by using a resistor from base to diode and using a capacitor between the two 10k values in the blameless from diode to ground and to the power rail?..

To Lineup, I am getting a 230mv drop across each of the the emitter dgeneration resistors using 2n5551 in the diffderential pair, should i decrease the emitter resistor values or will 230mv be fine??. originally without the degeneration resistors my bias originally set at 25mv would climb to 46mv..For the current mirror I use 2n5401..


for the second stage vas it is bootstrapped, i tried a conventional ccs on this one, while it did seem to give better bass the low end didnt seem to have as much musicality and sounded a bit too homogenous and lifeless. alsa the fact that everytime i powered down the capacitor bank would stay charged overnight and i consistenly had to use bleeder resistors every time i made a change, with the bootstrapped vas all stored charge was drained through the circuit after power off..


As for hughs designs I think all the aksa were all resistor biased differentials, as i only ever saw two to92s in the pics, up until the lifeforce which is said to be an aksa55 upgrade, based on a glance at the photo and the green led next to the transistor i think maybe its a ccs,..


With the current mirror in the differential I end up with the same dc offset i had with the resistor balanced differential, which is -25mv left and -25mv right, im interested to know why the negative dc offset?? during the building of the amplifier I hfe matched all transistors except the output pairs..

Thanks
Colin

 

[ilimzn]

lllmzn, the capacitor you speak of, is this similar to what is done in the blameless amplifier design for the differential CCS by using a resistor from base to diode and using a capacitor between the two 10k values in the blameless from diode to ground and to the power rail?..

No, that capacitor is used to provide an AC bootstrap to the resistor feeding the reference for the CCS.

What I am talking about is an additional cap that can be added to the current mirror. In the current mirror, the bases of the two BJTs and one of the BJT's collector connect together. Connect one end of the cap to this point. The emitters of the BJTs in the current mirror connect via degeneration resistors to a supply rail. Connect the other end of the cap to this point.
The cap value will need to be quite big as the impedance of this node is quite low (a couple of kohms). Use a good quality elecrolytis of at least 220uF (low leakage and low ESR) bypassed by a good foil cap.
This cap restricts the current mirror action only to DC and very low frequency, so it will still perform the LTP DC balancing function. For AC, though, the VAS side of the mirror becomes a current source of 1/2 LTP tail current.