TGM Amplifier ?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
My first thread in this forum... how to begin

I want to build an amplifier and know how it works. I have never done this before but I do enjoy starting something new, learning and researching and designing. Eventually building. But I can't start with a plan that will last longer than a few weeks in case I run out of steam or get distracting by the summer... I'm looking for some encouragement from the forum to pull me through ;-)

My excuse is that I'm making a Home Theatre system (DIY HT speakers done already) so I need 5 channels and it has to be musical. I want the future option to bridge channels to allow drive my stereo speakers too. I was tempted to buy AKSA 55 but it's no longer for sale and you can't build it yourself anymore - only fully assembled. Hugh was very kind about it. I have started from AKSA-like schematics found in this forum - I understand they are not the 'real' ones but they are at least in the public domain.

So I read many posts in this forum and my interest settled clearly on a discrete SS amp (although very tempted by Susan Parker Zeus amplifier). My interest quickly drawn to the GEM, DX, AKSA bootstrap family. I'm trying to understand how they work, last week I read about p-n junctions, this week about bootstrapping, long-tailed pairs and Miller capacitance. If I copy one of these excellent amplifier designs I can benefit from all that has gone before, but it will not be mine and that will make me lazy to understanding how they work.

I have to learn about design and see if you guys have the patience to go over old ground again with yet another design that looks so very familiar. I will build it; first step is to order parts, beg/borrow a dual bench supply (or I have to build it) and scope.

EDIT:

Here is the final incarnation of the amp, starts at post 349, with photos at post 354

http://www.diyaudio.com/forums/solid-state/140461-tgm-amplifier-18.html#post2893495

Final schematic is attached at post 356

http://www.diyaudio.com/forums/solid-state/140461-tgm-amplifier-18.html#post3259963
 

Attachments

  • tgm_1v0.gif
    tgm_1v0.gif
    77.6 KB · Views: 6,106
Last edited:
What's up. bigun... Sorry to say, that won't work.. but
(attached) is as close to your schema that will.

1. not enough current across T1/2 , 10k for R3 solved that.
(1.4ma per device)
2 . no degeneration on T1/2 .. too much gain , also R9/1
too high. ( amp produced a square wave with 1v input).

3. I used 100pF for C5 (safe) and r15 way too low.. 100R
is better.

Many refinements can be added to this classic topology
like a constant current source for the input pair,
negative feedback compensation, input RC filter...etc.


OS
 

Attachments

  • runs.gif
    runs.gif
    14.5 KB · Views: 5,232
doesn't work?

Thanks very much for taking the trouble to look at this OS. These are just the things I want to learn, but I'm worried I'm going to be sent off to buy a big book on the topic shortly, so let me ask a question...

My aim is to build something as simple (in terms of part count) as the ASKA 55 so I started with a schematic similar to it (the real one being unavailable). So I assume that after some learning and refining of the schematic I then build something on proto-board and play with it to see what affects it's performance (most importantly by listening to it). If I can get to a decent result (for me this means smooth hi frequencies) like this then I'll be happy. How would you compare this approach with the option of building the DX (proven design) where I don't have to ask a dozen questions on this forum on a daily basis and then playing with the prototype of that as an alternative way to reach my goal ??

By the way - why do you want separate emitter resistors on the LTP ?

You also recommended different devices from AKSA - is this because you have them set up in your modelling software or because they are easier to obtain ?

Many thanks...
 
doesn't work?

it's not your fault ..68k would of worked at + - 70 rails, but
not at 35v..

you should download...

http://www.linear.com/designtools/software/ltspice.jsp
build your own..
You also recommended different devices from AKSA

aska is a "master" ,BUT... there are certain combinations
of components that just will not work , real or simulated.
a mje340/bd139/ or the 2sc XXX have similar characteristics,
but to make a ideal circuit you would have to choose
one of them (especially for main VAS).

OR , as others suggest, jusy build it, burn many transistors,
and learn "the hard way"...
OS
 
It's a sunny morning and I have a week off work, time to 'go to amplifier school' :D

OS,

being an iMac person I can't use that spice software but I've found something that looks to be equivalent at http://www.macspice.com/ and the I'll start to look at this LTP section further. I had drawn it with only 0.5mA running through and I don't know why this is bad (yikes). Also, I am to learn from reading about SS vs tubes that the secrets are in the distortion characteristics. There are people who say even order harmonics are better than odd and others who say this is nonsense, but either way I should like to learn how the design can impact distortion at some point - i.e. what happens when the design is less 'ideal'.

Nigel,

The extra bootstrap capacitor was shown that way as I have read it is a 'known improvement' to the AKSA amplifier, using special high-quality Blackgate capacitors. Perhaps not the starting point for me - wait until I have something running before I try this. The resistors with the fuses have a purpose which I can't remember ! - there's a post discussing this somewhere between Pavel and AKSA...
 
Disabled Account
Joined 2007
Bigun said:
I had drawn it with only 0.5mA running through and I don't know why this is bad (yikes).


Hi,
You'll want to look at the device specs to see where (current draw wise) it will operate best. For the 2N5401 it's in the range of 2-10mA. You need to keep an eye on dissipation too, not to run the device too hot. 3-4mA at your rails would be fine.
 
Macspice should work good,BG.
just find the .models and copy/paste.
I did finish your amp.. added a current source.
with aksa's 22p ..the amp has a very good loop gain response
(won't oscillate) with 10-22R for the 2 tail resistors (t1/2)
you have enough total open loop gain 45db+
and 27k for input and feedback give enough closed loop gain.

(attached) does very well and should also sound nice. :)
edit: you might want to increase C3 (22u) to 100u,
less offset at output.

OS
 

Attachments

  • tmg.gif
    tmg.gif
    19.4 KB · Views: 4,762
RE: schematic in post #1

Looks OK, but I think some mods are needed.

First, add an RC filter to the negative rail too, anyway you'll get high PSRR.
I'm not sure D1 is needed.
Then R1 must be = R9.
C4B would blow. Omit that.
A 100n...1u foil cap in paralel with R15 would be a good idea. R15's value is too low.
Add resistors is serier with C11 and C13 (1r...10r).
Use some lag compensation as OStripper suggested is his schema.
Omit L1 if the amp is stable driving capacitive load without the inductor. Or add a low value resistor in paralel with the inductor.
Use CFP Vbe multiplier or the bias will vary with rail variation.

I cannot agree with OStripper related to emmiter degeneration. I don't think they are needed in this circuit.
Current source and current sink for the LTP are not needed. Keep it simple.
An RC is needed in the input for lowing TIM (series R, paralel C).

Good luck!
 
by fjr -Current source and current sink for the LTP are not needed. Keep it simple.

I suppose you could ditch the CCS if you were using
independant power supplies for each amp. But BG said
these would be part of a HT system (3-6 amps).

the current source I added does -60db /1k, with none
PSRR turns to CRAP. So, maybe even a better one
(LED, cap, 2 resistors) would be in order.

The sound of this topology would remain the same.
ASKA himself improved this design in later amps,
(aska100) by adding the CCS AND current mirrors.
OS
 
Disabled Account
Joined 2007
Re: RE: schematic in post #1

fjr said:
Looks OK, but I think some mods are needed.

First, add an RC filter to the negative rail too, anyway you'll get high PSRR.
An RC is needed in the input for lowing TIM (series R, paralel C).

:up:

ostripper said:


the current source I added does -60db /1k, with none
PSRR turns to CRAP.


As mentioned, an RC filter will improve PSRR dramatically. That 100 ohm R (on one rail!) is a wee bit too big though, more like 10-22 ohm plus a 100uF or more cap.
 
sorry guys, this is probably snails pace for you. But first time ever using MacSpice this weekend - I have managed (after finishing the floor tiling) to get the LTP set up and obtain a dc analysis (see attached).

I am starting to understand how the LTP is the 'heart' of the amplifier, it determines many of it's properties as well as being a very elegant design element.

I can see that the total current through the LTP is 0.75mA but the imbalance is still a bit high (0.45mA T1 and 0.3mA T2).

I still plan to use AKSA as my 'guiding principle' so no constant current source. However, I read a posting in a forum that it uses 68k as the default load for the LTP. Looking at the limited data I have for 2N5401 I don't see why the lower current (0.5mA) would be a problem (does it hurt linearity or frequency response ?)

I understand that including two separate load resistors for each T1/2 aids performance by 'swamping the emitter resistance variations with current' (paraphrased from Rod Elliot).

I haven't figured out how to get spice to do an AC analysis (yet)
 

Attachments

  • ltp1.gif
    ltp1.gif
    45.8 KB · Views: 2,470
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.